约束函数估值策略辅助的微粒群算法

A Constraint Approximation Assisted PSO for Computationally Expensive Constrained Problems

针对约束函数计算费时的优化问题,提出了一种基于分类器的预测微粒群算法。通过构造一个分类器对种群个体进行约束条件满足与否的估计判断,从而减少约束函数的计算时间,缩短整个优化时间。在13个标准测试函数上的测试结果表明,本文提出的方法可以大大减少约束函数的实际计算次数。

A classifier-based particle swarm optimization is proposed for computationally expensive constrained problems in this paper.A classifier can be constructed in order to judge whether the individual is satisfied with the constraints or not,which will help reduce the computation time of constrained function,and decrease the total optimization time.The experimental results on the 13 standard test functions show that the proposed method can greatly decrease the number of calculation for constraint functions.