Abstract: In the light of the lack of effective methods to generating test case for multiple paths coverage based on com- plex system, we proposed a novel evolutionary generation approach of test case for multiple paths coverage. First, gener- is algorithm was improved; the ability of parent evolutionary selection, individual evolution and adaptive update method of populations were improved in the evolution of population, which cound solve the algorithm for early slow convergence or premature effectively. And then, according to the features and requirements of multiple paths coverage, fitness func- lion based on path-match was designed so that in one run, it was able to generate multiple test data to cover multiple target paths. Finally, the proposed approach was applied into several benchmark programs. The experimental results show that the proposed approach can improve the efficiency of test data generation effectively.

Key words: Complex system,Multiple paths coverage,Test case generating,Genetic algorithms,Fitness function