Software Test Cases Generation Based On Improved Particle Swarm Optimization

With the widely use of computer software in industries, the software system is getting so large that the test of software becomes very hard, but the software system development cycle becomes shortest. The software testing is an important part of software quality assurance, and occupies nearly half of the the software development time.Therefore, improving the efficiency of software testing is the focus of software development, and automatic software testing emerges as the times and fast develops.Automatic software test case generation method is the key of software testing, which can simplify the software testing process, reduce the cost of software testing and improve the quality of software testing. Recently many researchers have conducted a lot of research about the software test case generation technology, and achieved certain results.Particle swarm optimization (PSO) has a few parameters and fast convergence and easy to implement, so it has been used for test case generation technology. However, particle swarm optimization algorithm easily to a local optimal for the diversity of the population reduced, and test case generation algorithm is mostly calculate one by one and then take the union of each path, so the process is repeated and the calculated resources are not reuse. There is the evaluation of all paths through the construction of fitness function for multiple paths coverage, although the efficiency is improved, but the multi path function structure is very complex. This paper proposes a heuristic method of test case generation based on improved particle swarm optimization algorithm, which joined the self group of active feedback mechanism (SAF) and Gauss mutation operator (G) that can adjust the diversity of the population, to prevent falling into the local optimal. Then, the improved algorithm is applied to the software test case generation algorithm, which uses one path coverage branch function superposition method to construct the fitness function and introduces the multi-path parallel test thought. It can realize the computing resources reuse and multi path test case generation. The experiment results and analysis of improved particle swarm optimization algorithm show that test case generation algorithm is very efficient and less generation iterations.Finally, we have developed a tool based on SAF-GPSO test case generation algorithm, and checked it through the triangle judging procedures. The tool realizes the method in this thesis by path cover test cases generating and all-path coverage test case generation function.