深圳大学学报理工版

针对粒子群优化(particle swarm optimization, PSO)算法无法在提高收敛速度的同时避免早熟的缺陷,提出基于正态分布衰减惯性权重粒子群优化(normal distribution decay inertial weight particle swarm optimization, NDPSO)算法.以正态分布曲线作为惯性权重的衰减策略曲线,通过引入控制因子对粒子的位置进行改善,使得NDPSO算法能很好的在优化过程中平衡全局搜索和局部搜索能力.使用8个标准函数测试分别对粒子群优化(particle swarm optimization, PSO)、线性权重衰减粒子群优化(linear decay inertial weight particle swarm optimization, LDWPSO)、指数权重衰减粒子群优化(exponential decay weight particle swarm optimization, EXPPSO)、收缩因子粒子群优化(constriction factor particle swarm optimization, CFPSO)、高斯分布衰减惯性权重粒子群优化(Gaussian decay inertial weight particle swarm optimization, GDIWPSO)、基于动态加速度系数的粒子群优化(particle swarm optimization based on dynamic acceleration coefficients, PSO-DAC)、性权重自适应粒子群优化(inertia weight adaptive particle swarm optimization, 简称PSO-LH)算法以及NDPSO算法进行仿真,分析他们的收敛速度和收敛精度.结果表明,NDPSO算法不管在单峰函数问题还是多峰函数问题上,总体性能都优于其他算法.

To avoid early maturity stagnation while increasing convergence speed of particle swarm optimization(PSO)algorithm, a normal distribution decay inertial weight particle swarm optimization(NDPSO)algorithm is proposed based on the inertia weight of which the decay strategy curve is normal distribution curve. By introducing the control factor to improve the position of the particle, the NDPSO algorithm can balance the global search and local search ability in the optimization process. The PSO, linear decay inertial weight particle swarm optimization(LDWPSO), exponential decay weight particle swarm optimization(EXPPSO), constriction factor particle swarm optimization(CFPSO), Gaussian decay inertial weight particle swarm optimization(GDIWPSO), particle swarm optimization based on dynamic acceleration coefficients(PSO-DAC), inertia weight adaptive particle swarm optimization(PSO-LH)and NDPSO algorithms are simulated using 8 standard function test evaluations, and their convergence speed and convergence precision are analyzed. The results show that the NDPSO algorithm outperforms other algorithms in terms of single-peak function or multi-peak function.

引言
1 粒子群优化算法及其改进算法
2 NDPSO算法
3 仿真实验
4 结语

图1 正态分布曲线图<br/>Fig.1 (Color online)Normal distribution picture

图1 正态分布曲线图
Fig.1 (Color online)Normal distribution picture

表1 八个标准测试函数<br/>Table 1 Eight benchmark test functions

表1 八个标准测试函数
Table 1 Eight benchmark test functions

图2 不同算法在8个测试函数中的收敛曲线图<br/>Fig.2 (Color online)Convergence curves of different algorithms in eight test functions

图2 不同算法在8个测试函数中的收敛曲线图
Fig.2 (Color online)Convergence curves of different algorithms in eight test functions

表2 不同算法在8个测试函数中的平均最优解1)<br/>Table 2 Average optimal solution of different algorithms in eight test functions1)

表2 不同算法在8个测试函数中的平均最优解1)
Table 2 Average optimal solution of different algorithms in eight test functions1)

[1] KENNDY J, EBERHART R C. Particle swarm optimization[C]// Proceedings of 4th IEEE International Conference on Neural Networks. Perth,Australia:IEEE, 1995: 1942-1948.
[2] GOLDBERG D E. Genetic algorithms in search optimization and machine learning[M]. Reading Mass, USA: Addison-Welsey, 1989: 95-99.
[3] JI Weidong, WANG Jianhua, ZHANG Jun. An improved real hybrid genetic algorithm[J]. Tehnicki Vjesnik-Tehnical Gazette, 2014, 21(5): 979-986.
[4] DORIGO, MANIEZZO M, COLORIN V. A ant system: optimization by a colony of cooperating agents[J]. IEEE Transactions on SMC, 1996, 26(1): 8-41.
[5] KARABOGA N. A new design method based on artificial bee colony algorithm for digital IIR filters[J]. Journal of the Franklin Institute-engineering and Applied Mathematics, 2009, 346(4):328-348.
[6] PANG Hui, LIU Fan, XU Zeren. Variable universe fuzzy control for vehicle semi-active suspension system with MR damper combining fuzzy neural network and particle swarm optimization[J]. Neurocomputing, 2018, 306:10-140.
[7] KENNEDY J. Small worlds and mega-minds: effects of neighborhood topology on particle swarm performance[C]// Proceedings of IEEE Congress on Evolutionary Computation. Piscataway, USA: IEEE, 1999: 1931-1938.
[8] KENNEDY J, Mendes R. Population structure and particle swarm performance[C]// Proceedings of IEEE Congress on Computational Intelligence. Hawaii, USA: IEEE, 2002: 1671-1676.
[9] VEDTERDTROM J S, RIGET J, KRINK T. Division of labor in particle swarm optimization[C]// Proceedings of the Congress on Evolutionary Computation. Honolulu, USA: IEEE, 2002: 1570-1575.
[10] BERGH F V D, ENGELBRECHT A P. A cooperative approach to particle swarm optimization[C]// Proceedings of IEEE Transactions on Evolutionary Computation. Portland, USA: IEEE, 2004: 225-239.
[11] DENG Xianli, WEI Bo, ZENG Hui, et al. A multi-population based self-adaptive migration PSO[J]. Acta Electronica Sinica, 2018, 46(8): 1858-1865.
[12] SHI Y, EBERHART R C. A modified particle swarm optimizer[C]// Proceedings of IEEE Congress on Evolutionary Computation. Piscataway, USA: IEEE, 1998: 69-73.
[13] YAN Chunman, LU Genyuan, LIU Yingting. et al. A modified PSO algorithm with exponential decay weight[C]// Proceedings of the 13th International Conference on Natural Computation, Fuzzy Systems and Knowledge Discovery. Changsha, China: IEEE, 2018: 239-242.
[14] CLERC M, KENNEDY J. The particle swarm-explosion, stability and convergence in a multidimensional complex space[J]. Proceedings of IEEE Transactions on Evolutionary Computation, 2002, 6(1): 58-73.
[15] ZHANG Xun, WANG Ping. Particle swarm optimization algorithm based on Gaussian function decreasing inertia weight[J]. Application Research of Computers, 2012, 29(10): 3710-3724.
[16] ZHOU Lingyun, DING Lingxin, PENG Hu. et al. Neighborhood centroid opposition-based particle swarm optimization[J]. Acta Electroica Snicia, 2017, 45(11): 2815-2823.
[17] SONG Huajun, LIU Feng, CHEN Haihua. et al. A stochastic maximum likelihood algorithm based on improved PSO[J]. Acta Electroica Sinica, 2017, 45(8): 1990-1994.
[18] 滕志军,吕金玲,郭力文,等.基于动态加速因子的粒子群优化算法研究[J].微电子学与计算机,2017,34(12):125-129.

备注

引言

1 粒子群优化算法及其改进算法

2 NDPSO算法

3 仿真实验

4 结语

补充材料文件请点击下载

期刊信息

备注

引言