Experimental validation of model-free predictive control based on the active vector execution time for grid-connected system Authors: Zakaria Lammouchi, Chouaib Labiod, Kamel Srairi, Mohamed Benbouzid, Abderrahmane Khechekhouche, Fahad Albalawi, Sherif S. M. Ghoneim, Enas Ali, Ahmed B. Abou Sharaf Published: 2024 ABSTRACT The application of Model-Free Predictive Control (MFPC) in power electronic systems has garnered increasing attention. In this paper, MFPC control based on replacing the classical factory model with an ultra-local model (ULM) is studied. Generally, the Integral Sliding Mode Observer (ISMO) is used to estimate the unknown part in the ULM where the non-physical factor in the ULM is selected with approximate values ​​ranging from the nominal value of the system which is contrary to the concept of MFPC control. In this research, an improved adaptive integral sliding mode observer (AISMO) based MFPC (AISMO-MFPC ) is proposed to estimate this factor with the unknown function in the ULM equation. The new observer design enables the estimation of this factor based on the current error, which allows for independent prposedcontrol of the system parameters.To obtain the lowest current ripple, the concept of active vector execution time (AVET ) has been incorporated into the proposed control where two vectors are selected in the sampling period to minimize the cost function instead of selecting a single vector. ULM is also used to calculate AVET which facilitates the implementation of the imposed control. The combination of the proposed AISMO-MFPC and AVET gives faster system response and reduces the current ripple and lower harmonics, especially in case of mismatch parameters. Finally, the effectiveness of the proposed control is confirmed under various conditions by the presented simulation and experimental results. Introduction Predictive control strategies have been developed over in recent years remarkably in power electronic, due to their crucial role in energy conversion, such as renewable energy systems, and grid-connected renewable energy systems1. Finite state model prediction control (FS-MPC) is one of the strategies model prediction control (MPC) that adopts the predictive model to achieve system control2, the FS-MPC is based on a cost function consisting of the error between the predicted variables and their reference values3–5. It is necessary to evaluate all vectors related of power electronic to determine the optimal effort6,7. Despite the good performance of MPC, this technique requires very accurate system parameters8,9, if these parameters contain measurement errors the performance of MPC can be greatly destroyed. A recent approach known as Model-Free Predictive Control (MFPC) has emerged as an alternative to dealing with parameter variations and solving problems found in MPC10,11.the authors of12 proposed a MFPC approach where motor current variations are measured and dynamically stored in a look-up table. Using these measurements, the approach predicts current variations. However, the frequency with which measurements are updated significantly impacts the performance of this method. MFPC has been proposed proposed in13, which uses a recursive least-squares (RLS) algorithm in the current prediction model. This approach recursively predicts the next-step current based on previous current samples. In14, the parameters of the autoregressive model with external inputs (ARX) are determined using an iterative least squares algorithm to control the current in the power converter. While this approach offers a significant advantage due to its high degree of flexibility and independence from specific system parameters and model phenomena15, it suffers from notable drawbacks, most notably its computational burden and dependence on the initial value. Model-Free Predictive Control (MFPC) using an ultra-local model (ULM ) is an approach that bridges the gap between completely model-free techniques and those that utilize models is proposed in16. An ULM is a simplified or highly localized model of the system that only captures very short-term or local behavior rather than the full dynamics of the system, the input directly articulates the derivatives of the system output through a linear model ULM3. In this framework, an unknown part summarizes all unknown system dynamics and potential disturbances, which necessitates estimation through online identification techniques6,16, which needs a complex mathematical process and necessitates employing high-speed and costly numerical processors. To overcome this problem, an observer-based MFPC scheme has been proposed in6, which estimates the unknown part of the system by a state observer. The extended state observer (ESO) has been applied in17,18 to determine the states of ULM. Although, the results in this work showed remarkably robustness of control, its drawbacks lie in that it is used as a sub-part rather than the main part of the prediction model in most researches19,20. In21, the ESO is proposed for ULM to simplify MFPC. In contrast, it finds that the unknown function in ULM has a non-linear character and does not have a constant value that may increase the estimation error. As discussed in22, the unknown part in ULM is estimated by using a sliding mode observer (SMO). However, the invariance principle is not preserved during the transition to the sliding manifold in the reaching phase. Fortunately, this issue can be effectively addressed through the utilization of an integral sliding mode observer (ISMO) based MFPC(ISMO-MFPC)23. ISMO is used to estimate the unknown part in the ULM where the non-physical factor in the ULM is selected with approximate values ​​ranging from the nominal value of the system. It is proved in24,25 that factor is chosen with approximate values ranging between \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\pm\:50\mathbf{\%}$$\end{document} of the nominal value of the system, which is contrary to the concept of MFPC control.This paper proposes an adaptive integral sliding mode observer (AISMO) based Model-Free Predictive Control (AISMO- MFPC) to estimate the unknown function and factor variation in ULM, where this factor depends on the error