遗传算法的爬山法改进

Improved genetic algorithm based on climbing

针对遗传算法在处理多峰优化问题时容易发生早熟现象的问题,采用了动态调整交叉概率值和变异概率值的方法,引入爬山法在迭代过程中进行局部寻优,仿真实验对比分析了标准遗传算法和改进遗传算法的性能.研究结果表明:改进后遗传算法的收敛速度较快,得到结果误差值比较小.研究结论证明在相应的进化阶段采用合理的概率值,利用爬山法对遗传算法局部寻优,可以避免早熟现象,提高遗传算法收敛速度和精度.

Genetic algorithms is prone to premature phenomenon when dealing with multimodal optimization problems. This paper proposed a dynamic adjustment probability method of crossover and mutation to solve the problem, and introduced climbing method into Genetic algorithms for local optimization in iterative process. The performance between standard genetic algorithm and improved genetic algorithm was compared through simulation experiments. The results show that the convergence speed is faster and error value is lower with the improved genetic algorithm. Research findings prove that obtaining appropriate probability value in the corresponding stage of evolution and using climbing for local optimization can avoid premature phenomenon of genetic algorithm and improve the convergence speed and accuracy.