The Design Of Train Bottom Bolt On Image Processing Research On Detection Technology

The bottom of the subway train is the most complex part of the subway structure and the most complicated parts.At present,the state detection of the subway train bottom is mainly manual detection means such as section inspection and factory inspection,which is not realtime and has the situation of missing inspection.Firstly,this paper analyzes and designs the train underbody bolt detection system based on image processing.Based on the system architecture design,the software and hardware schemes of the system are designed respectively.According to the hardware scheme design,the hardware system is built to obtain the original image of the underbody.Due to the image quality problem of the collected original image of the underbody,the collected image of the subway train underbody is repaired and corrected by image processing technology,including the correction of uneven illumination and distortion of the image.The high-quality image processed by image processing is spliced to obtain the complete image of the underbody.After the original image of the underbody is processed and corrected,a high-quality panoramic image of the underbody is obtained,which can meet the quality requirements of the target detection algorithm for the detected image.Therefore,the target detection algorithm is analyzed and studied through image processing,the Yolo algorithm and SSD algorithm are studied respectively,and the underbody bolts are detected respectively,By comparing and analyzing their respective advantages and disadvantages for bolt detection,on this basis,combined with the characteristics of the system,the SSD algorithm is optimized in terms of network structure,loss function and non maximum suppression,so as to achieve better bolt detection effect and meet the requirements of the system and subway for bolt detection.Finally,according to the designed scheme,the system is implemented and built in line 21 of Zhenlong depot of Guangzhou Metro.On this basis,the off-line test of the system is completed.The test results are analyzed and evaluated through relevant index parameters to verify that the system can meet the requirements of real-time detection of subway underbody.