current. In26, the MFPC control uses the sampling period to select one voltage vector for the all-candidate vectors according to minimize the cost function.The delay problem in the predictive controller may degrade the controller’s performance. Numerous efforts have been dedicated in literature to enhance switching sequences of the inverter such as the constant switching frequency has been proposed in27, or to improve the cost function such as the multi-objective cost function in7, in order to minimize output ripples.This paper introduces an approach termed Execution Time of Active Vector with Model-Free Predictive Control tailored for two-level inverters. The execution time AVET is calculated using the ULM equation. With the proposed control, there is good tracking and less computational burden as it uses only six voltage vectors instead of all eight candidate vectors in the cost function. In this paper, the main contribution is to introduce an adaptive integral sliding mode observer based MFPC (AISMO-MFPC ) is proposed to estimate the non-physical factor in the ULM equation with the unknown function, since this factor has no depend with the system parameters. While it varys with the change of the current error, This variation appears when the current values ​​change or during the mismach parameters, leading to a faster response and reducing the resulting current ripple and lower harmonics. Moreover, the integration of the concept of active vector execution time (AVET) in the proposed control contributes to reducing the computational time with better tracking where the cost function minimization process is done at this time with the least possible vectors for two-level inverter. Using ULM to calculate AVET also contributes to simplifying the implementation of the imposed control. The effectiveness and robustness of the proposed control are verified confirmed under various conditions by the presented simulation and experimental results .and compared with conventional methods. Conventional model-free predictive control Current prediction Figure 1 illustrated the power circuit of the system, the 2 L-VSI has three legs and each leg contains two switches. The two levelinverter coupled to an R-L load. The connection can simulate a grid-tied PV power generation system. Fig. 1Block diagram of a 2 L-VSI circuit feeding an R-L load. Block diagram of a 2 L-VSI circuit feeding an R-L load. The mathematical expression for this representation is written as follows28 :1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{v}}_{i}=R{\overrightarrow{i}}_{l}+L\frac{d{\overrightarrow{i}}_{l}}{dt}+\overrightarrow{E}$$\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}$$\:{\overrightarrow{v}}_{i}$$\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}$$\:{\overrightarrow{i}}_{l}$$\end{document} are the voltage vector generated by the inverter and the load current vector.The dynamics of Eq. (1) can be written as follows:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{\text{d}{i}_{l\alpha\:\beta\:}}{dt}=\frac{1}{L}\left({\overrightarrow{v}}_{i}-R{\overrightarrow{i}}_{l}-\overrightarrow{E}\right)$$\end{document} The output current prediction can be formulated using first-order Euler approximation as follows29:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)=\frac{{T}_{s}}{L}\left(-R{\overrightarrow{i}}_{l}\left(k\right)-\overrightarrow{E}\left(k\right)+{\overrightarrow{v}}_{i}\left(k\right)\right)+{\overrightarrow{i}}_{l}\left(k\right)$$\end{document} The predictive control based of MPC utilizes a mathematical model of the system to predict the future behavior of the output current. Nevertheless, this approach encounters various several style of uncertainty, including unmodeled dynamics and parameter mismatch. This paper employs the Ultra-Local Model (ULM) to address the issue of model uncertainty. Ultra-local model The variations in the input and output of system can be roughly described using a finite-dimensional ordinary differential equation30.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varGamma\:\left(t,y,\dot{y},\ddot{y},\dots\:{y}^{\left(n\right)},u,\dot{u},\ddot{u},\dots\:{u}^{\left(m\right)}\right)=0$$\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}$$\:\varGamma\:$$\end{document} is an unknown nonlinear function,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:y$$\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}$$\:u$$\end{document} are the output and input signalsof the system, and \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} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:m$$\end{document} are the control output and input orders of the system, respectively.The ULM principle can be used to simplify (4) as21 :5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{y}^{\left(n\right)}=F+\lambda\:u$$\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}$$\:{\:y}^{\left(n\right)}$$\end{document} is the derivative of order\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}of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\text{y}$$\end{document}. In this context, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:F$$\end{document} is defined as the unknown terms; it consists of the all-diverse type of uncertainty in the ULM, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\uplambda\:}$$\end{document} is a nonphysicalfactor .The value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:F$$\end{document} is constantly updated using system information (input and output), and is just an estimate, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:F$$\end{document} is replaced by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\widehat{F}$$\end{document} as an indication of the estimate using the algebraic definition technique31, the basic structure of ULM is shown in Fig. 2. Moreover, the algebraic identification technique is an online estimation approach, and it was used to estimate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\widehat{F}$$\end{document} in real-time. However, classical MFPC uses complex techniques and takes longer time to estimate the unknown part28,32. Equation (5) can be rewritten as