Research On Optimization And Application Of Particle Swarm Optimization In Software Engineering

Software testing is the key part in software engineering which impacts directly on the development prospects of software engineering.With the expansion of the scale of software at present, the traditional software testing method can’t meet the actual demand.Automated software test has become the focus of current research.In order to implement automated test,automated generation of test cases is the key.After the study of the researchers for many years,there are already certain achievements obtained in this field, but there is still a gap with the actual requirements,so the research to automated generation of test cases for the development of software test and even software engineering has certain scientific significance.This paper aims at the detailed research for the improved particle swarm optimization algorithm which is applied to automated generation of test cases.Firstly,the basic theory of the software test and the technology of test case generation are introduced,and some of the software testing tools which are widely used are introduced,then the basic principle of Particle Swarm Optimization algorithm(PSO) is introduced.Finally The convergence of the algorithm and some typical improved algorithm are analyzed.Based on this, a kind of hybrid particle swarm optimization algorithm with cooperation of multiple particle roles(MPRPSO) was proposed.The concept of particle roles was introduced into the algorithm to divide the population into three roles:exploring particle(EP),patrolling particle(PP) and local exploiting particle(LEP).In each iteration,EP was used to search the solution space by the standard particle swarm optimization algorithm,then PP which was based on chaos was made use of to strengthen the global search capability,and to replace some EPs to restore population vitality when the algorithm trapped in local optimum,finally,LEP was used to strengthen the local search to accelerate convergence by unidimensional asynchronous neighborhood search.The experimental result shows that new algorithm is better on the optimal performance with certain robustness.Then, this paper researches some key technologies of the software automated generation model of test case,and the technology of program instrumentation and branch function superposition methods are used.In order to optimize MPRPSO algorithm in the application of the automated generation of test cases, the concept of particle movement time is introduced in MPRPSO algorithm.That is,the EP in the iteration will be updated according to the testing principle, and will go forward more distance in the feasible direction which conforms to the requirements to speed up the convergence in automated generation of test case,on the basis of this a single-path automated generation model of test case which is based on improved MPRPSO algorithm is proposed. The example simulation experiment shows that improved MPRPSO algorithm generates test cases a little more quickly, also can realize the coverage of specified tests path better.Then, the concept of weight array is introduced in the generation problem of multi-path set of test cases.That is,once a test case is output,the weight of the path in this test case will decrease self-adaptively.The algorithm will be more likely to search the paths with bigger weight in the next iteration which enables the algorithm to dynamicallychange the search target path,and to output set of test cases with larger coverage.The simulation results indicate that the improved MPRPSO algorithm can complete the multi-path automated generation of test cases better, and has a certain effect.