DOI:10.16644/j.cnki.cn33-1094/tp.2016.07.016
摘 要: 煙花算法是最近提出的一種群體智能算法,效率較高,但是仍然容易陷入局部最優(yōu)解。為進(jìn)一步提高算法的性能做了兩點(diǎn)改進(jìn):①采用混沌初始化的方式,有利于初始解遍布整個(gè)解空間;②當(dāng)全局最優(yōu)解陷入停滯時(shí),自動(dòng)啟動(dòng)高斯擾動(dòng)模塊對(duì)全局最優(yōu)解擾動(dòng),有利于算法跳出局部最優(yōu)解。在多個(gè)具有不同特性的測(cè)試函數(shù)上的實(shí)驗(yàn)表明,改進(jìn)算法的性能優(yōu)于原始煙花算法。
關(guān)鍵詞: 煙花算法; 群體智能; 優(yōu)化算法; 混沌
中圖分類(lèi)號(hào):TP301.6 文獻(xiàn)標(biāo)志碼:A 文章編號(hào):1006-8228(2016)07-56-03
Improved fireworks algorithm based on Chaos initialization and Gaussian perturbation
Du Zhenxin
(School of Computer Information Engineering, Hanshan Normal University, Chaozhou, Guangdong 521041, China)
Abstract: FA (fireworks algorithm) is a newly proposed swarm intelligence algorithm; it has a high efficiency, but is still easy to fall into the local optimal solution. To further improve the algorithm's performance, this paper has done the improvement in two aspects: ① using chaos initialization to facilitate the initial solutions distribution throughout the solution space; ② when the global optimal solution falls into a standstill, the Gaussian perturbation module is automatically activated to perturb the global optimal solution, and help FA escaping the local optimal solution. The experiments on several test functions with different characteristics show that the performance of the improved algorithm is better than that of the original fireworks algorithm.
Key words: fireworks algorithm; swarm intelligence; optimization algorithm; chaos
0 引言
煙花算法是由Tan和Zhu[1]提出的一種群體智能優(yōu)化算法,具有良好的優(yōu)化性能,逐漸引起國(guó)內(nèi)外關(guān)注[2-5],但是仍然容易早熟收斂。本文在原始煙花算法基礎(chǔ)上,采用混沌初始化操作和高斯擾動(dòng)操作,提高了算法的性能。
3 實(shí)驗(yàn)
為了測(cè)試改進(jìn)算法的性能,本文算法與原始煙花算法FA進(jìn)行了對(duì)比試驗(yàn)。測(cè)試函數(shù)與文獻(xiàn)[1]中相同,F(xiàn)A與本文改進(jìn)算法的參數(shù)設(shè)置與文獻(xiàn)[1]相同,本文新增加的參數(shù)為:最小進(jìn)化速度閾值θ=0.01,最大全局極值擾動(dòng)次數(shù)d=10。表1是對(duì)比測(cè)試結(jié)果,其中FA的數(shù)據(jù)來(lái)自文獻(xiàn)[1]。
從表1可以看出,本文的改進(jìn)算法在所有測(cè)試函數(shù)上的結(jié)果全部好于或等于原始煙花算法,驗(yàn)證了本文改進(jìn)算法的有效性。
4 結(jié)束語(yǔ)
本文在兩個(gè)方面對(duì)原始煙花算法進(jìn)行了改進(jìn):①采用混沌初始化煙花的初始解;②當(dāng)全局最優(yōu)解接近陷于停滯時(shí),自動(dòng)啟動(dòng)高斯擾動(dòng)模塊,對(duì)當(dāng)前全局最優(yōu)解進(jìn)行多次高斯擾動(dòng),直到得到的擾動(dòng)值好于當(dāng)前的全局最優(yōu)解或者多次擾動(dòng)失敗退出擾動(dòng)模塊。這樣有利于全局最優(yōu)解跳出局部最優(yōu)解,促進(jìn)算法的進(jìn)化。實(shí)驗(yàn)結(jié)果表明本文的改進(jìn)是有效的。
參考文獻(xiàn)(References):
[1] Tan Y,Zhu Y C.Fireworks Algorithms for Optimization[C]//
Proceedings of Int. Conf. on Swarm Intelligence (ICSI2010),Part II. Beijing, China:Springer-Verlag,2010:355-364
[2] 張家琴.求解0/1背包問(wèn)題的煙花算法研究[J].武漢工程職
業(yè)技術(shù)學(xué)院學(xué)報(bào),2011.23(3).
[3] 杜振鑫.煙花算法中爆炸半徑的改進(jìn)研究[J].計(jì)算機(jī)時(shí)代,
2013.1:28-29
[4] 譚營(yíng).煙花算法引論[M].科學(xué)出版社,2015.
[5] 王培崇,高文超,錢(qián)旭,茍海燕,汪慎文.應(yīng)用精英反向?qū)W習(xí)的
混合煙花爆炸優(yōu)化算法[J].計(jì)算機(jī)應(yīng)用,2014.34(10):2886-2890
[6] Zheng S,Andreas J and Tan Y.Enhanced Fireworks
Algorithm[C]//IEEE International Conference on Evolutionary Computation. Cancun, MEXICO: IEEE,2013:2069-2077
[7] Alatas B. Chaotic bee colony algorithms for global
numerical optimization[J].Expert Systems with Applications,2010.37(8):5682-5687