Short Term Traffic Flow Depth Prediction Algorithm Based On Weather Factors

With the development of economy,the numbers of vehicles are rising.A series of traffic problems such as traffic congestions and traffic accidents also arise,which severely restricted the road capacity.Intelligent Transportation Systems rely on monitoring and inducing the running status of vehicles,and optimizing the distribution of traffic flow in road networks,can effectively alleviate road traffic congestions and traffic accidents.It is currently recognized as the most effective way to solve traffic problems around the world.As one of the key technologies in the field of intelligent transportation research,traffic flow prediction has become a hot topic at home and abroad.Most current traffic flow prediction algorithms only consider normal forecasting,without considering the impact of weather factors,traffic accidents and other unexpected factors on the forecast results.This paper proposes a short-term traffic flow prediction model that fusion weather factors.Firstly,the weather and traffic data collected are preprocessed such as missing,denoising,and normalization;Secondly,correlation analysis is used to obtain the most relevant weather factors related to traffic flow data;Finally,construct a short-term traffic flow prediction model based on deep belief network and support vector regression,and the weather data and traffic flow data are used as predictive model training samples.According to the differences in the distribution of traffic flow data in time,in order to improve the prediction accuracy,traffic flow on workdays and weekends is predicted separately.The prediction model was validated by actual traffic flow and weather data provided by the United States Traffic Data Research Laboratory and the University of Utah MesoWest.The results show that the prediction model proposed in this paper is compared with the unfused weather factors model,the average prediction accuracy is improved by 4%,the effectiveness of the proposed traffic flow prediction model with fusion weather factors is verified,which provides theoretical support for timely and reliable traffic guidance and control.