Abstract: The characteristics of genetic algorithm and BP neural networks are compared. As evolutionary algorithm neural network and genetic algorithm have same goal but they have different methods. The necessity of the combination genetic algorithm and neural networks is expounded. This paper puts forward a kind of improved genetic algorithm to optimize BP neural network weights, using the global random searching ability of genetic algorithm to make up the question that neural network is easy to fall into local optimal solution. At the same time, the crossover method of genetic algorithm is changed. The same generation does not cross. The parent and son are crossed. Genetic algorithm premature loss of evolutionary ability is averted.

Key words: genetic algorithm, neural networks, optimization, weight, optimal solution

摘要： 对遗传算法和BP神经网络的特点进行了比较，作为进化算法神经网络与遗传算法的目标相近而方法各异。阐述了遗传算法与神经网络结合的必要性。提出了一种改进的遗传算法优化BP神经网络的权值，用遗传算法的全局随机搜索能力弥补了神经网络容易陷入局部最优解的问题。同时，在遗传算法中改变传统的同代交叉机制，采用父代与子代进行交叉，避免了遗传算法过早丧失进化能力。

关键词: 遗传算法, 神经网络, 优化, 权值, 最优解