基于神经网络和智能预测控制的机器人算法

The algorithm for robot using neural network and intelligent predictive control

提出了一种新颖的控制方案,即采用神经网络预测控制器来控制机器人的力/位置。这种控制器能任意逼近机器人这种不确定对象,不用知道系统的精确结构,同时由于预测控制的加入,使系统在线计算方便,控制质量提高了。通过对机器人的仿真结果可以看出,传统的PID对于不确定对象不能很好的解决,而采用本文设计的控制器,系统的鲁棒性和快速性都得到了改善,并且具有较好的控制效果。

In this paper, a new control scheme that uses NN predictive controller to control force/position of robot is proposed. The controller can approach uncertainty randomly without knowing exact structure. With prediction the on-line computation is convenient, and the control quality is improved. Simulation shows the robustness and rapidness of system are improved and the system has a good control effect by using NN predictive controller.