基于云模型的自适应并行模拟退火遗传算法  

Adaptive parallel simulated annealing genetic algorithms based on cloud models

作　　者：董丽丽[1] 龚光红[1] 李妮[1] 孙勇[1] 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191

出　　处：《北京航空航天大学学报》2011年第9期1132-1136,共5页Journal of Beijing University of Aeronautics and Astronautics

基　　金：国家自然科学基金资助项目(61004089);教育部博士点基金资助项目(20091102120013)

摘　　要：针对遗传算法收敛速度慢,容易"早熟"等缺点,提出了一种改进的遗传算法,即基于云模型的自适应并行模拟退火遗传算法(PCASAGA,Adaptive Parallel Simulated An-nealing Genetic Algorithms Based On Cloud Models).PCASAGA使用云模型实现交叉概率和变异概率的自适应调节;结合模拟退火避免遗传算法陷入局部最优;使用多种群优化机制实现算法的并行操作;使用英特尔推出的线程构造模块(TBB,Threading Building Blocks)并行技术,实现算法在多核计算机上的并行执行.理论分析和仿真结果表明:该算法比其他原有的或改进的遗传算法具有更快的收敛速度和更好的寻优结果,并且充分利用了当前计算机的多核资源.Due to the shortcomings of genetic algorithms such as the low convergence rate and premature convergence, an improved genetic algorithms was proposed, called adaptive parallel simulated annealing genetic algorithms based on cloud models （PCASAGA）. PCASAGA applied cloud models to the adaptive regulation of the crossover probability and mutation probability. Simulated annealing was combined to prevent genetic algorithms from local optimum. Multi-species optimization mechanism was used to realize algorithm parallel operation. Intel＇s threading building blocks （TBB） parallel technology was also used to realize algorithm parallel execution on multi-core computers. Theoretical analysis and simulation results verify that PCASAGA has better convergence speed and optimal results than original or improved genetic algorithms, and it takes full advantage of the current computers multi-core resources.

关 键 词：遗传算法 模拟退火 云模型 自适应 并行 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]