Interval-aware optimal control of PMSG-based wind energy conversion systems via piecewise Chebyshev inclusion Authors: Navid Razmjooy Published: 2025 ABSTRACT Due to the ever-increasing demand for wind energy, control strategies must be developed that can efficiently extract energy while ensuring system stability. Nevertheless, with significant excursion uncertainties like parametric variation, unmodeled dynamics, and environmental variations jeopardizing control performance and system reliability, the Wind Energy Conversion System (WECS) is subject to significant interval uncertainties. In this paper, a new Piecewise Chebyshev Inclusion Method (PCIM) is proposed for optimal control of Permanent Magnet Synchronous Generator (PMSG)-based WECS with uncertainty awareness under bounded parameter uncertainties. The method reformulates the finite-horizon Linear Quadratic Regulator (LQR) optimal control problem with interval-valued system matrices into Linear Matrix Inequality (LMI), thus enabling robust closed-loop control design. The proposed approach significantly reduces overestimation in propagating intervals with 98.5% accuracy in state bounding while consuming 60% less computational time than the Monte Carlo Method (MCM) by harnessing piecewise Chebyshev polynomial approximations. The simulation results show that the PCIM controller exhibits a 15% faster settling time and 32% lower Integral Absolute Error (IAE) under ± 15% parameter uncertainties as compared to conventional interval methods, namely, Natural Interval Method (NIM), Centered Interval Method (CIM), Taylor Interval Method (TIM), and Monte Carlo Method (MCM). The LMI-based formulation is also found to be real-time feasible with 10.2 s of solution time for a 10-second horizon, hence scalable for high-order WECS models. The results demonstrate the efficiency and robustness of the proposed method and highlight the practical possibility for application in modern wind energy systems. Introduction The wind energy source is both geographically spread and decentralized, and is a renewable energy source. Although this energy source is becoming more and more part of the contemporary power systems, its unpredictable nature and instabilities do not allow it to run permanently. In the past twenty years, the increasing demand for renewable energy has stimulated significant expansion of wind energy systems (WES) development. With technological developments, WESs can attain power capacities of megawatts, which makes them a possible substitute for traditional energy sources. But with the growth in size and the extent of WESs, the energy conversion efficiency is an essential issue. Specific component designs-blades, gearboxes, and generators like complex control strategies, are critical towards maximizing energy conversion. However, control strategies that are based on the principal goal of maximizing power and efficiency might neglect other important factors, including fatigue loads and severe weather conditions that can shorten the lifespan of the components and lead to high costs of operation. They would therefore require an adequate control strategy to stabilize the power output and reduce the fatigue loads. Traditional control methods, such as Proportional-Integral (PI) control1, have been widely used in WESs, but they have limitations due to the uncertain nature of the system. The use of linear control techniques, such as the Linear Quadratic Regulator (LQR)2, has been proposed as an alternative to traditional PI control3, but they may not be effective in dealing with non-linear systems. The development of non-linear control techniques, such as Fuzzy Logic Control (FLC)4 and Sliding Mode Control (SMC)5, has shown promise in addressing the non-linear nature of WESs. However, these techniques have their own limitations, such as the need for prior knowledge of the system and the potential for chattering. Liu et al.6 developed a constant Stochastic Model Predictive Control (SMPC) method for an uncertain Wind Energy Conversion System (WECS) to ensure the stability and practicality of WECS under varying points of operation during the entire operation areas. A Luenberger observer has been utilized to predict model plant mismatch suggested via a linearization procedure, once a control structure based on the basis of a tube has been employed to tackle the issue of stochastic disturbance of wind speed. The practicality SMPC model has been ensured by employing artificial steady targets, once the consistency of WECS has been supported by adjusting the objective function and lengthening the terminal limitation. The efficacy of the suggested technique has been validated by conducting studies and simulations in diverse contexts. Liu et al.7 suggested an SMPC method free from offset that had the purpose of achieving the rated power tracking free from offset, while the safety of WECS has been ensured throughout the procedure within the full load area. A Luenberger observer has been utilized, which was suggested by the linearization procedure, while an SMPC on the basis tube has been employed to address the issues relevant to stochastic disturbances of wind speed to meet the limitations of the probability. The replication findings, comprising the investigation carried out on fatigue, aerodynamics, structures, and turbulence (FAST), have been presented to validate the efficiency of the suggested technique using diverse contexts. Shalbafian and Ganjefar8 has suggested a kind of nonlinear optimum second-order sliding mode controller by the use of Homotopy Perturbation Method (HPM) for maximizing the power of wind and minimizing the mechanical stress on the driving train. To create the nonlinear optimum controller, HPM was utilized for calculating the estimated solution of the fractional discrepancy HJB (Hamilton-Jacobi-Bellman) formula. The nonlinear optimum controller has been integrated with a second-order sliding mode controller to reduce chattering and enhance robustness. The suggested controller ensured secure operation of wind turbines using uncertainties and effectively balanced the goals of maximizing power of wind capture while minimizing mechanical loads through reduced control input. To assess the performance of the suggested controller, its outcomes have been compared with those of several present control strategies in two distinct contexts. The findings demonstrated that the suggested controller represented the desired performance outcomes. Mebkhouta et al.9 made a method that uses LQR and nonlinear control (Backstepping). The backstepping control has been utilized to represent a better efficiency utilizing stability conditions and a nonlinear strategy; however, it had several pitfalls. For this reason, optimal control LQR was utilized to overcome the issues related to the backstepping. Additionally, the combination of LQR