改进RBF网络PID算法及在气动力伺服系统中的应用

针对气动力伺服系统的非线性、时变性和不确定性,在已有RBF神经网络PID控制算法的基础上,提出了一种改进的控制算法。在RBF网络参数调整中引入动量因子,考虑参数变化的经验积累,减小系统振荡;同时,采用LM(Levenberg-Marquardt)算法代替梯度下降法对算法中PID参数进行实时在线调整,加快其响应速度。最终通过MATLAB仿真和基于LabVIEW的实物验证实验,测试了改进算法在气动力伺服系统中的控制效果。实验结果表明,改进算法的快速性和鲁棒性明显提高,在气动力伺服系统中具有良好的控制效果,且在工业现场具有实用性。

Improved RBF Neural Network PID Control Strategy and Its Application in Pneumatic Force Servo System

Focusing on nonlinear, time-varying, uncertainty during pneumatic force servo system control process, this study proposes an improved RBF neural network PID control algorithm. It introduces momentum factor into RBF network parameters adjustment, to consider the experience in parameters change process and decrease the system oscillation. Besides, LM (Levenberg-Marquardt) instead of gradient descent method is adopted for real-time online PID parameter adjustment, to speed up its response. Finally, through MATLAB simulation and physical verification experiments based on LabVIEW, he control effects of the improved algorithm in pneumatic force servo system is verified. The results show that the improved algorithm obviously enhances its rapidity and robustness, and the algorithm is practicability in pneumatic force servo system’s control under industrial environment.