Study On The Behavioral Portrait Of Inland Ships Crossing Bridge Based On AIS

The construction of bridges in the inland navigation waters of China is becoming more and more perfect.Ships often need to pass bridges in the inland waterway and even pass through multiple bridges.When a ship is driving on an inland waterway,its behavior is not only affected by other ships,but also by bridges.Ship behavior is closely related to shipping safety.It is important to find the characteristics of ship crossing bridge behavior,mining law of ship navigation and predicting ship trajectory.Based on the construction of the label system of ship bridge-crossing behavior in the water area of the bridge,we studies the characteristics of ship bridge-crossing behavior and paints a portrait of the ship in the scene of ship passing through the navigable hole of the bridge.Furthermore,the Apriori algorithm for mining association rules and class-association rules(CAR)classifier are used to realize the prediction model of ship bridge-crossing behavior portrait,and new ship cases are used to verify the validity of the model.Main research work includes:1)Processing AIS data based on waterways of the bridge area.Based on the AIS data of the ascending ship in Wuhan section of the Yangtze River in dry season,the AIS data of ships in the waterways of the preparation area and the waterways of the bridge area were sampled by the ray method.After data cleaning and standardized processing,the independently driven ships are selected as the research object while encounter ships and follow-up ships are excluded.2)Constructing a ship portrait label system based on user portrait theory.According to the user portrait theory and the label system construction method,the dynamic and static attributes characteristics of the ship are analyzed,and the factual label system framework and the model label system framework of the ship's bridge-crossing behavior are determined.3)Constructing a portrait model of ship bridge-crossing behavior portrait model based on ship bridge-crossing behavior characteristics.We analyze ship behavior by extracting static characteristic information and dynamic data information such as ship speed,course Angle,ship size and ship type from AIS data,and use clustering andclassification technology to process multiple ship behavior characteristics in the waterways of preparation area and bridge area.According to the established model label system of ship bridge-crossing behavior,the corresponding label is assigned to each ship's behavior characteristics,so as to establish a ship label library and depict a relatively complete ship bridge-crossing behavior portrait.4)Constructing the prediction model of the ship bridge-crossing behavior portrait based on association rules.The ship bridge-crossing behavior association rules are preliminarily mined by Apriori algorithm,then the rules are pruned and ranked according to the given condition attributes and prediction attributes,so that the CAR classifer is constructed to predict ship bridge-crossing behavior.Finally,ship bridge-crossing behavior of 613 test sets were predicted and compared with the actual results.The prediction accuracy was 76.83%.