基于粒子滤波神经网络的倒立摆控制系统

Inverted Pendulum Control Based on Particle Filter Neural Network

为了提高倒立摆的控制精度,采用粒子滤波算法估计网络权值,改进神经网络算法,用训练好的神经网络代替倒立摆系统,得到基于粒子滤波神经网络控制的倒立摆系统。仿真结果表明,粒子滤波神经网络控制器仅需较小的控制就能获得较好的控制效果,其保持倒立摆平稳所移动的调整位移、角度变化幅度更小,调整速度和角速度也更小。

In order to improve control precise,particle filter is used to estimate weight of network to improve its algorithm,then the trained network takes the place of inverted pendulum. Simulation results indicate the controller based on particle filter neural networkcan acquire better control effect with smaller control,it can keep balance of inverted pendulum under smaller variation of displace,angular,speed and angular speed.