Deep reinforcement learning based controller with dynamic feature extraction for an industrial claus process

•The DRL-based controller integrating with the process dynamics is developed.•The dynamic feature is extracted from the historical data using the Seq2seq network.•The controller was trained by interacting with the Seq2seq model.•The standard deviation of the control variable in the industrial Claus...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Journal of the Taiwan Institute of Chemical Engineers Ročník 146; s. 104779
Hlavní autoři: Liu, Jialin, Tsai, Bing-Yen, Chen, Ding-Sou
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier B.V 01.05.2023
Témata:
ISSN:1876-1070, 1876-1089
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Abstract •The DRL-based controller integrating with the process dynamics is developed.•The dynamic feature is extracted from the historical data using the Seq2seq network.•The controller was trained by interacting with the Seq2seq model.•The standard deviation of the control variable in the industrial Claus process can be reduced by up to 55% using the proposed DRL-based controller. The significant time delay between the manipulated and controlled variables introduces challenges in the task of system identification when implementing model predictive control (MPC) for an industrial process. Recently, deep reinforcement learning (DRL) with model-free characteristics has attracted considerable attention from the process control community. However, the model-free assumption in DRL is based on the property of the Markov decision process (MDP), in which all state variables must be observed. This assumption is not true for an industrial process. In this study, the sequence-to-sequence (Seq2seq) network was employed to build a surrogate model based on the industrial Claus process data. Meanwhile, the hidden state output from the encoder of the Seq2seq network, which represents the dynamic feature of the process, connects to the DRL-based controller to compensate for the partial observabilities of a real process. The results show that the standard deviation of the control variable, which refers to the H2S to SO2 concentration ratio in the tail gas, can be reduced by up to 55% using the proposed DRL-based controller comparing with the current control strategy. [Display omitted]
AbstractList •The DRL-based controller integrating with the process dynamics is developed.•The dynamic feature is extracted from the historical data using the Seq2seq network.•The controller was trained by interacting with the Seq2seq model.•The standard deviation of the control variable in the industrial Claus process can be reduced by up to 55% using the proposed DRL-based controller. The significant time delay between the manipulated and controlled variables introduces challenges in the task of system identification when implementing model predictive control (MPC) for an industrial process. Recently, deep reinforcement learning (DRL) with model-free characteristics has attracted considerable attention from the process control community. However, the model-free assumption in DRL is based on the property of the Markov decision process (MDP), in which all state variables must be observed. This assumption is not true for an industrial process. In this study, the sequence-to-sequence (Seq2seq) network was employed to build a surrogate model based on the industrial Claus process data. Meanwhile, the hidden state output from the encoder of the Seq2seq network, which represents the dynamic feature of the process, connects to the DRL-based controller to compensate for the partial observabilities of a real process. The results show that the standard deviation of the control variable, which refers to the H2S to SO2 concentration ratio in the tail gas, can be reduced by up to 55% using the proposed DRL-based controller comparing with the current control strategy. [Display omitted]
