基于RBF算法的前馈惯性稳定平台PID控制回路的设计  

Design of Feedforward Inertial Stabilization Platform's PID Control Loap Based on RBF Single Neuron

作　　者：刘义[1] 郭建都[1] 郭黎霞[1] 

机构地区：[1]中国北方车辆研究所,北京100072

出　　处：《车辆与动力技术》2011年第1期31-34,共4页Vehicle & Power Technology

摘　　要：在数字惯性稳定平台系统中,由于无法实现前馈惯性稳定平台控制回路的全补偿条件,控制回路的稳定精度和跟踪能力仍提高有限.针对这一问题,采用以RBF单神经网络自动校正比例、微分、积分3个参数的算法来提高控制回路性能,研究结果表明:采用该算法的控制回路在跟踪能力上明显优于传统PID控制回路,该算法在阶跃响应时不仅超调量满意,而且上升时间与调节时间非常理想,跟踪误差小于千分之一,同时对于惯性稳定平台控制回路参数的摄动以及输出扭矩的干扰都有很强的自适应能力.Because feedforward compensation condition can＇t all be satisfied in digital inertial stabilization platform system, stable precision and tracking ability of the platform＇s control circuit can＇ t be improved effectively. To solve the problem, author uses RBF single neuron to correct proportion, integral and differential coefficient automatically to improve the control circuit performance. The simulation results show that the tracking ability of the control circuit that used new method is obviously better than conventional PID control circuit. When inputting a step signal in the control circuit, the overshoot of the method is satisfied and the rise time and the steady time are perfect. The track error is smaller than one in a thousand. The platform＇ s control circuit has favorable adaptability to the change of it＇ s parameters and the disturbance on torque.

关 键 词：惯性稳定平台 PID 控制回路 前馈补偿 RBF单神经 

分 类 号：TJ810.376[兵器科学与技术—武器系统与运用工程] TP391.77[自动化与计算机技术—计算机应用技术]