计算机应用与软件
計算機應用與軟件
계산궤응용여연건
COMPUTER APPLICATIONS AND SOFTWARE
2014年
11期
275-278
,共4页
约束优化%免疫克隆优化%精英库机制%灾变算子%局部收敛%参数选择
約束優化%免疫剋隆優化%精英庫機製%災變算子%跼部收斂%參數選擇
약속우화%면역극륭우화%정영고궤제%재변산자%국부수렴%삼수선택
Constrained optimisation%Immune clone optimisation%Elite bank mechanism%Cataclysm operator%Local convergence%Pa-rameters selection
针对约束优化问题提出一种基于精英库机制的改进型免疫克隆优化算法ICOAEB( Immune clonal optimization algorithm based on elite bank)。该算法利用精英库机制动态存储迭代过程中父代优势个体,实现优秀个体的多代记忆,从而提高算法寻优能力;并利用灾变算子扰动算法运行过程从而摆脱迭代缓慢的状态,避免局部收敛。通过对五个约束优化函数的测试,实验结果表明ICOAEB的求解精度和稳定性较高,可以较好地解决约束优化问题。最后针对影响算法性能的两项重要参数选择问题给出了相关的实验及分析。
針對約束優化問題提齣一種基于精英庫機製的改進型免疫剋隆優化算法ICOAEB( Immune clonal optimization algorithm based on elite bank)。該算法利用精英庫機製動態存儲迭代過程中父代優勢箇體,實現優秀箇體的多代記憶,從而提高算法尋優能力;併利用災變算子擾動算法運行過程從而襬脫迭代緩慢的狀態,避免跼部收斂。通過對五箇約束優化函數的測試,實驗結果錶明ICOAEB的求解精度和穩定性較高,可以較好地解決約束優化問題。最後針對影響算法性能的兩項重要參數選擇問題給齣瞭相關的實驗及分析。
침대약속우화문제제출일충기우정영고궤제적개진형면역극륭우화산법ICOAEB( Immune clonal optimization algorithm based on elite bank)。해산법이용정영고궤제동태존저질대과정중부대우세개체,실현우수개체적다대기억,종이제고산법심우능력;병이용재변산자우동산법운행과정종이파탈질대완만적상태,피면국부수렴。통과대오개약속우화함수적측시,실험결과표명ICOAEB적구해정도화은정성교고,가이교호지해결약속우화문제。최후침대영향산법성능적량항중요삼수선택문제급출료상관적실험급분석。
We propose an improved immune clone optimisation algorithm which is based on elite bank mechanism ( ICOAEB ) for constrained optimisation problem.This algorithm utilises the elite bank mechanism to dynamically store the superior individuals of father generation during the iteration process, implements the multi-generation memory of superior individuals, and thereby improves algorithm’ s optimisation ability.ICOAEB also uses the cataclysm operator to disturb the running process of the algorithm so as to get rid of the slow iteration state and to prevent local convergence of the algorithm.By testing five constrained optimisation functions, the experimental results show that the solution accuracy and stability of ICOAEB is higher, and it can well solve the constrained optimisation problem.Finally, correlated experiment results and analyses are given for the selection of two important parameters which can affect the performance of the algorithm.