摘要
针对标准粒子群算法在处理非线性约束优化问题时存在收敛速度慢、精度低和易陷入局部最优的缺点,设计了一种新型混合粒子群算法,该算法采用可行性原则处理约束条件,避免惩罚函数法中惩罚因子选取的困难;引入基本复合形法产生初始可行群体,加快粒子群收敛速度;引入遗传算法的交叉和变异策略,避免粒子群陷入局部最优;在迭代末期的优解附近,进行改进复合形算法的寻优,提高最优解的精度.通过算法测试基准函数的优化计算,结果显示,新型混合粒子群算法有较好的优化性能,并在核动力设备优化设计中有很好的应用.
The standard particle swarm optimization has the shortcomings of slow convergence speed and poor accuracy of convergence,while easily falling into the local optimum when dealing with nonlinear constraint optimization problems.To overcome these difficulties,a new kind of hybrid particle swarm optimization algorithm was designed;it adopts the feasibility principle to handle constraint conditions,avoiding the difficulty of choosing a punishment factor when using the penalty function method.The basic complex algorithm was introduced to the hybrid particle swarm optimization algorithm to produce an initial feasible group,accelerating particle swarm convergence speed.The crossover and mutation strategy in a genetic algorithm was introduced to keep the particle swarm from falling into the local optimum.An improved complex algorithm was employed for obtaining better results when achieving iteration times in order to improve the accuracy of the optimal result.Testing the benchmark function through the optimization calculation shows that the new hybrid particle swarm optimization algorithm has better optimization performance,and it has been satisfactorily applied in the optimal design of nuclear power components.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2012年第4期534-538,共5页
Journal of Harbin Engineering University
关键词
粒子群算法
复合形算法
遗传算法
核动力设备
优化设计
particle swarm optimization
complex algorithm
genetic algorithm
nuclear power components
optimal design
