NSGA-Ⅱ面向多目标测试数据模型的生成

Generation of NSGA-Ⅱ for Multi-objective Test Data Model

测试数据的自动化生成的实现是软件测试自动化的重要研究项目。当前很多研究人员使用多种方法实现测试数据的自动生成,但生成的测试数据一般仅实现最大覆盖率的测试标准。在测试数据生成问题上,希望生成的测试数据能够达到最大的覆盖率,同时也希望生成的测试数据集越小越好,可以降低执行时间,同时提高执行效率。文中从覆盖标准和内存消耗两个方面对测试数据进行评估,采用多目标优化算法NSGA-Ⅱ,实现同时满足最大分支覆盖率和最大内存分配的测试数据的自动生成。实验表明,NSGA-Ⅱ算法生成的测试数据比其他多目标优化算法能更好地满足两个目标。

The realization of the test data automatic generation is an important research project of software test. At present many research- ers use a variety of methods to realize the automatic generation of test data, but the generated test data are generally only to achieve maxi- mum coverage rate. In test data generation, hope that generated test data can achieve maximum coverage rate and generated test data set is as small as possible, therefore reducing the execution time and improving the execution efficiency. To evaluate testing data from two as- pects of coverage standard and memory consumption, adopt multi-objective optimization algorithm, NSGA- 11, to achieve automatic gen- eration of test data which satisfies the biggest branch coverage rate and the maximum memory allocation at the same time. The experimen- tal results show that the NSGA- II to generate test data is better to meet the two goals than the other multi-objective optimization algo- rithms.