论文部分内容阅读
在面向对象软件集成测试中,类的测试顺序同系统的测试桩复杂度密切相关,但由于描述类间依赖的关系复杂,衡量测试代价的目标多样,测试序列难以确定.鉴于在软件测试领域,智能优化算法的应用已获得良好的效果,因此可以将智能优化算法融入类集成测试序列生成问题,形成多目标优化类集成测试序列.对已有基于智能优化算法的多目标类集成测试序列生成技术进行比较研究.对已有类集成测试序列生成技术进行分类,包括典型的基于线性加权的多目标优化算法和基于帕累托模型的多目标优化算法,并概括描述相关研究进展;通过Markov过程、优化理论等对应用于多目标类集成测试序列中的智能优化算法进行理论分析,重点分析其全局收敛性及优缺点;对各算法的比较实验结果表明基于粒子群算法和帕累托最优模型的多目标优化算法均可以生成较优的类集成测试序列.“,”Class integration test order (CITO) is closely related to stubbing complexity for programs in object-oriented software testing.However,complicated dependent relationships between classes and diverse interdependent objectives measuring stubbing cost cause the difficulty to determine test order.Considering the fact that intelligent optimization algorithms have been successively applied in other fields of software testing,approaches based on multi-objective intelligent optimization in CITO have been proposed.This paper aims at comparing these different approaches.First,it categorized approaches in CITO and discussed two typical multi-objective optimization algorithms based on linear weighting method and Pareto model,as well as the advances of CITO.Then,it analyzed these multi-objective intelligent optimization algorithms by the theory of Markov process and theoretical optimization,especially compared their global convergence,advantages and disadvantages in generating test order.Finally,the results of comparative experiments show that the class integration test order determination methods based on Particle Swarm Optimization (PSO) and Pareto model can generate optimum solution.