摘要： 针对基本遗传算法全局搜索能力差和收敛速度慢，且在求解多峰函数时仅能得到部分最优解的缺点，提出一种基于梯度优化的自适应小生境算法。该算法利用当前种群适应度和种群代数来设计交叉算子和变异算子，有效地保持了种群的多样性，改善全局搜索能力，加快了收敛速度，应用改进的梯度优化算子保证进化向最优解方向靠近，提高了计算峰值的精确度。对Shubert函数的仿真试验证明，该算法能改善全局搜索能力，加快算法收敛速度并提高计算精度。

关键词: 小生境遗传算法, 自适应, 梯度优化, 非均匀变异算子

Abstract: To deal with low efficiency and low convergence speed in searching the global optimum, and only gaining several the optimum while it is used in multimodal-function-optimization, an adaptive Niche Genetic Algorithm(NGA) based on gradi-optimization is proposed in this paper. The adaptive crossover operator and mutation operator are used to guarantee the population diversity, improve searching the global optimum and convergence speed. The gradi-optimization is used to improve the precision of the optimum. Simulation results in the Shubert show that this method is nice at improving on searching the global optimum, convergence speed and its superiority in precision.

Key words: Niche Genetic Algorithm(NGA), adaptive, gradi-optimization, non-uniform mutation eperator

中图分类号: 

• TP18