Online estimation of circuit parameters of permanent magnet synchronous motor

Permanent Magnet Synchronous Motor (PMSM) is preferred in the industry because of their high torque, high power density, small in size and high efficiency. Permanent magnet motors are often preferred for high performance applications such as servo applications, electrical vehicles and robotics. Since the PMSM circuit parameters change during operation depending on temperature and current, exact circuit parameters need to be known in applications such as Sensorless Control and Direct Torque Control. In this study, first of all, the structure and mathematical models of PMSM have been examined. PMSM control methods, Pulse Width Modulation methods used in the controls are explained. Factors affecting PMSM equivalent circuit parameters; the effects of temperature, current and magnetic saturation were investigated. In the study, firstly, the Field Oriented Control of the PMSM was simulated and the Adaptive Data Fusion (ADF) method based on the Normalized Least Mean Squares algorithm was used to estimate the equivalent circuit parameters of the PMSM. The applied method is based on the estimation of the circuit parameters that prove the voltage equations in the d-q equivalent circuit model of PMSM. The proposed method was tested by simulation studies for three different motors. In the simulation studies, first of all, the actual circuit parameters of each motor are estimated. Then, the stator resistance was increased by 10% and then by 30%, assuming increased winding resistance with temperature. Finally, for each motor the circuit parameters are estimated, assuming inductances are reduced with the effect of current and magnetic saturation and the magnet flux is reduced with the effect of temperature, the value of these parameters is decreased by 10%. The curve of the estimated circuit parameters is given, mean values and the estimation errors are given in tabular form. When the results are examined, it is seen that the estimation errors are low and the estimation time is quite short.