Research On Automatic Generation Of Test Data Based On Hierarchic Genetic Algorithm

With the expansion of software projects and the development of software technology,software testing has an increasingly important role in the early and late stage of softwaredevelopment. In software testing, the selection of test data directly impact if the errors ofsoftware can be detected as expected, so the selection of test data plays an important role.The generation of test data based on genetic algorithm had been widely studied. But thegenetic algorithm has slow convergence, local convergence and other issues.This Paper researched and analysed on the test data automated generation methodbased on base genetic algorithm. To solve the slow convergence, local convergence andother issues, this paper proposed a method that is generation test data based on hierarchicalgenetic algorithm. This method uses Hierarchical genetic algorithm. Hierarchical geneticalgorithm improves base genetic algorithm by dividing the initial population to layer sonpopulation. It could solve the base genetic algorithms’s issue. At the same time, thismethod’s fitness function was designed using Huffman encoding. That also can improve theefficiency of generation test data. Above all, this method can solve the the slowconvergence, local convergence issue in base genetic algorithm and improve the efficiencyof generation of test data.After proposing above method, this paper set up the test tool model based onhierarchical genetic algorithm. The test tool model includes path analyzer, predicateinstrumentationer, test data generator. Then this paper developed this tool. At last, this paerdid experiments using the basic program and the automation test project based on Seleniumto verify the advantage of this tool. The results show that the hierarchical genetic algorithmto generate optimal test of speed and quality compared to the basic genetic algorithm hasgreatly improved, more suitable for the automatic generation of test cases.