四旋翼飞行器控制系统稳定性优化设计仿真

摘要

针对四旋翼飞行器高度非线性、多输人多输出、强耦合欠驱动等易受外界干扰的控制系统.传统的PID控制方法对高度非线性易受扰动的控制系统,系统快速性得不到体现且易出现抖振的问题,因此提出改进RBF神经网络自适应PID控制算法.改进RBF神经网络模型的径向基函数中引入自适应向量参数,加快了径向基函数的收敛速度,从而提高模型控制的准确度和加快系统收敛.采用梯度下降法训练网络中心矢量、基宽向量、网络权值和在线调整PID参数.通过对建立的仿真模型实验,仿真结果表明,改进RBF神经网络算法与传统PID控制算法的控制器相比,对可变模型具有更强的自适应能力,准确性和快速性得到较大的提高,具有更强的稳定性.%In the paper,a self-adaptive PID control algorithm of the improved radial basis function (RBF) neural network was proposed.The parameter adaptive of the vector was introduced in the improved RBF neural network model,speeding up the convergence rate,so as to enhance the accuracy of model control and speed up convergence system.The Gradient descent method was used to train network centric vector,base width vector,the network weights and adjust PID parameters online.The simulation model experiment show that compared with the traditional PID algorithm of control,the improved algorithm of RBF neural network controller has better adaptive ability of variable model,and the speed and accuracy are greatly improved.