Sensorless Control Of Synchronous Reluctance Motor Based On Kalman Filter

In recent years,with the implementation of carbon peaking and carbon neutrality goals and the rise in rare earth permanent magnet prices,synchronous reluctance motor(SynRM)has received increasing attention from researchers.SynRM has the advantages of high efficiency similar to permanent magnet synchronous motors,as well as low cost and simple structure similar to induction motor.Nowadays,it has been widely applied in the general equipment industry,including compressors,fans,conveyors,and other industries.Sensorless control technology is an indispensable key function for SynRM drives,which reduces the cost and improves the reliability of the drive system effectively.However,the SynRM sensorless control technology suffers from the following difficulties:(1)The fundamental frequency model-based sensorless control method is affected by uncertain system noise and measurement noise,such as current detection errors,parameter perturbations,load disturbances,low signal-to-noise ratio at low speed,resulting in a decrease in position estimation accuracy,and then deteriorates the control performance of SynRM sensorless drive in low-speed regions and complex operating conditions in turn.(2)Due to the modelling error caused by dynamic inductance in the extended back electromotive force model(EEMF)of SynRM and the dynamic position disturbance in the active back electromotive force model(AEMF),the dynamic performance of the position estimation method based on Kalman filter is severely degraded.(3)The magnetic saturation and cross-coupling effects in SynRM resulte in a nonlinear mathematical relationship between inductance parameters and current.In many industrial applications,the inductance cannot be accurately obtained,leading to a serious decline in the sensorless control performance of SynRM due to the mismatch of inductance parameters.This article conducts in-depth research on SynRM sensorless control technology and breaks through the aforementioned core technical difficulties,which is of great significance for expanding the application of SynRM sensorless control in the domestic manufacturing industry.To solve the problem of traditional Kalman filter(KF)-based position estimation performance degradation caused by uncertain time-varying system noise such as magnetic saturation and cross-coupling effects of SynRM,disturbances,and low signal-to-noise ratio at low speed,the SynRM sensorless control method based on positive-definited Sage-Husa Kalman filtering(PDSH-KF)is proposed in this paper.First,the influence of the magnetic saturation effect on the EEMF model under dynamic conditions is analyzed and integrated into the improved EEMF(IEEMF)model,enhancing the modeling accuracy of the motor under dynamic conditions.Second,traditional KF is applied to estimate IEEMF,and the mechanism of the decrease in position estimation accuracy when the theoretical noise does not match the actual value is analyzed.Last,an IEEMF estimation method based on PDSH-KF is proposed,which estimates the mean and covariance matrix of system noise online in the iterative process of KF.The adaptive adjustment of the statistic properties can ensure that the KF always works in the optimal state,thereby improving the position estimation accuracy of the SynRM sensorless control system under the uncertain system noise.To overcome the problems of phase lag in position estimation,complex calculation of fading factors,and dynamic position disturbances in existing position estimation methods based on adaptive fading Kalman filtering,a SynRM sensorless control method based on lag compensation-enhanced adaptive quasi-fading Kalman filter(LC-AQFKF)is proposed.First,the adaptive quasi-fading factor is derived based on the relationship between theoretical innovation and actual innovation,which optimizes the structure of the observer and reduces computational complexity.Second,the transfer function between the estimated AEMF and the actual one is derived by analyzing the frequency domain characteristics of AQFKF in estimating the AEMF of SynRM.The transfer function reveals the mechanism of phase lag in AFKF position estimation methods,and a compensation strategy for estimating the phase lag of AEMF is proposed.Finally,a dual dynamic position compensation method is proposed to compensate for position estimation errors caused by AEMF model disturbances and center frequency deviation in second-order generalized integration under dynamic operating conditions.Since the AEMF is a sinusoidal AC signal,the AEMF observer designed based on KF has the characteristic of low-pass filtering,which makes the phase lag and amplitude attenuation of estimated AEMF more severe with increasing frequency.Thus,the traditional KF-based position estimation method cannot balance the high accuracy and strong anti-interference ability when SynRM operates at medium to high speeds.To overcome the above issues,the sensorless control based on resonant Kalman filter(RKF)is proposed.First,the generalized integral resonance perturbation estimator is introduced into the KF,compensating for the AEMF estimation error caused by the low-pass filtering characteristics of the KF in real-time,thereby improving the accuracy of position estimation and the anti-interference performance of the control system.Second,the frequency domain characteristic of the RKF observer is theoretically analyzed,and the stability of the proposed method is proved based on the Lienard-Chipard stability criterion.Last,to overcome the overcompensation problem caused by differential operation in existing dynamic position compensation methods,an improved dynamic position compensation method is proposed,compensating for the dynamic position disturbance in the AEMF model in real-time where a second-order tracking differentiator is employed,and then the position estimation accuracy of the method in dynamic conditions is improved.The KF-based position estimation method is designed by using a single IEEMF or AEMF model,which essentially selects the direction of the minimum rotor magnetic circuit reluctance as the d-axis.When inductance parameter mismatch occurs,the estimated position of the KF observer based on the single axis-oriented AEMF model is more sensitive to inductance parameter deviation as the current vector angle increases.To suppress the influence of inductance parameter mismatch on the SynRM sensorless control system,the sensorless control based on biaxial oriented AQFKF is proposed.First,the minimum inductance-oriented AEMF model of the SynRM is proposed based on the selection of the direction with the maximum rotor magnetic circuit reluctance as the d-axis.The observers designed for estimating the two AEMFs exhibit different parameter sensitivity characteristics at different current vector angle ranges.Second,the AQFKFs and position extraction methods are designed for the two AEMF models,and hysteresis switching and fusion methods based on the current vector angle are designed to get the estimated rotor position and speed.By combining two AQFKFs in areas where inductance parameters are not sensitive,the position estimation error during inductance mismatch can be effectively reduced,thereby enhancing the robustness of SynRM sensorless control to inductance parameter mismatch.