Research On Acoustic Emission Faults Detection Technology For Simulated Bridge Structure

Real-time load conditions of the bridge monitoring and fault diagnosis will play an important role of the industry, as the ratio of bridge in transportation junction increased year by year. Poor working conditions and time varying load conditions lead to the bridges easily damaged. Usually the bridges distributes wide, and the bridges are unattended, maintenance of the bridges are inconvenience. Therefore the study of bridge condition monitoring and fault diagnosis has practical significances to control of the maintenance costs and to optimize the maintenance strategy.The paper analyzes the principle of acoustic emission phenomenon which is caused by structure deformation, describes the causes of several common fault acoustic emission signal. Characteristic signals of early failure are weak, time-varying and non-stationary. Proposes to use wavelet transform combined with neural network to diagnosis the bridge faults.In this paper, proposing use wavelet threshold de-noising to eliminate noise interference, which based on analyzing continuous wavelet transform and its discretization, and with the use of Matlab to simulat and verify it. Using BP neural network to identify the various fault conditions of the bridge. To reduce the BP neural network structure, selecting the method of statistical analysis to extract characteristic features from signal which after wavelet threshold de-noising,as the input of the network. Design a bridge structure fault type recognition system Based on wavelet threshold de-noising and neural network classifier. Using the recognition system to design a bridge structure fault type, the result of the analysis is accurate.