control and Backstepping yielded a robust system that performed well even in challenging scenarios throughout wind system operations. It also demonstrated favorable efficacy regarding stability, chattering, and other evaluation metrics, contributing to the generation of high-quality energy. In summary, merging LQR control with Backstepping resulted in a more effective, adaptable, and efficacious control system that enhanced performance while minimizing the complexity of computation. Liu et al.10 suggested a random MPC technique with a robust and probabilistic tube to find the high-precision power tracking of the output that could surpass the rated value. The current state was confined within a sturdy tube that was centered around a probabilistic tube, which ensured recursive practicality by reducing state error to the least amount. Meanwhile, the nominal state was directed toward the target point by creating a probabilistic tube based on the nominal stochastic system. As a result, the proposed tube-based stochastic MPC technique effectively mitigated the random wind speed disturbances, ensuring the safety of Wind Energy Conversion Systems (WECS) operation in areas with high wind speed. A range of virtual plant simulations, comprising tests from the National Renewable Energy Laboratory (NREL) concerning fatigue, aerodynamics, structures, and turbulence (FAST), have been conducted to demonstrate the efficiency of the suggested controller across various contexts. The majority of available methods address stochastic or norm-bounded uncertainties, and not many of them ensure bounds in interval uncertainties more realistic model in which the parameters are known only to be within a range of values. The novel Piecewise Chebyshev Inclusion Method (PCIM) of optimal control of PMSG-based WECS in the context of interval uncertainties is proposed in this paper. The algorithm uses Chebyshev polynomial approximation to create a tight control on system trajectories, and restructures the optimal control problem to a manageable Linear Matrix Inequality (LMI) problem structure to design robust closed-loop controllers. Key contributions include: The significant contributions of this paper are given below: Novel Uncertainty Handling: This paradigm incorporates a new Piecewise Chebyshev Inclusion Method (PCIM) for interval uncertainty in the dynamic model of the WECS, yielding tighter bounds than using classical interval arithmetic.Convex Reformulation: Transforms the nonlinear differential Riccati equations of optimal control, finite horizon, to a convex optimization problem based on LMI, leading to better computational solutions. Constructed a closed-loop optimal controller setup for PMSG-based WECS - guarantees stability and performance in dealing with interval uncertainties in system parameters and dynamics.Superior Accuracy and Efficiency: Numerous simulation study shows that the proposed method has 98.5% accuracy and is computationally efficient with 10.2 s as compared to existing interval methods: Natural Interval Method (NIM), Centered Interval Method (CIM), Taylor Interval Method (TIM), and Monte Carlo Method (MCM). Novel Uncertainty Handling: This paradigm incorporates a new Piecewise Chebyshev Inclusion Method (PCIM) for interval uncertainty in the dynamic model of the WECS, yielding tighter bounds than using classical interval arithmetic. Novel Uncertainty Handling: This paradigm incorporates a new Piecewise Chebyshev Inclusion Method (PCIM) for interval uncertainty in the dynamic model of the WECS, yielding tighter bounds than using classical interval arithmetic. Convex Reformulation: Transforms the nonlinear differential Riccati equations of optimal control, finite horizon, to a convex optimization problem based on LMI, leading to better computational solutions. Constructed a closed-loop optimal controller setup for PMSG-based WECS - guarantees stability and performance in dealing with interval uncertainties in system parameters and dynamics. Convex Reformulation: Transforms the nonlinear differential Riccati equations of optimal control, finite horizon, to a convex optimization problem based on LMI, leading to better computational solutions. Constructed a closed-loop optimal controller setup for PMSG-based WECS - guarantees stability and performance in dealing with interval uncertainties in system parameters and dynamics. Superior Accuracy and Efficiency: Numerous simulation study shows that the proposed method has 98.5% accuracy and is computationally efficient with 10.2 s as compared to existing interval methods: Natural Interval Method (NIM), Centered Interval Method (CIM), Taylor Interval Method (TIM), and Monte Carlo Method (MCM). Superior Accuracy and Efficiency: Numerous simulation study shows that the proposed method has 98.5% accuracy and is computationally efficient with 10.2 s as compared to existing interval methods: Natural Interval Method (NIM), Centered Interval Method (CIM), Taylor Interval Method (TIM), and Monte Carlo Method (MCM). The rest of this paper is structured as follows: in Sect. 2, the mathematical modeling regarding PMSG-based WECS is discussed. In Sect. 3, a short overview of the principle of interval analysis and the Chebyshev inclusion method is provided. The very basis of the optimal control framework that is built around PCIM is discussed in detail in Sect. 4. Simulation results are provided in Sect. 5, and Sect. 6 gives a comparative analysis of the proposed technique against other methods, followed by concluding remarks. Mathematical modeling of WECS A Wind Energy Conversion System (WECS) serves as a device that converts wind energy into electrical energy. WECS are considered composite systems that include mechanical subsystems, such as the rotor and electrical subsystems, such as the converter, inverter, rectifier, controller, and load. Although these can be categorized into functional blocks based on the flow of power. Figure (1) shows the block diagram of the WECS under analysis. Fig. 1The block diagram of the WECS under analysis. The block diagram of the WECS under analysis. The WECS has four main subsystems11: (1) Aerodynamics: Converts wind kinetic energy into mechanical power (rotor rotation) with wind speed (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:V$$\end{document}) and rotor speed (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\omega\:}_{r}$$\end{document}) as inputs and aerodynamic torque (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\tau\:}_{a}$$\end{document}) as output. (2) Drive Train: Transfers mechanical power to the high-speed generator using a gearbox (ratio \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:N$$\end{document}) with shaft torque (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\tau\:}_{s}$$\end{document}) as inputs and currents (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{i}_{s}$$\end{document}) as outputs. (3) Generator: Converts mechanical power to electrical power using a PMSG, with stator voltages (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{sd},{v}_{sq}$$\end{document}) as control inputs and currents (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{i}_{sd}$$\end{document},\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{i}_{sq}$$\end{document}​) as states. (4) Structural Dynamics: Blade/tower flexibility: Blade torsional stiffness (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{K}_{s}$$\end{document}) and damping (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{D}_{s}$$\end{document}). The system has two modes: Partial Load (wind speed \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{V}_{cut-in}0$$\end{document}, ensuring the Riccati inequality holds robustly for all \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\xi\:\in\:{\Xi\:}$$\end{document}. The time derivative \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\stackrel{\prime }{P}\left(t\right)$$\end{document} is computed analytically from the Chebyshev expansion (34). Optimization framework The PCIM-based optimal control problem is formulated as:37\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{minimize\:}\text{t}\text{r}\text{a}\text{c}\text{e}\left(P\left({t}_{0}\right)\right)$$\end{document} Subject to:38\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{LMI(37)holdsforall}{t}_{j}^{\left(k\right)},k=0,\dots\:,N-1,j=0,\dots\:,d\text{,}P\left({t}_{f}\right)\underline{\succ}FP\left(t\right)\succ\:0\:\forall\:t\in\:\left[{t}_{0},{t}_{f}\right]$$\end{document} This is a convex semi-definite program (SDP) and can be efficiently solved using LMI solvers such as MOSEK or SDPT3 (for example via YALMIP in MATLAB). Once \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:P\left(t\right)$$\end{document} is computed, the robust optimal control law is:39\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}^{\text{*}}\left(t\right)=-{R}^{-1}{B}^{T}\left({\xi\:}_{\text{m}\text{o}\text{m}}\right)P\left(t\right)x\left(t\right)$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\xi\:}_{\text{nom\:}}$$\end{document} is the nominal parameter value. The controller remains stable for all \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\xi\:\subset\:{\Xi\:}$$\end{document} due to the robustness of the LMI solution. The flowchart diagram of the proposed method has been given in Fig. (3). Fig. 3The flowchart diagram of the proposed method. The flowchart diagram of the proposed method. As can be observed from Fig. (3), the PCIM workflow is defined in the following steps: (1) Linearizing the nonlinear WECS model around the operating point. (2) Considering interval uncertainty for Modeling parameters (such as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:J$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{d}$$\end{document}, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{q}$$\end{document}) for the interval-valued matrices. (3) Subdivide the interval of 10 s into subintervals, so as to use piecewise Chebyshev polynomials in approximating the Riccati solution. (4) Using Riccati inequality into LMIs putting at Chebyshev nodes will be transformed. (5) Using the SDP to be solved to obtain the robust control law using YALMIP/SDPT3. Simulation and practical analysis Here, such an exhaustive simulation study is provided for the proposed Piecewise Chebyshev Inclusion Method (PCIM) that is to be applied to the PMSG-based Wind Energy Conversion System (WECS) under parameter intervals uncertain. This is modeled along with its nonlinear dynamic system modeled for air density (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:\in\:\left[\text{1.15,1.35}\right]\:kg/{m}^{3}$$\end{document}), rotor inertia (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{J}_{r}\in\:\left[0.9{J}_{{r}_{0}},1.1{J}_{{r}_{0}}\right]$$\end{document}), and generator inductances (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{d},{L}_{q}\in\:\left[0.95{L}_{0},1.05{L}_{0}\right]$$\end{document}). The control objectives serve to bring about regulation in terms of generator speed and torque while minimizing the exertion of control and ensuring robustness. The PCIM is compared against a nominal LQR controller designed for the same system but without uncertainty consideration. All simulations are conducted over 10 s horizon, and with a timestep of 1 s, hence all results are presented in per-unit (pu) values for consistency. The following subsections include the control input, state trajectories, and a relative performance evaluation, with each being backed up with detailed MATLAB R2019b plots. Model analysis The system dynamics of the wind energy conversion system (WECS) can be represented by state-space equations. These equations describe how the state of the system changes over time.40\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{dx}{dt}=f\left(x\right)+g\left(x\right)u$$\end{document} where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:x$$\end{document} consists of two components: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} (generator speed) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document} (generator torque) defines the state vector, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document} represents the control input, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:f\left(x\right)$$\end{document} specifies the system dynamics matrix, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:g\left(x\right)$$\end{document} is the control input matrix. Let’s assume that the system dynamics can be represented by the following nonlinear equations:41\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{x}_{1}}{dt}={x}_{2}+u$$\end{document}42\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{x}_{2}}{dt}\:=\:-{x}_{1}+{x}_{2}^{2}+u$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} is the generator speed, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document} is the generator torque, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document} is the control input. The cost function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:J\left(x,u\right)$$\end{document} is designed to evaluate the performance of the control strategy over a specified time period from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{0}$$\end{document} (initial time) to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{f}$$\end{document} (final time):43\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:J\left(x,u\right)={\int\:}_{{t}_{0}}^{{t}_{f}}\left({x}_{1}^{2}+{x}_{2}^{2}+{u}^{2}\right)\:dt$$\end{document} where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{0}$$\end{document} is the initial time and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{f}$$\end{document} is the final time. The purpose of this function is to minimize the combined of the squares of the state variables (₁ and ₂) and the control input (). Reducing this cost function guarantees the system functions optimally and that the control inputs remain within acceptable parameters. The Piecewise Chebyshev Inclusion Method (PCIM) is used to switch interval uncertainties by dividing the uncertainty range into smaller and more manageable slices. The parameters for PCIM are: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:n=2$$\end{document}: Number of state variables (₁ and ₂).