Study On The Method For Functional Validation Of On-board Equipment Of Train Control System Based On DMI Image Recognition

As a unified technology platform for Chinese high-speed railway, CTCS-3 Train Control System provides an important guarantee for the safe and efficient operation of railway. In order to verify whether a CTCS-3 train control system can meet the requirements of related technical specifications, as well as to ensure the cross-the-line-operation of trains, it is necessary to carry out a series of testing. The black-box testing method based on data-driven mechanism is a common approach for the laboratory testing of on-board equipment. It consists of three steps, namely test data generation, testing execution and results analysis. Currently most of the work is done manually being characteristic of heavy workload and low efficiency. Therefore, it is necessary to carry out the research on the key methods of automatic testing. As the key technology of testing execution, the extraction of DMI (Driver Machine Interface) information and functional validation of on-board equipment are the basis of automatic testing.Previously Beijing Jiaotong University has done some research on this issue. An automatic test platform has been built and the recognition method of DMI information has been proposed. But the previous work didn’t cover the recognition method of plan area, and the deep analysis of DMI information. This paper focuses on the method of functional validation for CTCS-3 on-board equipment based on DMI image recognition. The main works are as follows:1. Giving an overview of overall structure of CTCS-3 train control system and automatic test platform for on-board equipment. A method of functional validation for on-board equipment based on DMI image recognition is proposed and the DMI regions to be recognized are analyzed.2. Designing the method of DMI image recognition. After locating the border and segmentation of DMI regions, the recognition method for different kinds of DMI elements is deeply studied. The minimum distance classifier and template matching are used for classifying icons and artificial neural network are used for characters. Due to the complexity of the plan area, the deceleration point and MRSP(Most Restrictive Speed Profile) are divided by diffident thresholds, and vertexes of MRSP are recognized by Harris corner detection. The accuracy and factor of recognition for each DMI element are analyzed.3. Modeling the process of test execution which represents the change of the state of the CTCS system and defining the calculation method of system state. By comparison of actual system state formed with DMI information and the expected system state formed by test sequence, the functions of on-board equipment can be validated. And an example of the functional validation method based on DMI image recognition is given.4. Designing a functional validation system for on-board equipment and implementing it using MFC and OpenCV (Open Source Computer Vision Library) in the integrated development environment of Visual Studio 2010. The system can realize the recognition of DMI information and the functional validation for on-board equipment using recognition result and test sequence.