一种带智能补偿器的神经网络控制器及其应用

A Neural Networks Controller with intelligent Compensator and Its Application

为非线性控制系统设计了一种带智能补偿器的神经网络控制器.神经网络用来提供主要的控制输入信号,网络的训练可离线进行,网络的使用采用联想记忆方法.经神经网络控制器控制后,系统的实际输出值与期望值之间偏差不大,因此可将误差模型方程线性化.针对这一线性化模型,可以设计一个智能补偿器来补偿由于训练精度、干扰及时延等因素引起的系统误差.利用以上设计方法,本文对战斗机的“眼镜蛇”机动飞行控制进行了仿真研究,仿真结果表明所设计的控制系统具有良好的性能.

A neural networks controller (NNC) with intelligent compensator (IC) is designedfor nonlinear control system. The main input control signals are provided by the NNC. Thenetworks can be trained off-line, and the implementation can be made by way of associativememory. After the system is controlled by the NNC, the errors between the actual output andthe expected are so small that the error equation can be linearized. To the linear model, the ICis designed to compensate for the system errors caused by the networks training, the distur-bance and the time-delay, etc. Finally, the control method is applied to the 'Cobra' fightermaneuver in the simulation study, and the simulation results demonstrate that designed controlsystem is fairly good in performance.