| Dalian Port is an important hub of the national central port and comprehensive threedimensional transportation network.While the shipping economy is thriving and developing,it has also become a huge carbon emitter,facing severe emission reduction pressure.It is urgent to further accelerate the decarbonization process of Dalian Port.In recent years,with the continuous improvement of ship technology and oil quality,ship emissions have undergone significant changes.Existing research is not only far from now but also cannot fully reflect the characteristics of carbon emissions pollution from ships in Dalian Port Area.Therefore,it is necessary to conduct in-depth research on the calculation of carbon emissions from port ships.Based on AIS data and Lloyd’s Register data,this thesis calculates carbon dioxide emissions around several core issues of carbon dioxide emissions from ships in the sea area of Dalian Port,studies the spatial and temporal distribution of carbon emissions using geographic information system,spatial analysis and other technologies,and explores the model of in-depth learning to achieve highly accurate spatial and temporal prediction of carbon emissions.Strive to obtain information that is conducive to the carbon emission reduction strategy of port waters from the spatio-temporal analysis and prediction of multi-source data,and help the country achieve the goals of "carbon peak" and "carbon neutrality".This thesis focuses on the carbon dioxide emissions from ships in Dalian Port,and carries out the following research:(1)Research on the calculation of ship carbon emissions in Port of Dalian.The measurement and estimation of ship carbon emissions in Port of Dalian is the premise and basis of carbon emission related research.A complete set of Data cleansing methods has been established through data processing of time-space Big data such as AIS data and Lloyd’s Register data,and then carbon emissions of Port of Dalian have been calculated based on data driven technology.(2)Research on temporal and spatial distribution of ship carbon emissions in Port of Dalian.Based on the calculation of carbon emissions,by analyzing the characteristics of different ship types,emission sources,and operating conditions,we can grasp the carbon emission patterns of different ships;By analyzing the characteristics of ship carbon emissions in different months,days,and hours,grasp the temporal emission patterns.By using spatial pattern analysis technology to study the spatial distribution characteristics of carbon emissions and grasp the spatial patterns of carbon emissions.(3)Research on the future trend of ship carbon emissions in Port of Dalian.Based on the data Rasterisation and spatial identification methods in GIS,the carbon emission inventory is used to build a spatio-temporal dataset,and the dataset is divided,normalized,and constructed a sliding time window.Using the Tensor Flow neural network framework to construct a neural network model,this study investigates the accuracy and effectiveness of three-dimensional convolutional models,Convolutional Neural Networks Long Short Term Memory(Conv LSTM)models,and Self Attention Conv LSTM(SA-Conv LSTM)models with self-attention mechanism in predicting carbon emissions in this spatiotemporal sequence.To sum up,this paper takes the carbon emissions of ships in Port of Dalian area as the research object,uses the bottom-up power method based on ship activities,builds a carbon emissions calculation model based on AIS data and Lloyd’s Register database,and calculates the carbon emissions volume.The spatial pattern method was applied to study the spatiotemporal distribution characteristics of carbon emissions.We studied the construction methods of spatiotemporal datasets,implemented different spatiotemporal prediction models using Tensor Flow,and explored the effectiveness of deep learning spatiotemporal sequence models for prediction.This work can provide reference and guidance for future carbon emission prediction research,and improve the accuracy and reliability of carbon emission prediction. |
