Study On The Automatic Test Data Generation Based On Genetic Algorithm And Particle Swarm Optimization Algorithm

Software testing is an important means to ensure software reliability. It plays avery important role in the software development cycle. Test data generation is the keyto realize the automation of software testing. This technology can greatly reduce thetime and cost for software development, so the search for effective test data automaticgeneration method is still a problem to be solved.Particle swarm algorithm and genetic algorithm as optimization algorithms,through simulating individual adaptability, use transformation rules constructed tosearch the optimal solution in the solution space. They are very good methods forautomatic test data generation. At present, only the single algorithm and its operationsare often not effective development and detection algorithm property. According tothis problem, this paper combines with two algorithms in the case of improving thealgorithm’s performance, so as to enhance the performance of the algorithm andimprove the efficiency of automatic test dataIn this paper, the main work is as follows:(1) Through analysis of genetic algorithm, selection operator in genetic algorithmis introduced into the optimal preservation strategy and has an improvement.automatictest data generation tool of the dynamic variable parameter is established. The tool notonly can dynamically input parameters of genetic algorithm through visualizationinterface, but also can input corresponding fitness function according to the differentpath selection, overcomeing the defect of modifying the source code for fitnessfunction(2) Through analysis of particle swarm algorithm, the particle velocity items ofevolution equations are reduced. Algorithm completes the evolution process onlythrough the particle position update. Algorithm combines with the fitness and particleaggregation degree to formulate adaptive adjustment strategy. Algorithm adoptesdifferent methods to set inertia weight to make the particle adaptive change inertiaweight dynamicly. Thereby this can balance global exploration and local improvementand improve the convergence velocity and precision of the algorithm.(3) This paper designs a hybrid optimization algorithm of genetic algorithm andparticle swarm algorithm, and applies to automatic test data generation for path. Thispaper takes improved particle swarm algorithm as an important operator of geneticalgorithm, so as to achieve the optimal solution according to improved genetic algorithm. This paper validates the effectiveness of algorithm through the5benchmark programs and4industrial programs.