【摘要】 提出了一种改进的差分算法(Improved DE,IDE),通过在现有变异策略DE/current-to-p-best/1的基础上加上群体中心对个体的差分,从而加快了整体算法的收敛速度;并采用一种简化的模拟退火选择策略,以兼顾收敛速度和寻找全局最优解的能力,避免了过快收敛而陷入局部最优解。此算法虽未采用自适应策略,经测试其优化结果比加入自变异策略的J.Zhang′s差分进化算法(JADE)更好。该算法无论经数学测试、还是应用于天线阵列的优化,都有非常好的鲁棒性。更多还原

【Abstract】 In this paper,an improved differential evolution(IDE) algorithm is proposed.In order to accelerate the convergence of the algorithm,a difference of the population center with respect to individuals is added to the mutation operator of DE/current-to-p-best/1.To avoid the trap of a local optimum and get balance between the convergence and the global search ability,a simplified simulated annealing(SA) selection strategy is adopted.Though there is no adaptive strategy in IDE,better performances than J.Zhang’s self-adaptive differential evolution(JADE) are obtained in the test by benchmark functions.The high robustness of IDE is verified both in the test of mathematical functions and application in array pattern synthesis.更多还原