火箭炮发射架位置伺服系统中存在燃气流冲击力矩、参数时变、外部扰动等非线性因素的作用,针对传统PID控制难以在此复杂工况下取得良好控制效果的缺点,文中设计了基于传统PID控制的RBF神经网络监督控制器,完成了神经网络离线样本的选取和训练算法的改进,利用神经网络自学习、自整定能力增强系统的自适应能力。仿真结果表明此复合控制策略可以有效提高系统的控制品质。

For existence of both disturbance of gas-fired impact and variation of parameters of rocket launcher, traditional PID controller may not meet high precision request, a novel supervised radial basis function neural network (RBF NN) controller based on PID control was designed. Off-line sample of NN was chosen and improvement of arithmetic was proposed. Adaptation of system was enhanced by abilities of self-learning and self-tuning of NN. Simulation shows this control strategy can improve the performance of rocket launcher servo system.