Acoustic Emission Monitoring And Pattern Recognition For Bridge Cable Corrision Damage

Nowadays, the cable-stayed bridge is one of the main types of modern large-span bridges because of its outstanding capability of spanning. The cable is one of the most important components of the stayed bridge. Recently the main bridge cable is made of high strength steel,wire or strand. However, because it is exposured in the outside environment for a long-term, is prone to erosion chloride or other corrosive media. Over many years, the cable structure will have no aversion to a certain degree of damage, the mechanical peoperties will certainly be decayed, and the life will gradully shorten. In extreme cases, the cable can cause the bridge to collapse suddenly rupture, causing a catastrophic accident. Therefore, it is very necessary to monitor the corrosion of the bridge. Acoustic emission technology as a high sensitivity of NDT techniques for monitoring bridge cable corrosion damage proved to be an effective method. Then analyzed by means of signal processing technology to achieve recognition of acoustic emission sources, and thus a reasonable evaluation of corrosion damage to the cable.This article major study contents are as follows:Firstly, analyse the Strand in chloride environment electrochemical corrosion experiments realistic approximation analog cable corrosion environments, use the analysis algorithm (PCA) to extract acoustic emission monitoring data based on principal component eigenvalues, Applications based on global optimization were optimized clustering algorithm for large data monitoring information obtained mining, statistical analysis, classification. In the end, find similar injury phase is characterized by the law and to analyze the sound source characteristics of the cable in different stages of corrosion damage process.Secondly, analyse the propagation characteristics of acoustic emission in the cable structure, due to the non-stationary characteristics of acoustic emission signals, this paper will. choose the best time-frequency analysis technology:HHT, which is based on empirical mode decompostion (EMD) when comparing the time-frequency analysis knologies. Meanwhile, it will be applied to all kinds of acoustic emission monitor technology. By studying the HHT algorithm, extension method will be applied to resolve the endpoint effect of EMD; Envelope fitting algorithm which use the constraint-based spline interpolation will be applied to resolve the overshoot and undershoot problem of the traditional EMD in using cubic spline fitting; For the model aliasing effects of EMD, this paper will use the ensemble empirical mode decomposition to solve; Finally, analyse the Hilbert spectrum characteristic of different corrosion damage stage.Thirdly, apply the self-organizing feature map neural network technology to the cable corrosion damage, SOFM is one of the competition neural network which based on an unsupervised learning method, the map through the input mode self-organizing training and judgment, ultimately transforming data into different types. By processing the acoustic emission monitoring data, the adaptive clustering result was visually presented on a two-dimensionl map on the output layer, finally achieve the recognition of different sound sources.