基于模式搜索的类电磁算法求解约束优化问题

针对约束优化问题,提出了一种基于模式搜索的类电磁算法。引入了粒子的违反度函数,将约束优化问题转化为双目标无约束优化模型来求解;提出了双目标模型中粒子的电荷和受力的计算公式,引导不可行粒子转化为满足约束条件的粒子;为了提高算法的搜索能力,结合模式搜索算法改进种群中的粒子,为类电磁算法提供了有效的局部信息。与以往算法仿真结果相比,新算法具有性能好、较稳定的优点。

Electromagnetism-like method based on pattern search for constrained optimization problem

An electromagnetism-like method(EM) is proposed for solving constrained optimizations on the basis of pattern search(PS).The violation degree function is introduced,and the new technique for constraints handling is adopted to transform the constrained optimization problem into a bi-objective unconstrained optimization model.The computational equations of the charge and force exerted on the particles are presented for the bi-objective model,which will lead the infeasible particles to transform into feasible ones.In order to enhance the exploratory ability,the pattern search is incorporated to improve the particle which provides effective local information for the EM.Compared with the simulation results of the existing algorithms,the proposed algorithm has the advantages of good performance and favorable stability.