Online Adaptive Neural Network Based Position Control of Linear Reluctance Motor
Year: 2026
Type: conference-paper
Abstract: This paper presents an Online Adaptive Neural Network controller for improving an LRM position tracker performance. The accuracy of motion using conventional controllers is limited by the nonlinear magnetic characteristics of a linear reluctance motor (LRM). In the first the LRM dynamic model implemented on MATLAB R2024a, continuously updates its neural weights depending on instantaneous error to account for nonlinearity and parameter changes. The LRM model is developed using the d-q flux framework to ensure accuracy in electromagnetic representation. Three reference trajectories: linear, nonlinear, and trapezoidal were used to assess performance. When compared to the conventional PI controller, simulation results showed improvements in RMSE, steady-state accuracy, and transient response. OANN demonstrates exceptional tracking capability under all evaluated situations.
Authors: Sarah Hareer, Diyah Kammel Shary
Institution: Southern Technical University
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7203833152