Digital ship operation is just around the corners with the rapid development of Artificial Intelligence(AI)and Industrial Internet of Things(IIoT)technologies.Real time condition monitoring and Predictive Maintenance(...Digital ship operation is just around the corners with the rapid development of Artificial Intelligence(AI)and Industrial Internet of Things(IIoT)technologies.Real time condition monitoring and Predictive Maintenance(PdM)of marine diesel engines are crucial to realize the success of ship digital operations.The study investigates the PdM in two-stroke marine diesel engines using Machine Learning(ML)and Machine Learning Operations(MLOps)based on engine operational data.Practical data with labeled engine scuffing incidents are collected from a shipping company.The real scuffing incidents are predicted based on the expected operational behavior modeling method and a customized framework.Three case studies are conducted based on 2 different vessels for the purpose of model validations and further investigation.During the expected behavior modeling procedure,comparisons among different ML models accounting for various parameters(E.g.,targets,operational features,moving average types and widths)are conducted and sensitivity studies are performed in order to identify the best solutions for engine PdM in shipping practice.Based on the study,the model effectiveness and efficiency are demonstrated and a limited generalization ability of the expected behavior modeling method with ML has been realized,which can facilitate the alarming and scheduling of maintenance events for vessels.The models and findings from this research work can be easily adapted for possible future use in ship operations.展开更多
Artificial intelligence(AI)relies on data and algorithms.State-of-the-art(SOTA)AI smart algorithms have been developed to improve the performance of AI-oriented structures.However,model-centric approaches are limited ...Artificial intelligence(AI)relies on data and algorithms.State-of-the-art(SOTA)AI smart algorithms have been developed to improve the performance of AI-oriented structures.However,model-centric approaches are limited by the absence of high-quality data.Data-centric AI is an emerging approach for solving machine learning(ML)problems.It is a collection of various data manipulation techniques that allow ML practitioners to systematically improve the quality of the data used in an ML pipeline.However,data-centric AI approaches are not well documented.Researchers have conducted various experiments without a clear set of guidelines.This survey highlights six major data-centric AI aspects that researchers are already using to intentionally or unintentionally improve the quality of AI systems.These include big data quality assessment,data preprocessing,transfer learning,semi-supervised learning,machine learning operations(MLOps),and the effect of adding more data.In addition,it highlights recent data-centric techniques adopted by ML practitioners.We addressed how adding data might harm datasets and how HoloClean can be used to restore and clean them.Finally,we discuss the causes of technical debt in AI.Technical debt builds up when software design and implementation decisions run into“or outright collide with”business goals and timelines.This survey lays the groundwork for future data-centric AI discussions by summarizing various data-centric approaches.展开更多
基金The work of this paper was conducted under the WP3 of the SLGreen project:“Digital Twins for engine condition monitoring and wear prediction”The project is funded by the Innovation Fund Den-mark(IFD)under File No.3149-00017B,the Danish Maritime Fund,and the Lauritzen FondenThe use of the Amazon SageMaker tool is financially supported by AWS.
文摘Digital ship operation is just around the corners with the rapid development of Artificial Intelligence(AI)and Industrial Internet of Things(IIoT)technologies.Real time condition monitoring and Predictive Maintenance(PdM)of marine diesel engines are crucial to realize the success of ship digital operations.The study investigates the PdM in two-stroke marine diesel engines using Machine Learning(ML)and Machine Learning Operations(MLOps)based on engine operational data.Practical data with labeled engine scuffing incidents are collected from a shipping company.The real scuffing incidents are predicted based on the expected operational behavior modeling method and a customized framework.Three case studies are conducted based on 2 different vessels for the purpose of model validations and further investigation.During the expected behavior modeling procedure,comparisons among different ML models accounting for various parameters(E.g.,targets,operational features,moving average types and widths)are conducted and sensitivity studies are performed in order to identify the best solutions for engine PdM in shipping practice.Based on the study,the model effectiveness and efficiency are demonstrated and a limited generalization ability of the expected behavior modeling method with ML has been realized,which can facilitate the alarming and scheduling of maintenance events for vessels.The models and findings from this research work can be easily adapted for possible future use in ship operations.
文摘Artificial intelligence(AI)relies on data and algorithms.State-of-the-art(SOTA)AI smart algorithms have been developed to improve the performance of AI-oriented structures.However,model-centric approaches are limited by the absence of high-quality data.Data-centric AI is an emerging approach for solving machine learning(ML)problems.It is a collection of various data manipulation techniques that allow ML practitioners to systematically improve the quality of the data used in an ML pipeline.However,data-centric AI approaches are not well documented.Researchers have conducted various experiments without a clear set of guidelines.This survey highlights six major data-centric AI aspects that researchers are already using to intentionally or unintentionally improve the quality of AI systems.These include big data quality assessment,data preprocessing,transfer learning,semi-supervised learning,machine learning operations(MLOps),and the effect of adding more data.In addition,it highlights recent data-centric techniques adopted by ML practitioners.We addressed how adding data might harm datasets and how HoloClean can be used to restore and clean them.Finally,we discuss the causes of technical debt in AI.Technical debt builds up when software design and implementation decisions run into“or outright collide with”business goals and timelines.This survey lays the groundwork for future data-centric AI discussions by summarizing various data-centric approaches.
