基于差分进化与优胜劣汰策略的灰狼优化算法  

Gray wolf optimization algorithm based on differential evolution and survival of fitness strategy

作　　者：朱海波 张勇[1] Zhu Haibo;Zhang Yong(School of Electronics and Information Engineering,University of Science and Technology Liaoning,Anshan 114051,China)

机构地区：[1]辽宁科技大学电子与信息工程学院,辽宁鞍山114051

出　　处：《南京理工大学学报》2018年第6期678-686,共9页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(61473054)

摘　　要：为了改善灰狼优化算法收敛速度慢、寻优精度低、易早熟等缺陷,提出1种改进的灰狼优化算法。在基本灰狼优化算法的基础上,引入差分进化机制生成1个变异种群,通过其动态缩放因子和交叉概率因子避免算法陷入局部最优。引入优胜劣汰的生物竞争淘汰策略,根据比较进化变异后狼群个体适应度值淘汰m只狼,同时随机生成与被淘汰狼数量相同的狼。采用典型的单峰与多峰函数对该文算法进行测试。仿真结果表明,该文算法的综合性能优于粒子群优化(PSO)和人工蜂群(ABC)等其他对比算法,提高了局部搜索的效率和精度。将该文算法应用于冷凝器实际控制参数整定优化问题中,并与遗传算法(GA)、PSO和工程整定(ZN)法进行比较。仿真结果表明,该文算法整定的参数输出响应的调整时间和上升时间减小,最大超调量降低且稳定性好。An improved gray wolf optimization algorithm is proposed for the slow convergence speed,low optimization accuracy and premature convergence of the gray wolf optimization algorithm.Based on the basic gray wolf optimization algorithm,a population of individual variation is generated by the differential evolution mechanism,and the local optimization is avoided by a dynamic scaling factor and a crossover probability factor.After variation,m wolfs are eliminated by comparing the individual fitness value,and the same number of wolves are randomly generated based on survival of the fitness.This algorithm is tested by typical unimodal or multimodal benchmarks functions.The simulation results show that the optimization performance of this algorithm is better than that of the particle swarm optimization(PSO)and artificial bee colony(ABC)algorithm etc.,and the efficiency and accuracy of local search are improved.This algorithm is applied to the optimization of the actual control parameters of a condenser and compared with the PSO,genetic algorithm(GA)and engineering turning(ZN)methods.The simulation results show that the adjustment time and rise time of the output response for the adjusted parameter are reduced,the maximum overshoot is reduced,and the stability is good.

关 键 词：差分进化 优胜劣汰策略 灰狼优化算法 典型单多峰函数 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]