摘要
针对标准差分进化(DE)算法在高维复杂函数优化中易早熟收敛,进而导致搜索精度降低甚至优化失败的问题,提出一种基于混沌和高斯局部优化的混合差分进化算法(CGHDE).该算法在进化初期利用混沌的遍历性,可有效地避免算法陷入局部最优;而在进化后期,采用高斯搜索又可有效地提高收敛精度.实验表明,CGHDE算法对函数维度的敏感性大大低于标准DE算法,并且寻优能力强、稳定性好、搜索精度高,特别适合于工程中高维复杂函数的优化问题.
Chaos and Gaussian local optimization based hybrid differential evolution(CGHDE) is proposed to solve the premature convergence and low precision of standard differnential evolution(DE) when applied to high-dimensional complex engineering problems.By means of the randomicity of chaotic local search,the CGHDE algorithm tends to explore in a wide search space to overcome the premature in the earlier evolution phase,and then performs exploitation to refine the optimum by using Gaussian optimization to improve the output in the later run phase.Simulations show that,CGHDE algorithm is not as sensitive to function dimensions as standard DE and has the advantages of powerful optimizing ability,more stability,higher optimizing precision and suitable for high-dimensional complex functions optimization.
出处
《控制与决策》
EI
CSCD
北大核心
2010年第6期899-902,共4页
Control and Decision
基金
国家自然科学基金项目(60872021)
上海市重点学科和科委重点实验室项目(S30108
08DZ2231100)
关键词
混沌优化
高斯优化
差分进化
遗传算法
Chaos optimization
Gaussian optimization
Differnential evolution
Genetic algorithms