Research On Damage Detection Of The Girder For High Speed Railway Cable-stayed Bridge Based On Time Series |
| Posted on:2018-03-09 | Degree:Master | Type:Thesis |
| Country:China | Candidate:J Zhang | Full Text:PDF |
| GTID:2322330536962292 | Subject:Engineering Mechanics |
| Abstract/Summary: | PDF Full Text Request |
| With the development of high-speed railway in our coumtry,the research of the high speed railway bridge identification has become a hot research field and the related theory and technology are also in constant development and innovation.In order to solve the problem of damage detection for cable-stayed bridge,a new method based on the time series analysis of stay cable acceleration response combined with neural network under the excitation of high speed train is proposed.The main research work in this paper is as follows:(1)A shell finite element model of the lab cable-stayed bridge aimed to damage identification is established.Dynamic response of each substructure of a cable-stayed bridge caused by high-speed train is obtained by numerical method and the rules of each response under different speeds are discussed.(2)The basic theory and time domain feature about time series such as autocorrelation function and partial autocorrelation are elaborated.How to set up ARMA model is introduced in detail.Sample length and sampling interval are determined according to Shannon sampling theorem.The parameter and order number of ARMA model are determined according to long autoregressive model method and AIC criterion.It proved to be feasible that using the acceleration response of the stay cable to establish the ARMA model through numerical simulation.(3)The combination of first three order autoregressive coefficient algebra of ARMA model is used to be the input of BP neural network training and testing.The size of network is reduced by optimizing the output methods of neural network according to the characteristics of the shell finite element model of main girder.The damage location and degree of girders are indicated through the output of the BP neural network and the results show that the method of the ARMA model based on acceleration response of stay-cables combined with BP neural network can effectively identify the damage of main girder in different conditions. |
| Keywords/Search Tags: | cable-stayed bridge, time series, neural network, damage identification |
PDF Full Text Request |
Related items |