Abstract: To deal with the inefficiency generation test data existing in Basic Particle Swarm Optimizer(BPSO) for multi-path to the target branch node, automatic generation method of test data for software structure is proposed based on Multiple-population PSO(MPPSO) algorithm, which includes modified individual migration strategy and the restrictive function of predicate redefined, and uses the search method of parallelism. Experimental results show that the iterative number and execution time of MPPSO algorithm is respectively 1/7 and 1/3 in contrast to BPSO, and it has obvious advantages in the efficiency of the execution and search.

Key words: software structure test, Multiple-population Particle Swarm Optimizer(MPPSO), migration, predicate, restrictive function, multi-path

摘要： 软件测试中到达待覆盖的目标分支为多条独立路径时，应用基本粒子群优化算法生成测试数据的效率较低。为此，提出多种群粒子群优化算法的软件结构测试数据自动生成方法，包括改进个体迁移策略、重新定义谓词约束函数与采用并行性的搜索方法。实验结果表明，该方法的迭代数和执行时间分别为基本粒子群算法的1/7和1/3，在执行和搜索效率上有着明显的优势。

关键词: 软件结构测试, 多种群粒子群优化, 迁移, 谓词, 约束函数, 多路径

CLC Number: 

• TP311