Fully Data-Driven Inductance Reconstruction for Position-Sensorless Control of Switched Reluctance Motors

초록

This study proposes a fully data-driven rotor position estimation method for sensorless control of switched reluctance motors (SRMs) under diverse operating conditions. The entire control process is performed online, spanning from initial startup to full-cycle position estimation without any prior motor information. For the initial startup and low-speed operation, a threshold-triggered slope drive (TTSD) method is proposed. While driving using TTSD, inductance extrema are computed and stored to construct an unsaturated inductance profile. Subsequently, an inductance vector method is used for estimating rotor position and is performed concurrently with the reconstruction of the energized phase inductance using an unsaturated inductance profile. Through this method, the rotor position is accurately estimated over the entire cycle, regardless of high-load operating conditions. Furthermore, an adaptive method for compensating the reconstructed inductance is also introduced to achieve robust estimation even under bus voltage fluctuations, without requiring a voltage sensor. The proposed method can be implemented through a simple online estimation process with minimal memory usage. Experimental results confirm its capability to achieve accurate rotor position estimation and robust operation without relying on model parameters, offline data, or additional hardware.

키워드