Image-based Detection Of Abnormal Condition For High-speed Maglev Long Stator Track

The high-speed maglev train is a new type of transportation which uses the electromagnetic force to realize the non-contact "zero height" rapid operation of the vehicle along the track,in which the maglev track plays an important role.It not only interacts with the vehicle suspension guidance system to generate electromagnetic suction to achieve the stable suspension guidance of the vehicle,but also it is the long stator component of the train synchronous linear traction system which interacts with the vehicle suspension electromagnet to generate the linear traction force.Its state directly affects the traction and braking performance of the vehicle,especially the abnormal conditions such as scratches,cracks,shedding of the tooth surface epoxy layer,cable hanging and missing installation bolts.However,these orbital conditions are difficult to be detected in time due to the small size in the rail transit system which is an elevated form of long distance.If the treatment cannot be found in time,it will gradually lead to deterioration of the electrical performance of the motor and even cause an accident.In order to ensure the stability,safety and comfort of vehicle operation,it is necessary to pay attention to the maintenance of high-speed maglev track and accelerate the development of fast and effective detection system.Referring to the experience of wheel-rail traffic,an image detecting system and method is suitable for being used to detect the above abnormal conditions.Supported by the national key scientific research project "High-speed maglev track line comprehensive detection system",an image-based mounted high-speed maglev track detection system was studied and designed in this paper.The algorithm research was carried out on the detection of abnormal conditions such as abrasion,cracking and shedding of the long stator tooth surface epoxy layer.Solution for abnormal conditions such as cable hanging and missing mounting bolts were proposed,and some experimental verifications was carried out.Firstly,an image-based piggyback detection system scheme is designed according to requirements,which includes an image acquisition device and a ground information processing system.For the image acquisition device,the camera,lens and light source are calculated,analyzed and selected.The structure of the device was designed and analyzed.The experimental results of image acquisition show that the device can obtain high-quality long stator track images.For the ground information processing system,the process combination of algorithm modules such as image processing,location segmentation,feature extraction and classification was designed.Secondly,due to the influence of the installation environment,space and light,the image processing technology was studied in this paper.The acquired orbital image was transformed into an orthogonal projection by perspective transformation,and the adaptive histogram equalization was used to alleviate the uneven illumination problem existing in the image,and the contrast of the image was improved.On this basis,based on the gray,spatial and shape features of the track image,an image cutting method based on thresholding is proposed to realizes the location cutting of tooth surface image,cable image and bolt image.In order to further detect the condition of the tooth surface epoxy layer,an image enhancement algorithm based on sliding window is proposed,which solves the problem of less feature samples,and a multi class classification method based on nonlinear support vector machine is designed to solve several anomalous classification problems such as tooth surface epoxy layer abrasion,crack and shedding.Thirdly,aiming at the detection problem of cable hanging,according to the feature of cable covering the far end tooth surface,a detection algorithm based on projection integral and theoretical projection curve for similarity measurement was proposed.Aiming at the detection problem of installation bolt missing,according to the bolt as standard part with uniform specification,the bolt detection method based on the template matching was proposed in the paper.The above two algorithms are preliminary schemes,which need further experiments and in-depth study.Finally,the experimental device and the test system were designed and tested,and the effectiveness of the above algorithm was partially verified.The research results lay a foundation for the rapid and effective detection of track lines,and provide maintenance guarantee for further application of high-speed maglev system.