基于GWO-WNN模型的城市轨道交通短时进站客流预测

Short-term Passenger Flow Forecast for Urban Rail Transit based on GWO-WNN Model

城市轨道交通短时进站客流具有较强的非线性与随机性。针对该问题,提出一种基于灰狼优化算法(Grey Wolf Optimizer, GWO)与小波神经网络(Wavelet NeuralNetwork,WNN)的城市轨道交通短时进站客流预测模型(GWO-WNN模型)。在构建小波神经网络的基础上,利用灰狼优化算法对初始权值和小波因子进行全局搜索寻优,有效避免了小波神经网络预测结果不稳定及极易陷入局部最小值的缺点。结合案例分析,将单纯WNN模型预测结果与GWO-WNN模型的预测结果比较分析,结果表明GWO-WNN模型相比单纯的WNN模型,在绝对误差和绝对误差百分比方面有极大的提升,均方根误差也可以佐证优化后的WNN模型预测精度更高。

Driven by the short-time passenger data characteristics of nonlinearity and randomness, this study proposes a short-term passenger flow forecast model for urban rail transit based on GWO-WNN model. Firstly, this study builds a wavelet neural network. Then, the grey wolf optimizer is used to search for the initial weight and wavelet factor globally, which effectively avoids the disadvantages of unstable results and falling into local minimums easily. Experimental results show that GWO-WNN model has a significant improvement in absolute error and absolute error percentage compared with the simple WNN model, and the Root Mean Square Error can also prove that GWO-WNN model has higher prediction accuracy.