Franz M. Rohrhofer , Stefan Posch , Clemens Gößnitzer , José M. García-Oliver , Bernhard C. Geiger
{"title":"Utilizing neural networks to supplant chemical kinetics tabulation through mass conservation and weighting of species depletion","authors":"Franz M. Rohrhofer , Stefan Posch , Clemens Gößnitzer , José M. García-Oliver , Bernhard C. Geiger","doi":"10.1016/j.egyai.2024.100341","DOIUrl":null,"url":null,"abstract":"<div><p>Artificial Neural Networks (ANNs) have emerged as a powerful tool in combustion simulations to replace memory-intensive tabulation of integrated chemical kinetics. Complex reaction mechanisms, however, present a challenge for standard ANN approaches as modeling multiple species typically suffers from inaccurate predictions on minor species. This paper presents a novel ANN approach which can be applied on complex reaction mechanisms in tabular data form, and only involves training a single ANN for a complete reaction mechanism. The approach incorporates a network architecture that automatically conserves mass and employs a particular loss weighting based on species depletion. Both modifications are used to improve the overall ANN performance and individual prediction accuracies, especially for minor species mass fractions. To validate its effectiveness, the approach is compared to standard ANNs in terms of performance and ANN complexity. Four distinct reaction mechanisms (H<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span>, C<span><math><msub><mrow></mrow><mrow><mn>7</mn></mrow></msub></math></span>H<span><math><msub><mrow></mrow><mrow><mn>16</mn></mrow></msub></math></span>, C<span><math><msub><mrow></mrow><mrow><mn>12</mn></mrow></msub></math></span>H<span><math><msub><mrow></mrow><mrow><mn>26</mn></mrow></msub></math></span>, OME<span><math><msub><mrow></mrow><mrow><mn>34</mn></mrow></msub></math></span>) are used as a test cases, and results demonstrate that considerable improvements can be achieved by applying both modifications.</p></div>","PeriodicalId":34138,"journal":{"name":"Energy and AI","volume":"16 ","pages":"Article 100341"},"PeriodicalIF":9.6000,"publicationDate":"2024-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2666546824000077/pdfft?md5=74cf715196106974d71e060a8c29f244&pid=1-s2.0-S2666546824000077-main.pdf","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Energy and AI","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2666546824000077","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}
引用次数: 0
Abstract
Artificial Neural Networks (ANNs) have emerged as a powerful tool in combustion simulations to replace memory-intensive tabulation of integrated chemical kinetics. Complex reaction mechanisms, however, present a challenge for standard ANN approaches as modeling multiple species typically suffers from inaccurate predictions on minor species. This paper presents a novel ANN approach which can be applied on complex reaction mechanisms in tabular data form, and only involves training a single ANN for a complete reaction mechanism. The approach incorporates a network architecture that automatically conserves mass and employs a particular loss weighting based on species depletion. Both modifications are used to improve the overall ANN performance and individual prediction accuracies, especially for minor species mass fractions. To validate its effectiveness, the approach is compared to standard ANNs in terms of performance and ANN complexity. Four distinct reaction mechanisms (H, CH, CH, OME) are used as a test cases, and results demonstrate that considerable improvements can be achieved by applying both modifications.
人工神经网络(ANN)已成为燃烧模拟中的一种强大工具,可取代需要大量记忆的综合化学动力学表格。然而,复杂的反应机制给标准的人工神经网络方法带来了挑战,因为多物种建模通常会导致对次要物种的预测不准确。本文介绍了一种新颖的方差网络方法,该方法可应用于表格数据形式的复杂反应机理,而且只需为完整的反应机理训练一个方差网络。该方法采用了自动保存质量的网络架构,并根据物种损耗采用了特定的损耗加权。这两项修改都用于提高 ANN 的整体性能和单个预测的准确性,尤其是对小物种质量分数的预测。为了验证该方法的有效性,我们将其与标准自动数值网络的性能和复杂性进行了比较。四个不同的反应机理(H2、C7H16、C12H26、OME34)被用作测试案例,结果表明,通过应用这两种修改,可以实现相当大的改进。