Sensorless Control Of Induction Machines Based On An Adaptive Extended Kalman Filter With Maximum Likelihood Criterion

Speed sensorless control technology has attracted wide attention because it saves system cost and enhances system reliability.In recent years,with the development of high performance processors,Extended Kalman Filter(EKF)has attracted much attention in the field of speed sensorless control.However,the robustness of EKF to model uncertainties and internal and external disturbances is poor,which leads to estimation bias and system divergence in practical application.In order to improve the adaptive ability of EK F,this paper mainly studies the speed estimation method of induction motor based oin MaximumLikelihood Criterion and Adaptive Extended Kalman filter(MLC-AEKF),and the robustness and robustness of the system are studied theoretically and verified experimentally.The main research contents are as follows:Firstly,the fifth-order mathematical model of an induction motor is constructed,and the stability of the motor itself in the full speed range is analyzed based on the zero-pole diagram obtained from the state equation.Secondly,the basic principle of EKF is discussed in detail,and the speed identification link of induction motor based on discrete EKF is established and applied to the rotor field orientation control system of induction motor.The influence of sampling period on system performance is analyzed.Thirdly,the shortcomings of EKF are analyzed and its adaptive schemes are discussed.The basic principle of Imaximum likelihood criterion is introduced through comparative analysis.A residual covariance estimator based on maximum likelihood criterion is constructed and the adaptive mechanism of residual covariance to the uncertainty of EKF nmodel is studied.The sliding window data of residual covariance estimator is exponentially attenuated to improve the performance of residual covariance adaption.Then,the mathematical model of induction motor based on maximum likelihood criterion adaptive EKF is established,and the robust mechanism of the system is studied.Fourthly,the speed sensorless control system of an induction motor based on maximum likelihood adaptive EKF is simulated and validated by using MA'TLAB/Simulink sotftware.Theeffects of sliding window length and exponential attenuation factor on the steady-state and dynamic performance of the system are studied by simulation,and the specific parameters are determined.Finally,an experimental platform based on TI's TMS320F28335 microprocessor is built,and the basic performance and algorithm validity of the adaptive EKF speed identification method based on maximum likelihood criterion are verified by experiments.The simulation and experimental results show that the proposed speed identification method based on MLC-AEKF in this paper can enhance the adaptability of the system to the model mismatch and the changeable working conditions of induction motors effectively,compared with the extended Kalman filter algorithm.The speed identification error can be reduced when the motor parameters changed at low speed.Moreover,the estimation errers are restrained when the internal and external uncertainties occur,and the steady and dynamic performance of sensor control system under various working conditions are improved.