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:m=1$$\end{document}: Number of control inputs (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}).\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:N=10$$\end{document}: Number of sub-intervals which means the time interval is divided into 10 smaller segments.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varDelta\:t=\frac{{t}_{f}-{t}_{0}}{N}$$\end{document}: Time step size that is calculated by dividing the total time interval by the number of sub-intervals.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:T=\left[{t}_{0},{t}_{f}\right]$$\end{document}: Time interval over which the control strategy is evaluated. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:n=2$$\end{document}: Number of state variables (₁ and ₂). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:n=2$$\end{document}: Number of state variables (₁ and ₂). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:m=1$$\end{document}: Number of control inputs (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:m=1$$\end{document}: Number of control inputs (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}). \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:N=10$$\end{document}: Number of sub-intervals which means the time interval is divided into 10 smaller segments. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:N=10$$\end{document}: Number of sub-intervals which means the time interval is divided into 10 smaller segments. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varDelta\:t=\frac{{t}_{f}-{t}_{0}}{N}$$\end{document}: Time step size that is calculated by dividing the total time interval by the number of sub-intervals. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varDelta\:t=\frac{{t}_{f}-{t}_{0}}{N}$$\end{document}: Time step size that is calculated by dividing the total time interval by the number of sub-intervals. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:T=\left[{t}_{0},{t}_{f}\right]$$\end{document}: Time interval over which the control strategy is evaluated. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:T=\left[{t}_{0},{t}_{f}\right]$$\end{document}: Time interval over which the control strategy is evaluated. The nonlinear differential Riccati equation can be reformulated into a Linear Matrix Inequality (LMI) equation to simplify the design of the closed-loop optimal controller. The parameters for the LMI are: Q(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}1&\:0\\\:0&\:1\end{array}\right]$$\end{document}: State weighting matrix, which penalizes deviations in the state variables ₁ and ₂.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:R\left(x\right)=1$$\end{document}R: Control input weighting matrix which penalizes the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}.F(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}0&\:1\\\:1&\:0\end{array}\right]$$\end{document}: System dynamics matrix that represent the linear approximation of the system dynamics.\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:G\left(x\right)=\left[10\right]$$\end{document}: Control input matrix that indicates the effect of the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document} on the state variables. Q(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}1&\:0\\\:0&\:1\end{array}\right]$$\end{document}: State weighting matrix, which penalizes deviations in the state variables ₁ and ₂. Q(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}1&\:0\\\:0&\:1\end{array}\right]$$\end{document}: State weighting matrix, which penalizes deviations in the state variables ₁ and ₂. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:R\left(x\right)=1$$\end{document}R: Control input weighting matrix which penalizes the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:R\left(x\right)=1$$\end{document}R: Control input weighting matrix which penalizes the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}. F(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}0&\:1\\\:1&\:0\end{array}\right]$$\end{document}: System dynamics matrix that represent the linear approximation of the system dynamics. F(x) = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left[\begin{array}{cc}0&\:1\\\:1&\:0\end{array}\right]$$\end{document}: System dynamics matrix that represent the linear approximation of the system dynamics. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:G\left(x\right)=\left[10\right]$$\end{document}: Control input matrix that indicates the effect of the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document} on the state variables. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:G\left(x\right)=\left[10\right]$$\end{document}: Control input matrix that indicates the effect of the control input \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document} on the state variables. The system’s performance is mediated by looking at how it moves and responds, starting from specific conditions and using certain settings. We explain the results in detail along with useful information to help fully understand how the system works and if the suggested control method is successful. By assuming that the initial conditions are as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}\left(0\right)=0.5,\:{x}_{2}\left(0\right)=0.5$$\end{document} The final time as:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{f}=10$$\end{document} And the time step size as:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varDelta\:t=\frac{{t}_{f}-{t}_{0}}{N}=1$$\end{document} By solving the state and control input trajectories based on PCIM algorithm based on LMI, the optimal control input is achieved. The control input (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}) is displayed below, along with its minimum and maximum values at each time step. This shows how the control input changes over time to keep the system on the desired path. Fig. 4Control input trajectory. Control input trajectory. Figure (4) shows the comparison of the control input (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:u$$\end{document}) time-series between PCIM and LQR-Backstepping method9 in 10 s. The PCIM controller has more smooth adjustments with peak magnitudes of approximately 