Research On The Online Parameter Identification Of Permanent Magnet Synchronous Motor

The internal parameters of permanent magnet synchronous motor will inevitably change due to the variation of temperature and magnetic saturation in actual operation.For vector control,real-time and accurate acquisition of motor parameters can not only serve as an important basis for motor fault detection,but also online optimize controller design,such as accurate estimation of torque,Sensorless Control,Current Loop Parameter Setting and so on.The limitation of off-line measurement is caused by the higher requirement of experimental conditions and the universality problem.In this paper.Interior permanent magnet synchronous motor(IPMSM)is taken as the research object,and the full-rank online identification of its internal parameters is studied as follows:The influence of temperature rise and magnetic saturation on motor parameters is analyzed and the dynamic mathematical model of permanent magnet synchronous motor is deduced in rotating coordinate system,based on the above relevant derivation,the influence of parameter variation on the performance of vector control system is analyzed in detail.The internal parameters of the motor will change dynamically under different operating conditions.In order to solve the inherent under-rank problem in system parameter identification based on motor model,the traditional full-rank identification strategy is studied,extended kalman filter and recursive least square method with forgetting factor are deduced and analyzed.Combining the characteristics of the two identification methods and the operating conditions of the motor,an online full-rank permanent magnet synchronous motor identifier is constructed,the non-ideal factors of inverters and digital delay are analyzed in detail and the open-loop compensation is carried out to improve the full rank identification accuracy.For the limitation of traditional identification strategy,a full rank identification scheme with external DC component injection is proposed.Selecting a specific mean filter sampling value,the effects of equivalent dead-time effect,rotor position disturbance,magnetic saturation and under-rank problem on parameter identification are overcome.Design parameter identifier based on recursive least square method with forgetting factor and adaline neural network,which improves the universality and accuracy of the full rank identification strategy and saves the computing resources.The full-rank identification strategy is compared in the IPMSM towing test platform,and the identification parameters are applied to sensorless control and torque observation for verification.The results show that the full-rank identification optimization strategy considering the periodic disturbance of the voltage source inverter can identify the internal multi-parameter of the motor more accurately without obtaining any relevant parameters of the motor and the inverter.