The Pantograph Slider Cracks Fault Detection Based On Curvelet Transform

Pantograph is an important device of electric locomotive to get current from catenary, and its sliders contact with catenary directly. The pantograph coordinates with the catenary and they are responsible for the work of transmitting the power to electric locomotive. Therefore, the state of pantograph has a direct impact on the electrified railway's running safety, its failure may lead to severe pantograph-catenary accidents and hinder the development of electrified railway. At present, China High-speed railway has developed quickly, and it proposes higher requirements to the safety and reliability of the pantograph. Therefore, the implementation of real-time on-line monitoring to pantograph has very great significance.The current pantograph fault detection techniques are mainly the artificial detection, on-line fixed-point detection, image processing-based detection etc. All of them achieve the detection by sensors, infrared equipments, cameras and other assistive devices. However, the existing studies on pantograph fault detection are concentrating on slider abrasion detection, few researches were on bad state of the slider surface such as crack detection. But the crack of the slider would cause more serious pantograph-catenary accidents easily.In this paper, the image of pantograph slider is taken as the sampling object, image processing is adopted as the method, a new algorithm of pantograph slider crack fault detection based on Curvelet transform is proposed to achieve pantograph slider crack fault detection. Firstly, a series of image pretreatments such as image de-noising, image enhancement, edge detection and image segmentation are performed on the collected image, the slider part is intercepted from the whole pantograph image by these processes; then the slider image is decomposed by the Curvelet transform, the crack image features are distinguished from other intrinsic features of the slider image by analyzing the characteristics of Curvelet decomposition coefficients; at last, threshold processing is used to deal with the Curvelet decomposition coefficient matrixes. Thus, the crack characteristics are extracted, and the detection of the pantograph slider crack fault is achieved finally. Aiming at the problem of exiting differences between the practical pictures and experimental pictures, image fusion and image enhancement are utilized to make up for the deficiency of definition.Experiments are tested in the MATLAB Environment in this paper, pantograph slider extraction algorithms and crack recognition algorithms are achieved by programming, and they are used to analyze and verify a group of pictures which have different exposure rates and definitions. The experimental results show that the algorithms in this paper can extract the pantograph slider crack fault effectively.In the end of the paper, the software of pantograph slider crack detection is designed. It achieves the whole process of pantograph slider crack detection by using the algorithm proposed here.