城市公共自行车系统(PBS)服务点自行车数量的再平衡是解决“租还车难”问题的关键,对服务点租还需求量的短时预测则是PBS再平衡的基础;通过分析PBS租还需求的内外关联影响因素,提出基于深度学习理论的LSTM (Long-Short Term Memory,长短时间记忆)单元的循环神经网络(Recurrent Neural Network,RNN)服务点租还需求量预测模型,并通过区域PBS平均出行OD,对预测模型的输入特征进行合理优化,实现PBS服务点租还需求量的短时预测;以杭州市下沙PBS服务区为实验对象,选取三组不同的输入时间步长对预测模型进行实践验证,结果显示:在选取的模型结构与输入特征下,采用循环神经网络对服务点租还需求量进行预测能够比传统前馈神经网络在结果上更加接近实际值,并且精度较为满意,表明了该预测方法可行有效.%The problem of rebalancing bicycles between service points in urban public bicycle system (PBS) is a key issue in determining the service quality of PBS.Predicting the short-term ride demand of PBS service points plays an important role in the rebalancing problem.By analyzing the external and internal ride demand factors of PBS,the ride demand predicting model of service points based on recurrent neural network with long-short term memory unit,which comes from deep learning domain,is proposed.The features of this model are extracted and optimized by calculating the mean OD of from the PBS historical data.And finally achieved the short-term ride demand prediction of PBS service points.The experiment in Hangzhou PBS show that using the recurrent neural network to predict the ride demand for service points can be closer to the actual value than the traditional feed forward neural network in the result.And the accuracy of recurrent neural network is more satisfactory,which indicates the predicting method is feasible and effective.
展开▼
