Nonlinear System Identification of Variable-Pitch Propellers Using a Wiener Model

David Grasev, Miguel A. Mendez · 2026

Read Nonlinear System Identification of Variable-Pitch Propellers Using a Wiener Model by David Grasev, Miguel A. Mendez (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-02503.txt.

Introduction

Grasev and Mendez identify a compact model for a variable-pitch propeller powertrain, covering the full chain from control signals to thrust. Their Wiener-style model passes the inputs through simple linear dynamics for the motor and pitch actuator, then maps the result through a fixed nonlinear relation to thrust. The model reproduces measured behavior accurately while staying interpretable and light enough for real-time digital twins and control.

Catalog record

CREATORDavid Grasev, Miguel A. Mendez
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36673
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.02503

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Nonlinear System Identification of Variable-Pitch Propellers Using a Wiener Model David Grasev, Miguel A. Mendez Abstract This work presents the system identification of a variable-pitch propeller (VPP) powertrain, encompassing the full actuation chain from PWM signals to thrust generation, with the aim of developing compact models suitable for real-time digital twinning and control applications. The identification is grounded in experimental data covering both static and dynamic responses of the system. The proposed model takes the form of a Wiener-like architecture, where the PWM inputs are first processed through linear first-order dynamics describing the motor and pitch actuation, and the resulting states are then mapped via a static nonlinear relation to the generated thrust. This structure naturally arises under the assumptions that the electronic actuation operates on a much faster time scale than the mechanical response, and that the contribution of the aerodynamically induced torque is negligible in the tested regime. The resulting parsimonious representation is shown to reproduce the measured dynamics with good accuracy while remaining interpretable and computationally light, thereby providing a practical basis for integration in control-oriented digital twin frameworks. keywords propellers, variable-pitch propellers, machine learning, nonlinear system identification 1 Introduction This paper addresses the system identification of a variable-pitch propeller (VPP) system, encompassing the full actuation chain—from pulse-width modulation (PWM) signals for RPM and pitch commands to thrust generation. The work was conducted within the Re-Twist project (REinforcement TWInning SysTems) at VKI, aimed at developing physics-based digital twins through the integration of system identification, data assimilation, and reinforcement learning. Section 1.1 introduces the Reinforcement Twinning (RT) framework underlying this project. Section 1.2 highlights the opportunities and relevance of VPPs within this context. Section 1.3 reviews existing modeling and identification approaches, and Section 1.4 states the objectives of the present study. 1.1 The Reinforcement Twinning (RT) framework Reinforcement Twinning (RT) is a recently proposed framework coupling digital twins with control agents through parallel model-based and model-free learning. An adaptive digital twin is continuously updated from real-time data via system identification and data assimilation, while a control agent learns an optimal policy through reinforcement learning or other optimization methods. The twin provides physical consistency and sample-efficient learning, whereas the model-free branch enhances robustness and exploration. Their interaction is bidirectional: the twin acts as a virtual testbed for control strategies, and discrepancies between simulated and real responses drive both model updates and policy refinement. This mutual learning merges the efficiency of model-based and the adaptability of model-free control. RT was demonstrated in wind turbine control, drone trajectory optimization, and cryogenic tank management Schena et al. 2024 . Enhanced variants with stronger corrections were tested for flapping-wing drones Poletti et al. 2025 , and a multi-fidelity Bayesian optimization formulation for attitude control of a two-propeller balance achieved improved learning efficiency Yannick Lecomte and Mendez 2025 . The present work constitutes a first step toward applying RT to VPPs, emphasizing simplified powertrain models and their control performance, thereby laying the groundwork for adaptive digital twins of VPP systems. 