Research On Traffic Status Recognition Based On Ship Big Data And Visualization Technology

In recent years,with the rapid development of water transportation in China,the traffic situation of inland waterways has become increasingly complicated,and the navigation efficiency and safety of ships have been seriously threatened.In order to relieve the traffic pressure of inland waterway,it has become an inevitable requirement for the intelligent development of China's shipping industry to efficiently obtain and intelligently analyze the traffic information of inland waterway,and then to effectively distinguish and predict the traffic status of inland waterway.The rapid development of big data technology and ship Automatic Identification System(Automatic Identification System,AIS)widespread popularization,made in a timely manner for ships in inland waterway traffic data has become possible,vast amounts of vessel traffic data for inland waterway traffic flow forecasting and traffic state identification provides data security.How to extract the characteristic parameters that can reflect the traffic condition of inland waterway from these data and build the prediction model and identification model of vessel traffic flow based on this is the premise of water traffic induction and control.Based on the above background,this paper focuses on the characteristics of inland waterways and uses channel AIS big data to conduct research on channel traffic flow prediction and traffic state recognition.Firstly,a method for extracting the traffic flow parameters of ships in inland waterway based on mass AIS data is designed.According to the characteristics of massive AIS data,the big data processing platforms Hadoop and Spark were used for effective management and preliminary preprocessing.Combined with the characteristics of ship traffic flow,the parameters of ship traffic flow in multiple segments were extracted at specific time intervals,and fault identification and repair were carried out.Then the traffic flow parameters after repair were smoothed by Savitzky-Golay filter,which effectively improved the quality of traffic flow prediction data.Finally,the spatial and temporal correlation analysis of each segment traffic flow is carried out to confirm the advantages of multi-segment traffic flow prediction.Secondly,according to the time series characteristics of ship traffic flow parameters,the traditional time series analysis model ARIMA and the deep learning models such as LSTM,SAE and CNN+LSTM based on multi-segment comprehensive analysis are designed and implemented for ship traffic flow prediction.Through the comparative analysis of the experiment,it can be seen that the CNN+LSTM multi-segment traffic flow prediction model based on the improved moving average has a high accuracy in both short-term and long-term prediction.Thirdly,according to the characteristics of the inland waterway transportation status,the fuzzy set theory is applied to traffic state identification,then designed and implemented fuzzy c-means clustering traffic state discrimination model combining the theory of fuzzy sets and clustering thought.And the discriminant results are visualized by various methods,Intuitively show the target segment time traffic status in the future.Based on the theory of ship traffic flow,this paper makes a full study on the identification and prediction of the traffic status of inland waterway.Inland waterway vessel traffic flow prediction model and the traffic state identification model are established.It has been verified by experiment the result can be more accurately reflect the state of inland waterway transportation in the future time,provides reference for traffic induced and navigation resources allocation.