為提高狼群算法的收斂速度,在此提出了一種稱為三級(jí)領(lǐng)導(dǎo)式和微粒進(jìn)化方程的自適應(yīng)狼群算法,人為地把灰狼分成兩類,領(lǐng)導(dǎo)層三只灰狼:如αβ和δ,剩下的為猛狼w。在游走搜索階段隨機(jī)設(shè)定一個(gè)獵物位置,利用狼群與獵物之間的距離來(lái)指導(dǎo)游走搜索獵物;在召喚階段,利用三個(gè)領(lǐng)導(dǎo)層灰狼作為頭狼來(lái)引導(dǎo)猛狼向獵物靠近,避免了傳統(tǒng)狼群算法只有一只頭狼引導(dǎo)整個(gè)狼群就容易陷入局部最優(yōu)的情況;在圍攻獵物階段利用慣性因子來(lái)表示以往奔襲的經(jīng)驗(yàn)、學(xué)習(xí)因子與隨機(jī)數(shù)之間的乘積來(lái)表示猛狼自身經(jīng)驗(yàn)的認(rèn)識(shí)與總結(jié)、迭代影響因子來(lái)表示整體狼群經(jīng)驗(yàn)的認(rèn)識(shí)與調(diào)整,綜合起來(lái)狼群粒子奔襲速度加快收斂速度和跳出局部最優(yōu),從而找到真實(shí)的整體最優(yōu)值。本次選取的8個(gè)測(cè)試函數(shù)對(duì)應(yīng)的對(duì)比性實(shí)驗(yàn)結(jié)果表明:該方法較為精確地實(shí)現(xiàn)尋找到了測(cè)試函數(shù)的最優(yōu)值且較早地快速收斂到最優(yōu)解,在后期也平穩(wěn)收斂到真實(shí)的最優(yōu)值,該算法適用于多維多波峰函數(shù)求極值問(wèn)題。 

【文章來(lái)源】：計(jì)算機(jī)工程與應(yīng)用. 2019,55(15)北大核心CSCD

【文章頁(yè)數(shù)】：10 頁(yè)

【部分圖文】：

狼群等級(jí)（優(yōu)勢(shì)度自上而下降低）

其他方法快得多。與其他兩種算法相比，時(shí)間消耗多出了0.002~0.250s，主要是用于判斷當(dāng)前三個(gè)領(lǐng)導(dǎo)頭狼的適應(yīng)函數(shù)值，再?zèng)Q定由哪個(gè)作為當(dāng)前尋優(yōu)的領(lǐng)導(dǎo)者而引起的計(jì)算時(shí)間。以及對(duì)于同規(guī)模函數(shù)消耗的時(shí)間要多些，主要是在攻擊階6005004003002001001002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao140120100806040200F11(x1,x2)0Parameterspace5005005000500x2x1圖2第一次實(shí)驗(yàn)F11函數(shù)和搜索迭代圖121086421002003004005000IterationBestscoreobtainedsofar/108ObjectivespaceGWO[1]GWO[2]GWOChenChao14121086420F13(x1,x2)0Parameterspace55505x2x1圖3第一次實(shí)驗(yàn)F13函數(shù)和搜索迭代圖450400350300250200150100501002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao5004003002001000F14(x1,x2)0Parameterspace1001001000100x2x150505050圖4第一次實(shí)驗(yàn)F14函數(shù)和搜索迭代圖302520151051002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao3.02.52.01.51.00.50F18(x1,x2)/1080Parameterspace55505x2x1圖5第一次實(shí)驗(yàn)

btainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao140120100806040200F11(x1,x2)0Parameterspace5005005000500x2x1圖2第一次實(shí)驗(yàn)F11函數(shù)和搜索迭代圖121086421002003004005000IterationBestscoreobtainedsofar/108ObjectivespaceGWO[1]GWO[2]GWOChenChao14121086420F13(x1,x2)0Parameterspace55505x2x1圖3第一次實(shí)驗(yàn)F13函數(shù)和搜索迭代圖450400350300250200150100501002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao5004003002001000F14(x1,x2)0Parameterspace1001001000100x2x150505050圖4第一次實(shí)驗(yàn)F14函數(shù)和搜索迭代圖302520151051002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]GWOChenChao3.02.52.01.51.00.50F18(x1,x2)/1080Parameterspace55505x2x1圖5第一次實(shí)驗(yàn)F18函數(shù)和搜索迭代圖00.040.080.120.160.20F20(x1,x2)0Parameterspace55505x2x11.61.82.02.22.42.62.83.03.23.41002003004005000IterationBestscoreobtainedsofarObjectivespaceGWO[1]GWO[2]


本文編號(hào)：3409560