摘要
为避免粒子群算法(PSO)早熟的缺点,设计了一种双种群进化粒子群算法(DE-PSO)。DE-PSO是基于PSO,引入选择、交叉及差分变异操作,并结合合理有效的粒子评价方法及越界处理方法之后形成的。将DE-PSO应用于两个地下水管理模型算例,第一个算例DE-PSO解的总抽水量分别比遗传算法(GA)、模拟退火算法(SA)和PSO减少了64、256、207m3/d,第二个算例DE-PSO解的总治理成本分别比GA、SA和PSO减少了57.74、151.93、76.59万元。两个算例中DE-PSO都表现出稳定的进化趋势,寻优效率好于GA、SA和PSO,可以有效求解地下水管理模型问题。
To solve the premature convergence problem of Particle Swarm Optimization （ PSO ）, a new algorithm named DE-PSO （Double-population Evolution-Particle Swarm Optimization ） was designed. DE-PSO introduced selection, crossover and differential mutation into PSO, and adopted a new evaluation method to evaluate swarms and a new control method to ensure all swarms can fly inside search space. DE-PSO was applied to solve two groundwater management cases. In the first case, DE-PSO produced a design with respectively 64, 256 and 207 m3/d less pumping rate than those of GA ,SA and PSO; in the second case, DE-PSO produced a design with respectively ￥ 577,400, ￥ 1,519,300 and ￥765,900 less remedial design cost than those of GA, SA and PSO. Two case studies indicated that DE-PSO could evolve steadily, the searching efficiency was better than GA, SA and PSO. DE-PSO could solve groundwater management model problems effectively.
出处
《地质学刊》
CAS
2012年第1期37-43,共7页
Journal of Geology
基金
国家自然科学基金项目(J0830522)资助
关键词
地下水管理模型
粒子群算法
双种群
差分变异
Groundwater management model
Particle swarm optimization
Double population
Diffecential variation
