共享单车用户出行目的地预测与分流技术研究

Travel Destination Prediction and Diversion Technology of Shared Bike Users

共享单车能够提升出行便利程度、缓解交通压力。在分析用户出行规律及时空范围内交通情况的基础上,采用深度学习预测模型DPNNst对共享单车出行目的地进行预测,能够整合交通资源、降低出行成本,有助于实现共享发展的目标。设计研究共享单车用户出行目的地的预测模型,综合卷积神经网络、长短期记忆网络、全连接神经网络等多种神经网络的计算优势,构建深度学习预测模型DPNNst,在配合LSTM训练的基础上,保证预测结果的精准度,从而引导用户实现高效还车,达到提升城市交通容纳空间的效果。

Shared bikes can improve travel convenience and relieve traffic pressure.The study analyzes the users’travel rules and space-time traffic situation,conducts DPNNst of in-depth learning prediction model,and predicts the travel destination of shared bike to integrate transportation resources,reduce travel cost,and realize the objective of shared development.Then the study designs the prediction model of travel destination of shared bike users;utilizes the calculation advantages of various neural networks,such as convolutional neural networks,long and short term memory network,and fully connected neural network,etc.;constructs in-depth learning prediction model DPNNst;predicts the accuracy of the results based on the combination of LSTM training to guide users to achieve efficient bike return and improve the effect of urban transportation capacity.