基于LSTM-SVR的地铁客流量预测研究  

Research on Passenger Flow Forecast of Hangzhou Metro Based on LSTM-SVR

作　　者：谢玫秀 邓晓林 谭淞 XIE Meixiu;DENG Xiaolin;TAN Song(Guizhou University School of Electrical Engineering,Chengdu 610401,China)

机构地区：[1]成都文理学院,成都610401

出　　处：《移动信息》2024年第8期245-247,共3页MOBILE INFORMATION

摘　　要：在日常生活中,交通管理部门需要精准、及时地预测地铁客流量,以帮助出行者规划出行路线,节约出行时间,缓解交通拥堵问题。文中从组合不同模型的角度出发,尝试探索了更准确的数据预测。首先,时间序列数据特征适用于LSTM模型处理,借助该模型可以实现极强的特征提取功能;其次,支持向量回归(SVR)是一种基于SVM的回归算法,具有高维的表达能力,在数据拟合上具有显著优势。文中结合LSTM与SVR构建组合模型,并借助实际的仿真实验数据验证了组合模型的有效性。另外,为使结果更具说服力,文中还比较分析了最近邻回归(KNN)模型的实验数据。In daily life,traffic management departments need to accurately and timely predict subway traffic counting to help travelers plan travel routes,save travel time,and alleviate traffic congestion problems.This paper attempts to explore more accurate data forecasting from the perspective of combining different models.First,the characteristics of time series data are suitable for LSTM model processing,and with the help of this model,strong feature extraction can be realized;secondly,support vector regression(SVR)is a regression algorithm based on SVM,which has high-dimensional expression ability and has significant advantages in data fitting.In this paper,a combined model is constructed by combining LSTM and SVR,and the effectiveness of the combined model is verified with the help of actual simulation experimental data.In addition,in order to make the results more convincing,the experimental data of nearest neighbor regression(KNN)model is also compared and analyzed.

关 键 词：客流量预测 特征提取 混沌粒子群 组合模型 

分 类 号：TP306.1[自动化与计算机技术—计算机系统结构]