The Research Of Railway Communication Cable Fixtures Detection Algorithm

Communication cables,which transmit signal,are fixed to one side of the wall by fixtures on practical railway.Damage to the fixture will cause the cable to fall off,the maintenance of fixtures is a key task to ensure the safety of vehicle operation.Traditionally,this task is carried out by railway workers who search for defective fixtures that need to be replaced along the railway.To detect the railway equipment,a multi-functional railway comprehensive inspection vehicle using computer vision algorithms was developed to replace the complicated manual inspection.In recent years,researchers have made many efforts on the automatic detection of catenary and rails.However,the research on cable fixtures is still blank.The development of an effective automatic cable fixtures detection algorithm not only reduces the time for railway inspections,but also improves the safety of vehicle operation.Aiming at the task of cable fixture defective inspection,the works of this paper are as follows1.Summing up the domestic and foreign research related to the automatic detection for railway operation equipment.Analyzing the advantages and disadvantages of using deep learning and local feature algorithm at fixture detection respectively.To provide theoretical support for the research on the cable fixture defective detection.2.Aiming at the detection for defective cable fixture on railway,a defective fixture detection algorithm based on local features was proposed.The algorithm consists of two parts: fixture location and classification,defective fixture inspection.Firstly,Yolov3-tiny was used to locate and classify the fixtures.Second,HOG and SVM strategy were used to detect defective fixtures.The experimental results shows that the detection accuracy and the recall rate of the proposed algorithm are 93.4% and96.6% respectively on the multiple-types fixture data set.Theoretical analysis and experimental result show that the algorithm can reach the requirements of practical application.3.Aiming at the real-time requirement of the detection task of multiple-type defective fixtures,a faster detection algorithm has been proposed.On the basics of Yolov4-tiny,firstly,to modify its CSP(Cross Stage Partial)architecture to reduce the repeated gradient information,so that the modified architecture achieves a better performance on the communication cable fixtures dataset.Meanwhile,modifies its classification loss function by introducing a heuristic penalty factor,on the imbalanced dataset,make network to pay more attention on the minor classes in the progress of backpropagation.Experimental result shows that,the proposed algorithm has increased recall and precision by 3.3% and 2.4% compared with the baseline algorithm,respectively.Compared with HOG+SVM algorithm,the detection speed of this algorithm reduced by 40%.Theoretical analysis and experimental results show that the proposed algorithm can significantly reduce the time of detection,guarantee the detection performance at the meantime.