ArticleNumber 104779
Author Liu, Jialin
Tsai, Bing-Yen
Chen, Ding-Sou
Author_xml – sequence: 1
  givenname: Jialin
  orcidid: 0000-0002-3180-6600
  surname: Liu
  fullname: Liu, Jialin
  email: jialin@thu.edu.tw
  organization: Research Center for Smart Sustainable Circular Economy, Tunghai University, No. 1727, Sec.4, Taiwan Boulevard, Taichung, Taiwan
– sequence: 2
  givenname: Bing-Yen
  surname: Tsai
  fullname: Tsai, Bing-Yen
  organization: Department of Chemical and Materials Engineering, Tunghai University, No. 1727, Sec.4, Taiwan Boulevard, Taichung, Taiwan
– sequence: 3
  givenname: Ding-Sou
  surname: Chen
  fullname: Chen, Ding-Sou
  organization: Green Energy and System Integration Research and Development Department, China Steel Corporation, No. 1, Chung Kang Road, Hsiao Kang, Kaohsiung, Taiwan
BookMark eNqFkMtOwzAQRS1UJErpF7DxD6TYThonCxaoPKVKbGBtOeMJOHKdynaB_j0JRSxYwGxmNNK50j2nZOJ7j4Scc7bgjJcX3aJLFnAhmMiHTyFlfUSmvJJlxllVT35uyU7IPMaOjSOZEMspcdeIWxrQ-rYPgBv0iTrUwVv_Qhsd0VDofQq9cxjou02v1Oy93ligLeq0C0jxIwUNyfaeDhlUe2q92cUUrHYUnN5Fug09YIxn5LjVLuL8e8_I8-3N0-o-Wz_ePayu1hnkLE8Zr3QjAHRZANNC1CIXQoMoi0IWYHTFDKBk0FYGRcNRNq3kS8O4zNuSVYj5jOSHXAh9jAFbtQ12o8NecaZGZ6pTX87U6EwdnA1U_YsCm_TYa-hn3T_s5YHFodabxaAiWPSAxgaEpExv_-Q_ASEDjds
CitedBy_id crossref_primary_10_1016_j_jtice_2024_105864
crossref_primary_10_1016_j_jtice_2025_106119
crossref_primary_10_1016_j_jtice_2024_105407
crossref_primary_10_1016_j_supflu_2025_106542
crossref_primary_10_1016_j_jmsy_2025_07_002
crossref_primary_10_1016_j_jtice_2024_105430
crossref_primary_10_1007_s10489_024_05943_8
Cites_doi 10.1016/j.jprocont.2018.11.004
10.3115/v1/W14-4012
10.1016/j.compchemeng.2020.107016
10.1002/aic.17306
10.1162/neco.1997.9.8.1735
10.1016/j.compchemeng.2019.106649
10.1016/j.fuel.2013.05.070
10.1038/nature16961
10.1016/S0004-3702(98)00023-X
10.3115/v1/D14-1179
10.1016/j.compchemeng.2014.01.019
10.1016/j.jtice.2021.104200
10.1016/j.jtice.2021.04.062
10.1016/j.jtice.2022.104498
10.1038/nature14236
10.1021/acs.iecr.1c04984
10.1002/aic.14523
10.1109/TII.2019.2952429
10.1207/s15516709cog1402_1
10.1016/j.jtice.2022.104445
10.1016/j.compchemeng.2017.10.008
10.1016/j.jtice.2021.06.050
10.1002/aic.16689
10.1016/j.engappai.2018.07.003
10.1038/nature14539
10.1016/j.compchemeng.2022.107987
10.1016/j.jclepro.2021.125915
10.1016/j.compchemeng.2021.107280
10.1016/j.compchemeng.2020.107077
10.1016/j.compchemeng.2019.05.029
10.1021/acs.iecr.1c04668
10.1016/j.jprocont.2021.06.004
ContentType Journal Article
Copyright 2023 Taiwan Institute of Chemical Engineers
Copyright_xml – notice: 2023 Taiwan Institute of Chemical Engineers
DBID AAYXX
CITATION
DOI 10.1016/j.jtice.2023.104779
DatabaseName CrossRef
DatabaseTitle CrossRef
DatabaseTitleList
DeliveryMethod fulltext_linktorsrc
Discipline Economics
Engineering
EISSN 1876-1089
ExternalDocumentID 10_1016_j_jtice_2023_104779
S1876107023001086
GroupedDBID --K
--M
.DC
.~1
0R~
1B1
1~.
1~5
4.4
457
4G.
