Studies On Rotor Time Constant Identification Of Induction Motors For Multiple Scenarios

Facing with high requirement of control performance,identification of parameters is required for increased accuracy of motor control.At the same time,speed sensorless control technology is an important way to reduce costs and improve system reliability.Therefore,in this thesis,to identify rotor time constant online,the feasibility and strategy of parallel identification is studied.Firstly,for scenarios with rotor speed sensors,rotor time constant is often deviated from its nominal value due to operating conditions of induction motors and magnetic saturation,which is solved by introducing particle filter algorithm to improve the identification accuracy.Further,for speed sensorless scenarios,rotor speed is considered as a parameter to be identified in control design.To deal with the deep coupled identification of rotor time constant and rotor speed,the parallel identifying strategy of rotor speed and rotor time constant is proposed respectively for middle-high rotor speed operating conditions and in low rotor speed operating conditions.In this way,the parallel indetification of both parameters is realized in whole rotor speed operating conditions.The main contributions and innovations of this dissertation are summarized as follows:(1)To modify the parameters of motors during simulation,the s-domain and s-function models of induction motor are established.According to the information of nameplate and stator blueprint,the motor model is formulated to obtain the parameters.The instantaneous physical information and electromagnetic distribution information are obtained.By injecting voltage signal into static motor,the parameter estimation is achieved by recursive least square method offline.A filter is designed based on second-order Butterworth filter with phase angle compensation to eliminate the signal interferences.The accuracy is improved by second-order corrected flux linkage observer.(2)An online identification strategy of rotor time constant is proposed based on particle filter algorithm.The relationships among state variables are analyzed to investigate the factors which influence the identification.Simulation results show that the proposed particle filter algorithm can effectively improve the convergence speed and the accuracy of identification,meanwhile the flux is also accurately observed when starting up the motor.(3)A speed sensorless control strategy is proposed based on online identification of rotor time constant.By theoretical deduction and analysis,the smaller identification value of rotor time constant will lead to larger rotor speed.Considering the variation and the identification of rotor time constant is slower than rotor speed’s,the motor operating state can be determined by the ratio of d-axis and q-axis component of stator current to adjust the cut-off frequency.By intermittently integrating the rotor time constant identification system into the rotor speed identification system,the rotor speed observation is corrected online by rotor time constant identification.In this way,the control accuracy is increased both in middle-high rotor speed region and in whole load torque region.(4)Due to the strong coupling between rotor speed and rotor time constant in the indirect vector control system of induction motor,it is difficult to realize their parallel identification.The parallel identification based on multi-innovation extended Kalman filter based on forgetting factor(FMI-EKF)was proposed to reduce the error bandwidth of the Kalman filter parallel identification system and improve its reliability.Meanwhile in order to avoid data saturation,a forgetting factor was introduced to reduce the dependence of the system on historical data.The simulation results show that the proposed parallel identification is suitable in the range of full speed and full rotor time constant,which has enhanced capability against disturbances.And compared with the traditional Kalman filter system,the accuracy of the parallel identification is significantly improve by the proposed FMI-EKF.