Spatio-Temporal Characteristic Analysis And Traffic Speed Prediction Of Vessels Based On AIS

Vessel traffic flow can effectively reflect the overall state of vessel movement within a certain time and space.Analyzing and predicting vessel traffic flow parameters in busy waters can not only provide a basis for resource management and allocation of waterways and ports,but also assist maritime authorities in obtaining accurate and efficient time-space dynamic information of waterways,providing a basis for vessel dispatching and command,improving navigation efficiency,and related port water area planning.In recent years,with the popularization and application of AIS(Automatic Identification System),the massive amount of AIS data has provided sufficient data support for studying vessel traffic flow.In view of the shortcomings in the current research field of vessel traffic flow,this thesis uses AIS data from the Nanjing section of the Yangtze River waterway to obtain the speed parameters of vessel traffic flow,and on this basis,analyzes the spatio-temporal characteristics of vessel traffic speed in the target area.After determining the spatio-temporal relation,a vessel traffic speed prediction model based on the spatio-temporal characteristics is constructed.The main research contents are as follows:(1)AIS data preprocessing: Select the AIS data of the target area from the original AIS data and further divide the target area into 7 sub segments.Preprocess AIS data which is unrelated to the vessel traffic speed and other outlier AIS data.(2)Calculation of vessel traffic speed: After preprocessing AIS data for all segments of the target area,calculate the vessel traffic speed for each segment using the above AIS data.The vessel traffic speed is calculated using a spatial average speed and a time interval of 60 minutes.(3)Constructing vessel traffic speed prediction model: This article proposes a TCN-RBFNN composite model based on Temporal Convolutional Network(TCN)and Radial Basis Function Neural Network(RBFNN),which can effectively obtain the spatio-temporal characteristics of vessel traffic flow speed.Additionally,TCN-RBFNN composite model is used to construct the vessel traffic flow speed prediction model using data from target area.(4)Spatio-temporal characteristics analysis and vessel traffic speed prediction in target area: Conduct spatio-temporal characteristics analysis of vessel traffic flow speed in different sub regions.Furthermore,the predictive performance and generalization ability of the TCN-RBFNN composite model were validated using ship traffic flow speed data in the upstream directions of regions A,D,and F region.LSTM,Prophet,and TCN models were also introduced for comparative experiments.This thesis constructs a TCN-RBFNN composite model that comprehensively considers spatio-temporal characteristics to predict vessel traffic flow speed based on the characteristics both in temporal and spatial dimensions.The experimental results show that the TCN-RBFNN composite model outperforms other comparative models in performance evaluation indicators,verifying the excellent performance of the TCN-RBFNN composite model proposed in this thesis in terms of prediction accuracy,generalization ability,and stability,At the same time,the effectiveness of constructing a prediction model considering the spatio-temporal characteristics of vessel traffic flow was also verified.