The Ultrasonic Guided Wave Broken Rail Monitorinc Algorithm Based On Variational Mode Decomposition

Rail transit is one of the most important means of transportation, and broken rail is the core problem affecting traffic safety. Exposed steel rail in railway transportation is often subject to the erosion of various kinds of natural phenomena. Train is made up of many kinds of mechanical and electrical devices. Its operation will produce and even introduce all kinds of noise. To accurately check track working conditions, and filter various kinds of noise in the environment through good algorithm is very important.In the paper, the signal denoising algorithm in the ultrasonic guided wave broken rail monitoring system is studied. ultrasonic signal in the system is a kind of nonlinear and nonstationary. weak signal, and is easily affected by noise. In order to achieve better denoising effect, the empirical mode decomposition (EMD) algorithm, ensemble empirical mode decomposition (EEMD) algorithm are applioed to filter and decompose the signals. It is found that endpoint effect and modal aliasing appear for EMD algolithm in the process of solving,which reduces EMD decomposition efficiency and signal-to-noise (SNR). EEMD, to which white gaussian noise is added on the basis of the EMD added, inhibits the modal aliasing phenomenon to some extent, but it is complicated. Besides, improper selection of some of the parameters still causes modal aliasing, and results in failure of EEMD algorithm. To respond to the above deficiencies, variational mode decomposition (VMD) algorithm is used for ultrasonic guided wave signal denoising. According to the comparison results of EMD and EEMD, VMD is simple to implement, and reduces the algorithm complexity by 30%?50%; SNR of VMD increased by 6%?31%, and the mean square error reduced by 9%?19%;for weak signals, the signal amplitude attenuation after EMD and EEMD reconstruction is very big, and the signals are submerged by noise; the signal amplitude attenuation after VMD is close to the original signal, and increased by 23%?48% relatively to that of EMD and EEMD. VMD algorithm has the best performance, optimal denoising effect, especially for especially serious noise , and can increase SNR greatly.The algorithm modeling and simulation is conducted on the actual signals collected in the system under the severe environment through MATLAB. Finally, the algorithm is verified with upper monitor system developed based on C#.NET. according to the results, VMD algorithm can effectively filter out the noise and conduct signal reconstruction, and obtain higher SNR in the case of relatively low algorithm complexity. Thus, it has a certain reference value for the design of ultrasonic guided wave broken rail detection system.