Research On Online Identification Of AC Servo Motor Parameters Based On Model Reference Adaptive System

"Industry 4.0" led by intelligent manufacturing has become a new vane of technological development in the world.Robots and High-end CNC machine tools are widely used in traditional industries and high-tech fields.As one of the key components,servo systems directly determine the robots and High-end CNC machine tools' performance.The design of controllers in servo control systems often require precise motor parameters.However,as the motor operating conditions and load change,the stator inductance,stator resistance,rotor permanent magnet flux linkage and rotational inertia of motor will change accordingly.And the generally adopted improvement in controlling method cannot completely eliminate its influence.Therefore,it is especially important to identify the parameters of the motor.To counter the problems above,the online identification of servo motorparameters and error compensation of identification system will be studied in this paper.In this paper,the under-rank problem of servo motor multi-parameter identification is firstly analyzed and a step-by-step parameter identification algorithm is proposed.By applying a negative current to the d-axis,the order of the system model is improved.The model reference adaptive method is applied to the motor for online identifying its electrical parameters and mechanical parameters.Based on Popov hyper-stability theory and Landau discrete recursive algorithm,the adaptive law of motor electrical parameters and the adaptive law of rotational inertia are designed separately.The effectiveness of the algorithm is verified by Matlab simulation platform,and the identification speed,accuracy,dynamic process and selection of adaptive law parameters are analyzed.In the actual discrete system,there are various kinds of interference which all have influence in identification accuracy.For the system error of deadtime effect,phase error and current feedback error,this paper deeply analyzes the cause of these errors and their influence on motor parameter identification.For the deadtime effect,an adaptive deadtime compensation algorithm based on Lyapunov theorem is proposed,which can effectively suppress the error caused by the deadtime effect.Besides,a method of leading phase compensation is applied to regulate the delay of encoder communication and the phase error caused by the vector control delay.For the current feedback error,software algorithm is adapted to remove the zero drift of the current sensor and the offset error.The error suppression algorithm is simulated by Matlab,which proves the efficiency of the above algorithm in suppressing system error and improving the identification accuracy.Finally,by designing the hardware platform based on DSP+FPGA,the speed loop control algorithm,identification algorithm and error compensation algorithm are completed on the DSP,while the current loop control algorithm is completed on the FPGA.The effectiveness of the system error compensation algorithm and the online identification algorithm is verified by experiments,and the experimental results are analyzed.