Eq. (6) :6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{y}^{\left(n\right)}=\widehat{F}+\lambda\:u$$\end{document} Fig. 2Structure of the ultra-local model. Structure of the ultra-local model. The concept of an ultra-local model has been incorporated into Model-Free Predictive Control theory to simplify the complexity associated with mathematical model. The following relation of the first-order (n = 1) ULM model is defined as21:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{y}=\widehat{F}+\lambda\:u$$\end{document} Conventional ISMO-MFPC Substituting the dynamic equation of system (2) into the ULM model, the first-order ULM of system (7) can be written as in32:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{\overrightarrow{i}}_{l}}{dt}=\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}$$\end{document} From Eq. (8), the structure of ULM is independent parameters of model. Also, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\widehat{F}$$\end{document} are time-varying unknown functions. In MFPC, the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\widehat{F}\:$$\end{document}under goes continuous updates that are determined by utilizing input and output measurements as shown in Fig. 3. Fig. 3Conventional ISMO-MFPC scheme. Conventional ISMO-MFPC scheme. According to the Eq. (8), the derivative current during sampling period \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{T}_{s}$$\end{document} can be expressed using first-order Euler approximation of discrete signal29 as :9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{\overrightarrow{i}}_{l}}{dt}\approx\:\frac{{\overrightarrow{i}}_{l}\left(k+1\right)-{\overrightarrow{i}}_{l}\left(k\right)}{{T}_{s}}=\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}$$\end{document} The current prediction at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left(k+1\right)$$\end{document}can be written as:10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{i}}_{l}\left(k\right)+{T}_{s}\left(\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}\right)$$\end{document} The current references utilized in the cost functions in the α-β frame are derived at (k + 1). it can be estimated using the 2nd order formula of the Lagrange extrapolation as follows33:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{\alpha\:,\beta\:}^{\text{*}}\left(k+1\right)=3{\overrightarrow{i}}_{\alpha\:,\beta\:}^{\text{*}}\left(k\right)-3{\overrightarrow{i}}_{\alpha\:,\beta\:}^{\text{*}}\left(k-1\right)+3{\overrightarrow{i}}_{\alpha\:,\beta\:}^{\text{*}}\left(k-2\right)$$\end{document} The cost function of the MFPC can bewriting to obtain optimal vector among the eight feasible ones with a 2-level VSI, which directly compares between the reference vector and candidate vector as17:12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\left|i=1:8\right.}=abs\left({i}_{\alpha\:}^{\text{*}}-{i}_{l\alpha\:}\left(k+1\right)\right)\:+abs\left({i}_{\beta\:}^{\text{*}}-{i}_{l\beta\:}\left(k+1\right)\right)$$\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}$$\:{i}_{l\alpha\:}\left(k+1\right)$$\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}$$\:{i}_{l\beta\:}\left(k+1\right)$$\end{document} are the components of load current prediction respectively in the α-β frame. From the expression of the cost function in (12), the traditional ISMO-MFPC control uses the sampling period to select one voltage vector from the all-candidate vectors, as shown in Fig. 5 .The proposed controluses only six candidates to minimize the cost function instead of all eight possibilities. Fig. 4 Simulation Bloc of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:\lambda\:$$\end{document}. Simulation Bloc of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:\lambda\:$$\end{document}. ISMO observer However, this technology uses complex techniques and is more time-consuming to estimate the unknown part. An observer ISMO-based MFPC (ISMO-MFPC) scheme has been proposed in21, which is estimated as the unknown part of the system, The ISMO design for the ULM was presented as follows:13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left|\begin{array}{l}{\overrightarrow{\varvec{e}}}_{\varvec{i}}={\overrightarrow{\varvec{i}}}_{\varvec{l}}-\overrightarrow{{\widehat{\varvec{i}}}_{\varvec{l}}}\\\:\frac{\varvec{d}\overrightarrow{{\widehat{\varvec{i}}}_{\varvec{l}}}}{\varvec{d}\varvec{t}}=\widehat{\varvec{F}}+\varvec{\lambda\:}{\overrightarrow{\varvec{v}}}_{\varvec{i}}+\overrightarrow{{\varvec{u}}^{\varvec{{\prime\:}}}}\\\:\frac{\varvec{d}\widehat{\varvec{F}}}{\varvec{d}\varvec{t}}={\overrightarrow{\varvec{e}}}_{\varvec{i}}+{\varvec{K}}_{\varvec{i}}{\int\:}_{0}^{\varvec{t}}\varvec{s}\varvec{g}\varvec{n}\left({\overrightarrow{\varvec{e}}}_{\varvec{i}}\right)\varvec{d}\varvec{\tau\:}\end{array}\right.$$\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}$$\:{\varvec{i}}_{\varvec{l}}$$\end{document} is the actual load current and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\widehat{\varvec{i}}}_{\varvec{l}}$$\end{document} is the estimated current, the important signal \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\overrightarrow{{\varvec{u}}^{\varvec{{\prime\:}}}}$$\end{document} denotes the input of ISMO. The coefficient of ISMO \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{K}}_{\varvec{i}}$$\end{document} should be chosen as positive values where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\varvec{K}}_{\varvec{i}}$$\end{document} is chosen such that the state vector is rapidly pushed in a neighborhood of the sliding surfaces. On the other hand, the factor \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varvec{\lambda\:}$$\end{document} in the ULM is chosen with approximate values ranging between \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\pm\:50\mathbf{\%}$$\end{document} of the nominal value of the system34, which is contrary to the concept of MFPC control. This paper proposes an adaptive integral sliding mode observer (AISMO) to estimate the unknown function and the factor variation in ULM, where this factor depends on the error current. Proposed AISMO–MFPC-(AVET) control To improve control performance by reducing the output current ripple signal and the robustness in case parameter mismatch. An