Traffic Flow Forecast Of Wuhan Yangtze River Bridge Based On Kalman Filter

With the proposed "One Belt and One Road" strategy in our country and the advancing of trade globalization,the water-borne transportation will become increasingly busier.As an important study objects and research hotspot in traffic engineering,the research of ship traffic flow forecasting is of important reference value for the future planning of the navigation channel and the layout of the navigational aids for the port channel as well as the macro-economic trend.Based on the current research on it,combined with the research experience and methods in other fields,the paper put forward a forecasting method to the forecast of actual traffic flow in the Wuhan Yangtze River Bridge.The second chapter introduces the theory of ship traffic flow,and formes a ship traffic flow forecasting system.In the third chapter,the KF theory,the EKF algorithm and the MI-EKF method are introduced.It reveals the forecasting process and principles of the KF.The fourth chapter analyzes the characteristics of ship traffic flow in Wuhan Yangtze River Bridge from two aspects of ship length distribution and ship speed distribution by using statistical methods.The study found that the cross-section of the Yangtze River traffic flow correlation is weak.It is difficult to predict different sections of traffic flow data;at the same time the length of the traffic flow and the speed distribution of the ship are in normal distribution and the probability distribution of the number of ships in each length is given.Finally,in the fifth chapter,KF algorithm and MI-EKF algorithm are used to forecast and analyze the monthly daily average daily vessel traffic flow and the daily real-time vessel traffic flow of the Wuhan Yangtze Bridge.Comparing the forecast data with the actual data,the MI-EKF algorithm can effectively predict the ship traffic flow of the Wuhan Yangtze River Bridge segment.The improved algorithm has obvious improve in forecasting stability and forecasting accuracy.It has a better effect on daily average daily traffic flow forecasting,and short-term traffic flow forecast needs to pay attention to the measurement noise and system noise control,and requires high data acquisition and processing.