基于GWO-LSTM的网约车需求短时预测模型

Short-term Forecasting Model of Demand for Network Booking Taxi Based on GWO-LSTM

为平衡网约车供需,支持车辆调度,该文研究了灰狼优化算法(GWO)优化参数的长短期记忆神经网络(LSTM)在网约车出行需求短时预测中的应用。研究了网约车出行需求时空特性,进行了影响因素的相关性分析;提出GWO改进LSTM的网约车需求短时预测模型;以实际数据验证了模型有效性并与其他模型进行对比。结果表明,相较于传统的LSTM网络及BP神经网络,该优化模型平均绝对误差分别提升了36.89%和52.12%,均方根误差分别提升了34.45%和48.16%;优化效果显著。

In order to balance the supply and demand of network booking taxi and support vehicle scheduling,this paper studies the application of the long-short term memory(LSTM)neural network with the optimized parameters of the Gray Wolf optimization(GWO)algorithm in the short-term forecast of network booking taxi travel demand.The spatial and temporal characteristics of the network booking taxi demand are studied,and the correlation analysis of the influencing factors is carried out.The short-term prediction model of network booking taxi demand based on GWO improved LSTM is proposed,and the validity of the model is verified by the actual data,and compared with other models.The results show that compared with the traditional LSTM network and BP neural network,the mean absolute error of the optimization model has been improved by 36.89%and 52.12%,and the root mean square error has been improved by 34.45%and 48.16%respectively.The optimization effect is significant.