论文部分内容阅读
针对基本粒子群算法易陷入局部极值呈现早熟性、收敛速度慢、精度低等问题,提出一种约减的自适应粒子群优化算法(RAPSO)的软件结构测试数据自动生成方法.对基本粒子群进化方程进行约减,提出基于惯性权重的自适应调整方案,将惯性权重直接作用于粒子的位置,以分支函数叠加法作为适应值函数.RAPSO去掉了PSO进化方程的粒子速度项而使原来的二阶微分方程简化为一阶微分方程,仅由粒子位置控制进化过程.针对三角形判定程序的结构测试数据自动生成进行实验.实验结果表明,该方法可以更高效地自动生成测试数据.
In order to solve the problem that basic particle swarm optimization (PSO) is apt to fall into the local extremum, the premature convergence, the slow convergence speed and the low precision, a reduced generation PSO (Automatic Particle Swarm Optimization) software structure test data automatic generation method is proposed. The evolutionary equation is reduced to some extent, and an adaptive adjustment scheme based on inertia weight is proposed, in which the inertia weight is directly applied to the particle position and the branch function superposition method is used as the fitness function.RAPSO removes the particle velocity term of PSO evolutionary equation, The second order differential equation is simplified as a first order differential equation and the particle position is used to control the evolutionary process. The structure test data of triangulation program is automatically generated and tested. The experimental results show that this method can automatically generate test data more efficiently.