基于遗传算法与小波神经网络的客流预测研究

针对轨道交通短时客流具有动态性、非线性、不确定性的特点,提出一种基于遗传算法与小波神经网络的轨道交通短时客流预测方法.该方法利用具有全局搜索最优的遗传算法优化小波神经网络,有效的避免了神经网络易陷入局部最小值的缺陷.在分析轨道交通短时客流的特征上,利用实测数据对模型进行验证.结果表明,相比遗传算法优化的BP神经网络模型,单一的小波神经网络模型其预测精度更高,误差更小,能在实际中应用.

Passenger Flow Prediction Based on Genetic Algorithms and Wavelet Neural Networks

Aiming at characteristics of dynamic,nonlinearity,and uncertainty of rail transit short-term passenger flow,the paper presents a short-term passenger flow forecasting method based on genetic algorithms and wavelet neural networks.This method utilizes the genetic algorithm which has global search optimal to optimize wavelet neural network,which can avoid the defects of neural network that easy to fall into local minimum.Based on the analysis of the characteristics of rail transit short-term passenger flow,real data is used to testify precision of the model.The result show that the forecast model based on genetic algorithms and wavelet neural networks has higher accuracy and lower error than the model based on genetic algorithms and back propagation neural networks and wavelet neural networks.There are certain practical value.