多尺度多方法组合的网约车需求预测方法研究  

Research on online car⁃hailing demand prediction by combining multi⁃scale and multi⁃method

作　　者：丁夏蕾 郭秀才[1] 程勇[1] DING Xialei;GUO Xiucai;CHENG Yong(College of Electrical and Control Engineering,Xi’an University of Science and Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学电气与控制工程学院,陕西西安710054

出　　处：《现代电子技术》2022年第3期96-102,共7页Modern Electronics Technique

基　　金：国家重点研发计划“物联网与智慧城市”专项:智慧城市的群体态势辨识与服务计算的基础理论及关键技术研究(2018YFB2100800)。

摘　　要：为了解决单一时间序列模型对网约车需求预测精度不高,而导致网约车供需不平衡的问题,提出一种多尺度多方法组合的网约车需求预测方法。对网约车影响因素进行分析并提取出主要特征,利用历史平均模型、ARIMA时间序列模型、LSTM长短期记忆神经网络进行网约车需求预测,分别提取出网约车数据的周期性规律、差分变化规律和其他复杂规律,为了最大限度发挥三种单一模型的优势,使用三种单一模型建立组合模型,最后用灰狼优化算法对组合模型的权重进行寻优。通过在真实数据集上对单一模型和组合模型的预测精度进行比较,结果表明,组合模型在五种评价标准下均优于单一模型,更好地发挥出单一模型的预测优势,预测精度更高。同时在不同的适应度函数下进行实验,验证了模型的鲁棒性。采用组合预测模型预测精度更高,更适合作为网约车需求预测的有效模型。The single time series model fails to have high prediction accuracy for online car⁃hailing demand,which leads to the imbalance of supply and demand of online car⁃hailing.Therefore,an online car⁃hailing demand prediction combining multi⁃scale and multi⁃method is proposed.The influencing factors of online car⁃hailing demand are analyzed and the main features of the demand are extracted.The historical average model,autoregressive integrated moving average(ARIMA)time series model and long short⁃term memory(LSTM)neural network model are used to predict the online car⁃hailing demand,and the periodicity laws,differential change laws and other complex laws of online car⁃hailing data are extracted respectively.A combined model is established to maximize the advantages of the three models.The gray wolf optimization(GWO)algorithm is used to optimize the weight of the combined model.The prediction accuracy of the three models are compared with that of the combined model by the real data set.The results show that the combined model is better than each of the single model under the five evaluation criteria,can better exerts the prediction advantages of each of the single model,and has higher prediction accuracy.Experiments were conducted under different fitness functions to verify the robustness of the combined model.It is cocluded that the combined model has higher prediction accuracy and is more suitable as an effective model for online car⁃hailing demand prediction.

关 键 词：组合预测模型 网约车需求预测 灰狼算法 LSTM ARIMA 时间序列 深度学习 

分 类 号：TN911.1-34[电子电信—通信与信息系统] TP181[电子电信—信息与通信工程]