Liqiang Zhu

Recognition algorithm for the disengagement of cement asphalt mortar based on dynamic responses of vehicles

Disengagement of emulsified cement asphalt mortar will increase the dynamic action between the vehicle and the track; as a consequence, the rate of cement asphalt mortar disengagement will increase further. This is a serious threat for the safe operation of high-speed railways and the service life of rail equipment. In this study, a vertical coupled model for the vehicle–China Railway Track System II-type slab track with cement asphalt disengagement was established. The cement asphalt mortar was divided into units in order to simulate the arbitrary length of disengagement. Under different conditions, the effects of the cement asphalt mortar disengagement on the dynamic characteristics of the coupled model were analyzed. The results show that when the length of disengagement exceeds 0.65 m under the condition of horizontal complete disengagement, the dynamic responses of the system increase much sharply than the condition of horizontal partly disengagement. Because of the difficulty in identifying defects in the track substructure, a novel method was proposed to rapidly identify the cement asphalt mortar disengagement based on the dynamic responses of the coupled system and particle swarm optimization–support vector machines. The feature vectors were extracted from the acceleration of the wheelset, which were used as training samples in support vector machines. The classification results show that the recognition algorithm based on the acceleration of the wheelset and support vector machines is effective. The location of the track plate with the cement asphalt mortar disengagement at lengths of 0.65 m, 1.3 m, and 1.95 m can be identified with an acceptable accuracy. The robustness of the proposed algorithm under different vehicle speeds, track spectrums, and signal–noise ratios was verified. Recognition of defects in the track substructure using sensors mounted on in-service vehicles has the potential to provide a valuable tool for ensuring the safe operation of railways and for developing a maintenance plan.

High-speed railway clearance surveillance system based on convolutional neural networks

In this paper, the convolutional neural networks with the pre-trained kernels are applied to the video surveillance system, which has been built along the Shanghai-Hangzhou high-speed railway to monitor the railway clearance scene and will output the alarm images with the dangerous intruding objects in. The video surveillance system will firstly generate the images which are suspected of containing the dangerous objects intruding the clearance. The convolutional neural networks with the pre-trained kernels are applied to process these suspicious images to eliminating the false alarm images, only contain the trains and the empty clearance scene, from other suspicious images before the final output. Experimental result shows that, the process of each test image only takes 0.16 second and has a high accuracy.

The Estimation Approach of Rail Thermal Stress Based on Vehicle-Track Dynamic Responses

On account of the difficulty in measuring the true rail thermal stress, a new method is proposed to rapidly evaluate rail thermal stress based on the dynamic responses of vehicle-track coupling system. A vertical slab-track model with the rail longitudinal thermal stress is established. The natural vibration characteristic of the track under the influences of the rail longitudinal stress distribution has been calculated individually. The changed vibration modal function and frequency of track are used in calculating the dynamic response of vehicle-track coupling model. It can be seen from the dynamic responses that vibration frequency is shifted and amplitude peak is enhanced by thermal stress. In this paper, the digital features are extracted from the responses of wheelset and rail in time and frequent domain, which are used as training samples in SVM. Then, the rail longitudinal stress can be estimated with the prediction model. The numerical simulation results indicate that, by utilizing the train as the excitation source of the dynamic system, the rail longitudinal stress can be estimated with measurable response of wheel and rail at an acceptable accuracy.

Jointless track monitoring system based on Fiber Bragg Grating sensors

Jointless track is an essential part of high speed railway. But the damage such as the buckled rail and the broken rail occasionally happened. In this paper a system is designed for monitoring the status of jointless track based on analyzing the principle of FBG (Fiber Bragg Grating) sensors and the structure of double columns end thorns. The layout of the FBG sensors, including displacement sensors, strain sensors and temperature sensors, which are installed on the end thorns at K761 position on the up line from Suzhou East Station to Bengbu South Station and Huaihe River Bridge on the China high speed line Beijing-Shanghai, is described in detail. Also the remote monitoring system software is introduced in this paper. The results of field tests show that the correlation among temperature, strain and displacement is well and sensors are stable and reliable during a long term remote monitoring.

A mosaic method for large perspective distortion image

An image mosaic method focusing on images with large perspective distortion is proposed. Correction of perspective distortion based on perspective transformation model is applied first in this method to reduce the inconvenience of the distortion. To the corrected images, feature extraction, feature matching relied on the features in the overlapped area, and weighted fusion are applied to obtain a seamless mosaic image. The experimental result shows that this method is effective to the images with large perspective distortion.