Adaptive Differential Evolution Algorithm And Its Application To Automatic Generation Of Software Test Cases

Differential Evolution has become one of the most popular random search algorithms because of its simple structure,rapid convergence,strong robustness and easy implementation,and has been successfully used to solve various application problems.The performance of differential evolution algorithm depends on the use of parameter control and mutation strategy.Reasonable design of parameter control method and mutation strategy can effectively improve the performance of differential evolution algorithm.In this thesis,the parameter control method,mutation strategy and application of differential evolution algorithm in software testing are studied.The main contents and innovations of this thesis include:First of all,a differential evolution with multi-factor ranking based parameter adaptation is proposed.The proposed algorithm includes two mechanisms(parameter storage mechanism and parameter allocation mechanism).The parameter storage mechanism according to the ranking of fitness value,fitness difference,and parameter size in the evolution process to select the optimal parameter storage,and this mechanism can guide the next generation of the population to generate reasonable parameters.The parameter allocation mechanism adopts a hierarchical strategy,individuals are sorted and stratified according to fitness.The purpose of this mechanism is to find appropriate parameters according to the characteristics of individuals at different levels,which effectively improves the local search ability of the algorithm.In this thesis,the proposed algorithm is tested on the benchmark function,and the results show that the two mechanisms are effective.The parameter storage and allocation mechanism proposed in this thesis can effectively improve the performance of differential evolution algorithm.Secondly,a differential evolution algorithm for adaptive selection of mutation strategy pools is proposed.This algorithm proposes an improved mutation strategy and uses two different strategy pools.The proposed mutation strategy is based on the improvement of DE/current-to-rand/1,which uses poor individuals and good individuals to add a certain direction.It could enhance the local search ability of the population.At the same time,this algorithm adopts two kinds of strategy pool,the individual chooses the strategy pool suitable for the current stage according to the success probability of the previous generation so that the population adapts to select the mutation strategy and balances the exploration and detection ability.The proposed algorithm is also tested on the benchmark function,and the results show that the algorithm is competitive with other algorithms.Finally,the adaptive multi-strategy differential evolution algorithm proposed in this thesis is also applied to the automatic generation of software test cases,and compared with the traditional algorithm,the experiment shows that the optimized algorithm can improve the efficiency and performance of test case generation.