This study developed a new methodology for analyzing the risk level of marine spill accidents from two perspectives,namely,marine traffic density and sensitive resources.Through a case study conducted in Busan,South K...This study developed a new methodology for analyzing the risk level of marine spill accidents from two perspectives,namely,marine traffic density and sensitive resources.Through a case study conducted in Busan,South Korea,detailed procedures of the methodology were proposed and its scalability was confirmed.To analyze the risk from a more detailed and microscopic viewpoint,vessel routes as hazard sources were delineated on the basis of automated identification system(AIS)big data.The outliers and errors of AIS big data were removed using the density-based spatial clustering of applications with noise algorithm,and a marine traffic density map was evaluated by combining all of the gridded routes.Vulnerability of marine environment was identified on the basis of the sensitive resource map constructed by the Korea Coast Guard in a similar manner to the National Oceanic and Atmospheric Administration environmental sensitivity index approach.In this study,aquaculture sites,water intake facilities of power plants,and beach/resort areas were selected as representative indicators for each category.The vulnerability values of neighboring cells decreased according to the Euclidean distance from the resource cells.Two resulting maps were aggregated to construct a final sensitive resource and traffic density(SRTD)risk analysis map of the Busan–Ulsan sea areas.We confirmed the effectiveness of SRTD risk analysis by comparing it with the actual marine spill accident records.Results show that all of the marine spill accidents in 2018 occurred within 2 km of high-risk cells(level 6 and above).Thus,if accident management and monitoring capabilities are concentrated on high-risk cells,which account for only 6.45%of the total study area,then it is expected that it will be possible to cope with most marine spill accidents effectively.展开更多
Road accidents are one of the most relevant causes of injuries and death worldwide,and therefore,they constitute a significant field of research on the use of advanced algorithms and techniques to analyze and predict ...Road accidents are one of the most relevant causes of injuries and death worldwide,and therefore,they constitute a significant field of research on the use of advanced algorithms and techniques to analyze and predict traffic accidents and determine the most relevant elements that contribute to road accidents.The research of road accident prediction aims to respond to the challenge of offer tools to generate a more secure mobility environment,and ultimately,save lives.This paper aims to provide an overview of the state of the art in the prediction of road accidents through machine learning algorithms and advanced techniques for analyzing information,such as convolutional neural networks and long short-term memory networks,among other deep learning architectures.Furthermore,in this article,a compendium and study of the most used data sources for the road accident forecast is made.And a classification is proposed according to its origin and characteristics,such as open data,measurement technologies,onboard equipment and social media data.For the analysis of the information,the different algorithms employed to make predictions about road accidents are listed and compared,as well as their applicability depending on the types of data being analyzed,along with the results obtained and their ease of interpretation and analysis.The best results reported by the authors are obtained when two or more analytic techniques are combined,in such a way that analysis of the obtained results is strengthened.Among the future challenges in road traffic forecasting lies the enhancement of the scope of the proposed models and predictions by the incorporation of heterogeneous data sources,that include geo spatial data,information from traffic volume,traffic statistics,video,sound,text and sentiment from social media,that many authors concur that can improve the precision and accuracy of the analysis and predictions.展开更多
基金This research was supported by a grant[KCG-01-2017-01]through the Disaster and Safety Management Institute funded by the Ministry of Public Safety and Securitythe National Research Foundation of Korea(NRF)grant[No.2018R1D1A1B07050208]funded by the Ministry of Science and ICT of Korea Government.
文摘This study developed a new methodology for analyzing the risk level of marine spill accidents from two perspectives,namely,marine traffic density and sensitive resources.Through a case study conducted in Busan,South Korea,detailed procedures of the methodology were proposed and its scalability was confirmed.To analyze the risk from a more detailed and microscopic viewpoint,vessel routes as hazard sources were delineated on the basis of automated identification system(AIS)big data.The outliers and errors of AIS big data were removed using the density-based spatial clustering of applications with noise algorithm,and a marine traffic density map was evaluated by combining all of the gridded routes.Vulnerability of marine environment was identified on the basis of the sensitive resource map constructed by the Korea Coast Guard in a similar manner to the National Oceanic and Atmospheric Administration environmental sensitivity index approach.In this study,aquaculture sites,water intake facilities of power plants,and beach/resort areas were selected as representative indicators for each category.The vulnerability values of neighboring cells decreased according to the Euclidean distance from the resource cells.Two resulting maps were aggregated to construct a final sensitive resource and traffic density(SRTD)risk analysis map of the Busan–Ulsan sea areas.We confirmed the effectiveness of SRTD risk analysis by comparing it with the actual marine spill accident records.Results show that all of the marine spill accidents in 2018 occurred within 2 km of high-risk cells(level 6 and above).Thus,if accident management and monitoring capabilities are concentrated on high-risk cells,which account for only 6.45%of the total study area,then it is expected that it will be possible to cope with most marine spill accidents effectively.
基金the Universidad Nacional de Colombia,funding call“Convocatoria Nacional Para el Apoyo a Proyectos de Investigacion y Creacion Artistica de la Universidad Nacional De Colombia 2017-2018”,project code 41614。
文摘Road accidents are one of the most relevant causes of injuries and death worldwide,and therefore,they constitute a significant field of research on the use of advanced algorithms and techniques to analyze and predict traffic accidents and determine the most relevant elements that contribute to road accidents.The research of road accident prediction aims to respond to the challenge of offer tools to generate a more secure mobility environment,and ultimately,save lives.This paper aims to provide an overview of the state of the art in the prediction of road accidents through machine learning algorithms and advanced techniques for analyzing information,such as convolutional neural networks and long short-term memory networks,among other deep learning architectures.Furthermore,in this article,a compendium and study of the most used data sources for the road accident forecast is made.And a classification is proposed according to its origin and characteristics,such as open data,measurement technologies,onboard equipment and social media data.For the analysis of the information,the different algorithms employed to make predictions about road accidents are listed and compared,as well as their applicability depending on the types of data being analyzed,along with the results obtained and their ease of interpretation and analysis.The best results reported by the authors are obtained when two or more analytic techniques are combined,in such a way that analysis of the obtained results is strengthened.Among the future challenges in road traffic forecasting lies the enhancement of the scope of the proposed models and predictions by the incorporation of heterogeneous data sources,that include geo spatial data,information from traffic volume,traffic statistics,video,sound,text and sentiment from social media,that many authors concur that can improve the precision and accuracy of the analysis and predictions.