adaptive integral sliding mode observer based MFPC (AISMO-MFPC ) is proposed to estimate the non-physical factor in the ULM equation and the unknown function. Moreover, incorporating the concept of Active Vector Execution Time (AVET) into the proposed control helps reduce the computation time with better traceability as the cost function minimization process is done with fewer possible vectors for a two-level inverter than conventional control. Adaptive AISMO observer This paper proposes an adaptive observer AISMO to observer the unknown function and factor variation in ULM where this factor is depended of error current, the factor changes as the current error change. In21, the sgn function was proposed in the ISMO design. This function introduces abrupt changes in the control inputs, which may cause the system to switch quickly between states due to the discontinuous nature of the signal function.To overcome this problem, this paper proposes replacing the sgn function by a continuous approximation function, such as the sigmfunction35, which takes values in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left]-1\:;1\right[$$\end{document}. Thus, the sigm function in the improved AISMO helps to achieve the desired performance and reduce the sensitivity to disturbances, the proposed observer AISMO is designed as follows:14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left|\begin{array}{l}{\overrightarrow{e}}_{i}={\overrightarrow{i}}_{l}-\overrightarrow{{\widehat{i}}_{l}}\\\:\frac{d\overrightarrow{{\widehat{i}}_{l}}}{dt}=\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}+\overrightarrow{{u}^{{\prime\:}}}\\\:\begin{array}{l}\frac{d\widehat{F}}{dt}={\overrightarrow{e}}_{i}+{K}_{i}{\int\:}_{0}^{t}Sigm\left({\overrightarrow{e}}_{i}\right)d\tau\:\\\:\frac{d\lambda\:}{dt}=-{\delta\:}_{1}\lambda\:\left({\delta\:}_{2}\left(\lambda\:-1\right)-\frac{{\delta\:}_{3}\left|{\overrightarrow{e}}_{i}^{*}\right|}{1+min\left({\delta\:}_{4\:},P\right)}\right)\end{array}\end{array}\right.$$\end{document} In order to calculate the evaluation of\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:\lambda\:$$\end{document}, the error \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{e}}_{i}^{*}$$\end{document} (i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{e}}_{i}^{*}$$\end{document} =\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}^{*}-\overrightarrow{{\widehat{i}}_{l}}$$\end{document}) is used, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}^{*}$$\end{document} is the reference of the current. The function \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} represents the power of the output observation error; it is given by the following equation36:15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:P=\frac{1}{T}{\int\:}_{\text{m}\text{a}\text{x}(0,t-T)}^{t}{\left|{\overrightarrow{e}}_{i}^{*}\right|}^{2}d\tau\:$$\end{document} As demonstrated in (14), it is composed of four meticulously selected parameters, namely \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\updelta\:}}_{1},\:{{\updelta\:}}_{2},{{\updelta\:}}_{3}$$\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}$$\:{{\updelta\:}}_{4\:}$$\end{document}. On the other hand, the parameter \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\updelta\:}}_{3\:}$$\end{document} can be assigned high values, primarily intended to saturate the integral term of P.The design of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document} in (14) allows the observer gain to take high values, which leads to the paths of the proposed observer converging exponentially and quickly to the trajectory of the system. As a result, the observation error quickly disappears, after which the values of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document} decrease to the predetermined value. The scheme of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document} is presented in Fig. 5. Fig. 5 Conventional ISMO-MFPC and Proposed AISMO-MFPC-(AVET) in one sampling period, (a): Conventional ISMO-MFPC, (b): AISMO-MFPC-(AVET). Conventional ISMO-MFPC and Proposed AISMO-MFPC-(AVET) in one sampling period, (a): Conventional ISMO-MFPC, (b): AISMO-MFPC-(AVET). On the other hand, the important signal \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}^{{\prime\:}}$$\end{document}denote the input of AISMO observer, in order tocaculate \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\overrightarrow{{u}^{{\prime\:}}},\:$$\end{document}the Lyapuno stability theorem is used to design \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\overrightarrow{{u}^{{\prime\:}}}$$\end{document}. MFPC-(AVET) control In MFPC-(AVET) Control, two vectors are selected at one sampling period. The active vector that uses the execution time of the active vector to select it, while the second is a zero vector as it is applied after the active vector. The active vector is executed at a certain time (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document}), and the zero vectors are enabled at the period’s remaining time (tZ) to mitigate the changing direction. Thus, a sample period is split up as18:16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{T}_{s}={t}_{a}+{t}_{z}$$\end{document} Therefore, the one period is spilt into (t(k) to t(k + \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document})) and(t(k + \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document}) to t(k + 1)), as shown in Fig. 4. (a): Conventional ISMO-MFPC, (b): AISMO-MFPC-(AVET). At the instant t(k +\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document}), the current predictions \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+{t}_{a}\right)$$\end{document}are obtained by directly applying the six feasible vectors (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{v}}_{i\left|1\dots\:6\right.}$$\end{document}). All other six feasible vectors are shown in Table 1. Table 1Six feasible active voltage vectors. