Long-term Prediction of Traffic Volume Based on Clustering Weighted Markov Chainsy
- Resource Type
- Conference
- Authors
- Li, Runmei; Dong, Shuyun; Xie, Kaibing
- Source
- 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) Intelligent Transportation Systems (ITSC), 2022 IEEE 25th International Conference on. :3466-3471 Oct, 2022
- Subject
- Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Engineering Profession
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Transportation
Solid modeling
Fluctuations
Clustering algorithms
Markov processes
Predictive models
Prediction algorithms
Data models
- Language
This paper presents a long-term traffic volume prediction approach using a novel Clustering Weighted Markov Chains Prediction (CWMCP) model which is composed of four modules: the Markov property test module, traffic volume states classification module, transition probability matrix construction and long term traffic volume forecasting module. By adopting the sequential sample clustering algorithm, the historical traffic volume data are clustered into intervals according to the similarity degree of the same time period for each day. Each interval represents a state of the Markov chains. The weighted Markov chain is adopted to solve problem that multiple maximum values of forecasting probability vectors may appear for calculating the transition probability. Based on the law of large numbers, the long-term traffic volume prediction is realized by finding the corresponding state of the maximum probability. Experimental results show that the proposed CWMCP model can predict not only the traffic volume with high accuracy but also the fluctuation range of the traffic volume.