液压系统中粒子群优化神经网络权值的控制算法  

Control algorithm based on particle swarm optimization neural network weights in hydraulic system

作　　者：王茂林[1] 叶小红[1] 王洪光[1] 李少远[1] 姜贵林[1] 

机构地区：[1]哈尔滨工程大学机电工程学院,黑龙江哈尔滨150001

出　　处：《应用科技》2014年第3期51-54,共4页Applied Science and Technology

基　　金：国家自然科学基金资助项目(51175099)

摘　　要：针对非线性、时变等缺陷导致传统的控制器控制效果较差、不适应电液伺服系统的现象,提出了用于电液伺服控制的基于粒子群优化算法对神经网络的权值进行学习训练的PSO-NN算法。结合电液伺服系统实例分析,用MATLAB仿真得到了输入阶跃信号和正弦信号时,PSO-NN算法的输出曲线以及适应度曲线;为了展示PSO-NN算法的效果,用BP算法仿真了对应输入阶跃信号和正弦信号的输出。仿真结果表明:在电液伺服系统的控制中,PSO-NN算法性能优于BP算法,系统输出具有更好的收敛性和对输入的跟随性,从而证明PSO-NN算法对于电液伺服系统的控制是合适并有效的。Considering the control nonlinearity and uncertainties and other defects of intelligent control electro -hy-draulic servo system , PSO-NN algorithm is proposed to use particle swarm optimization algorithm to train neural network weights .Combined with a specific instance of electro-hydraulic servo system , when the system input is step and sine signal , the output and fitness curves of PSO-NN algorithm are simulated by using MATLAB software .In order to demonstrate the result of PSO-BP algorithm, BP algorithm is simulated further .The simulation results show that PSO-NN algorithm is superior to the BP algorithm due to the better output convergence and input following per -formance of the electro-hydraulic servo system , and that PSO-NN algorithm is suitable and effective for electro-hy-draulic servo system control .

关 键 词：粒子群优化算法 BP神经网络 PSO-NN算法 权值训练 电液伺服系统 

分 类 号：TP217.31[自动化与计算机技术—检测技术与自动化装置]