A Method to Improve Flux Estimation under Gaussian Noise in Sensorless Stator Flux Control of Induction Motors

摘要

This paper proposes a programmable Low Pass Filter (LPF) to improve the performance of sensorless stator flux orientation control for induction motors. Commonly, a pure integrator based on the voltage model is used to estimate stator flux linkages for its simplicity. However, due to noise errors such as strong Gaussian noise variance, a drift and a dc offset are easily produced by the integrator when the back electromotive force (emf) is relatively smaller than Gaussian noise variance. Thus, the pure integrator is replaced by the programmable LPF to reduce those errors. The performance of the induction motor control mainly depends on the accuracy of the estimated flux. When the back emf is very small and noisy, an enhanced method of the programmable LPF is applied to improve the flux estimation. Such method is verified and implemented by a simulation tool called Simulink.