基于SIMILARITY的演化路径测试数据生成方法研究

Study on SIMILARITY-based evolutionary method for path test data generation

针对基于SIMILARITY的演化路径测试数据生成方法中的问题,通过大量试验确定了遗传算法的基本参数后,在编码方式、选择方式和交叉方式的8种参数组合以及24~45 bit的8种串长情形下,利用三角形分类程序对该方法生成测试数据的效果进行了研究.结果表明,最优组合也与原结果有一个数量级的差距.

To overcome the problem confronted in the research of an evolutionary test data generation method for path testing,an experimental studying on the evolutionary method using a triangle classifier had been implemented in this paper.The elementary parameters for genetic algorithm were determined by a lot of experiments,8 combinations of encoding,selection and crossover operations,8 species of string length（24~45 bit） were also been investigated.The results showed that there was a magnitude of gap between the optimal combination and the original results.