PID神经网络在飞控解耦中的仿真研究  

Application of PID Neural Network in Simulation of Flight Decoupling Control

作　　者：程波[1] 何贵波 李天[1] 

机构地区：[1]沈阳飞机设计研究所,沈阳110035 [2]空军第二飞行学院,西安710306

出　　处：《电光与控制》2012年第6期41-44,57,共5页Electronics Optics & Control

摘　　要：在进行飞行控制律设计时,为了获得好的控制品质,有必要对各输出量进行解耦,并且使控制器能根据被控对象动力学特性的改变而自适应调整。为此,介绍了两种PID神经网络解耦控制结构和一种神经网络辨识器,并分别给出了PID神经网络解耦控制器和辨识器的前向算法及反向传播算法;最后分别应用两种解耦方法针对飞机横侧向线性模型进行计算仿真,得到了两种解耦结果;证明了PID神经网络解耦控制的有效性,并比较了两种结构的优缺点。In design of the flight control law, it is necessary to design a self-adaptive decoupling controller to overcome the nonlinear and varied dynamic characteristics in order to get good performance. Two kinds of PID neural network decoupling controllers and an neural network identifier were introduced, and the feedforward and back propagation algorithms were presented. The the two controller were applied to aircraft linearity model for simulation, and two kinds of decoupling result were obtained, which proved the effectiveness of PID neural network decoupling controllers. The advantages and disadvantages of the two controllers were also illuminated.

关 键 词：飞行控制律 自适应调整 神经网络 解耦控制 线性模型 

分 类 号：V24[航空宇航科学与技术—飞行器设计] TP183[自动化与计算机技术—控制理论与控制工程]