[ { "id": "authors:cyjra-0jy93", "collection": "authors", "collection_id": "cyjra-0jy93", "cite_using_url": "https://resolver.caltech.edu/CaltechAUTHORS:20180312-114520324", "type": "article", "title": "A Bayesian Learning Method for Structural Damage Assessment of Phase I IASC-ASCE Benchmark Problem", "author": [ { "family_name": "Oh", "given_name": "Chang Kook", "clpid": "Oh-Chang-Kook" }, { "family_name": "Beck", "given_name": "James L.", "clpid": "Beck-J-L" } ], "abstract": "Rapid progress in the field of sensor technology leads to acquisition of massive amounts of measured data from structures being monitored. The data, however, contains inevitable measurement errors which often cause quantitative damage assessment to be ill-conditioned. The Bayesian learning method is well known to provide effective ways to alleviate the ill-conditioning through the prior term for regularization and to provide meaningful probabilistic results for reliable decision-making at the same time. In this study, the Bayesian learning method, based on the Bayesian regression approach using the automatic relevance determination prior, is presented to achieve more effective regularization as well as probabilistic prediction and it is expanded to provide vector outputs for monitoring of a Phase I IASC-ASCE simulated benchmark problem. The proposed method successfully estimates damage locations as well as its severities and give considerable promise for structural damage assessment.", "doi": "10.1007/s12205-018-1290-1", "issn": "1226-7988", "publisher": "Korean Society of Civil Engineers", "publication": "KSCE Journal of Civil Engineering", "publication_date": "2018-03", "series_number": "3", "volume": "22", "issue": "3", "pages": "987-992" }, { "id": "authors:fhhbc-gxr28", "collection": "authors", "collection_id": "fhhbc-gxr28", "cite_using_url": "https://resolver.caltech.edu/CaltechAUTHORS:OHCjem08", "type": "article", "title": "Bayesian Learning Using Automatic Relevance Determination Prior with an Application to Earthquake Early Warning", "author": [ { "family_name": "Oh", "given_name": "Chang Kook", "clpid": "Oh-Chang-Kook" }, { "family_name": "Beck", "given_name": "James L.", "clpid": "Beck-J-L" }, { "family_name": "Yamada", "given_name": "Masumi", "clpid": "Yamada-Masumi" } ], "abstract": "A novel method of Bayesian learning with automatic relevance determination prior is presented that provides a powerful approach to problems of classification based on data features, for example, classifying soil liquefaction potential based on soil and seismic shaking parameters, automatically classifying the damage states of a structure after severe loading based on features of its dynamic response, and real-time classification of earthquakes based on seismic signals. After introduction of the theory, the method is illustrated by applying it to an earthquake record dataset from nine earthquakes to build an efficient real-time algorithm for near-source versus far-source classification of incoming seismic ground motion signals. This classification is needed in the development of early warning systems for large earthquakes. It is shown that the proposed methodology is promising since it provides a classifier with higher correct classification rates and better generalization performance than a previous Bayesian learning method with a fixed prior distribution that was applied to the same classification problem.", "doi": "10.1061/(ASCE)0733-9399(2008)134:12(1013)", "issn": "0733-9399", "publisher": "American Society of Civil Engineers", "publication": "Journal of Engineering Mechanics", "publication_date": "2008-12", "series_number": "12", "volume": "134", "issue": "12", "pages": "1013-1020" }, { "id": "authors:nr6rs-94825", "collection": "authors", "collection_id": "nr6rs-94825", "cite_using_url": "https://resolver.caltech.edu/CaltechEERL:EERL-2007-04", "type": "monograph", "title": "Bayesian Learning for Earthquake Engineering Applications and Structural Health Monitoring", "author": [ { "family_name": "Oh", "given_name": "Chang Kook", "clpid": "Oh-Chang-Kook" } ], "abstract": "Parallel to significant advances in sensor hardware, there have been recent developments\nof sophisticated methods for quantitative assessment of measured data that\nexplicitly deal with all of the involved uncertainties, including inevitable measurement\nerrors. The existence of these uncertainties often causes numerical instabilities\nin inverse problems that make them ill-conditioned.\nThe Bayesian methodology is known to provide an efficient way to alleviate this illconditioning\nby incorporating the prior term for regularization of the inverse problem,\nand to provide probabilistic results which are meaningful for decision making.\nIn this work, the Bayesian methodology is applied to inverse problems in earthquake\nengineering and especially to structural health monitoring. The proposed\nmethodology of Bayesian learning using automatic relevance determination (ARD)\nprior, including its kernel version called the Relevance Vector Machine, is presented\nand applied to earthquake early warning, earthquake ground motion attenuation estimation,\nand structural health monitoring, using either a Bayesian classification or\nregression approach.\nThe classification and regression are both performed in three phases: (1) Phase\nI (feature extraction phase): Determine which features from the data to use in a\ntraining dataset; (2) Phase II (training phase): Identify the unknown parameters\ndefining a model by using a training dataset; and (3) Phase III (prediction phase):\nPredict the results based on the features from new data.\nThis work focuses on the advantages of making probabilistic predictions obtained\nby Bayesian methods to deal with all uncertainties and the good characteristics of\nthe proposed method in terms of computationally efficient training, and, especially,\nvi\nprediction that make it suitable for real-time operation. It is shown that sparseness\n(using only smaller number of basis function terms) is produced in the regression\nequations and classification separating boundary by using the ARD prior along with\nBayesian model class selection to select the most probable (plausible) model class\nbased on the data. This model class selection procedure automatically produces\noptimal regularization of the problem at hand, making it well-conditioned.\nSeveral applications of the proposed Bayesian learning methodology are presented.\nFirst, automatic near-source and far-source classification of incoming ground motion\nsignals is treated and the Bayesian learning method is used to determine which ground\nmotion features are optimal for this classification. Second, a probabilistic earthquake\nattenuation model for peak ground acceleration is identified using selected optimal\nfeatures, especially taking a non-linearly involved parameter into consideration. It is\nshown that the Bayesian learning method can be utilized to estimate not only linear\ncoefficients but also a non-linearly involved parameter to provide an estimate for\nan unknown parameter in the kernel basis functions for Relevance Vector Machine.\nThird, the proposed method is extended to a general case of regression problems\nwith vector outputs and applied to structural health monitoring applications. It\nis concluded that the proposed vector output RVM shows promise for estimating\ndamage locations and their severities from change of modal properties such as natural\nfrequencies and mode shapes.", "publisher": "Earthquake Engineering Research Laboratory", "publication_date": "2007-09-17" } ]