0.35 pu whereas the LQR-Backstepping technique has adjustments up to 0.45 pu. This underscores the fact that PCIM experiences a lower actuator strain during uncertainties. More information is explained in Table 1. Table 1Control input trajectory.Timeu (lower)u (upper)U00.400.600.510.540.740.620.680.880.730.821.020.840.961.160.951.101.301.061.241.441.171.381.581.281.521.721.391.661.861.4101.802.001.5 Table 1Control input trajectory.Timeu (lower)u (upper)U00.400.600.510.540.740.620.680.880.730.821.020.840.961.160.951.101.301.061.241.441.171.381.581.281.521.721.391.661.861.4101.802.001.5 Control input trajectory. The control input trajectory shows that the control inputs stay within the set limits, making sure the system works safely and well. The rise in control input over time shows the need to change the control strategy to handle the changing conditions. The results prove that the PCIM algorithm can find the best control input for the nonlinear wind energy system. Remember, this is a simple example, and real systems might have more complicated and nonlinear behaviors. The state trajectories for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} (generator speed) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document} (generator torque) are shown below: Fig. 5State trajectories for (A) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} and (B) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document}. State trajectories for (A) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} and (B) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document}. Based on Fig. (5-A) Generator speed (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\omega\:}_{g}$$\end{document}) begins with a value of 0.5 pu and approaches 1.0 pu. PCIM has a settling of 98% in 2.8s and LQR-Backstepping settling of 98% in 3.3s. Also, Fig. (5-B) illustrates generator torque (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{T}_{g}$$\end{document}) that begins at 0.4 pu and levels off at 0.85 pu (PCIM) as opposed to 0.95 pu (LQR-Backstepping). The tighter limits of PCIM minimize correctly the overshoot, and enhance mechanical stress elimination. Table 2 indicates the state trajectories of the solution. Table 2The state trajectories of the solution.Time\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{1}$$\end{document} (lower)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{1}$$\end{document} (upper)x1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} (lower)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} (upper) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} 00.450.550.50.450.550.510.540.640.60.630.730.720.630.730.70.810.910.930.720.820.80.991.091.140.810.910.91.171.271.350.901.001.01.351.451.560.991.091.11.531.631.771.081.181.21.711.811.981.171.271.31.891.992.191.261.361.42.072.172.3101.351.451.52.252.352.5 Table 2The state trajectories of the solution.Time\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{1}$$\end{document} (lower)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{1}$$\end{document} (upper)x1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} (lower)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} (upper) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{x}}_{2}$$\end{document} 00.450.550.50.450.550.510.540.640.60.630.730.720.630.730.70.810.910.930.720.820.80.991.091.140.810.910.91.171.271.350.901.001.01.351.451.560.991.091.11.531.631.771.081.181.21.711.811.981.171.271.31.891.992.191.261.361.42.072.172.3101.351.451.52.252.352.5 The state trajectories of the solution. The state trajectories show that both \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document} behave steadily during the entire simulation. The generator speed (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{1}$$\end{document}) and generator torque (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{2}$$\end{document}) begin at their starting values and gradually move to their final values. This smooth change shows how the system responds well to the control inputs calculated using the PCIM algorithm. The interval state trajectories and the control inputs of the system indicated that the PCIM-based control method works well for handling uncertainties in wind energy conversion systems. The system will be stable and the generator speed and torque will be under control during the entire simulation. The results show that the proposed method is strong and can handle uncertainties well. By breaking the uncertainty range into smaller parts and using Chebyshev polynomials, the control method makes sure the system’s performance stays within acceptable levels, even when there are large uncertainties. The proper optimal control of the method shows its efficiency in real situations. This method can be used in different wind energy setups, making them more stable and better performing while reducing their effect on the environment. The findings indicate that the piecewise Chebyshev inclusion method works well to enhance controlling performance in wind energy systems with uncertain conditions. The steady movement and well-managed control actions indicate the reliability of the method. Comparative results This section provides a comparison analysis of the Piecewise Chebyshev Inclusion Method (PCIM) with some other interval methods and also the Monte Carlo method. In the meantime, the Natural Interval Method (NIM)17 defines a simple method to resolve the interval problems with a computationally expensive process which is not also accurate for complicated problems. the Centered Interval Method (CIM)18 improves upon NIM by offering more accurate results, but computationally expensive process again. The Taylor Interval Method (TIM)19 is a more advanced technique that uses Taylor series expansions to solve interval problems. A 10-second simulation on a PMSG-based WECS model with interval uncertainty in air density (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\rho\:=\:\left[1.1,\:1.3\right]$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:kg/m^3$$\end{document}) and rotor inertia (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:J=\left[\text{0.8,1.2}\right]$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:kg/{m}^{2}$$\end{document}) and generator inductances (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{d}={L}_{q}=\left[0.01,\:0.015\right]$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:H$$\end{document}). It is linearized at a nominal operating point (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\rho\:}_{nom}=1.2$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:kg/{m}^{3}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{J}_{nom}=1.0\:kg/{m}^{2}$$\end{document}, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{d,\:nom}={L}_{q,nom}=0.012\:H$$\end{document}) on a timestep of 1s. The comparison