Automatic Generation Of Test Data Based On Heuristic Search Algorithm

In recent years, as the complexity of the software gradually increased and thescale became huge, software testing is more and more important. The limitation ofmanual testing is broken by automation software test technology, so it enhances theefficiency of generating test data. This paper discusses about heuristic algorithm forautomatic generation of software test data, and takes the triangle problem as anexample to verify that heuristic algorithm has a higher efficiency in the automatic testdata generation.First, this paper introduces automatic testing technology and focuses on automatictest data generation-path oriented. Then, it elaborates heuristic algorithms in detail.Heuristic algorithms, including genetic algorithms, simulated annealing algorithm, antcolony algorithm and Tabu search algorithm is a set of proposed rules guiding thesearch direction. This paper focuses on genetic and simulated annealing algorithm.Genetic algorithms abstracted away from the biological evolutionary process, basedon the sense of probability, random iterative evolution. The genetic algorithm is aglobal optimization search mode, with a wide range of practical. However, the geneticalgorithm has its own shortcomings, such as lack of local search ability, slowconvergence and stability is also poor. Simulated annealing algorithm is an extensionof the local search. In view of the detection of genetic algorithm, this paper combinessimulated annealing algorithm with genetic algorithm to avoid the prematurephenomenon.Subsequently, this paper makes detailed analysis about how to use geneticannealing algorithm for automatic test, found the application model, and build a testenvironment. And then encode the chromosome, improve the fitness functionstructure. Then it proposes cartridge-based method on data generation automaticallyabout path test. Finally, take a typical example to achieve the model, and the resultswere analyzed to verify the superiority of the genetic annealing algorithm on thegeneration of test data automatically.