基于RNN的神经网络辨识液气联合工作液压作动器模型  

The Identification and the Model of the Hydraulic & Pneumatic Actuator Based on RNN Recurrent Neural Networks

作　　者：孙宝强[1] 叶金杰[1] 潘紫微[1] 

机构地区：[1]安徽工业大学机械工程学院,安徽马鞍山243002

出　　处：《机械工程师》2009年第4期41-43,共3页Mechanical Engineer

基　　金：安徽省教育厅自然科学基金项目(2004kj056zd)

摘　　要：建立了一种回归神经网络辨识非线性液压作动器系统数学模型的辨识方法,研究了基于回归神经网络内部状态反馈的辨识算法,利用辩识实验获得的过程输入/输出数据动态调整神经网络权值。仿真结果辨明:回归神经网络描述的液压作动器系统数学模型具有较高精度,算法全局逼近能力良好。The method of identitying was built to identify mathematical model of nonlinear hydraulic actuator system. The algorithm of identifying was researched base on inner state feedback of recurrent neural networks. The weights of neural networks were dynamic adjusted by input/output data of process that it was acquired by experiments of identifying. The result of simulation shows that mathematical model of neural networks of hydraulic-gas Actuator system has better precision and algorithm has ability to approximate error of global.

关 键 词：回归神经网络 系统辩识 液压作动器 动态BP算法 

分 类 号：TD422.5[矿业工程—矿山机电]