Intelligent Generation of Test Cases for a Parallel Testing System: A Case Study on Railway Systems

Currently, the testing approach for complex systems relies on manual test case generation, resulting in problems, such as low efficiency and incomplete test case development. To address these challenges and improve the quality of test case generation for complex systems, this article proposes an intelligent method for generating test cases in parallel testing systems. Within the domain of railway systems, characterized by inherent intricacies, the centralized traffic control system serves as a pertinent example. This typical large and intricate railway transport system presents significant challenges, particularly in ensuring safety and conducting comprehensive functional testing. The design of the parallel testing system is detailed using the artificial systems, computational experiments, and parallel execution methodology, where an artificial system is built in a data-driven way that realistically replicates the test environment of a real system. Computational experiments were conducted on a test case library using the bidirectional encoder representations from transformers (BERT) model of natural language processing. BERT's next sentence prediction task was used for the associative learning of test case pairs. Finally, the test case intelligent generation software completes the parallel execution of the final testing task, which can intelligently generate the best relevant test cases based on the results of the computational experiments of the BERT model. This approach serves as a valuable tool for test engineers, enabling them to streamline test case formulations and enhance the efficiency of testing in complex systems.