{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/101628"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/101628","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Learning low-dimensional feature dynamics using convolutional recurrent autoencoders","abstract":"Model reduction of high-dimensional dynamical systems alleviates computational burdens faced in various tasks from design optimization to model predictive control. One popular model reduction approach is based on projecting the governing equations onto a subspace spanned by basis functions obtained from the compression of a dataset of solution snapshots. However, this method is intrusive since the projection requires access to the system operators. Further, some systems may require special treatment of nonlinearities to ensure computational efficiency or additional modeling to preserve stability. In this work we propose a deep learning-based strategy for nonlinear model reduction that is inspired by projection-based model reduction where the idea is to identify some optimal low-dimensional representation and evolve it in time. Our approach constructs a modular model consisting of a deep convolutional autoencoder and a modified LSTM network. The deep convolutional autoencoder returns a low-dimensional representation in terms of coordinates on some expressive nonlinear data-supporting manifold. The dynamics on this manifold are then modeled by the modified LSTM network in a computationally efficient manner. An offline training strategy that exploits the model modularity is also developed. We demonstrate our model on three illustrative examples each highlighting the model's performance in prediction tasks for systems with large parameter-variations and its stability in long-term prediction.","abstract_html":"Model reduction of high-dimensional dynamical systems alleviates computational burdens faced in various tasks from design optimization to model predictive control. One popular model reduction approach is based on projecting the governing equations onto a subspace spanned by basis functions obtained from the compression of a dataset of solution snapshots. However, this method is intrusive since the projection requires access to the system operators. Further, some systems may require special treatment of nonlinearities to ensure computational efficiency or additional modeling to preserve stability. In this work we propose a deep learning-based strategy for nonlinear model reduction that is inspired by projection-based model reduction where the idea is to identify some optimal low-dimensional representation and evolve it in time. Our approach constructs a modular model consisting of a deep convolutional autoencoder and a modified LSTM network. The deep convolutional autoencoder returns a low-dimensional representation in terms of coordinates on some expressive nonlinear data-supporting manifold. The dynamics on this manifold are then modeled by the modified LSTM network in a computationally efficient manner. An offline training strategy that exploits the model modularity is also developed. We demonstrate our model on three illustrative examples each highlighting the model's performance in prediction tasks for systems with large parameter-variations and its stability in long-term prediction.","abstract_has_math":false,"creators":["Gonzalez, Francisco Javier"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Aerospace Engineering","degree_department":null,"school":null,"contributors":["Balajewicz, Maciej"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2018,"date_issued":"2018-09-27T16:18:03Z","date_published":"2018-09-27T16:18:03Z","updated_at":"2026-07-22T22:24:40Z","subjects":["Nonlinear model reduction","deep learning","dynamical systems","data-driven modeling","convolutional neural networks","recurrent neural networks"],"languages":["en"],"rights":["Copyright 2018 Francisco J. Gonzalez"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/101628","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Balajewicz, Maciej"]},{"key":"dc:creator","label":"Author","values":["Gonzalez, Francisco Javier"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2018-09-27T16:18:03Z","2018-07-20","2018-08"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Aerospace Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Nonlinear model reduction","deep learning","dynamical systems","data-driven modeling","convolutional neural networks","recurrent neural networks"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2018 Francisco J. Gonzalez"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/101628"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Model reduction of high-dimensional dynamical systems alleviates computational burdens faced in various tasks from design optimization to model predictive control. One popular model reduction approach is based on projecting the governing equations onto a subspace spanned by basis functions obtained from