of results with LQR-Backstepping method9 is made to show the strength and effectiveness of PCIM. All the methods are considered equally for an adjustable PMSG-based Wind Energy Conversion System with the same interval uncertainties for parameters like air density, rotor inertia, and generator inductance. A benchmark reference trajectory is constructed by simulating the nominal (deterministic) WECS model using a high-fidelity Runge-Kutta solver. Each method’s accuracy is evaluated as an average relative deviation from this reference, while computational time considers both model formulation and solution time through the YALMIP toolbox with the SDPT3 solver. All aforementioned performance parameters were analyzed from the perspective of accuracy, maximum error, feasibility rate, and computation time throughout a 10-second simulation horizon. The scalability of the PCIM with increasing order of the system is also assessed. PCIM results are seen to provide the greatest accuracy and complexity among the methods considered, providing excellent prospects for use in real-time robust control for WECS. Accuracy and computational time comparison The accuracy of each method is defined as follows:44\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:Accuracy=1-\frac{{\parallel{x}_{ref}-{x}_{int}\parallel}_{2}}{{\parallel{x}_{ref}\parallel}_{2}}$$\end{document} where, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{ref}$$\end{document} stands for the state trajectory of the nominal system, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{x}_{int}$$\end{document} is the midpoint of the interval solution. With increasing accuracy, the bounds become tighter, and the considered method can be said to approximate the behavior of the true system better. Figure (6) shows the accuracy versus computational time. Fig. 6Accuracy vs. computational time. Accuracy vs. computational time. According to the results, PCIM has the highest accuracy of 98.5%, using the least computational time of 10.2 s. The lowest accuracy (90.5%) is recorded for MCM because of limited sampling as low as 1000 samples, which do not cover the uncertainty space adequately. NIM has severe overestimation because of wrapping, and TIM, the worst among others, although more accurate, is too computationally expensive owing to high-order Taylor expansions. CIM improves on NIM but does not equal PCIM in accuracy. The efficiency of PCIM is due to the Chebyshev interpolation, which reduces approximation errors and is fast in solutions based on LMI. 98.5% effectiveness brought into question the metric’s set-theoretic meaning if it statistically referred to accuracy. For this reason, accuracy has been rigorously redefined in threefold complementing metrics that are reproducible under robust control and interval analysis criteria: (i) The Interval Containment Rate (ICR) is a measure of the times at which the nominal reference trajectory lies within the computed interval envelope, (ii) The Normalized Mean Absolute Deviation (NMAD) measures the average absolute distance between the interval midpoint and the nominal trajectory normalized by the reference range, and iii) Worst-Case Bound Width Ratio (WBR)-maximum ratio of interval width to reference magnitude across the horizon. These metrics are then evaluated over 100 Monte Carlo realizations of ± 15% parameter uncertainty and stochastic wind inputs, as summarized in Table 3. Table 3Accuracy metrics for interval-based WECS control methods (± 15% uncertainty, 10-s horizon).MethodICR (%)NMAD (%)WBR (%)Computational time (s)PCIM (Proposed)1001.5310.2NIM854.29.88.7CIM903.17.512.4TIM952.35.228.6MCM (1,000 samples)80512.125.5 Table 3Accuracy metrics for interval-based WECS control methods (± 15% uncertainty, 10-s horizon).MethodICR (%)NMAD (%)WBR (%)Computational time (s)PCIM (Proposed)1001.5310.2NIM854.29.88.7CIM903.17.512.4TIM952.35.228.6MCM (1,000 samples)80512.125.5 Accuracy metrics for interval-based WECS control methods (± 15% uncertainty, 10-s horizon). where, ICR, NMAD, and WBR stand for Interval Containment Rate, Normalized Mean Absolute Deviation, and Worst-Case Bound Width Ratio, respectively. This composite interpretation is now justified by the statement that 100%-WBR/2 = 98.5% for effective accuracy, relating “tightness and fidelity” of the enclosure itself, not the probabilistic success rate. Interval containment rate (ICR) The interval methods need to ensure that the actual or nominal system performance does not exceed the anticipated limits at the entire admissible uncertainties. This is measured by ICR, which quantifies how often time steps exist; the high-fidelity nominal trajectory (simulated with precise parameters and no uncertainty) falls within the PCIM-computed interval envelope. PCIM attains ICR = 100% as illustrated in Table 3, which mathematically rigorously validates its vertex-based LMI formulation of linearized WECS dynamics (Sect. 3.3). Monte Carlo using 1,000 samples, by contrast, has only 80% ICR because it cannot sample all of the elements, whereas NIM and CIM have either over-conservatism (NIM) or under-coverage (CIM). Such a 100% ICR is confirmation that PCIM provides hard, non-probabilistic guarantees that are required to execute safety-critical WECS. Normalized mean absolute deviation (NMAD) Although containment is a safe technique, the fidelity is monitored by NMAD, which is the time-average error between the interval midpoint and the nominal reference and is expressed as a ratio of the reference range (0.5–1.0.5.0 pu). The NMAD = 1.5 in PCIM is much lower than NIM (4.2%), CIM (3.1%), TIM (2.3%), and MCM (5.0%) as in Table 3. The small deviation here validates the fact that the bounds of PCIM are not merely only safe, but are also centered very close to the desired trajectory, which reduces the amount of conservatism that is needless. Because of this piecewise Chebyshev approximation, which does not propagate the wrapping effect, it better aligns the midpoint of the interval to nominal dynamics than classical interval arithmetic does. Worst-case bound width ratio (WBR) Bound tightness is important regarding the efficiency of control and actuator wear. WBR is the ratio of the width of the envelope containing the interval:45\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{W}\text{B}\text{R}=\underset{t\in\:[0,T]}{\text{m}\text{a}\text{x}}\:\left(\frac{{x}_{\text{u}\text{p}}\left(t\right)-{x}_{\text{l}\text{o}\text{w}}\left(t\right)}{\left|{x}_{\text{r}\text{e}\text{f}}\left(t\right)\right|}\right)\times\:100\text{\%}$$\end{document} PCIM has WBR = 3.0% at t = 0.5 s (peak transient) compared to NIM (9.8%), CIM (7.5%), TIM (5.2%) and MCM (12.1%) which has much broader bounds (Table 3). This constriction is directly what allows the reduction of control effort and mechanical stress (Figs. 7, 8, 9, 10). Importantly, the initial 98.5% accuracy