基于组合预测模型的铁路客运量预测研究

Research on Railway Passenger Volume Forecast Based on Combined Forecast Model

为提高铁路客运量的预测精度,对组合预测模型的权重分配方法及组合方式的预测效果进行研究。综合考虑预测误差及其均方差的影响,构建基于Logit模型的权重分配模型以解决组合预测模型的权重分配问题,并提出模型求解算法。以北京市铁路客运量预测为例,研究BP神经网络、霍尔特线性趋势指数平滑法和ARIMA模型的多种组合方案的预测效果,并验证基于Logit模型的权重分配模型的优势。研究结果表明:线性与非线性预测模型组合的预测精度优于线性与线性预测模型的组合,其中,B-H-A模型的组合预测效果最好,误差低至0.606%。另外,通过与等分权重法对比,基于Logit模型的权重分配模型赋值的权重能提高组合预测模型的预测精度,且适用性更好。

In order to improve the prediction accuracy of railway passenger volume, the weight distribution method of the combined prediction model and the prediction effect of the combined method are studied.Considering the influence of the forecast error and its mean square error, a weight distribution model based on the Logit model is constructed to solve the weight distribution problem of the combined forecasting model,and a model solving algorithm is proposed. Taking Beijing’s railway passenger traffic forecast as an example,the forecast effects of multiple combinations of BP neural network, Holt linear trend exponential smoothing method and ARIMA model are studied, and the advantages of the weight distribution model based on the Logit model are verified. The research results show that the prediction accuracy of the combination of linear and nonlinear prediction models is better than the combination of linear and linear prediction models. Among them,the combined prediction effect of the B-H-A model is the best, with an error as low as 0.606%. In addition,by comparing with the equal weight method, the weight assigned by the weight distribution model based on the Logit model can improve the prediction accuracy of the combined prediction model, and has better applicability.