Fault State Intelligent Recognition Of Insulator And Isoelectric Line Based On OpenCV

The support and suspension devices, as the core of high speed rail overhead catenary, whose state determines the stability of the contact suspension system directly. Once faults occur, the current receiving quality will be reduced, even the safety of train operation will be affected. At present, the state detecting and monitoring system of support and suspension device has been widely applied in railway, and compared to the traditional manual inspection its efficiency is improved. However, a large number of inspection images generated by the system still need manual interpretation, which causes limited automation. Therefore, it is necessary to study intelligent fault recognition algorithm based on inspection image.Combined with the characteristics of the inspection image of catenary suspension state detecting and monitoring system, this paper analyzed the successful application of the image processing technology in detecting of catenary geometry, wear and fault recognition of the insulator in power system, then a fault state recognition method of insulator and isoelectric line was proposed based on image processing. Firstly, the images were pre-processed. The image denoising and enhancement was completed using median filtering and multi-scale Retinex. The separation of target and background in the complex environment and light conditions were achieved by fusion and improvement of local adaptive and global binarization algorithm. Then the fault state of insulator and isoelectric line were recognized separately.In the insulator fault recognition, according to the characteristics of the distribution of insulator strings, the recognition method based on the relationship between the position of closed and connected areas was proposed to achieve precise positioning of the edge and corner of insulator. Secondly, local insulator image was accurately extracted by affine transformation. Finally, the are of insulator was corrected by adaptive piecewise affine, and the fault state of insulator was recognized with the band spacing rules of the gray statistical curves as characteristic criterion.In the isoelectric line fault recognition, ROI was specified to position and extract isoelectric line region indirectly with the accurate information of the steady arm endpoint, which was obtained by the improvement of the probabilistic Hough transform line detection. Then, the fault state of isoelectric line was recognized by detecting the existence and the number of the closed region formed between isoelectric line and the steady arm bearing.Finally, this paper verifies the reliability and feasibility of insulator and isoelectric line fault state recognition algorithm and system, through testing inspection images of catenary suspension state detecting and monitoring devices under different conditions of several railway lines.