M. Asadi, M. Mohseni, M. Kashani, Michael Fernández, Mathew Smith
{"title":"结合机器学习的焊接仿真训练元模型,用于快速探索各种焊接顺序场景","authors":"M. Asadi, M. Mohseni, M. Kashani, Michael Fernández, Mathew Smith","doi":"10.1115/pvp2019-93672","DOIUrl":null,"url":null,"abstract":"\n Distortion is a common problem in welded structures, and therefore welding standards require a mitigation plan to be in place before welding. When dealing with multiple welds, an optimal intermittent weld sequence can effectively minimize the distortion by counter-balancing the transient distortion during welding. However, the process of finding an effective weld sequence is a challenging task given a large number of possible combinations, i.e. several thousand for a few welds. As an acceptable approach, welding simulation tools allow engineers to optimize a welding sequence without the need for multiple physical samples. Despite efficient simulation tools and powerful supercomputers, yet simulation tools have been limited by CPU time and therefore not mature for practical designs. To this end, we constructed and integrated an inexpensive low-fidelity machine learning (ML) algorithm with the expensive high-fidelity simulation. This ML model was then trained to increase the fidelity by a wisely chosen train set of simulation to construct a meta-model for active exploration of various weld sequence scenarios. As opposed to existing ML algorithms that require an extensive data set to train, our algorithm picks relatively small training set to construct a meta-model. We present an example of our algorithm implemented in a real welded structure project.","PeriodicalId":23651,"journal":{"name":"Volume 6B: Materials and Fabrication","volume":"1 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2019-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":"{\"title\":\"Welding Simulation Integrated With Machine Learning to Train a Meta-Model for Fast Exploration of Various Weld Sequence Scenarios\",\"authors\":\"M. Asadi, M. Mohseni, M. Kashani, Michael Fernández, Mathew Smith\",\"doi\":\"10.1115/pvp2019-93672\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"\\n Distortion is a common problem in welded structures, and therefore welding standards require a mitigation plan to be in place before welding. When dealing with multiple welds, an optimal intermittent weld sequence can effectively minimize the distortion by counter-balancing the transient distortion during welding. However, the process of finding an effective weld sequence is a challenging task given a large number of possible combinations, i.e. several thousand for a few welds. As an acceptable approach, welding simulation tools allow engineers to optimize a welding sequence without the need for multiple physical samples. Despite efficient simulation tools and powerful supercomputers, yet simulation tools have been limited by CPU time and therefore not mature for practical designs. To this end, we constructed and integrated an inexpensive low-fidelity machine learning (ML) algorithm with the expensive high-fidelity simulation. This ML model was then trained to increase the fidelity by a wisely chosen train set of simulation to construct a meta-model for active exploration of various weld sequence scenarios. As opposed to existing ML algorithms that require an extensive data set to train, our algorithm picks relatively small training set to construct a meta-model. We present an example of our algorithm implemented in a real welded structure project.\",\"PeriodicalId\":23651,\"journal\":{\"name\":\"Volume 6B: Materials and Fabrication\",\"volume\":\"1 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-11-15\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"2\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Volume 6B: Materials and Fabrication\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1115/pvp2019-93672\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Volume 6B: Materials and Fabrication","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1115/pvp2019-93672","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Welding Simulation Integrated With Machine Learning to Train a Meta-Model for Fast Exploration of Various Weld Sequence Scenarios
Distortion is a common problem in welded structures, and therefore welding standards require a mitigation plan to be in place before welding. When dealing with multiple welds, an optimal intermittent weld sequence can effectively minimize the distortion by counter-balancing the transient distortion during welding. However, the process of finding an effective weld sequence is a challenging task given a large number of possible combinations, i.e. several thousand for a few welds. As an acceptable approach, welding simulation tools allow engineers to optimize a welding sequence without the need for multiple physical samples. Despite efficient simulation tools and powerful supercomputers, yet simulation tools have been limited by CPU time and therefore not mature for practical designs. To this end, we constructed and integrated an inexpensive low-fidelity machine learning (ML) algorithm with the expensive high-fidelity simulation. This ML model was then trained to increase the fidelity by a wisely chosen train set of simulation to construct a meta-model for active exploration of various weld sequence scenarios. As opposed to existing ML algorithms that require an extensive data set to train, our algorithm picks relatively small training set to construct a meta-model. We present an example of our algorithm implemented in a real welded structure project.