Improved Full-Order Adaptive Observer for Sensorless Induction Motor Control in Railway Traction Systems Under Low-Switching Frequency

In the applications of medium- to high-power railway traction systems, the switching frequency and controller sampling rate are relatively low due to the requirements for low losses, high reliability, and limited processor power. These limitations greatly hinder the sensorless control of induction motors, especially in the high-speed operation range. In this paper, a full-order adaptive observer based on an improved discrete model is proposed, which significantly improves the system stability and control performance over the entire speed range. The feedback gain matrix of the presented control scheme is developed in a discrete time domain, ensuring that all the poles of the closed-loop observer are within the stable region. Moreover, a speed estimation mechanism based on Lyapunov's approach is designed in the synchronous rotating reference frame to further improve the precision of speed observation. Rigorous simulation studies and experimental tests of a benchmark induction motor in railway traction system demonstrate that the improved adaptive observer achieves superior sensorless control performance over the conventional adaptive observer.