摘要
研究了磁流变阻尼器的磁滞效应对控制效果的影响,提出了采用神经网络预测的方法来减小磁滞效应对振动控制的不利影响.通过过去时刻和当前时刻的结构响应来预测将来时刻的结构响应,并用线性最优控制(LQR)算法经将来时刻结构响应计算得到主动控制力,为磁流变阻尼器产生该主动控制力赢得时间.研究结果表明:通过神经网络预测能有效补偿磁滞时间,使控制效果接近无磁滞效应的控制效果.
The adverse effects of magnetic hysteresis of magnetorheological （MR） dampers on control result were investigated. Neural network prediction method is proposed to reduce the disadvantage of magnetic hysteresis. The future structural response is predicted on the basis of the past and the present structural response through neural network and the active control force is calculated by the future structural result with linear quadratic regulator control algorithm. As a consequence, enough time is won to compute and generate the control force. The simulation results imply that neural network prediction can successfully compensate for the magnetic hysteresis of MR dampers and the control result is almost the same as that of no magnetic hysteresis.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2007年第3期110-112,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金重大国际合作资助项目(50520130296)
关键词
磁流变阻尼器
磁滞效应
神经网络
预测控制
magnetorheological dampers
magnetic hysteresis
neural network
prediction control
