Test Case Generation Of Path Coverage Based On Genetic Algorithm

Nuclear power design software involves intensive computing tasks, required high precision and complexity. So, to ensure the high quality continuation of the nuclear power design software, reducing software error in the development, it needs to widespread adopt software testing to ensure nuclear power design software has high quality. The usual software testing technology mainly rely on artificially-design, it has requirements for higher technology, and it will inevitably make mistakes; and due to the complexity of the logic and the data structure of nuclear power design software, leading to the input data set many complex structure, nested, huge amount of data, so lead to low efficiency. During the work of nuclear power design software, we mainly adopt the basic path testing method, but it is complex to generate the basic path, and it may exist unreachable paths and so on. As a result it will increase blindness of test case generation, and reduce the efficiency of test case generation. For the above, this article, from the whole process of test cases generation, propose a improved technology of test case generation, it mainly to study three aspects:1) The basic path generation. The technology transform program control flow graph to tree graph by combining the relational expression of nodes and edges, to generate basis path set with character that the length of each path is the smallest, it can simplify the process. Path reuse can provide convenience for the later work that make the infeasible path to feasible path.2) The accessibility of path detection. The paper propose an approach that record the define-use data information of program control flow graph, and record the function summary of call site, then update its call context data flow information, finally judge the conflict branch of program conditional statement to detecting the paths accessibility, reduce the blindness of testing, lay a good foundation for the effective test cases.3) The test case data automatically generation. Using Exponential transform the original fitness function and “information sharing and save” strategy, and keeping the “communication” and save the effective information in the process that is the better fitness value of the individual information, to guide the evolution. The method improves the probability to get the optimal solutionBased on the classic triangle program compare, verify the correctness and effectiveness of this technology. The technology is provided by this paper is successful applied to nuclear power calculation software testing process, it can reduce the work for software testers, improving the path coverage of software testing. Enriches the academic system of the test case data automatically generated, has a certain practical value and theoretical significance.