{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/fleet-prognosis-with-physics-informed","title":"Fleet Prognosis with Physics-informed Recurrent Neural Networks","arxiv_id":"1901.05512","date":"2019-01-16","proceeding":null,"authors":["Renato Giorgiani Nascimento","Felipe A. C. Viana"],"abstract":"Services and warranties of large fleets of engineering assets is a very\nprofitable business. The success of companies in that area is often related to\npredictive maintenance driven by advanced analytics. Therefore, accurate\nmodeling, as a way to understand how the complex interactions between operating\nconditions and component capability define useful life, is key for services\nprofitability. Unfortunately, building prognosis models for large fleets is a\ndaunting task as factors such as duty cycle variation, harsh environments,\ninadequate maintenance, and problems with mass production can lead to large\ndiscrepancies between designed and observed useful lives. This paper introduces\na novel physics-informed neural network approach to prognosis by extending\nrecurrent neural networks to cumulative damage models. We propose a new\nrecurrent neural network cell designed to merge physics-informed and\ndata-driven layers. With that, engineers and scientists have the chance to use\nphysics-informed layers to model parts that are well understood (e.g., fatigue\ncrack growth) and use data-driven layers to model parts that are poorly\ncharacterized (e.g., internal loads). A simple numerical experiment is used to\npresent the main features of the proposed physics-informed recurrent neural\nnetwork for damage accumulation. The test problem consist of predicting fatigue\ncrack length for a synthetic fleet of airplanes subject to different mission\nmixes. The model is trained using full observation inputs (far-field loads) and\nvery limited observation of outputs (crack length at inspection for only a\nportion of the fleet). The results demonstrate that our proposed hybrid\nphysics-informed recurrent neural network is able to accurately model fatigue\ncrack growth even when the observed distribution of crack length does not match\nwith the (unobservable) fleet distribution.","url_abs":"http://arxiv.org/abs/1901.05512v1","url_pdf":"http://arxiv.org/pdf/1901.05512v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"fleet-prognosis-with-physics-informed","repo_url":"https://github.com/PML-UCF/pinn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"graph-regression","task_name":"Graph Regression"},{"task_slug":"graph-to-sequence","task_name":"Graph-to-Sequence"},{"task_slug":"physics-informed-machine-learning","task_name":"Physics-informed machine learning"},{"task_slug":"prognosis","task_name":"Prognosis"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}