Theory And Method Of Automatically Generating Test Data Based On Adaptive Genetic Algorithm

With the increasing functionality of software,software becomes more and more complex,so the efficiency of test data generation becomes very low in software testing.In order to improve the efficiency of test data generation,targeted theories and methods need to be adopted.At present,the test data generation method based on genetic algorithm has been widely studied.However,due to the particularity of test data generation problems,the efficiency of generating test data using traditional genetic algorithms is still very low.Therefore,finding a more effective method for test data generation has always been a very urgent issue.In view of this,this paper separately focuses on path coverage testing and mutation testing to study how to improve the performance of the algorithm to improve the efficiency of test data generation.Firstly,based on path coverage testing,a method of test data generation based on individual kernel density adaptive genetic algorithm is proposed.Focusing on the connection between individuals in the search space,the probability of selection,crossover and mutation in genetic operations are dynamically changed by calculating the kernel density between individuals,thereby accelerating the convergence speed of the algorithm,and applying the improved algorithm to test data generation to improve the efficiency of test data generation.Secondly,based on mutation testing,a method of test data generation based on mutation feature collecting for adaptive genetic algorithm is proposed.Firstly,the mutation branches are generated according to the weak mutation testing criteria and inserted into the source programs;then,according to the clustering principle of mutant similarity analysis,the mutants are classified;then,based on the weak mutation test criteria,the mathematical model of mutation test data generation problem is established;then,using an adaptive genetic algorithm to solve the established model.In the process of solving,collecting the execution time of the killed mutants,and using this information as the basis for selecting More effective test data.Finally,based on the comparison experiment,the effectiveness of the method in test data generation was verified.The research method proposed in this paper effectively improves the efficiency of test data generation and further enriches the theory of software testing.Therefore,it has important theoretical significance and application values.