Research On The Spatial-temporal Analysis Of Ship CO2 Emissions In Port Based On AIS Data

With the prosperous development of maritime transportation,the carbon pollution generated by ships is growing at a faster rate.Carrying out the study of carbon emission from port ships is conducive to the green,low-carbon and efficient development of China’s ports.The carbon emission characteristics of ships contain multiple dimensions such as time,space,ship type and working conditions,and there is a certain coupling relationship between the data of each dimension.At present,the existing carbon emission characteristics of ships are often analyzed only from a single dimension,ignoring the multidimensionality and co-occurrence in time of the carbon emission process.Therefore,it is necessary to find an effective spatio-temporal analysis technique for ship emission data,to jointly analyze ship emission in multiple dimensions,and to obtain multidimensional ship carbon emission characteristics to help ship carbon emission reduction research.In this thesis,we first measure the carbon emissions from port ships.The ship activity state is divided according to the AIS data,the ship main engine power is determined using polynomial fitting,and the ship equipment information such as emission factors is supplemented to complete the calculation of ship carbon emissions.Subsequently,a rule of carbon emission information discretization is specified,and the study area is divided into a grid,with the grid number indicating the spatial location;the time is divided into several intervals,with the interval number indicating the time;the ship type and activity state are also numbered separately,and the carbon emission data of the ship after discretization is a data set of ship type,time,speed and the number of the regional grid,which is also the carbon emission label data set.To extract multidimensional ship carbon emission features,this thesis uses non-negative tensor decomposition method for port ship carbon emission spatio-temporal features.Tensor is a multidimensional data structure,which has the advantages of preventing the spatial structure and internal potential information of the original data from being destroyed,etc.Non-negative tensor decomposition can effectively extract the potential components.Firstly,a mapping rule is prescribed to construct a three-dimensional ship carbon emission tensor using carbon emission labeling dataset with three dimensions of region,time,ship type and activity state respectively.Then the influence of the tensor decomposition rank on the decomposition results is discussed to determine the optimal value of the decomposition rank to extract the three-dimension ship carbon emission features.The operating ship in Tianjin port in May 2021 is taken as the case study,in order to verify the effectiveness of the feature extract method proposed in this thesis.The thesis analyzes the differences between the features obtained when the tensor decomposition rank is 5、6 and 7,and finally chooses the decomposition rank of 6.By decomposing the three-dimensional carbon emission tensor of Tianjin port,six main carbon emission features are obtained,which are five carbon emission features mainly for ships in docking status and one feature for tug-assisted container ships and cargo ships entering and leaving the port area.These features determine the ship type and activity status of the main CO₂ emitting ships in the Tianjin port area,and the area and time of the respective activities of these ships.Without prior knowledge,the characteristics obtained by the method in this thesis are consistent with the actual situation,which verifies the effectiveness of the proposed method.Compared with one-dimensional analysis,the carbon emission feature extraction method proposed in this thesis can obtain more detailed carbon emission information of ships.This study focuses on the multi-dimension of ship carbon emissions,and proposes an effective spatio-temporal analysis method for ship carbon emissions.The multi-dimensional analysis of ship emissions provides more accurate carbon emission information.A beneficial exploration is made in the temporal and spatial analysis of port ship carbon emissions.