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{i}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{2}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{3}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{4}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{5}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{6}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{i}$$\end{document} 100110010011001101According to (8), the derivative current during \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document} can be expressed using first-order Euler approximation of discrete signal29 as :. 17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{\overrightarrow{i}}_{l}}{dt}\approx\:\frac{{\overrightarrow{i}}_{l}\left(k+{t}_{a}\right)-{\overrightarrow{i}}_{l}\left(k\right)}{{t}_{a}}=\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}$$\end{document} Table 1Six feasible active voltage vectors. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{i}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{2}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{3}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{4}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{5}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{v}_{6}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{S}_{i}$$\end{document} 100110010011001101According to (8), the derivative current during \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document} can be expressed using first-order Euler approximation of discrete signal29 as :. Six feasible active voltage vectors. According to (8), the derivative current during \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document} can be expressed using first-order Euler approximation of discrete signal29 as :. The current prediction at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left(k+{t}_{a}\right)$$\end{document}write as:18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+{t}_{a}\right)={\overrightarrow{\widehat{i}}}_{l}\left(k\right)+{t}_{a}\left(\widehat{F}+\lambda\:{\overrightarrow{v}}_{i}\right)$$\end{document} As shown in Fig. 4. Clearly, the predicted current at (k + 1) can be obtained by directly applying the zero vectors \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{(V}}_{i,\left|7\:or\:8\right.})\:$$\end{document}between t=(k + \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document}) and t= (k + 1) as :\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{d{\overrightarrow{i}}_{l}}{dt}\approx\:\frac{{\overrightarrow{i}}_{l}\left(k+1\right)-{\overrightarrow{i}}_{l}\left(k+{t}_{a}\right)}{{t}_{z}}=\widehat{F}+\lambda\:{\overrightarrow{v}}_{i\left|7\:or\:8\right.}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{\widehat{i}}}_{l}\left(k+{t}_{a}\right)+{t}_{z}\left(\widehat{F}+\lambda\:{\overrightarrow{v}}_{i\left|7\:or\:8\right.}\right)$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{\widehat{i}}}_{l}\left(k+{t}_{a}\right)+{t}_{z}\left(\widehat{F}+{\uplambda\:}0.0\right)$$\end{document}19\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{\widehat{i}}}_{l}\left(k+{t}_{a}\right)+{t}_{z}\widehat{F}$$\end{document} AVET calculation Combining Eqs. (18) and (19), one can obtain the execution time of the active vector. The predictedcurrent in (k + 1) can beexpressed as :\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{\widehat{i}}}_{l}\left(k+{t}_{a}\right)+{t}_{z}\widehat{F}$$\end{document}20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:={\overrightarrow{\widehat{i}}}_{l}\left(k\right)+{t}_{a}\left(\widehat{F}+{\uplambda\:}{\overrightarrow{v}}_{i\left|1,0.6\right.}\right)+{t}_{z}\widehat{F}$$\end{document} Where the zero vectors is enabled at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{z}$$\end{document}24 :\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{z}={T}_{s}-{t}_{a}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)={\overrightarrow{\widehat{i}}}_{l}\left(k\right)+{t}_{a}\left(\widehat{F}+{\uplambda\:}{\overrightarrow{v}}_{i\left|1,0.6\right.}\right)+\left({T}_{s}-{t}_{a}\right)\widehat{F}$$\end{document}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:=\:{\overrightarrow{\widehat{i}}}_{l}\left(k\right)+{t}_{a}\widehat{F}+{t}_{a}{\uplambda\:}{\overrightarrow{v}}_{i\left|1,0.6\right.}+{T}_{s}\widehat{F}-{t}_{a}\widehat{F}$$\end{document}21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right)=\:{\overrightarrow{\widehat{i}}}_{l}\left(k\right)+{t}_{a}{\uplambda\:}{\overrightarrow{v}}_{i\left|1,0.6\right.}+{T}_{s}\widehat{F}$$\end{document} From (21), the execution time of the active vector can be expressed as:22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}=\frac{{\overrightarrow{i}}_{l}\left(k+1\right)-{\overrightarrow{\widehat{i}}}_{l}\left(k\right)-{T}_{s}\widehat{F}}{{\uplambda\:}{\overrightarrow{v}}_{i\left|1,0.6\right.}}$$\end{document} Theoretically, by applying two vectors in one period an active vector at time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}$$\end{document} and a zero vector at tZ, the output current follows the reference at time (k + 1) with nearly zero-state errors10, so :23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\overrightarrow{i}}_{l}\left(k+1\right){=\overrightarrow{i}}_{l}^{*}$$\end{document} Finally, the execution time expression (22) can be rewritten as follows:24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{t}_{a}=\frac{{i}_{l}^{*}-{\overrightarrow{\widehat{i}}}_{l}\left(k\right)-{T}_{s}\widehat{F}}{{\uplambda\:}{v}_{i\left|1,0.6\right.}}$$\end{document} From (11) and (21), the cost function of the proposed technique can be rewrite to deduce the optimal vector at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left(k+1\right)$$\end{document} for the six possible active vectors28:25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{G}_{\left|i=1:6\right.