| With the continuous development of China's railway industry, a higher railway transportation safety requirement should be satisfied. Nowadays, railway maintenance in our country is still in the manner of the manual detection along the railway line for maintaining or inspect railroad tracks, In that process, the defect detection relies mainly on personal subjective judgment, it is difficult to ensure the objectivity and accuracy of the results of the object, and the requirements on the experience and accountability of the trackmen are relatively high. The traditional inspection already began to expose the weakness in the process of the development of the railway, it is hoped that there is a safe, accurate, rapid railway maintenance method to ensure the safty of the railway operations, therefore, developing a kind of automation equipment, which could be applied in the actual railway road safety inspection, will be put on the agenda.The main work is as follows:According to China's railway maintenance standards and the related experience in railway maintenance at home and abroad, the paper analyses the existing problems and railway road maintenance targets, in order to achieve the goals of improving the detection efficiency, the working environment of the railway maintenance, the quality of railway maintenance and the safe of the railway operations, A automation method of railroad safety inspection based on video image processing theory was proposed.hi view of characteristics of railway defect, the related theories, used by the automated inspection scheme based on image processing, was proposed. Analyzing feasibility with examples was done. Analyzing the whole composition of railway defect detection system based on image processing and the relationship between the various components, the corresponding theory and method of building the system were proposed.Research on sleeper defect detection key techniques and theoretical support based on rough neural network, focusing on analyzing the main content and the sleeper detection goal request, this paper put forward the related theory of feature extraction, feature discretization, feature reduction, and rough neural network model building involved in defect detection system. Analyzing feasibility with examples was done. Analyzing the characteristics of the target area and disease of sleeper in detail, the related theories and methods of region and diseases extraction were put forward, and the feasibility of the method and theory with examples was proved.Based on the above studies, the railway safety automatic detecion system will be designed. The system consists of the five parts:tracks inspection subsystem, sleeper detection subsystem, testing ballast subsystem, fasteners detection subsystem, gaskets detection subsystem. In this paper, the research focuses on a detailed theoretical discussion and examples on the sleeper detection subsystem. The example results show that this design approach is effective for sleeper defect detection; the development of the system also has great application value. |
