基于改进遗传算法的嵌入式软件时序测试数据自动生成方法

Automatic generation method of embedded software timing test data based on improved genetic algorithm

为了提高嵌入式软件时序测试数据自动检测和生成的精度、加快其收敛性,提出基于改进遗传算法的嵌入式软件时序测试数据自动生成方法。构建嵌入式软件时序测试数据的结构参数分析模型,采用动态编译和输出稳定性测试的方法,进行嵌入式软件时序测试数据检测和跟踪识别,通过混合属性数据特征和数据层次聚类分析的方法,建立嵌入式软件时序测试数据的遗传进化学习模型,得到静态软件源文件用例集,通过嵌入式软件时序测试数据知识图谱的分类和层次性架构的方法,建立嵌入式软件时序测试数据的特征图谱,采用改进的自适应遗传算法,提高选中最优解的概率;构建嵌入式软件时序测试数据的特征分布集,结合概率密度特征分布,实现对嵌入式软件时序测试数据自动生成。测试表明,该方法进行嵌入式软件时序测试数据生成的数据表达能力较好,实现数据快速建模和特征分析,提高软件测试能力。

In order to improve the accuracy of automatic detection and generation of time-series test data of embedded software and speed up its convergence, an automatic generation method of time-series test data of embedded software based on improved genetic algorithm is proposed. Build the structural parameter analysis model of embedded software time series test data, use the methods of dynamic compilation and output stability test to detect and track the embedded software time series test data, establish the genetic evolutionary learning model of embedded software time series test data through the methods of mixed attribute data characteristics and data hierarchical clustering analysis. Through the classification and hierarchical structure of the knowledge map of embedded software time series test data, the feature map of embedded software time series test data is established, and the improved adaptive genetic algorithm is adopted to improve the probability of selecting the optimal solution. The feature distribution set of embedded software timing test data is constructed, and the automatic generation of embedded software timing test data is realized by combining the probability density feature distribution. The test results show that the method has good data expression ability for time series test data generation of embedded software, realizes rapid data modeling and feature analysis, and improves software testing ability.