}=abs\left({i}_{\alpha\:}^{*}-{i}_{l\alpha\:}\left(k+1\right)\right)\:+abs\left({i}_{\beta\:}^{*}-{i}_{l\beta\:}\left(k+1\right)\right)$$\end{document} Figure 6 displays the proposed AISMO-MFPC-(AVET) control flow diagram. First, the system’s output currents and voltages are measured. Second, the unknown function and the\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document}factor are estimated using AISMO. This step ensures the optimal value of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\lambda\:$$\end{document}. The execution time of the active vector is calculated in the third step using six possible voltage vectors through the Eq. (24), after which the prediction current and cost function can be obtained according to the Eqs. (21) and (25). Finally, optimal voltages are selected through the predictive control principle. After that, PWM technology is used to complete the implementation of the proposed control37, as shown in Fig. 7. Fig. 6Flow diagram of the proposed AISMO-MFPC-(AVET). Flow diagram of the proposed AISMO-MFPC-(AVET). Fig. 7Bloc diagram of proposed control AISMO-MFPC-(AVET). Bloc diagram of proposed control AISMO-MFPC-(AVET). Design of the AISMO The design of AISMO observer comprises two phases. The first one involves selecting the surface. To ensure that the tracks of the system converge with the sliding surfaces within a specified time and remain on them, the control law is formulated in the second phase. In this paper, the current components are chosen as control variables; note that the integrated sliding mode function is expressed as follows35:26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:S={e}_{i}+{K}_{i}{\int\:}_{0}^{t}Sigm\left({e}_{i}\right)d\tau\:$$\end{document} The Lyapunov stability theory is applied to design and guarantee the stability of AISMO observer. From the principle of Lyapunov criterion, it can determine the input signal\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}^{{\prime\:}}$$\end{document}in (14), and they can be calculated by using following steps. To fulfill the stability criterion \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}<0$$\end{document}35, a Lyapunov function \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} is defined and chosen as follows:27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:V=\frac{1}{2}\left({S}^{2}+{\widehat{e}}_{f}^{2}\right)$$\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}$$\:{\widehat{e}}_{f}$$\end{document} is the error can be defined as: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{e}_{f}={F}^{*}-\widehat{F}$$\end{document}, then, this \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{F}^{*}$$\end{document} is the best estimation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:F$$\end{document}. The derivative of (27) can be defined as:28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S\dot{S}-\dot{\widehat{F}}\left({F}^{*}-\widehat{F}\right)$$\end{document} The calculation of\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}$$\end{document} proceeding by using the derivation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{S}$$\end{document} (26) in (28) as :29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S{\dot{e}}_{i}+S.{K}_{i}.Sigm\left({e}_{i}\right)-\dot{\widehat{F}}\left({F}^{*}-\widehat{F}\right)$$\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}$$\:{e}_{i}{=i}_{l}-{\widehat{i}}_{l}$$\end{document}, the derivation is used to calculate\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}$$\end{document} as follows :30\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S\left(\frac{{di}_{l}}{dt}-\frac{{d\widehat{i}}_{l}}{dt}\right)+S{.K}_{i}.Sigm\left({e}_{i}\right)-\dot{\widehat{F}}\left({F}^{*}-\widehat{F}\right)$$\end{document} By putting the expression of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{{di}_{l}}{dt}$$\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}$$\:\frac{{d\widehat{i}}_{l}}{dt}$$\end{document} as:31\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S\left(F+\lambda\:{v}_{i}-\widehat{F}-\lambda\:{v}_{i}-{u}^{{\prime\:}}\right)+S{.K}_{i}.Sigm\left({e}_{i}\right)-\dot{\widehat{F}}\left({F}^{*}-\widehat{F}\right)$$\end{document} After simplifying,32\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S\left(F-\widehat{F}\right)-S{u}^{{\prime\:}}+S.{K}_{i}.Sigm\left({e}_{i}\right)-\dot{\widehat{F}}\left({F}^{*}-\widehat{F}\right)$$\end{document} We add and subtract the function \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{F}^{*}$$\end{document} in the first part of (32) and at the same time, substitute \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{\widehat{F}}$$\end{document} by \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:S$$\end{document} as in (26). Therefore, (32) turns to:33\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S\left({e}_{f}-{u}^{{\prime\:}}+{K}_{i}Sigm\left({e}_{i}\right)\right)$$\end{document} Assuming \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{e}_{fmax}$$\end{document} is the upper bound of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{e}_{i}$$\end{document}, i.e., |\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{e}_{f}$$\end{document}| <\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{e}_{fmax}$$\end{document}x. The integral sliding mode input signal\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}^{{\prime\:}}$$\end{document}is chosen as follows :34\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{u}^{{\prime\:}}={\frac{\left|S\right|}{S}e}_{fmax}+{K}_{i}.Sigm\left({e}_{i}\right)+\eta\:.S.Sigm\left(S\right)$$\end{document} Finally, from (33) and (34), the expression of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}$$\end{document}can be rewritten as:35\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\dot{V}=S{e}_{f}-\left|S\right|{e}_{fmax}-S.\eta\:.Sigm\left(S\right)$$\end{document} Since \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$-S.\eta\:.Sigm\left(S\right)$$\end{document} is always negative and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Se}_{f}$$\end{document}<\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\left|S\right|e}_{fmax}$$\end{document}, as shown in the previous equations, the derivation of the Lyapunov function is negative, the stability conditions in Eq. (35) make sure that the designed observer is stable. Results analysis In order to evaluate the performance and robustness of the conventional ISMO-MFPC and proposed AISMO-MFPC-(AVET) control, the simulations and experiments have been executed in this section. The system parameters utilized in this paper are detailed in Table 2. 