RBF神经网络在静电悬浮位置控制中的应用

静电悬浮位置控制系统具有非线性、时变性的特点，传统的控制方法不能有效抑制扰动的影响。针对该问题提出了一种神经网络与PID相结合的RBF-PID控制策略。以球形样品为例分析其受力情况，推导出静电悬浮位置控制系统的机理模型。搭建基于RBF-PID控制器的静电悬浮位置控制系统，并根据仿真结果实时在线调整控制参数。仿真结果表明，当样品带电量从10–9 C突变至3×10–9 C时，RBF-PID控制器只需0.12 s即可使样品达到稳定状态。实验结果表明，当样品处于加热状态时，实时调整参数后系统的平均绝对误差为0.0416 mm，控制效果比传统PID控制策略提高了70%。所提出的控制方法辨识精度高，具有比传统PID方法更强的鲁棒性和稳定性。 

RBF Neural Network in Electrostatic Levitation Position Control

With the characteristics of nonlinearity and time variation, the position control system of electrostatic levitation facility cannot suppress the disturbance effectively by using traditional control methods. To solve this problem, a RBF-PID control strategy combining neural network and PID control method was proposed. Firstly, the mechanical analysis on the spherical sample was researched, and the mechanism model of the position control system of electrostatic levitation was deduced. Then, the position control system of electrostatic levitation was constructed, based on the RBF-PID controller. The control parameters were adjusted in real-time according to the simulation results. The simulation results show that it takes 0.12 s to make the sample stable when its surface charge suddenly changes from 10–9 C to 3×10–9 C. The experimental results show that when the sample is in the heated status, the mean absolute error of control system is 0.0416 mm by adjusting the parameters in real time, and the control effect is 70% better than that of traditional PID controller .The proposed RBF-PID control strategy has a high identification accuracy, and it has stronger robustness and stability compared with the traditional PID control strategy.