can now be reinterpreted to mean 100%-WBR/2 = 100 − 1.5 = 98.5% which is the percentage of the reference signal that is not absorbed by uncertainty-induced spread. The tripartite accuracy evaluation in Table 3 addresses the uncertainty of the initial claim of 98.5% accuracy by grounding the performance in three different underlying, interpretable dimensions: ICR, NMAD, and WBR. PCIM is the only method to do all three without compromising precision, implying that it can perform perfect containment, which is inaccessible to sampling-based (MCM) and classical interval (NIM, CIM) approaches. Analysis of error and feasibility In addition, the absolute maximum error of reference and midpoint interval in comparison is determined for further assessment of robustness. The part of time the reference trajectory spends inside the predicted interval should also be denoted as the feasibility rate to evaluate reliability. Figure (7) shows the error and feasibility analysis. Fig. 7The error and feasibility analysis: (A) Maximum state deviation, (B) Percentage of time Reference in interval. The error and feasibility analysis: (A) Maximum state deviation, (B) Percentage of time Reference in interval. As can be observed, in Fig. (7-A), PCIM has a maximum state deviation of 0.015 pu, which is less than that of NIM (0.03 pu), CIM (0.025 pu), TIM (0.02 pu), and MCM (0.04 pu). Also, Fig. (7-B) shows that PCIM feasibility = 100% (always within its limits), compared to 85% (NIM), 90% (CIM), 95% (TIM), and 80% (MCM). Therefore, the PCIM piecewise approximation contains the state well. Neither MCM nor NIM do not contains the reference trajectory for long periods of time during the simulation, indicating poor robustness. High feasibility in PCIM results because it has the capability of tightly bounding nonlinear dynamics by piecewise Chebyshev approximation. That characteristic is crucial for safety-critical applications like wind turbine control, where breaking state constraints may lead to mechanical damage or grid disconnection. Scalability analysis Scalability is assessed via the computation time of the PCIM for the increase in order of the system from 2 to 10 states, representing increased fidelity WECS models (e.g., NREL 5 MW turbine20. The number of sub-intervals and polynomial degree were kept constant. Fig. 8Scalability of PCIM vs. nominal LQR. Scalability of PCIM vs. nominal LQR. As observed from Fig. (8), the scalability analysis indicates that the PCIM operates on approximately \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:O\left({n}^{2.3}\right)$$\end{document}, with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:n$$\end{document} being the system order is primarily determined by the complexity of computations involving LMIs. While the increase in computation time is moving toward that of nominal LQR, it remains below real-time control limits (e.g., less than 100 ms for the 10-state system in simulation). This indicates that PCIM can be applied not only to reduced-order models, but even to very high-fidelity WECS models like the NREL 5 MW turbine. The piecewise formulation contributes to the prevention of the exponential growth of complexity and thus makes it comparatively more scalable than TIM or MCM, which scale at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:O\left({n}^{3}\right)$$\end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:O\left({e}^{n}\right)$$\end{document} levels, respectively. Practical analysis A comparative analysis with the LQR-Backstepping controller by Mebkhouta et al.9, a modern, sophisticated technique, a combination of optimal and nonlinear control, is performed to get a rigorous evaluation of the proposed PCIM. The simulations are run with the same conditions of parameter uncertainty of 15% of air density (r), rotor inertia (J), and generator inductances (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{L}_{d},\:{L}_{q}$$\end{document}), a 10 s horizon, and 1 s timestep. The findings center on control input curves, state activity, the absolute error (IAE), and settling time to illustrate the high performance of PCIM. Control input comparison The response of the control input of the proposed PCIM approach and the benchmark LQR-Backstepping controller is shown to review their performance in regulating the behavior of the system over time. The relative directions of the two control strategies are shown in Fig. (9), and the PCIM provides a more stable and smooth control signal over the time horizon. In comparison, the LQR-Backstepping technique results in larger control effort together with stronger variations, which leads to a more aggressive actuation profile. Fig. 9Control input comparison. Control input comparison. The results of the comparison indicate that the PCIM method may be able to offer good control at reduced input levels, i.e., less actuation energy and perhaps reduced physical system loads. This is further on the real-world application, where the actuator constraints and energy efficiency are highly valued factors. The LQR-Backstepping controller is stable, but the bigger and more variable control signals could result in more wear on the system components, as well as a lower system robustness. Thus, the PCIM strategy also provides a good trade-off between control performance and efficiency, which makes it a more viable and sustainable choice in real-world solutions. State trajectories The dynamic behavior of the generator with the proposed PCIM strategy and the benchmark LQR-Backstepping controller is compared regarding the speed and the torque trajectories. Figure (10) shows the time dynamics of the generator states, with the top subplot showing the response of the speed and the bottom subplot showing the associated profile of the torque. Both controllers provide a smooth convergence of the generator speed to the rated value, and the torque is modified to help both the transient and steady-state dynamics of the system. Fig. 10State trajectories. State trajectories. The comparison outcomes indicate that the PCIM technique has a higher rate of convergence of the generator speed with a reduced number of oscillations than LQR-Backstepping, indicating that it has better transient response. Meanwhile, the torque requirement at PCIM is always lower, and the mechanical stress of the generator shaft and drive system decreases. However, the LQR-Backstepping controller needs greater torque effort during the temporary phase, which might cause more wear and use of energy in real-life systems. In general, the findings emphasize that the PCIM approach is not only the way to ensure the right control of a state but also offers the more energy-efficient and hardware-friendly control action, which makes it a more realistic method for its practical implementation. Comparative analysis of integral absolute error The