基于NARMAX模型的阀控非对称缸神经网络预测控制  

Neural Network Predictive Control of Valve-controlled Asymmetric Cylinder Based on NARMAX Model

作　　者：袁磊 蒋刚 郝兴安[1] 刘思颂[1] 陈清平 徐文刚 YUAN Lei;JIANG Gang;HAO Xing-an;LIU Si-song;CHEN Qing-ping;XU Wen-gang(School of Mechanical and Electrical Engineering,Chengdu University of Technology,Chengdu,Sichuan 610059;Chengdu Lingchuan Special Industry Co.,Ltd.,Chengdu,Sichuan 610110)

机构地区：[1]成都理工大学机电工程学院,四川成都610059 [2]成都陵川特种工业有限责任公司,四川成都610110

出　　处：《液压与气动》2023年第1期86-93,共8页Chinese Hydraulics & Pneumatics

基　　金：四川省重大科技专项(2020ZDZX0019);四川省科技计划重点研发项目(2021YFG0075,2021YFG0076)。

摘　　要：针对非对称缸位置跟踪控制精度较差,提出了一种基于非线性自回归平均滑动离散模型(NARMAX)和量子粒子群算法的神经网络预测控制策略(QPSO-NNMPC)。利用NARMAX模型表示阀控非对称缸的动态模型,使用粒子群算法优化BP神经网络(PSO-BP)对阀控非对称缸系统在线预测,使用量子粒子群算法(QPSO)对目标函数非线性优化。仿真结果表明,在不同频率期望信号与变干扰力情况下,该控制策略具有良好的跟踪效果和鲁棒性。Aiming at the poor accuracy of asymmetric cylinder position tracking control,a neural network predictive control strategy(QPSO-NNMPC)based on nonlinear autoregressive mean-slip discrete model(NARMAX)and quantum particle swarm algorithm was proposed.The NARMAX model is used to represent the dynamic model of the valve-controlled asymmetric cylinder,the particle swarm optimization BP neural network(PSO-BP)is used to predict the valve-controlled asymmetric cylinder system online,and the quantum particle swarm optimization(QPSO)is used to optimize the objective function nonlinearly.The simulation results show that the control strategy has good tracking effect and robustness under the condition of different frequency desired signal and variable interference force.

关 键 词：阀控非对称缸 BP神经网络 粒子群算法 量子粒子群算法 位置跟踪控制 

分 类 号：TH137[机械工程—机械制造及自动化] TP273[自动化与计算机技术—检测技术与自动化装置]