1.2 Opportunities offered by variable-pitch propellers for UAVs In multirotors with fixed-pitch propellers, thrust is controlled solely via motor speed. VPPs introduce blade pitch as a second actuator, enabling thrust modulation through both RPM ( \omega ) and pitch angle ( \beta ). This added degree of freedom allows optimization of agility, efficiency, or endurance Chang et al. 2020 . Early work emphasized agility gains: varying pitch allows faster thrust modulation and reverse thrust, as shown in Cutler and How 2015 . Yet with advances in brushless motors and speed controllers, modern quadcopters already achieve high agility through RPM control, reducing the incentive for VPP use in lightweight UAVs. VPPs remain attractive for efficiency improvements. Pitch adjustment maintains favorable blade angles of attack, sustaining higher propulsive efficiency across wider operating ranges Fresk 2013 ; Fresk and Nikolakopoulos 2014 . Among pitch–speed combinations yielding equal lift, optimal ones can minimize power consumption and extend endurance Sheng and Sun 2016 . Additional benefits include reduced actuator mass in passive concepts Heinzen et al. 2015 and improved fault tolerance under motor failure Wang et al. 2020 . However, modeling VPP powertrains is challenging due to nonlinear coupling between RPM and pitch actuations and unsteady aerodynamic effects, motivating hybrid physics–data-driven modeling approaches. 1.3 Modeling and identification of variable-pitch propellers Propeller powertrain models traditionally combine electromechanical dynamics with quasi-steady aerodynamic formulations for thrust and torque, derived from Blade Element Momentum Theory (BEMT) Arellano-Quintana et al. 2018 ; Podsędkowski et al. 2025 ; Naoki et al. 2023 ; Chang et al. 2020 or stationary calibration Podsedkowski et al. 2020 ; Simmons 2022 ; Porter et al. 2016 ; Kulkarni and Sundaram 2024 ; Heinzen et al. 2015 . Linearized electromechanical models, often first-order, are widely used in control and identification studies Fresk 2013 ; Fresk and Nikolakopoulos 2014 ; Chang et al. 2020 ; Arellano-Quintana et al. 2018 ; Sheng and Sun 2016 ; Naoki et al. 2023 ; Michel et al. 2022 ; Sun et al. 2018 . For VPPs, thrust depends nonlinearly on both RPM and pitch. Sudden pitch changes cause torque variations and direct thrust feed-through Cutler and How 2015 , which single-state models cannot capture. Recent studies incorporated thrust feedback into actuator models to better represent these couplings Wu et al. 2024 . Some authors relaxed quasi-steady assumptions by introducing inflow lags Asper and Sirohi 2023 or adaptive estimation of aerodynamic coefficients during flight Ahsun et al. 2015 . Although full unsteady coupling remains underexplored, experimental studies confirm measurable unsteady effects in turbulent conditions Oo et al. 2023 . Nonetheless, quasi-steady models remain effective for hover and moderate maneuvers when properly calibrated Podsedkowski et al. 2020 ; Arellano-Quintana et al. 2018 ; Simmons 2022 ; Wu et al. 2024 ; Gauthier et al. 2024 , even though they fail in fast transients where induced circulation and vortex shedding dominate. 1.4 Objectives of the present study This preliminary study aimed to establish an experimental test bench enabling thrust and RPM measurements under controlled speed and pitch actuation, as well as varying loads. The setup, initially developed within Antoine Colonval 2025 ’s STP program, required refinement for systematic identification and control tests. The second goal was to evaluate the feasibility of representing the entire powertrain by an adaptive Wiener-type model, derived from first principles and identified using an adjoint-based procedure. Finally, the study includes initial model-based control experiments for RPM and pitch actuation with step thrust commands. The long-term objective is to build on these results toward an efficient, accurate framework for modeling and control of VPP powertrains in realistic UAV applications. 2 Dynamic Modeling of the Variable-Pitch Propeller We first present a general nonlinear model (Section 2.1 ), then reduce it under simplifying assumptions to the proposed Wiener model (Section 2.2 ). 2.1 A general electromechanical–aerodynamic model In hover (zero free-stream), simple aerodynamic models relate thrust T(t) to rotational speed \omega(t) and blade pitch \beta(t) as T(t)=\tfrac{1}{2}\,\rho\,A\,\omega(t)^{2}\,R^{2}\,C_{T}\bigl(\lambda(t),\,\beta(t)\bigr), (1) where \rho is air density, A=\pi R^{2} the disk area, and C_{T} the non-dimensional thrust coefficient, a static map of inflow ratio \lambda=v_{i}/(\omega R) (with induced velocity v_{i} ) and pitch \beta(t) . In maneuvers, thrust transients are modeled by assigning dynamics to \omega(t) , \beta(t) , and \lambda(t)

Read the complete text