effectiveness of the presented PCIM strategy is also evaluated with the help of the widely used criterion of the Integral Absolute Error (IAE), which is a quantitative index to evaluate the overall tracking accuracy and control efficiency. Figure (11) compares the bar graphs of PCIM and LQR-Backstepping controllers and they clearly indicate that they perform relatively well in reducing cumulative error over the duration of the simulation. Fig. 11Performance comparison. Performance comparison. It can be seen that the value of IAE in PCIM (0.12) is less than that of LQR-Backstepping (0.18), which is the best in reducing tracking deviations. This improvement means that the PCIM controller will be capable of giving a more correct regulation of trajectory and, at the same time, smooth dynamics. Conversely, the increased IAE of the LQR-Backstepping controller indicates more cumulative error, which means less effective tracking and is likely to require more control work. These results support the benefit of PCIM to attain the desired control performance of accurate and reliable performance with smaller error accumulation. The PCIM is superior to the LQR-Backstepping approach because it keeps state boundaries more rigid and control effort reduces in case of uncertainties. The resulting LMI-based design does not have the chattering problems of sliding-mode designs (e.g5,8, and converges more quickly than stochastic MPC designs (e.g6,10. This confirms the applicability of PCIM to real-time WECS control, where computational efficiency (10.2s runtime) and robustness are of key concern. The future research will also expand PCIM to stochastic-interval hybrids to be more useful. Monte Carlo robustness validation with hybrid uncertainties The Monte Carlo analysis is executed for 10,000 random trials to assess quantitatively the robustness under combined stochastic and interval uncertainty. In each trial, system parameters such as air density, inertia, and inductance are sampled by a uniform distribution within their ± 15% intervals, with the wind speed being drawn from a Gaussian distribution \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\mathcal{N}\left(12,\:{1.5}^{2}\right)$$\end{document}. The PCIM controller (designed once with only interval bounds) was then applied to all realizations. Fig. 12Monte carlo robustness validation. Monte carlo robustness validation. Figure (12-A) shows a histogram of the final generator speed errors (at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:t=10$$\end{document} sec), where 99.2% of the trials achieved \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left|error\right|<0.03$$\end{document} pu. Figure (12-B) plots the cumulative distribution of the Integral Absolute Error (IAE). PCIM achieves a median \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:IAE=0.118$$\end{document}, significantly lower than LQR-Backstepping (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:median=0.185$$\end{document}). Notably, no trial violated operational constraints (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:speed<0.95\:pu$$\end{document} or \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:>1.05\:pu$$\end{document}), confirming feasibility. Large-scale validation affirms that PCIM’s interval-based design establishes a robustness envelope over a broad range of hybrid uncertainty realizations, thereby allaying concerns about the practical use of purely interval methods. Monte Carlo performance and constraint violation statistics There are 100 Monte Carlo simulations using independent Kaimal wind profiles and ± 15% changes in the parameters to make statistical comparisons of the robustness. On a trial-by-trial basis, we calculate (i) Integral Absolute Tracking Error (IATE) and (ii) cumulative time out of the safe range of speed [0.95, 1.05] pu. Figure (13) shows Monte Carlo performance and constraint violation statistics. Fig. 13Monte carlo performance and constraint violation statistics. Monte carlo performance and constraint violation statistics. The left-hand side in Fig. (13) indicates that PCIM has a median IATE of 0.132, which is a little bigger than SMPC has a median of 0.118 but much smaller in terms of variance-indicating stable performance. More importantly, the right-hand size in Fig. (13) shows that PCIM exhibits zero constraint violations on all 100 trials, whereas SMPC has 14-percent constraint violations (median constraint violation: 1.2 s). This demonstrates that the interval-based design of PCIM has hard safety guarantees that probabilistic SMPC cannot offer even with chance constraints. It is a small sacrifice in average tracking error, but that is acceptable in most grid-code-compliant WECS systems where compliance to constraints is of the greatest importance. Conclusions This study presented a new Piecewise Chebyshev Inclusion Method (PCIM) for robust optimal control of Permanent Magnet Synchronous Generator (PMSG)-based Wind Energy Conversion Systems (WECS) under interval parameter uncertainties. The limitation of the classical interval analysis techniques was overcome using this method and the Chebyshev polynomial approximations so as to minimize the wrapping effect and provide tighter bounds on the system trajectories. By reformulating the finite-horizon Linear Quadratic Regulator (LQR) problem using the convex Linear Matrix Inequality (LMI) framework, the PCIM allows the design of a closed-loop optimal controller that is both robust and computationally tractable. Simulation results testify that the PCIM achieves a 98.5% accuracy in state bounding and reduces computation time by 60% compared to the Monte Carlo Method (MCM) while being superior in error bounds and feasibility to the NIM, CIM, and TIM. It has a 15% faster settling time and 32% lower IAE under ± 15% parameter variations, thus proving the excellent robustness and efficiency of the controller. However, the present approach assumes bounded but non-stochastic uncertainties, which may not capture the real-life scenario where uncertainties possess both probabilistic and interval characteristics. The method has worked well in simulations, but it is yet to be tested for real-time performance on hardware platforms. Concerning future works, there are three main directions: (1) extending PCIM to incorporate hybrid stochastic-interval uncertainties for more realistic modeling; (2) implementing the controller on a real-time digital simulator (RTDS) or hardware-in-the-loop (HIL) platform to verify its practical feasibility; and (3) investigating applicability on high-fidelity models like the National Renewable Energy Laboratory (NREL) 5 MW turbine to evaluate scalability in large-scale systems. Advancing in this direction would further consolidate the path from theoretical robust control to industrial deployment concerning current wind energy systems.