基于深度学习的地铁短时客流量预测

Short-term passenger traffic forecast based on deep learning

为了对地铁短时客流流量进行准确预测,建立了一种基于深度学习的地铁短时客流预测方法。通过分析地铁的客流数据,发现周一至周四、周五以及周末的发展模式存在一定的差异。依据该发现,基于深度学习的理论框架,建立了双向长短期记忆网络的地铁短时客流量预测模型。最后以广州体育西路地铁站数据为例进行预测分析,并将预测结果与决策树模型、支持向量机算法以及长短期记忆网络的预测结果进行对比分析。结果表明,双向长短期记忆网络全面优于其他预测算法,且该算法的平均预测精度超过90%,对地铁运力的合理配置等有一定的应用价值。

In order to accurately predict the short-term passenger flow of the subway,a short-term passenger flow forecasting method based on deep learning is established.By analyzing the passenger flow data of the subway,it is found that there are some differences in the development patterns from Monday to Thursday,Friday and weekend.According to the discovery,under the theoretical framework of deep learning,a passenger flow prediction model based on bi-directional long short-term memory network is established.Finally,the data of Guangzhou of Tiyuxilu Station is taken as an example for predictive analysis,and the prediction results are compared with the decision tree model,support vector machine algorithm and long short-term memory network.The results show that bi-directional long short-ierm memory network is better than other prediction algorithms,and the average prediction accuracy of the algorithm is more than 90%,which has certain application value for the rational allocation of subway capacity.