5VS
7-5
71M
8P~
8RM
AACTN
AAEDT
AAEDW
AAIAV
AAIKJ
AAKOC
AALRI
AAOAW
AAQFI
AAXUO
ABFNM
ABJNI
ABMAC
ABNUV
ABXDB
ABYKQ
ACDAQ
ACGFS
ACRLP
ADBBV
ADEWK
ADEZE
ADMUD
AEBSH
AEKER
AENEX
AFKWA
AFTJW
AGHFR
AGUBO
AGYEJ
AHPOS
AIEXJ
AIKHN
AITUG
AJBFU
AJOXV
AKURH
ALMA_UNASSIGNED_HOLDINGS
AMFUW
AMRAJ
AXJTR
BKOJK
BLXMC
DU5
EBS
EFJIC
EFLBG
EJD
ENUVR
EP2
EP3
FDB
FEDTE
FIRID
FNPLU
FYGXN
GBLVA
HVGLF
HZ~
J1W
KOM
M41
MO0
N9A
O-L
O9-
OAUVE
OZT
P-8
P-9
P2P
PC.
Q38
RIG
ROL
SDF
SES
SPC
SPCBC
SSG
SSZ
T5K
~G-
9DU
AATTM
AAXKI
AAYWO
AAYXX
ABWVN
ACLOT
ACRPL
ACVFH
ADCNI
ADNMO
AEIPS
AEUPX
AFJKZ
AFPUW
AIGII
AIIUN
AKBMS
AKRWK
AKYEP
ANKPU
APXCP
CITATION
EFKBS
~HD
ID FETCH-LOGICAL-c303t-18ab2cca64c0a2292322ac264474cda80dce70cf8de2b1e7bf715d0173f608ee3
ISICitedReferencesCount 9
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000951950000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1876-1070
IngestDate Sat Nov 29 06:58:29 EST 2025
Tue Nov 18 20:49:11 EST 2025
Fri Feb 23 02:37:57 EST 2024
IsPeerReviewed true
IsScholarly true
Keywords Deep reinforcement learning
industrial Claus process
Actor-critic networks
Sequence-to-sequence network
Twin delayed deep deterministic policy gradient algorithm
Language English
LinkModel OpenURL
MergedId FETCHMERGED-LOGICAL-c303t-18ab2cca64c0a2292322ac264474cda80dce70cf8de2b1e7bf715d0173f608ee3
ORCID 0000-0002-3180-6600
ParticipantIDs crossref_primary_10_1016_j_jtice_2023_104779
crossref_citationtrail_10_1016_j_jtice_2023_104779
elsevier_sciencedirect_doi_10_1016_j_jtice_2023_104779
PublicationCentury 2000
PublicationDate May 2023
2023-05-00
PublicationDateYYYYMMDD 2023-05-01
PublicationDate_xml – month: 05
  year: 2023
  text: May 2023
PublicationDecade 2020
PublicationTitle Journal of the Taiwan Institute of Chemical Engineers
PublicationYear 2023
Publisher Elsevier B.V
Publisher_xml – name: Elsevier B.V
References Mowbray, Smith, del Rio-Chanona, Zhang (bib0025) 2021; 67
Mokhatab, Poe, Speight (bib0041) 2006
Chou, Wu, Kang, Wong, Yao, Chuang, Jang, Ou (bib0024) 2020; 16
Elman (bib0027) 1990; 14
Cho K., Van Merriënboer B., Bahdanau D., Bengio Y. On the properties of neural machine translation: encoder-decoder approaches. arXiv preprint 2014; arXiv:1409.1259.
Fujimoto, Hoof, Meger (bib0035) 2018
Dogru, Wieczorek, Velswamy, Ibrahim, Huang (bib0018) 2021; 104
Mnih, Kavukcuoglu, Silver, Rusu, Veness, Bellemare, Ostrovski (bib0008) 2015; 518
Schulman J., Wolski F., Dhariwal P., Radford A., Klimov O. Proximal policy optimization algorithms. arXiv preprint 2017; arXiv:1707.06347v2.
Byun, Kim, Lee (bib0016) 2022; 167
Manenti, Papasidero, Bozzano, Ranzi (bib0040) 2014; 66
Kaelbling, Littman, Cassandra (bib0026) 1998; 101
Katharopoulos A., Vyas A., Pappas N., Fleuret F. Transformers are RNNs: fast autoregressive transformers with linear attention. arXiv preprint 2020; arXiv:2006.16236v3.
