| Permanent magnet synchronous motor has been widely applied in many industrial fields because of its high torque density and high power density.However,the high performance vector control technology of permanent magnet synchronous motor generally depend on accurate motor parameter information,so the parameter identification of permanent magnet synchronous motor is of great significance.The signal injection based method flexibly realizes the parameter extraction with state decoupling and under multiple working conditions,which has high application value.In order to further improve the accuracy of parameter identification and expand its generality,it is necessary to further study the key technical problems of parameter identification method based on signal injection,which is as follows:(1)the universality of the identification method need to be improved for the accurate parameter identification under as many as operation conditions;(2)the cross saturation and magnetic saturation should be considered,and the robustness of motor parameter identification should be improved;(3)the error effect in the control process of voltage source inverter should be overcome for improving the accuracy of parameter identification.This paper studies the above problems in order to further improve the effectiveness of permanent magnet synchronous motor parameter identification algorithm and promote its universal application.Offline parameter identification technology can obtain parameters in the static state of the motor,which is to realize the normal starting and operation control of the motor.However,the existing offline parameter identification methods mostly identify the parameters of static single working condition,which can not fully consider the variation trend of parameters under different working conditions,limiting its universality.Based on the requirements of different motor applications,two offline motor parameter identification strategies are proposed in this paper.Firstly,as for the motor system without shaft locking,the inductance identification algorithm is studied based on high-frequency square wave signal injection,considering the relationship between motor voltage and flux linkage.And,the polynomial fitting method is used to obtain the inductance surface.At the same time,by adjusting injection amplitude and frequency,the inductance values corresponding to different working conditions are identified under offline static state,and the numerical variation under different saturation states are obtained.Secondly,for the motor with locked rotor shaft,the inductance identification method based on sinusoidal signal injection is proposed.Through DC superimposed AC composite signal injection,the online saturation state simulation and inductance identification of the motor are realized.In addition,a stator resistance identification method based on slope current injection is studied,which avoids the influence of inverter nonlinearity on parameter identification and improves the robustness and accuracy of resistance identification.Online parameter identification technology can estimate the motor parameters in real time when the motor is running.However,the traditional online parameter identification algorithm is mostly based on voltage equation,which has problems such as model deficient rank and parameter coupling,which reduces the accuracy and robustness of parameter identification.In order to solve the problems of traditional methods,based on the physical characteristics of motor magnetic circuit,this paper studies the online parameter identification technology based on high-frequency equivalent impedance model.By injecting small amplitude high-frequency sinusoidal voltage,the motor parameters and motor state par ameters are decoupled,and the extraction of motor parameter information under multiple working conditions is realized.By analyzing the influence of harmonics on inductance identification during motor operation,a self-tuning strategy is proposed to select the appropriate amplitude and frequency of injection signal.In order to improve the identification accuracy of flux and resistance,the parameter sensitivity of voltage equation is analyzed,and the calculation of parameters is realized combined with the linear regression convergence strategy.Then,a self-learning and compensation strategy for motor rotor observation error identification based on minimum current vector scanning is proposed,which reduces the influence of rotor position observation error on parameter identification under online working conditions.Due to the modulation strategy and the nonideal characteristics of the driver elements,the output control signal is distorted,which generates additional errors to the parameter identification results.In order to further improve the accuracy of parameter identification algorithm,the nonideal characteristics of machine diver are analyzed in this paper.In order to reduce the influence of inverter nonlinearity on parameter identification,a nonlinear physical model of inverter is constructed.Combined with the coordinate transformation relationship of different rotor positions,the relationship between zero axis voltage error and angle is analyzed.Hence,a nonlinear self-learning algorithm of inverter considering zero axis voltage error is proposed.The source of sampling error in three-phase zero current clamping region and its influence on parameter identification are analyzed,and the compensation strategy of sampling error in zero current clamping area is studied.Based on the influence of different digital delay parameters in the identification of pulse width controller,the corresponding compensation methods of pulse width controller are given.The proposed algorithm is verified based on invertors and permanent magnet synchronous motor platforms with different power levels.Aiming at the online identification algorithm,the identification effects of steady-state and transient parameters of motor under different working conditions are verified.For offline parameter identification,the noise,motor stability and algorithm computation are tested and analyzed.The negative influence of the driver nonideal characteristics of the parameter identification is analyzed through experiments,and the com pensation effect of the proposed inverter nonlinear self-learning algorithm is verified.By comparing the identification results of the proposed algorithm with the traditional algorithm,the advantages of the proposed method in identification accuracy and robustness are verified. |
PDF Full Text Request