嵌入式实时以太网运动控制器设计与性能测评

针对标准以太网应用到运动控制领域存在的非确定网络延时问题,提出确定性调度配合延时补偿的方法提高系统性能。利用卡尔曼滤波算法提高时间同步精度,采用时隙划分调度策略保证数据交换的确定性,通过RBF (radial basis function)模糊神经网络PID算法对系统延时预测值进行补偿,使系统能跟随延时自适应调整,具有高精度多轴同步运动的能力。实验结果表明,卡尔曼滤波将时钟偏移标准差由11.6μs降至7.9μs,有效提高了同步精度,基于延时预测并补偿的控制算法比传统PID算法具有更短的响应时间,更高的稳态性能和同步跟随性能。

Design and performance test of embedded real-time Ethernet motion controller

Aiming at the problem that Ethernet used in the field of motion control leads to non-deterministic network delay,a method combining deterministic scheduling with delay compensation was presented to improve system performance.The accuracy of time synchronization was improved using Kalman filter algorithm.The time-slot division scheduling strategy was used to gua-rantee the certainty of data exchange,and the PID algorithm based on radial basis function fuzzy neural network was proposed to compensate the delay prediction value of system,so that the system could implement adaptive adjustment following the delay,with the performance of high precision multi-axis synchronous motion.The results of experiments indicate that the Kalman filter algorithm can reduce the standard deviation of clock offset from 11.6μs to 7.9μs,and effectively improve the time synchronization accuracy.The control algorithm based on delay prediction and compensation has shorter response time,better steady-state performance and synchronous following performance than traditional PID algorithm.