Convolutional neural network construction method for rail corrugation recognition

Abstract

The present invention relates to the monitoring of short-wave irregularities of rail transit wheels and rails, and in particular to a convolutional neural network construction method for rail corrugation identification, including the above-mentioned method for collecting and processing specific vibration information of rail transit vehicles, removing all In the frequency domain information, the frequency domain data in the frequency range that does not affect the specific vibration detection feature representation are obtained respectively to obtain optimized frequency domain information, and each optimized frequency domain information is expressed as a frequency spectrum with the same length of the frequency coordinate axis, and the The spectral data are used as training samples for a one-dimensional convolutional neural network. Compared with the traditional detection and identification using corrugation tester CAT, with this method, after training through the convolutional neural network, it is only necessary to collect vibration information and vehicle displacement information in the follow-up, and input the signal into the convolutional neural network containing specific training. The computer processing of the integrated neural network can quickly and efficiently identify rail corrugation.

Description

technical field [0001] The invention relates to the monitoring of short-wave irregularities of rail transit wheels and rails, in particular to a method for constructing a convolutional neural network for rail corrugation identification. Background technique [0002] Rail corrugation is the wavy wear of the rail, such as figure 1 shown. Due to the high load capacity and high operating density of the subway, complex line conditions (such as small curve radius and diversified track structure), frequent vehicle starting and braking, the wheel-rail interaction is intensified, and the wavy wear of the rail (referred to as rail corrugation) is serious. . [0003] Rail corrugation will bring a series of problems, such as causing abnormal vibration and noise pollution of the vehicle track, and reducing the fatigue reliability of the vehicle and track components. According to literature reports, the noise difference in the cab of a subway line with rail corrugation before and after...

AI Technical Summary

Problems solved by technology

The chord measurement method uses the rail itself as a moving reference system, so that the measurement reference benchmark is in a state of change with the unevenness of the rail height, causing the transfer function ratio (the ratio of the measured value to the actual value) to not be constant at 1, resulting in the chord measurement method often Can't really and reliably test rail corrugation

The inertial reference method usually uses the quadratic integral of the axlebox acceleration to characterize the corrugation value. Its disadvantage is that it is susceptible to interference from wheel wear, and due to the influence of the high-pass filter, the measurement error is large under low-speed conditions. Generally, only on large rail inspection vehicles

Machine vision methods often require sophisticated photoelectric camera equipment and complex image processing methods, and use complex pattern recognition technology for specific back-end processing, which is difficult and expensive in practical application

For example, using one of the traditional methods of corrugation tester CAT to test a subway line often takes as long as one month, which takes a long time. The reliability and versatility of traditional detection methods are not enough and the cost is high.

Images