4.1. Simulation results Using an adaptive observer, simulations were implemented in MATLAB software to validate the proposed AISMO-MFPC-(AVET) effect. The reference current was increased from 1 A to 3 A at 0.165s to assess the dynamic performance of both methods, as depicted in Fig. 8(a) and Fig. 8(b). Fig. 8Tracking performance and THD of the conventional ISMO-MFPC and AISMO-MFPC-(AVET). Tracking performance and THD of the conventional ISMO-MFPC and AISMO-MFPC-(AVET). during a step-change: from 1 A to 3 A. (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Figure 8(b) demonstrates the rapid dynamic tracking capability of AISMO-MFPC-(AVET) with a very short system transition (1.3ms) compared to the performance of the system with a conventional controller (2.8ms) as shown in Fig. 8(a). This fast transition is due to an adaptive observer that reduces the error between the two currents. Under accurate parameter settings, the proposed AISMO-MFPC-(AVET) exhibits superior steady-state performance compared to the conventional ISMO-MFPC.The accuracy of the controllers can be evaluated by the attained errors for the estimated current. The simulation resulted in a 9.5% current error in the conventional method and less than 6% error for the proposed method. Total Harmonic Distortion (THD) for conventional ISMO-MFPC is approximately 2.74%. Where the value of current in steady-state is 1 A as shown in Fig. 8(a), While in the proposed control, the THD value is about 1.82% as shown in Fig. 8(b). A sensitivity test to the load parameter mismatch was conducted to demonstrate the performance of the AISMO-MFPC-(AVET) controller supported by the adaptive observer. Figure 9 shows that when the inductance value is changed from L to – 30%(L)at the moment 0.155 (S), there is a noticeable increase in the tracking error value between the reference and the measured currents when using the traditional strategy compared to the proposed control. The advantage of the proposed method is due to adapting the factor \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\varvec{\lambda\:}\:$$\end{document}with the AISMO observer in case of mismatch parameters. This adaptation will be discussed in the next sections. Fig. 9Simulation results with variation load. (a): The conventional ISMO-MFPC, (b): AISMO-MFPC-(AVET). Simulation results with variation load. (a): The conventional ISMO-MFPC, (b): AISMO-MFPC-(AVET). Experimental results Figure 10 shows all the necessary experimental setups linked together to conduct tests on the proposed control. Details about the experimental parameters and the type of equipment needed are illustrated in Table 2. Fig. 10Experimental configuration. Experimental configuration. Table 2Experimental parameters.Equipment detailType and parameters of the experimentInductance valueResistance valuesampling time TSReference frequencyDC-link voltageController DSPIGBTGate DriverL = 16mHR = 1.3Ω100 µs50 Hz40 VTexas instruments TMS320F28379DFGH40N60IC IR2104 Table 2Experimental parameters.Equipment detailType and parameters of the experimentInductance valueResistance valuesampling time TSReference frequencyDC-link voltageController DSPIGBTGate DriverL = 16mHR = 1.3Ω100 µs50 Hz40 VTexas instruments TMS320F28379DFGH40N60IC IR2104 Experimental parameters. Inductance value Resistance value sampling time TS Reference frequency DC-link voltage Controller DSP IGBT Gate Driver L = 16mH R = 1.3Ω 100 µs 50 Hz 40 V Texas instruments TMS320F28379D FGH40N60 IC IR2104 The experimental setup comprises a variable load and three-phases 2 L-VSI powered by a DC power supply. The TMS320F28379D digital processor is utilized to implement competing AISMO-MFPC-(AVET) controllers. IR2104 integrated circuits were employed in these tests to produce switching signals for IGBTs. These ICs serve as high-voltage, high-speed power IGBT drivers, featuring both high and low side output channels. Figure 11 show the experimental results of the steady-state performance related to both the MFPC (Fig. 11(a)) and the proposed AISMO-MFPC-(AVET) controller (Fig. 11(b)) with a maximum value of current equal to 2 A. The proposed AISMO-MFPC-(AVET) shows acceptable performance compared to traditional ISMO-MFPC, which demonstrates its effectiveness relative to previous results. Fig. 11Experimental results to evaluate the performance of the two strategies with current reference (2 A), (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Experimental results to evaluate the performance of the two strategies with current reference (2 A), (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). The second test increased the reference current from 1 A to 3 A as shown in Fig. 12 to verify the performance of the two strategies with dynamic performance. Figure 12(b) shows the fast dynamic tracking of AISMO-MFPC-(AVET) with short latency compared to system performance with conventional controller (Fig. 12(a)). The response time for the proposed strategy was 1.5 milliseconds, whereas for the traditional strategy, it was 2.1 ms.Moreover, the current ripples of currents in conventional ISMO-MFPC are significantly distorted, indicating lower current predictions compared to AISMO-MFPC-(AVET). Fig. 12Experimental results to evaluate the performance of the two strategies with current reference increased from 1 to 3 A. (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Experimental results to evaluate the performance of the two strategies with current reference increased from 1 to 3 A. (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). The third test is based on the imposed control test for parameter mismatch as shown in Fig. 13. It can be seen that the tracking error between reference currents and measured currents is increased in case of parameter mismatch at ISMO-MFPC compared to AISMO-MFPC-(AVET), which indicates that the performance of AISMO-MFPC-(AVET) is insensitive for this test. The robustness of the proposed control is due to the use of the adaptive observer AISMO, which minimizes the tracking error between the reference and observed currents. Fig. 13Experimental results to evaluate the performance of the two strategies with parameter mismatch, (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Experimental results to evaluate the performance of the two strategies with parameter mismatch, (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Evaluate of factor λ The input