基于自适应变异粒子群优化算法的测试数据生成

Test data generation based on adaptive mutation particle swarm algorithm

针对粒子群优化算法中群体易出现过早收敛的不足,提出了粒子群优化算法的改进算法AMPSO(adaptive mutation particle swarm optimization)算法并应用于测试数据生成中。引入约简粒子群优化算法,提高算法搜索速度;在算法进化过程中增加自适应调整策略,定义适应度评价阈值判断群体早熟现象,构建一个改进的自适应变异算子提高粒子变异率;通过实验确定阈值比例系数。结合实验结果从收敛代数和收敛时间两方面对比分析,证明了所提方法不仅能够防止算法出现过早收敛的问题,而且提高了测试数据生成效率。

In the meta-heuristic algorithms, the particle swal^n optimization algorithm may fall into a precocity easily and waste resources compared with other algorithms, this paper proposed an improved particle swarm optimization algorithm AMP- SO algorithm and applied it to the process of test data generation. It introduced the reduced particle swarm optimization algo- rithm based on particle swarm optimization algorithm, in evolutionary process of algorithm the adaptive strategy defined fitness evaluation threshold to judge the precocious phenomenon and built an improved self-adaptive mutation operator to increase the variation rate, then determined threshold scale factor through experiments. It used four typical procedures in the experiment. Through analyzing the experimental data in two aspects of convergence generations and convergence time, the resuhs demon- strate the effectiveness of the proposed method compared with existing methods.