Enhanced adaptive flux observer for sensorless control of switched reluctance motor

In this paper, an enhanced adaptive flux observer (EAFO) is proposed to achieve sensorless control of Switched Reluctance Motors (SRMs) while reducing hardware costs. The method employs the Unscented Kalman Filter (UKF) to estimate the current based on state variables estimated by adaptive flux observer (AFO), which is then fused with the measured current. Compared to a full-state UKF, which requires the estimation of all states, the proposed approach focuses only on current observation, thereby improving computational efficiency. Moreover, the UKF-based method enhances the interference rejection performance of the adaptive flux observer in steady-state conditions. A Fourier series-based inductance model is used for flux estimation. During the initialization phase, a pulse injection method combined with the Newton-Raphson algorithm is applied to estimate the initial states. The method is implemented in Simulink to assess its performance under various current noise conditions. Simulation results demonstrate that, by incorporating the estimated current alongside the measured current, the system exhibits improved immunity to interference compared to relying solely on the measured current.