基于遗传算法的BP神经网络时间序列预测模型

神经网络能以任意精度逼近非线性函数 ,以神经网络为基础的时间序列预测模型能很好地反映非线性系统发展的趋势 ,但神经网络训练速度慢、易陷入局部极值。针对这种情况 ,用具有良好的全局搜索能力的遗传算法来改进神经网络时间序列预测模型 ,提出了一种将遗传算法和BP算法相结合的学习算法来训练BP神经网络 ,并将该神经网络时间序列预测模型应用于某时间序列的预测。

BP Network Sequence Prediction Model Based on Genetic Algorithm

A neural network can approximate to a nonlinear function with any accuracy. The prediction model based on BP neural network has the advantage to reflect the development trend of the nonlinear system. However, NN has a very slow study rate and easily gets into the local extremum. So we present a prediction model that is based on the genetic algorithm and BP network by using the genetic algorithm with good global searching ability to solve the problem. At last we apply the model to the prediction of the resource expense of Hubei province.