Lee, Shin, Realff (bib0002) 2018; 114
He, Shi, Tan, Song, Zhu (bib0004) 2021; 122
Haarnoja T., Zhou A., Abbeel P., Levine S. Soft actor-critic: off-policy maximum entropy deep reinforcement learning with a stochastic actor. arXiv preprint 2018; arXiv:1801.01290v2.
LeCun, Bengio, Hinton (bib0003) 2015; 521
Shin, Badgwell, Liu, Lee (bib0010) 2019; 127
Oh, Adams, Vo, Gbadago, Lee, Oh (bib0021) 2021; 149
Kang, Mirzaei, Zhou (bib0012) 2022; 130
Sutton, Barto (bib0017) 2018
Qin (bib0001) 2014; 60
Zhou, Xu, He, Zhang (bib0006) 2022; 138
Adams, Oh, Kim, Lee, Oh (bib0023) 2021; 291
Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A.N., Kaiser L., Polosukhin I. Attention is all you need. arXiv preprint 2017; arXiv:1706.03762v5.
Petsagkourakis, Sandoval, Bradford, Zhang, del Rio-Chanona (bib0014) 2020; 133
Razzaq, Li, Zhang (bib0038) 2013; 113
Zheng, Wu, Sun, Song, Zhou (bib0007) 2022; 138
Ma, Noreña-Caro, Adams, Brentzel, Romagnoli, Benton (bib0015) 2020; 142
Cho K., Van Merriënboer B., Gulcehre C., Bahdanau D., Bougares F., Schwenk H., Bengio Y. Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint 2014; arXiv:1406.1078.
Spielberg, Tulsyan, Lawrence, Loewen, Gopaluni (bib0011) 2019; 65
Silver, Huang, Maddison, Guez, Sifre, Van Den Driessche, Lanctot (bib0009) 2016; 529
Ma, Zhu, Benton, Romagnoli (bib0013) 2019; 75
Lillicrap T.P., Hunt J.J., Pritzel A. Heess N Erez T Tassa Y., Silver D., Wierstra D. Continuous control with deep reinforcement learning. arXiv preprint 2015; arXiv:1509.02971.
Moral, Ortiz-Imedio, Ortiz, Gorri, Ortiz (bib0039) 2022; 61
Campos, El-Farra, Palazoglu (bib0019) 2022; 61
Powell, Machalek, Quah (bib0020) 2020; 143
Cheng, Huang, Pang, Zhang (bib0022) 2018; 74
Hochreiter, Schmidhuber (bib0028) 1997; 9
Silver, Lever, Heess, Degris, Wierstra, Riedmiller (bib0036) 2014
Yu, Liu, Liu, Wang (bib0005) 2022; 132
Elman (10.1016/j.jtice.2023.104779_bib0027) 1990; 14
Moral (10.1016/j.jtice.2023.104779_bib0039) 2022; 61
Chou (10.1016/j.jtice.2023.104779_bib0024) 2020; 16
Mnih (10.1016/j.jtice.2023.104779_bib0008) 2015; 518
Zheng (10.1016/j.jtice.2023.104779_bib0007) 2022; 138
Hochreiter (10.1016/j.jtice.2023.104779_bib0028) 1997; 9
He (10.1016/j.jtice.2023.104779_bib0004) 2021; 122
Dogru (10.1016/j.jtice.2023.104779_bib0018) 2021; 104
10.1016/j.jtice.2023.104779_bib0030
Silver (10.1016/j.jtice.2023.104779_bib0009) 2016; 529
Qin (10.1016/j.jtice.2023.104779_bib0001) 2014; 60
Razzaq (10.1016/j.jtice.2023.104779_bib0038) 2013; 113
Byun (10.1016/j.jtice.2023.104779_bib0016) 2022; 167
Oh (10.1016/j.jtice.2023.104779_bib0021) 2021; 149
Campos (10.1016/j.jtice.2023.104779_bib0019) 2022; 61
Sutton (10.1016/j.jtice.2023.104779_bib0017) 2018
Shin (10.1016/j.jtice.2023.104779_bib0010) 2019; 127
Mowbray (10.1016/j.jtice.2023.104779_bib0025) 2021; 67
Spielberg (10.1016/j.jtice.2023.104779_bib0011) 2019; 65
10.1016/j.jtice.2023.104779_bib0029
Zhou (10.1016/j.jtice.2023.104779_bib0006) 2022; 138
Mokhatab (10.1016/j.jtice.2023.104779_bib0041) 2006
