基于神经网络的智能结构最优预测控制研究  

Optimal Predicted Control for Smart Structure Based on Neural Network

作　　者：雷永勤[1] 杜永峰[1] 

机构地区：[1]兰州理工大学防震减灾研究所,甘肃兰州730050

出　　处：《城市道桥与防洪》2013年第10期115-118,129,共5页Urban Roads Bridges & Flood Control

基　　金：国家自然科学基金(51178211)

摘　　要：介绍了径向基神经网络和序列最优控制算法的特点,针对结构振动控制中存在的时滞效应,提出了基于RBF神经网络的序列最优预测控制算法,与序列最优控制算法相比能有效地降低结构控制中的时滞效应的影响。结合一栋5层基础隔震框架结构,对该控制方法在控制过程中无时滞和有时滞情况下的控制效果进行仿真分析,仿真结果表明基于RBF神经网络的序列最优预测控制算法在无时滞的情况下与序列最优控制算法控制效果相当,在有时滞的情况下能有效克服时滞对控制系统的影响,控制效果仍较良好。The characters of Radial Basis Function Neural Network(RBFNN) and Sequential Optimal Control(SOC) algorithm are introduced. For the effect of time lag in the process of structural vibration control,Sequential Optimal Predicted Control(SOPC) algorithm is presented based on RBFNN. This algorithm can decrease the influence of time lag availably. A five-storey base isolated frame structure is used to simulate and analyze the control effect in the case of time lag and no time lag in the control process by this method. Simulation results show that control effect of SOPC algorithm based on RBFNN is equivalent with SOC algorithm in the case of no time lag, and can overcome the influence of time lag on the control system in the case of time lag. The control effect is still better.

关 键 词：智能结构 振动控制 时滞 神经网络 

分 类 号：TU352.12[建筑科学—结构工程]