Development Of Temperature-stress Detection System For High-speed Rail

Stress concentration is the main factor that leads to high-speed railway damage, which has muchconcealment and risk. Stress detection in high-speed railway is still lack of appropriate technology.Traditional detection methods don’t meet the rail stress testing of fast, accurate and real-timerequirements. Based on the above reason, this paper proposed a temperature-stress detection methodbased on the principle of Barkhausen noise. On the one hand, by using the electromagnetic method togenerate Barkhausen noise signal, allowing certain lift-off distance between the probe and the rail,and the coupling agent is not required. On the other hand, by using the method of extracting voltageamplitude of signal to identify the corresponding relationship between the eigenvalues andtemperature-stress, and make the appropriate calibration.The content of paper includes mechanical design, hardware design, software design, andexperimental verification of detection equipment. Mechanical part mainly includes design of detectionsensor and detection equipment. Hardware includes circuit design. Software includes design of userinterface and programming to realize logical function of equipment. Experimental verificationincludes the analysis of sampling data, validation of relationship between temperature-stress andBarkhausen signal. Finally, we proposed a data processing method of neural netwoks based on theresults of experiments. By establishing a BP neural network model with the stress as the output valuewhile temperature, mean value, RMS value, ring numbers, peak value, and ratio of peak and fullwidth of half peak value as the input values. We used MATLAB7.8.0Neural Network Toolbox tomodel and simulate neural network, then used the trained BP neural network to validate the testsamples. The results showed that if the neural network designed reasonably, training function andvarious parameters set correctly, the Network would have a high degree of accuracy andgeneralization ability. It was found in practical testing and result analysis that the data processingmethod could be feasible and effective on temperature-stress testing based on MBN, thus realizedcompensation for the temperature influence on stress testing.