整个算法分成三部分，第一个部分是神经网络整体结构的确定，然后是遗传算法对参数的优化，然后便是利用已经优化好的参数，利用神经网络进行预测。我们根据需要优化的权值以及阈值的数量确定算法个体的长度。个体通过适应度函数计算他们的适应度，根据适应度的大小，我们使用轮盘算法，确定他们的遗传，交叉还有变异等过程，优化以后的BP神经网络可以更好地收敛。-The whole algorithm is divided into three parts, the first part is to determine the overall structure of the neural networks, genetic algorithms and optimization of parameters, then that is a good use of the parameters have been optimized using neural networks to predict. We determine the length of the algorithm needs to be optimized according to the number of individual weights and thresholds. Individual fitness function is calculated by their fitness, according to the size of fitness, we use roulette algorithm to determine their genetic, crossover and mutation process as well as to optimize future BP neural network can better convergence.