基于BP神经网络模型的客流预测研究——以西安地铁小寨站为例  

Research on Passenger Flow Prediction Based on BP Neural Network Model-Take Xi'an Metro Xiaozhai Station as an Example

作　　者：张思瑶 杜晨阳 张懿槾 高婉琦 贺鹏飞 ZHANG Siyao;DU Chenyang;ZHANG yiman;GAO Wanqi;HE Pengfei(School of Traffic and Transportation,Xi'an Traffic Engineering Institute,Xi'an 710300,China;Operation Company of Xi'an Rail Transit Group Company Limited,Xi'an 710018,China)

机构地区：[1]西安交通工程学院交通运输学院,陕西西安710300 [2]西安市轨道交通集团有限公司运营分公司,陕西西安710018

出　　处：《河南科技》2025年第3期59-64,共6页Henan Science and Technology

基　　金：西安交通工程学院大学生创新创业训练计划项目“显而易见——基于人工智能的城市轨道交通可视化候车系统”(2023DC08);陕西省教育厅科学研究计划项目(22JC048);西安市科技计划项目(23GXFW0046)。

摘　　要：【目的】为了提高客流量预测的准确性,研究基于BP神经网络模型的客流量预测方法,为城市公共交通的调度和规划提供更为可靠的数据支持。【方法】采用BP神经网络模型,利用历史客流量数据作为训练样本,构建能够对未来客流量进行预测的模型。通过模型训练与验证,分析不同参数配置下的模型性能,并与传统预测方法进行了对比。【结果】结果表明,基于BP神经网络的预测模型在多个时间段的客流量预测中表现优异,预测误差显著低于传统方法。BP神经网络模型预测结果在仅使用均值进行预测的情况下,其准确度越接近于1精准度越高,即预测结果训练集为0.701,测试集为-0.906均接近于1。【结论】BP神经网络模型能够有效捕捉客流量的变化趋势,具有较高的预测精度,适用于复杂城市交通系统的客流量预测任务。未来的研究可进一步优化模型参数,并结合其他算法提高预测性能。[Purposes]In order to improve the accuracy of passenger flow prediction,this paper studies the passenger flow prediction method based on BP neural network model,aiming to provide more reliable data support for the dispatch and planning of urban public transportation.[Methods]Using the BP neural network model and using historical passenger flow data as training samples,a model capable of predicting future passenger flow was constructed.Through model training and verification,the model performance under different parameter configurations was analyzed and compared with traditional prediction methods.[Findings]The experimental results show that the prediction model based on BP neural network performs well in passenger flow prediction in multiple time periods,and the prediction error is significantly lower than the traditional method.When the prediction result of the BP neural network model is compared with the prediction only using the mean value,the closer the accuracy is to 1,the higher the accuracy is.That is,the training set of the prediction result is 0.701,and the test set is-0.906,both of which are close to 1.[Conclusions]Research has proven that the BP neural network model can effectively capture the changing trend of passenger flow,has high prediction accuracy,and is suitable for passenger flow prediction tasks in complex urban transportation systems.Future research can further optimize model parameters and combine with other algorithms to improve prediction performance.

关 键 词：BP神经网络模型 聚类分析法 移动平均法 客流预测 

分 类 号：U231.4[交通运输工程—道路与铁道工程]