神经网络PID控制在某火箭炮发射平台交流伺服系统中的应用  

Application of Neural Network PID Control in AC Servo System of a Rocket Launch Platform

作　　者：郝兴斌 程洪涛 丁超 HAO Xingbin;CHENG Hongtao;DING Chao(The No.32381 st Troop of PLA,Beijing 100000,China;Military Representative Office of Wuhan Regional Military Representative Bureau of Army Equipment Department in Xiangyang,Hubei 441000,China;Military Representative Office of Wuhan Regional Military Representative Bureau of Army Equipment Department in Wuhan,Wuhan 430000,China)

机构地区：[1]中国人民解放军32381部队,北京100000 [2]陆军装备部驻武汉地区军事代表局驻襄阳地区军事代表室,湖北襄阳441000 [3]陆军装备部驻武汉地区军事代表局驻武汉地区第一军事代表室,武汉430000

出　　处：《兵器装备工程学报》2021年第S01期146-150,共5页Journal of Ordnance Equipment Engineering

摘　　要：针对某舰载型火箭炮发射平台炮控系统存在非线性特性难以快速、精确控制的问题,提出了一种基于神经网络修正的PID控制算法。该算法利用单神经元的自学习和自适应能力,将PID控制算法中的误差比例系数、积分系数、微分系数作为权系数,根据位置误差进行在线学习、调节。MATLAB仿真结果表明,该算法的动态性能明显优于传统控制系统,且对负载扰动具有较好的鲁棒性。实物平台试验表明,等速跟踪控制无超调,速度平稳,跟踪误差满足技术指标要求。Aiming at the problem that the gun control system of a ship-based rocket launching platform has nonlinear characteristics and is difficult to quickly and accurately control it,a PID control algorithm based on neural network correction was proposed.The algorithm used the self-learning and self-adaptive capabilities of a single neuron,took the error proportional coefficient,integral coefficient,and differential coefficient in the PID control algorithm,and conducted online learning and adjustment according to the position error.Matlab simulation results show that the dynamic performance of the algorithm is significantly better than traditional control methods,and it has better robustness to load disturbances.The test results of the physical platform show that the constant velocity tracking control has no overshoot,the speed is stable,and the tracking error meets the requirements of technical indicators.

关 键 词：交流伺服系统 PID控制 神经网络 半实物仿真 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]