An Improved k-NN Nonparametric Regression-Based Short-Term Traffic Flow Forecasting Model for Urban Expressways

Abstract

In domestic major cities, the development of Urban Expressway is network-oriented. The traffic flow forecasting system is the important prerequisite and foundation of realizing real-time traffic management and control. However, the traffic flow forecasting research is mainly based on highways. Research and application of short-term traffic forecasting for urban expressway is severely insufficient. Therefore, the study of urban expressway flow forecasting is discussed and a short-term traffic flow forecasting system for urban expressway based on k-NN nonparametric regression model is proposed in this study. First, the study analyze the characteristics and needs of the urban expressway traffic flow, introduce k-NN nonparametric regression model and design the short-term traffic flow forecasting system based on k-NN overall. Then, the short-term urban expressway flow forecasting system based on k-NN is established in three aspects: the historical database, the search mechanism and algorithm parameters, and the forecasting plan. At last, a short-term traffic forecasting for urban expressway based on k-NN nonparametric regression model is developed in the VS2010 VC++ platform. Utilizing the Shanghai urban expressway section measured traffic flow data, the comparison of average and weighted k-NN nonparametric regression model is discussed and the reliability of the forecasting result is analyzed. Results show that the accuracy of the proposed method, under the 5-minute interval, is over 90 percent, which best proves the reasonableness of the proposed forecasting model based on k-NN nonparametric model.

Analysis of Traffic Characteristics in Beijing --Based on the Traffic Performance Index

“…The congestion statistics involved in this paper are the traffic performance index (TPI (2,4] basically smooth 0.2-0.5 times more than normal (4,6] light congestion 0.5-0.8 times more than normal (6,8] moderate congestion 0.8-1.1 times more than normal (8,10] severe congestion more than 1.1 times normal…”

“…This was very cutting-edge and innovative at that time [3]. Zhang et al (2010) used the K-nearest neighbor nonparametric regression method to forecast the special road conditions traffic flow with high accuracy [4]. To strengthen the scheduling ability of traffic management, Yang et al (2021) introduced a short-time method to forecast traffic flow, which is from the perspective of the implicit interaction of multi-lane traffic flow on road sections.…”

“…The value ranges from 0 to 10 and can be divided into five types. The corresponding values are as follows: [0,2] is smooth,(2,4] is basically smooth, (4,6) is light congestion,(6,8] is moderate congestion, and(8,10] is severe congestion (Table1).…”

PURP: A Scalable System for Predicting Short-Term Urban Traffic Flow Based on License Plate Recognition Data

“…Then, PURP constructs the spatio-temporal contexts of traffic flow prediction. Afterward, PURP builds traffic flow vectors [15] for the spatio-temporal contexts and identifies the top-similar spatio-temporal contexts by calculating the similarity between the vectors. Finally, the traffic flow values corresponding to these contexts are used to make predictions.…”