Interface type test guarantees the suitability and safety of interface between high-speed train and ATP (Auto Train Protection) onboard equipment, which makes the generation of test sequences much significant. However, test sequences are normally redacted manually, of which the requirements of availability, safety, redundancy and coverage of test cases usually fail to meet expected standards. Specifying optimal time spend and energy consumption of test sequences, we use a simple deep learning network to decide the potential set of test cases at decision points, and implement the modified genetic algorithm adapted to our objective to generate and optimize the test sequences. Experiments are conducted using the field data of type test for CRH3C high speed train with CTCS3-300T ATP onboard equipment. Comparing with the actual sequences used in type test indicates that, based on satisfying the requirements of test cases, the proposed strategy can effectively generate test sequences with optimized time spend and energy consumption by 10.63% and 27.16% better improvement respectively.
展开▼