the compression of a dataset of solution snapshots. However, this method is intrusive since the projection requires access to the system operators. Further, some systems may require special treatment of nonlinearities to ensure computational efficiency or additional modeling to preserve stability. In this work we propose a deep learning-based strategy for nonlinear model reduction that is inspired by projection-based model reduction where the idea is to identify some optimal low-dimensional representation and evolve it in time. Our approach constructs a modular model consisting of a deep convolutional autoencoder and a modified LSTM network. The deep convolutional autoencoder returns a low-dimensional representation in terms of coordinates on some expressive nonlinear data-supporting manifold. The dynamics on this manifold are then modeled by the modified LSTM network in a computationally efficient manner. An offline training strategy that exploits the model modularity is also developed. We demonstrate our model on three illustrative examples each highlighting the model's performance in prediction tasks for systems with large parameter-variations and its stability in long-term prediction.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2018-09-27 without embargo terms","The student, Francisco Gonzalez, accepted the attached license on 2018-07-20 at 14:23.","The student, Francisco Gonzalez, submitted this Thesis for approval on 2018-07-20 at 14:26.","This Thesis was approved for publication on 2018-07-20 at 16:16.","DSpace SAF Submission Ingestion Package generated from Vireo submission #12947 on 2018-09-27 at 10:50:24","Made available in DSpace on 2018-09-27T16:18:03Z (GMT). No. of bitstreams: 2 GONZALEZ-THESIS-2018.pdf: 1173299 bytes, checksum: 31551929f5965de5e8d4a8bcf6c40327 (MD5) LICENSE.txt: 4215 bytes, checksum: a194c799d59bc2c8e2993333c076a402 (MD5) Previous issue date: 2018-07-20"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Learning low-dimensional feature dynamics using convolutional recurrent autoencoders"]}]}],"canonical_facts":{"dc:contributor":["Balajewicz, Maciej"],"dc:creator":["Gonzalez, Francisco Javier"],"dc:date":["2018-09-27T16:18:03Z","2018-07-20","2018-08"],"dc:description":["Model reduction of high-dimensional dynamical systems alleviates computational burdens faced in various tasks from design optimization to model predictive control. One popular model reduction approach is based on projecting the governing equations onto a subspace spanned by basis functions obtained from the compression of a dataset of solution snapshots. However, this method is intrusive since the projection requires access to the system operators. Further, some systems may require special treatment of nonlinearities to ensure computational efficiency or additional modeling to preserve stability. In this work we propose a deep learning-based strategy for nonlinear model reduction that is inspired by projection-based model reduction where the idea is to identify some optimal low-dimensional representation and evolve it in time. Our approach constructs a modular model consisting of a deep convolutional autoencoder and a modified LSTM network. The deep convolutional autoencoder returns a low-dimensional representation in terms of coordinates on some expressive nonlinear data-supporting manifold. The dynamics on this manifold are then modeled by the modified LSTM network in a computationally efficient manner. An offline training strategy that exploits the model modularity is also developed. We demonstrate our model on three illustrative examples each highlighting the model's performance in prediction tasks for systems with large parameter-variations and its stability in long-term prediction.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2018-09-27 without embargo terms","The student, Francisco Gonzalez, accepted the attached license on 2018-07-20 at 14:23.","The student, Francisco Gonzalez, submitted this Thesis for approval on 2018-07-20 at 14:26.","This Thesis was approved for publication on 2018-07-20 at 16:16.","DSpace SAF Submission Ingestion Package generated from Vireo submission #12947 on 2018-09-27 at 10:50:24","Made available in DSpace on 2018-09-27T16:18:03Z (GMT). No. of bitstreams: 2 GONZALEZ-THESIS-2018.pdf: 1173299 bytes, checksum: 31551929f5965de5e8d4a8bcf6c40327 (MD5) LICENSE.txt: 4215 bytes, checksum: a194c799d59bc2c8e2993333c076a402 (MD5) Previous issue date: 2018-07-20"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/101628"],"dc:language":["en"],"dc:rights":["Copyright 2018 Francisco J. Gonzalez"],"dc:subject":["Nonlinear model reduction","deep learning","dynamical systems","data-driven modeling","convolutional neural networks","recurrent neural networks"],"dc:title":["Learning low-dimensional feature dynamics using convolutional recurrent autoencoders"],"dc:type":["text"],"thesis:degree_discipline":["Aerospace Engineering"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:40Z"}