Research On High Robustness Control Strategy Of Robot End Joint Servo System

The robot servo system is the power source and core moving part of the robot.The end joint motor has low inertia,low power and low torque characteristics,which is easy to be interfered by internal and external factors.Therefore,high-performance robot-end joint servo control algorithm is one of the core key technologies of robot servo systems.This research takes the end-joint motor as the research object to achieve high bandwidth,high robustness and high accuracy performance of the end-joint servo system.The robust current control strategy,robust current harmonic suppression algorithm,the high-precision on-line parameter identification method with considering the deadtime effect,and the robust speed control loop design considering friction and load inertia disturbance are built in detail.(1)A robustness current control algorithm is proposed to improve the dependence of deadbeat controller on parameters.The effect of resistance,inductance,and flux linkage mismatch on current loop response and stability are analyzed.In this research,an extended state observer is designed to improve the robustness and control accuracy of current loop.On this basis,an observer gain selection scheme based on the calculation of the maximum pole position is proposed to further expand the stability margin of the current loop in case of inductance error and eliminate the current static error in case of resistance and flux error.(2)The robustness of traditional notch filter is affected by current frequency and parameter error,a new adaptive notch filter is proposed to eliminate the specified current harmonic.The mathematical model of deadbeat current closed loop considering notch filter is established,and the stable state of current loop under different current frequency and parameter error is analyzed.A one-step prediction adaptive notch filter is proposed,which improves the stability region of the system by leading phase compensation.On this basis,a two-step predictive adaptive notch filter is proposed to improve the performance of harmonic suppression.Finally,the relationship between step size,convergence and stability is analyzed,and the value range of iterative step size is given.(3)The robust deadbeat current control algorithm can only improve the parameter robustness but can not improve the system response ability in case of parameter error,an online parameter identification algorithm considering the nonlinearity of the inverter is designed.The nonlinear characteristics of the inverter,motor resistance and inductance are regarded as the parameters to be identified.Firstly,the influence of parasitic capacitance on deadtime disturbance voltage is analyzed,a three-phase deadtime voltage model based on max-min coordinate system is proposed,and an online adaptive dead time identification algorithm which is designed for low frequency region.In addition,the deadtime voltage and resistance voltage drop can not be identified at the same time due to their coupling,a deadtime characteristic identification method based on off-line neural network is proposed.Combined with the on-line parameter identification strategy of model reference adaptive,the robustness and dynamic response ability of deadbeat current control algorithm is improved.(4)The controled motor has low inertia and is easily disturbed by load inertia,which reduces the dynamic performance and robustness of speed loop and considering the influence of friction on speed control accuracy,a robust speed loop control strategy is proposed.The speed closed-loop model based on two degree of freedom PI controller is established,and a system parameter adjustment free algorithm is designed in continuous domain and discrete domain,which realizes the controller parameter adjustment free when the system controller parameter changes within 20 times of inertia ratio.Finally,the equivalent differential equation of friction is established based on Lu Gre model,and a friction observer is designed by using discrete mapping to compensate the friction online to improve the closed-loop control accuracy at low speed.The correctness and effectiveness of the above algorithm have been verified by the established 400 W motor to verify the drag experimental platform.