无人驾驶履带车辆电子机械制动器的电流预测控制策略

Current Predictive Control of Electro-Mechanical Braking System for Unmanned Tracked Vehicles

面向制动负荷大的无人驾驶履带车辆,设计了一种基于永磁同步电机的电子机械制动系统,提出了一种针对制动执行全过程的控制策略.在制动空行程段,采用常规的位移、转速、电流三环PID控制,使其快速到达接触位置,快速消除制动器摩擦副间隙;在制动作动力增长段,为保证作动力跟随的精确性、快速性和稳定性,基于接触位置和压紧位置对应的作动力标定结果,内环采用永磁同步电机q轴电流的预测控制策略,保证电子机械制动作动器比例施加行车制动作动力.通过对电子机械制动作动器及作动过程的建模,分析了永磁同步电机q轴电流与制动作动全过程的对应关系,仿真验证了制动作动全过程控制策略中位移控制和扭矩控制动态切换的可行性,提高作动力控制精度和鲁棒性.结果表明该策略既能有效提高制动响应时间,保证制动作动系统能够在50 ms内实现制动间隙快速消除,又能解决电子机械制动作动器作动力控制的不均匀性问题,保证制动作动系统在受到扰动时仍具有较好的跟踪性能,预测误差不超过10%.

A new type of(EMB)based on permanent magnet synchronous motor(PMSM)was designed for the unmanned vehicles with heavy braking load.Firstly,a full braking process control strategy was proposed for the EMB actor.According to the calibration results of the actuating force in the contact position and the pressing po-sition,a conventional three-loop PID control was taken for displacement,rotation speed and current control to eliminate the gap of brake friction pair quickly in the idle stroke section.In the operating force increase section,a deadbeat predictive current control strategy was adopted in the inner ring based on internal model disturbance observer of the q-axis current for PMSM to make the EMB could apply service braking force in proportion,to ensure the feasibility,control accuracy and robustness of the dynamic switching in displacement control and torque control of the full braking process.And then,modeling the EMB actor and operate process,the corres-ponding relationship between the q-axis current of PMSM and the whole braking process was analyzed.Finally,a simulation test was carried out,verifying the feasibility,control accuracy and robustness of the dynamic switching in full braking process.The results show that the proposed strategy can improve braking response time effectively,make the braking system eliminate quickly the brake clearance within 50 ms,solve the problem of uneven braking force of EMB actor,and make the motion system keep good tracking performance in disturbed and the prediction error under 10%.