Abstract: In evolutionary test, design of fitness function is a crucial work. Traditional work can not design a fitness function with favorable guidance to all node-node-oriented tests. A system model of evolutionary test with node-node-oriented method is presented, and an improved method of fitness function is designed. The method doesn’t consider the execute order of nodes and control flow, but is based on the independence of nodes. Experimental result indicates that this new method can generate test data with less cost and more steady performance for cover criterion which has no data dependence in discrete nodes.

Key words: evolutionary test, genetic algorithm, fitness function

摘要： 传统方法不能对进化测试中所有面向节点-节点的测试类型都构造具有良好导向的适值函数。针对该问题，基于面向节点-节点进化测试系统模型，不考虑节点的执行顺序和控制流关系，从节点的独立性出发，提出一种改进的适值函数计算方法。实验结果表明，对离散节点之间没有数据依赖关系的覆盖准则，该方法代价较小、运行稳定。

关键词: 进化测试, 遗传算法, 适值函数

CLC Number: 

• TP311.56