Research On Test Case Prioritization Based On Multi-Objective Particle Swarm Optimization

With the acceleration of social information,computer software plays a more and more important role in all walks of life.More and more attention is paid to the quality of software.Software testing runs through the whole life cycle of software.It is an important method to ensure software quality.However,due to changes of customer requirements,expansion of software scale and software update frequently,the scale of regression test case set becomes larger and larger which increases the workload of regression testing.Restricted by manpower,time and other regression test cost,the traditional regression testing method is difficult to meet the current regression test task.Therefore,how to improve the efficiency of regression testing has become the concern of researchers and has very important research value and practical significance.Researchers have proposed many techniques to improve the efficiency of regression testing which mainly includes test case selection,test case simplification and test case prioritization.The first two methods will modify or delete test case,these irreversible operations can miss test cases that can detect errors.This paper mainly focuses on the test case prioritization Technology.Based on the analysis of the current situation of test case prioritization,we proposed a multi-objective test case prioritization framework based on the particle swarm optimization algorithm.How to overcome the disadvantages of particle swarm optimization algorithm is also studied.The previous multi-objective test case prioritization is mostly based on the method of weighting sum of multiple optimization objectives.The weight allocation of the method is greatly influenced by subjective thought which is not intelligent enough.Thesearch method based on evolutionary algorithm is feasible.However,the crossover and mutation mechanism of evolutionary algorithm is complex so that the efficiency will decline in the event of large-scale problems.In this paper,a test case prioritization framework based on multi-objective particle swarm optimization is designed and APSC and EET are chosen as optimization targets.Aiming at the specific situation of multi-objective test case prioritization,the particle encoding method is designed.Owing to the standard particle swarm optimization is based on continuous space to search and the binary discrete particle swarm optimization is more complex involves of encoding style and mapping mechanisms which are not suitable for multi-objective test cases prioritization.In this paper,a method of searching in discrete space is proposed.In non dominated sorting,in order to reduce time complexity of the algorithm,only solutions with the dominant number of 0 is chosen.The particle swarm optimization has many advantages,but its disadvantage is that it is easy to fall into local optimum.In this paper,we propose a global optimal solution dynamic adjustment algorithm based on Additional strategy,which can lead the particle swarm flight to the direction of excellent solution.Additional strategy is a typical greedy algorithm,which is characterized by feedback mechanism.Before updating the global optimal solution,using Additional strategy to compare the advantages and disadvantages of individual extreme value and global extreme value.That is to say,the statement covered by test cases that have been sorted is removed and then the statement coverage rate of the individual extreme value and global extreme value is calculated.Compare advantages and disadvantages of the current solution and the global optimal solution combine execution time,choose the better solution as new global optimal solution.By updating the global optimal solution dynamically,the search direction of the particle swarm is guided so as to avoid the particle swarm falling into local optimum.Finally,programs in the SIR library are selected to be used for contrast experiment.By comparing the distribution of Pareto optimal solution set and the APFD value of test case prioritization,it is proved that the proposed method can obtain better Pareto optimal solution set and higher error detection rate.