{"title":"基于神经网络的显式MPC化学反应器控制","authors":"Karol Kiš, Martin Klauco","doi":"10.2478/acs-2019-0030","DOIUrl":null,"url":null,"abstract":"Abstract In this paper, implementation of deep neural networks applied in process control is presented. In our approach, training of the neural network is based on model predictive control, which is popular for its ability to be tuned by the weighting matrices and for it respecting the system constraints. A neural network that can approximate the MPC behavior by mimicking the control input trajectory while the constraints on states and control input remain unimpaired by the weighting matrices is introduced. This approach is demonstrated in a simulation case study involving a continuous stirred tank reactor where a multi-component chemical reaction takes place.","PeriodicalId":7088,"journal":{"name":"Acta Chimica Slovaca","volume":null,"pages":null},"PeriodicalIF":0.9000,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"9","resultStr":"{\"title\":\"Neural network based explicit MPC for chemical reactor control\",\"authors\":\"Karol Kiš, Martin Klauco\",\"doi\":\"10.2478/acs-2019-0030\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Abstract In this paper, implementation of deep neural networks applied in process control is presented. In our approach, training of the neural network is based on model predictive control, which is popular for its ability to be tuned by the weighting matrices and for it respecting the system constraints. A neural network that can approximate the MPC behavior by mimicking the control input trajectory while the constraints on states and control input remain unimpaired by the weighting matrices is introduced. This approach is demonstrated in a simulation case study involving a continuous stirred tank reactor where a multi-component chemical reaction takes place.\",\"PeriodicalId\":7088,\"journal\":{\"name\":\"Acta Chimica Slovaca\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.9000,\"publicationDate\":\"2019-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"9\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Acta Chimica Slovaca\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.2478/acs-2019-0030\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"CHEMISTRY, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Acta Chimica Slovaca","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2478/acs-2019-0030","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"CHEMISTRY, MULTIDISCIPLINARY","Score":null,"Total":0}
Neural network based explicit MPC for chemical reactor control
Abstract In this paper, implementation of deep neural networks applied in process control is presented. In our approach, training of the neural network is based on model predictive control, which is popular for its ability to be tuned by the weighting matrices and for it respecting the system constraints. A neural network that can approximate the MPC behavior by mimicking the control input trajectory while the constraints on states and control input remain unimpaired by the weighting matrices is introduced. This approach is demonstrated in a simulation case study involving a continuous stirred tank reactor where a multi-component chemical reaction takes place.