Research Of Short-term Traffic Volume Prediction Based On Kalman Filtering

With the development of urbanization in the world and the popularity of cars, leading toenvironmental degradation, traffic accidents, as well as issues such as increasing traffic jams.In order to solve the above problems, the appropriate technologies have emerged IntelligentTransport Systems(ITS). In the field of intelligent transportation systems, road traffic forecastis the key, and the traffic flow on network traffic is the priority among priorities. However,based on the current research, traffic flow forecast model accuracy is not high, more complexmodeling, poor real-time performance and so on.In the early sixties and seventies of the20th century, foreign began to used forecastmodel for short-term traffic flow prediction field.And a lot of traffic flow forecasting models.For example, multiple linear regression model, time series model, neural network model, thehistorical trend model,Kalman filter model and so on. And this paper is focused on Kalmanfiltering in traffic flow forecast.This paper studies the urban road network traffic in a static stability with paroxysmalcharacteristics at the same time. From the perspective of spatial correlation,this paper use graycorrelation analysis method to analyze which parameters will affect the road. Besides, in thispaper in order to improve the prediction effect of Kalman filter model in the History Day, itbrings out that use several weeks corresponding to the ratio of the traffic flow instead of theoriginal data to establish Kalman filter prediction model base on History Day. Then by theexperimental simulation of the actual traffic flow, it calculates the model performance indexand carries out comparative analysis.Finally, Prediction model and the algorithm proposed in this paper are verified bysimulated data. Experiments show that: gray correlation analysis method could analyzedwhich parameters will affect the road; the prediction effect of the forecast model based onhistorical data and real-time data is better than the forecast model only based on real-time data,and shows that the model has strong adaptability and high accuracy of prediction.