資源描述:
《基于動(dòng)態(tài)評(píng)價(jià)選擇策略的改進(jìn)人工蜂群算法.pdf》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫(kù)。
1、JournalofComputerApplicationsISSN1001—90812015.07..10計(jì)算機(jī)應(yīng)用,2015,35(7):1969—1974CODENJYIIDUhttp://www.joca.ca文章編號(hào):1001—9081(2015)07—1969-06doi:10.11772/j.issn.1001-9081.2015.07.1969基于動(dòng)態(tài)評(píng)價(jià)選擇策略的改進(jìn)人工蜂群算法徐向平,魯海燕’,程畢蕓(江南大學(xué)理學(xué)院,江蘇無(wú)錫214122)(}通信作者電子郵箱luhaiyan@jiangnan,ed
2、u,e/1)摘要:針對(duì)標(biāo)準(zhǔn)人工蜂群(ABC)算法易陷入局部極值的問(wèn)題,對(duì)標(biāo)準(zhǔn)ABC算法的輪盤(pán)賭選擇機(jī)制進(jìn)行了修改,提出了一種基于動(dòng)態(tài)評(píng)價(jià)選擇策略的改進(jìn)人工蜂群(DSABC)算法。首先,根據(jù)到當(dāng)前為止一定迭代次數(shù)內(nèi)蜜源位置的連續(xù)更新或停滯次數(shù),對(duì)每個(gè)蜜源位置進(jìn)行動(dòng)態(tài)評(píng)價(jià);然后,利用所得的評(píng)價(jià)函數(shù)值為蜜源招募跟隨蜂。在6個(gè)經(jīng)典測(cè)試函數(shù)上的實(shí)驗(yàn)結(jié)果表明:與標(biāo)準(zhǔn)ABC算法相比,動(dòng)態(tài)評(píng)價(jià)選擇策略改進(jìn)了標(biāo)準(zhǔn)ABC算法的選擇機(jī)制,使得DSABC算法的求解精度有較大幅度提高,特別是對(duì)于兩種不同維數(shù)的Rosenbrock函數(shù),所得最優(yōu)
3、值的絕對(duì)誤差分別由0.0017和0.0013減小到0.000049和0.000057;而且,DSABC算法克服了進(jìn)化后期因群體位置多樣性丟失較快而產(chǎn)生的早熟收斂現(xiàn)象,提高了整個(gè)種群的收斂精度及解的穩(wěn)定性,從而為函數(shù)優(yōu)化問(wèn)題提供了一種高效可靠的求解方法。關(guān)鍵詞:群體智能;人工蜂群算法;動(dòng)態(tài)評(píng)價(jià)選擇策略;收斂精度;函數(shù)優(yōu)化中圖分類號(hào):TP301.6;TP18文獻(xiàn)標(biāo)志碼:AImprovedartificialbeecolonyalgorithmbasedondynamicevaluationselectionstrateg
4、yXUXiangping,LUHaiyan,CHENGBiyun(SchoolofScience,JiangnanUniversity,WuxiJiangsu214122,China)Abstract:ToovercometheproblemofeasilytrappingintolocaloptimaofstandardArtificialBeeCotony(ABC)algorithm,therouletteselectionstrategyofABCwasmodifiedandanimprovedABCbased
5、ondynamicevaluationselectionstrategy(DSABC)algorithmwasproposed.Firstly,thequalityofeachfoodsourcepositionwasevaluateddynamicallyaccordingtothetimesthatthefoodsourcepositionhadbeencontinuouslyupdatedorstagnatedwithinacertainnumberofiterationsSOfar.Then,onlooker
6、beeswererecruitedforthefoodsourceaccordingtotheobtainedvalueoftheevaluationfunction.Theexperimentalresultsonsixbenchmarkfunctionsshowthat,comparedwithstandardABCalgorithm,theproposeddynamicevaluationselectionstrategymodifiestheselectionstrategyofABCalgorithm,an
7、dgreatlyimprovesthequalityofsolutionofDSABCalgorithm,especiallyforfunctionRosenbrockwithdiferentdimensions,theabsoluteerrorofthebestsolutionreducesfrom0.0017and0.0013to0.000049and0.000057,respectively;Moreover,DSABCalgorithmcanavoidtheprematureconvergencecaused
8、bythedecreaseofpopulationdiversityatlaterstageandimprovetheaccuracyandstabilityofsolutions,thusprovidesanefficientandreliablesolutionmethodforfunctionoptimization.Keywords:s