coefficient in the proposed control method, factor λ, is typically considered a nonphysical constant and often determined through a trial-and-error process. The output current and the variation of factor λ are displayed in Fig. 14(a), providing insight into the performance and accuracy of the proposed adaptive AISMO, the reference current is increased from 1 to 2.5 A and the second test during the parameter mismatch at 0.25 s. Figure 14(b) presents the dynamic of the factor λ where it takes on high values when there are convergence errors between the reference and measured currents, especially at the initialization of the tracking, and when the reference current increased from 1 to 2.5 (A) at 0.16s, which leads to faster and better convergence time. After that, the value of λ decreases.a sensitivity test to the load parameter mismatch at 0.25s was conducted to demonstrate the performance of the AISMO-MFPC-(AVET) controller supported by the adaptive observer AISMO. Fig. 14Results of λ factor evaluation using adaptive AISMO for AISMO-MFPC-(AVET). (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Results of λ factor evaluation using adaptive AISMO for AISMO-MFPC-(AVET). (a): ISMO-MFPC, (b): AISMO-MFPC-(AVET). Sensitivity analysis In order to assess the performance and robustness of the AISMO-MFPC-(AVET) proposal parameter mismatch testing was conducted in this section. The THD and RMS are utilized as a measure to show the tracking accuracy of the reference current. Figure 15 presents a comparison of results among three methods: FS-MPC, ISMO-MFPC, and AISMO-MFPC-(AVET) under inductance mismatch by conducting inductance changes ranging from − 60% to + 60% from nominal inductance L. Fig. 15Sensitivity of FS-MPC, ISMO-MFPC and AISMO-MFPC-(AVET) techniques to parameter mismatch, (a): THD, (b): RMS Error. Sensitivity of FS-MPC, ISMO-MFPC and AISMO-MFPC-(AVET) techniques to parameter mismatch, (a): THD, (b): RMS Error. Figure 15(a) illustrated the test outcomes reveal that THDs are impacted by changes in inductance levels for two FS-MPC and ISMO-MFPC controllers. AISMO-MFPC-(AVET) produces output current with a lower THD as its value decreases to 0.92% when nominal value of the inductance and THD of 1.13% with a positive inductance mismatch of 33 mH. Figure 15(b) shows that when the inductance mismatch is less than the nominal values, the THD of the current increases and the root mean square current error (|RMS error|) also increases. The proposed strategy AISMO-MFPC-(AVET) showed less impact when there is a mismatch in the inductance value, which shows very good performance compared to two other strategies. The quantized results are summarized in Table 3. Table 3 presents acomparison results obtained by the three approaches, FS-MPC, ISMO-MFPC, and AISMO-MFPC-(AVET) in case nominal parameter and In second test, the inductance suddenly decreases by 60% (-60%(Ln)). In the case of nominal parameters, the values ​​of RMSEs and THD are acceptable in all strategies. In second test, the results confirmed that the FS-MPC control is sensitive under with the parameters mismatch. Furthermore, the AISMO-MFPC-(AVET) produces an output current with a lower error and lower total harmonic distortion level compared to conventional method ISMO-MFPC, the RMSE and THD in AISMO-MFPC-(AVET) are 0.82% and 2.03%, respectively. Thus, the results demonstrate the robustness of AISMO-MFPC-(AVET) against parameters mismatch. The last column of the table shows the complete parameter independence of the imposed method compared to other methods.Table 3Comparison of the results obtained by AISMO-MFPC-(AVET) and other methods.StrategiesRMSE (A)THD (%)MPC/ MFPCNominal LMismatch LNominalLMismatch LRequired ParametersFS‐MPC0.080%0.54%1.09%5.76%R, LISMO- MFPC0.084%0.27%1.32%3.48%\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda\approx1/L$$\end{document}AISMO-MFPC0.082%0.15%1.13%2.03%No required Table 3Comparison of the results obtained by AISMO-MFPC-(AVET) and other methods.StrategiesRMSE (A)THD (%)MPC/ MFPCNominal LMismatch LNominalLMismatch LRequired ParametersFS‐MPC0.080%0.54%1.09%5.76%R, LISMO- MFPC0.084%0.27%1.32%3.48%\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\lambda\approx1/L$$\end{document}AISMO-MFPC0.082%0.15%1.13%2.03%No required Comparison of the results obtained by AISMO-MFPC-(AVET) and other methods. Conclusion Model-free predictive control based on an adaptive observer using active vector execution time AISMO-MFPC-(AVET) is proposed in this paper to improve the model-free predictive control methods that use a ultra- local model. The design of the proposed adaptive observer is completely independent of the system parameters as it is based on the estimation of the unknown part and the non-physical factor in the ULM equation. To obtain the lowest current ripple, the concept of active vector execution time (AVET ) has been incorporated into the proposed control to minimize the cost function instead of selecting a single vector. ULM is also used to calculate AVET which facilitates the implementation of the imposed control.The comparative evaluation of AISMO-MFPC-(AVET) with ISMO-MFPC and conventional MPC approaches confirms the efficiency and effectiveness of the proposed control as well as its complete parameters independence, which enhances the robustness of the imposed control.The efficiency of the proposed control based on AISMO-MFPC-(AVET) has been verified experimentally using a TMS320F28379D digital processor with a two-level inverter. Experimental validation is a testament to the effectiveness of the AISMO-MFPC-(AVET) in real-time applications against parameters mismatch while providing superior control performance.Calculating some of the factors in the imposed observer and controller and estimating the unknown part in complex systems is a path for future research. The possibility of incorporating one of the artificial intelligence techniques into the imposed control provides a promising path for developing this method. Artificial neural network technology and reinforcement learning technology are advanced evaluations that improve control techniques’ performance. Which may lead to unprecedented levels of efficiency, accuracy, and robustness.