Test Data Generation based on Multiprocess Enhanced Multi-Population Genetic Algorithm

To improve the efficiency of test data generation, this paper proposes a method for test data generation based on a multi-process enhanced multi-population genetic algorithm(MPMGA). This method involves parallel execution of multiple populations by running multiple processes simultaneously, thereby accelerating the search process. In this approach, the problem of generating test data for multipaths is divided into several subproblems, and an independent population is created for each subproblem. What we have improved is that during the evolutionary process for test data generation, not only does each subpopulation operate in its own independent process, conducting its own evolutionary and selection operations, but also when a subpopulation fails to find a solution to its specific problem, it can migrate to other subproblems. Migrating subpopulations have tried to solve other subproblems. By utilizing Python’s multi-process approach, we can facilitate the exchange of information between processes and the sharing of evolutionary results. This enables efficient communication and collaboration among the different populations. The preliminary exploration results indicate that the proposed method achieves a high success rate in generating test data while maintaining low test costs. It is particularly effective in generating test data for difficult paths.