自适应调整控制参数的差异演化算法

Self-adapting control parameters in differential evolution

控制参数选取是包括差异演化在内的演化算法设计时所面临的一个重要问题,对算法的性能有着重大影响。针对差异演化算法参数选取问题,提出一种利用个体适应度作为参数调整决策依据,并结合一定的调整概率对F和CR进行自适应调整的方法,解决了手工设置控制参数的不便。同时利用交叉操作生成双子代个体与父代个体竞争形成新一代种群,加快了算法的收敛。对标准测试函数的仿真实验结果表明,该算法无论在最优解质量和收敛速度上都优于相关算法,尤其对于高维函数而言。

Parameters setting is an important problem of evolution algorithms, including differential evolution algorithm. It has an effect on the performance of evolution algorithms. According to the problem of parameters of differential evolution, a method is presented, which uses self-adaptive as a scientific evidence to adjust parameters and set F and CR combined with modulated probability. An algorithm is presented, which depends on the fitness of individual and modulated probability set the parameters F and CR automatically. This method can get the optimal control parameters for different optimization problem without user interaction. Moreover, two trial vectors are created by recombination for increased colony diversity and avoided premature con- vergence. These vectors compete with the parent individual to be the next generation. Experimental results indicate that the pro- posed algorithm is efficient and feasible. It is superior to other related methods such as DE, iDE, FADE, MPDE, DDE both on the quality of solution and on the convergence rate, especially for high dimension functions.