| With the large-scale opened and used of high-speed rail and EMU, large area layed of CWRhas become a necessity. The ability to detect the rail thermal stress accurately has an extremelyimportant significance for analyzing the stability of the CWR and the prevention of rail expansion.The main content of my topic is, to complete the machine assembling and commissioning of therail thermal stress detecting patrol equipment which is easy, friendly interface and can real-timeacquisition, processing and display data based on the research of the theoretical of Barkhausennoise and the combination of the causes and impact of the rail thermal stress. Therefore, this paperproposed the feasibility and reliability of detection of rail thermal stress based on the method ofBarkhausen.Firstly, based on the electromagnetic theory, this thesis analyzed the generation nature andtesting mechanism of Barkhausen signal, introduced the MBN signal characteristics, deeply studiedthe main physical factors of Barkhausen signal, including the excitation and properties of thematerial itself, focused in particular on the double effect of the temperature on the MBN signal.Secondly, on the basis of the preliminary work we completed the machine assembling andcommissioning of the rail thermal stress detecting patrol equipment which is easy-to operate andhas a friendly interface. Finally, a combination method of wavelet decomposition and neuralnetwork is used to analyze the MBN signal. After using the db5wavelet with six layers todecompose the MBN signal and reconstructing low-frequency signal at each layer, the mean andRMS value are extracted and followed by a discussion on the relationship between features and thevariation of the applied temperature and stress. A new neural network model is built by taking thetemperature, the mean and RMS values of MBN signals and the decomposition coefficients as theinput and the stress as the output. And we do comparison with the previous neural network model. |
PDF Full Text Request