PSO Based Diagnosis Approach for Surface and Components Faults in Railways

Railway transport is a type of transport which is commonly used today. Rail line must be robust due to the heavy structure of the railway vehicles. Components constituting the rail line are very important to prevent the disruption of transportation. In this study, faults are determined by monitoring the rail and fastening components constituting the railway. Test vehicle was used to get experimental data. The left and right rails were viewed from different angles by four cameras placed on the test vehicle. Status monitoring and fault detection were performed by applying image processing and particle swarm optimization methods to the images taken. Rail surface was determined by taking the right and left images of rail line from the right and left cameras from different angles. Images taken from the right and left cameras were assembled for the detection of faults in the rail surface. Image matching was performed during the detection of fastening components and the rail surface. Matching was performed for each image taken from the camera by taking into account the correlation coefficient. In the determination of rail surface, template image and similarity were measured by taking specific sections on the image respectively. After template image and similarity ratios of all sections taken from the image were calculated, the sectional image with the highest correlation coefficient was determined as the rail surface. Sections were taken randomly from the image during the detection rail component. The correlation coefficient of the template image was calculated with the sections taken. Correlation coefficient was used as the coherence function in the particle swarm optimization, and fastening components were determined. Condition monitoring was performed by combining the detection results obtained.