{"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/physics-guided-convolutional-neural-network","title":"Physics-guided Convolutional Neural Network (PhyCNN) for Data-driven Seismic Response Modeling","arxiv_id":"1909.08118","date":"2019-09-17","proceeding":null,"authors":[],"abstract":"Seismic events, among many other natural hazards, reduce due functionality\nand exacerbate vulnerability of in-service buildings. Accurate modeling and\nprediction of building's response subjected to earthquakes makes possible to\nevaluate building performance. To this end, we leverage the recent advances in\ndeep learning and develop a physics-guided convolutional neural network\n(PhyCNN) framework for data-driven seismic response modeling and serviceability\nassessment of buildings. The proposed PhyCNN approach is capable of accurately\npredicting building's seismic response in a data-driven fashion without the\nneed of a physics-based analytical/numerical model. The basic concept is to\ntrain a deep PhyCNN model based on available seismic input-output datasets\n(e.g., from simulation or sensing) and physics constraints. The trained PhyCNN\ncan then used as a surrogate model for structural seismic response prediction.\nAvailable physics (e.g., the law of dynamics) can provide constraints to the\nnetwork outputs, alleviate overfitting issues, reduce the need of big training\ndatasets, and thus improve the robustness of the trained model for more\nreliable prediction. The trained surrogate model is then utilized for fragility\nanalysis given certain limit state criteria (e.g., the serviceability state).\nIn addition, an unsupervised learning algorithm based on K-means clustering is\nalso proposed to partition the limited number of datasets to training,\nvalidation and prediction categories, so as to maximize the use of limited\ndatasets. The performance of the proposed approach is demonstrated through\nthree case studies including both numerical and experimental examples.\nConvincing results illustrate that the proposed PhyCNN paradigm outperforms\nconventional pure data-based neural networks.","url_abs":"http://arxiv.org/abs/1909.08118v1","url_pdf":"http://arxiv.org/pdf/1909.08118v1.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":"physics-guided-convolutional-neural-network","repo_url":"https://github.com/zhry10/PhyCNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"prediction","task_name":"Prediction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}