基于人工神经网络的结构磁流变阻尼器半主动控制-机械设计及理论专业论文.docxVIP

AbstractAbstract Abstract Abstract in recent years the semi。active control of structure with MR damper based Oll neural networks has been put forward，and has a broad prospect To evaluate the potential of MR dampers in con订ol applications and to take full advantages of the unique features of these devices，the accurate model to reproduce the behavior should be studied．However，it is difficult to accurately establish the mechanical model of MR dampers because ofthe complexity ofMR fluid， Over the last decade there has been considerable research on the development and application of artificial neural networks(ANN)．It has been shown that feedforward neural networks can arbitrarily approximate well any continous function． Based on test of the MR damper，this paper establishes the neural network model of the forward and inverse character of the MR damper．It is found in this paper that the ANN model call identify the inverse dynamical characteristics of MR damper rapidly and accurately． Concerning with the uncertainty mad nonlinear behaviors of structures，this paper also constructs a neural network with a single hadden layer trained by back propagation(BP)to approximate the inverse dynamics of a multi—degree—of-freedom structure subjected to earthquakes．The computer tests demonstrate its identifying ability． Based on BP network structure，the developing scheme of neural-indictor network and neural-control network are presented by training off-line the neural—indictor network and neural—control network with the idealistic—test．Computer simulation of a numerical example shows that the neural network models of the inverse dynamical characteristics of MR damper and the identification of structure with the neural network algorithm can reduce the response of structure effectively and take maximum advantage of the properties of MR damper． Key words：MR dmnper， semi·active control，neural network，identification 独创性声明本人声明所呈交的论文是我个人在导师指导下进行的研究工作 独创性声明 本人声明所呈交的论文是我个人在导师指导下进行的研究工作 及取得的研究成果