Robust Field-Oriented Control of Hybrid Stepper Motors Using a DRNN-Based Adaptive PID Controller
Year: 2026
Type: preprint
Abstract: Hybrid Stepper Motors (HSMs) are widely used in high-precision motion applications, including computer numerical control machines, robotics, three-dimensional printing systems, and aerospace positioning equipment, owing to their high torque density, excellent holding torque, and favourable cost-to-performance ratio. However, closed-loop operation under Field-Oriented Control (FOC) remains vulnerable to performance degradation caused by electrical parameter variations, mechanical uncertainties, and sudden load torque disturbances, which conventional fixed-gain Proportional–Integral–Derivative (PID) controllers cannot adequately compensate for. This paper proposes a Dynamic Recurrent Neural Network (DRNN)-based self-tuning PID controller for robust field-oriented control of hybrid stepper motors operating under combined parameter uncertainty and load disturbances. We first develop a complete nonlinear dynamic model of the HSM in the direct–quadrature synchronous reference frame, incorporating detent torque, back-electromotive-force coupling, and bounded parameter perturbations. A diagonal recurrent neural network with self-feedback hidden neurons is then designed to adaptively tune the outer speed-loop PID gains in real time, using an online gradient-descent learning algorithm with momentum derived from a chain-rule-based sensitivity analysis of the plant Jacobian. The proposed controller is integrated into a cascaded FOC architecture with zero d-axis current control and Space Vector Pulse Width Modulation, and is extensively evaluated in MATLAB/Simulink under five test scenarios: nominal speed tracking, speed reversal, sudden load disturbance, combined parameter uncertainty, and low-speed operation. Comparative results against a conventional fixed-gain PID controller, assessed using transient-response indices and integral error metrics, demonstrate the proposed scheme's superior tracking accuracy, disturbance rejection, and robustness.
Source: Research Square
Authors: Mike O. Ejunka, Patrick Odu, Victor O. Ekpang
Institution: University of Calabar
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Keywords: hybrid stepper motor, field-oriented control, adaptive PID control, dynamic recurrent neural network, self-tuning PID controller, parameter uncertainty, load disturbance rejection, nonlinear dynamic modeling, synchronous reference frame, space vector modulation, speed tracking, low-speed operation, sensitivity analysis
Open Access status: green
OpenAlex ID: W7225902679
