Short-term Urban Traffic Forecasting Based On Multi-kernel SVM Model

Nowadays with urban traffic congestion and jam more and more serious, the loss caused by traffic accidents, traffic pollution and energy consumption is becoming a pressing problem accordingly. Intelligent Transportation Systems (ITS) is recognized as a best way to sovle urban traffic problems.Dynamic traffic information services is the hub of each module in ITS, and real-time traffic information the blood of dynamic traffic services. As traffic information collection technology, Internet and wireless network technology mature, it makes real-time traffic information acquisition possible to lay an resources foundation for the analysis of real-time traffic conditions. How to quickly and accurately discover and predict future potential traffic conditions from huge number of real-time traffic data to assist in the dynamic navigation and route guidance for urban transportation planning and management department, travelers and participants, will be of great significance to easing traffic pressure, ensuring traffic safety, improving operating efficiency and reducing air pollution.In this paper, an intelligent analysis framework for travel speed gathered from floating car data (FCD) is established on the basis of dynamic traffic information platform. The main innovation lies belows:1:Implement a effective visualization of traffic flow and design an exploratory analysis method for traffic anomaly detection based on quantile.Standard normal deviation (SND) is now considered a very good method of traffic anomaly detection. But as the average and variance of sample data itself vulnerable to pollution, it can not reflect the difference between value of the sample center and the extreme to warp the judgment. In this article, an exploratory analysis method for traffic anomaly detection based on quantile is put forward which conducts strong anti-disturbance and robustness performance.2:Propose a hybrid multiple-kernel support vector machine model (MSVM) for conducting short-term traffic forecast. City road traffic system is characteristic of nonlinearity, uncertainty and spatial-temporal correlation, which makes the traffic system parameters description and knowledge extraction difficult, and results in current short-term traffic forecast methods can not obtain satisfactory accuracy. This paper presents a hybrid multiple-kernel support vector machine model (MSVM) for conducting short-term traffic forecast. With statistical analysis of large amounts of traffic conditions data samples, the proposed model not only has a capacity of recognizing and dealing with different types of input data separately, but also takes the advantages of global optimization, generalization and adaptability of support vector machine. Moreover, the parameters of the hybrid model get optimized with an improved particle swarm algorithm (PSO). Aiming at the linear correlation between real time and historical traffic condition, the nonlinear correlation between real time and previous time period, and also up and downstream traffic condition, the proposed model uses a linear kernel to extract the linear pattern of traffic flow and a nonlinear kernel to map the nonlinear pattern of residuals from the linear kernel. Both the historical regularity and time-variation characteristics of city road traffic are considered in the MSVM model so as to obtain the knowledge from the influential factors of real time traffic condition in order to improve forecast accuracy.