Research Of Deterministic Sampling Kalman Filter Based Sensorless Control Of Permanent Magnet Synchronous Motors

Whether the vector control of permanent magnet synchronous motor(PMSM)can be successfully realized depends on the detection accuracy of the rotor position.In many practical applications,additional position sensors can not be installed on the motor due to harsh environments,cost or system structure constraints.Taking this as the background we introduce the Cubature Kalman filter(CKF)Algorithm,The feasibility of using CKF algorithm to solve the velocity and rotor position estimation of PMSM without a position sensor is discussed in the paper.Detailed theoretical analysis,simulation research and experimental verification of the motor state estimation of PMSM vector control drive system based on CKF algorithm are provided.The Cubature Kalman filter,proposed by Canadian scholars,is a landmark achievement in the development history of Nonlinear Kalman filter estimation algorithm,and its appearance has created a brand-new and promising direction for the field of nonlinear filtering.Based on the invariant theory and the 3-order spherical-radial cubature criterion,CKF is used to approximate the Gauss weighted nonlinear function integral,so as to capture the statistical characteristics of the posterior probability density of the states.As a member of Kalman filter algorithm,CKF retains the typical simple,elegant and efficient recursive form of Kalman filter frame.CKF has higher estimation accuracy and nonlinear adaptability than EKF,which is based on Taylor expansion.Compared with UKF,CKF has more rigorous mathematical derivation and better numerical stability in high-dimensional nonlinear filtering problems.Compared with GHF,CKF has high computational efficiency while maintaining high estimation accuracy and numerical stability.In the field of electrical engineering,especially in the field of state estimation in the control of the motor without a position sensor,almost all the research is focused on the application of EKF,but this is,after all,the method of more than 50 years ago.Although in the field of nonlinear filtering in the last twenty years a number of excellent filtering algorithms have been developed.However,very rare studies,especially the latest appeared CKF,combine them with the state estimation problem of PMSM.In this paper,the CKF based position sensorless control of permanent magnet synchronous motor is studied comprehensively.Detailed derivation of CKF algorithm is given after introducing the theoretical basis of Bayesian filter frame,invariant theory and high dimensional numerical integration.At the same time,the square root decomposition algorithm SCKF is given from the point of view of numerical stability.CKF is introduced into the PMSM state estimation problem based on the state equation and observation equation of the motor in the stationary coordinate system.The proposed CKF based sensorless method is pre-validated by Matlab/Simulink,Simulation results show that the CKF algorithm has a good performance in terms of motor state estimation.In order to further verify the practical application value of Cubature Kalman filter,a PMSM drive platform based on TMS320F28377 D is constructed and a variety of typical experimental tests are carried out.Experimental results show that the proposed CKF based PMSM state estimation method has good steady-state and dynamic performance over a wide speed range operation.