Highway Toll Station Exit Flow Prediction Based On Pso-Lssvm-Gbrt Model
PROCEEDINGS OF THE 2019 31ST CHINESE CONTROL AND DECISION CONFERENCE (CCDC 2019)(2019)
摘要
The exit flow of highway toll stations is an important basis for the toll station operating department to set up toll gates and staff scheduling. In order to improve the prediction accuracy of the toll station's export flow, this paper considers the time-series correlation of the target toll station and temporal and spatial correlation between target toll stations and other toll stations on the road network. The LSSVM model optimized with PSO is used to predict the time-correlation level of the target toll station and combine it with the GBRT model to predict the temporal and spatial correlation between the target toll station and other toll stations, finally, combine the two model to get the prediction result. The model is evaluated using actual highway network charging data. and the average relative error percentage of the predicted target toll station exit flow is approximately 15.73%. Compared with LSSVM. PSO-LSSVM, GBRT, ARMA. BP neural network and other models, the proposed method has higher prediction accuracy.
更多查看译文
关键词
Exit flow, Prediction accuracy, The highway network charging data, Temporal and spatial correlation
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络