【摘要】 提出用种群发育停滞代数对变异概率和变异位数进行动态控制的改进遗传算法。该算法把种群没有更优个体产生看作种群发育停滞 ,将种群发育停滞代数定义为当前繁殖代序号与已得最优解的繁殖代序号之差 ;变异参数 (包括变异概率、变异位数 )初值与标准遗传算法 (SGA)相近 ;随着发育停滞代数的增长 ,增大变异参数 ;当有更优个体产生时 ,变异参数恢复到初值 ,种群发育停滞代数置 0 ;随种群发育停滞代数再次增长 ,变异参数再次增大 ,如此反复 ,直至算法结束。该算法在保持局部搜索能力的同时 ,提高了全局搜索能力及速度。用两个多极值函数(Camel函数、Shaffer’sF6函数 )对该算法进行测试 ,结果表明 ,与SGA及自适应遗传算法相比 ,该方法以相当强的鲁棒性收敛到全局最优解 ,且具有较高的收敛速度更多还原

【Abstract】 A better individual non-produced in a reproducing process was named as stasis of population upgrowth. Population upgrowth stasis generation count (PUSGC) was defined as the difference between the reproducing generation serial number and the generation serial number, at which the gained best individual can be born. The mutation parameters including mutation rate and mutation bit number were all initialized with values of standard genetic algorithm (SGA). With the PUSGC increasing, the mutation parameters were going up. While a better individual is produced, the mutation parameters come back to the initialized values, and the PUSGC is reset to zero. With the PUSGC increasing again, the mutation parameters improved again, the analogous reproduction process was cycling until the algorithm finished. This algorithm not only keeps the local searching capability but also improves the global searching ability and speed. Two multi-extreme functions including Shaffer’s F6 and Camel functions have been used to test the algorithm. Compared with SGA or adaptive genetic algorithm, the new method has better robusticity of convergence to the global optimal solution, and the convergence speed is much fast.更多还原