Manenti (10.1016/j.jtice.2023.104779_bib0040) 2014; 66
Lee (10.1016/j.jtice.2023.104779_bib0002) 2018; 114
Ma (10.1016/j.jtice.2023.104779_bib0015) 2020; 142
Kaelbling (10.1016/j.jtice.2023.104779_bib0026) 1998; 101
Adams (10.1016/j.jtice.2023.104779_bib0023) 2021; 291
Powell (10.1016/j.jtice.2023.104779_bib0020) 2020; 143
LeCun (10.1016/j.jtice.2023.104779_bib0003) 2015; 521
Cheng (10.1016/j.jtice.2023.104779_bib0022) 2018; 74
Yu (10.1016/j.jtice.2023.104779_bib0005) 2022; 132
10.1016/j.jtice.2023.104779_bib0037
10.1016/j.jtice.2023.104779_bib0031
10.1016/j.jtice.2023.104779_bib0032
10.1016/j.jtice.2023.104779_bib0033
Kang (10.1016/j.jtice.2023.104779_bib0012) 2022; 130
10.1016/j.jtice.2023.104779_bib0034
Ma (10.1016/j.jtice.2023.104779_bib0013) 2019; 75
Silver (10.1016/j.jtice.2023.104779_bib0036) 2014
Fujimoto (10.1016/j.jtice.2023.104779_bib0035) 2018
Petsagkourakis (10.1016/j.jtice.2023.104779_bib0014) 2020; 133
References_xml – volume: 167
  year: 2022
  ident: bib0016
  article-title: Multi-step lookahead Bayesian optimization with active learning using reinforcement learning and its application to data-driven batch-to-batch optimization
  publication-title: Comput Chem Eng
– reference: Katharopoulos A., Vyas A., Pappas N., Fleuret F. Transformers are RNNs: fast autoregressive transformers with linear attention. arXiv preprint 2020; arXiv:2006.16236v3.
– volume: 66
  start-page: 244
  year: 2014
  end-page: 251
  ident: bib0040
  article-title: Model-based optimization of sulfur recovery units
  publication-title: Comp Chem Eng
– volume: 75
  start-page: 40
  year: 2019
  end-page: 47
  ident: bib0013
  article-title: Continuous control of a polymerization system with deep reinforcement learning
  publication-title: J Process Control
– volume: 14
  start-page: 179
  year: 1990
  end-page: 211
  ident: bib0027
  article-title: Finding structure in time
  publication-title: Cogn Sci
– volume: 61
  start-page: 6106
  year: 2022
  end-page: 6124
  ident: bib0039
  article-title: Hydrogen recovery from coke oven gas. Comparative analysis of technical alternatives
  publication-title: Ind Eng Chem Res
– volume: 9
  start-page: 1735
  year: 1997
  end-page: 1780
  ident: bib0028
  article-title: Long short-term memory
  publication-title: Neural Comput
– reference: Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A.N., Kaiser L., Polosukhin I. Attention is all you need. arXiv preprint 2017; arXiv:1706.03762v5.
– volume: 130
  year: 2022
  ident: bib0012
  article-title: Robust control and training risk reduction for boiler level control using two-stage training deep deterministic policy gradient
  publication-title: J Taiwan Inst Chem Eng
– volume: 143
  year: 2020
  ident: bib0020
  article-title: Real-time optimization using reinforcement learning
  publication-title: Comp Chem Eng
– reference: Lillicrap T.P., Hunt J.J., Pritzel A. Heess N Erez T Tassa Y., Silver D., Wierstra D. Continuous control with deep reinforcement learning. arXiv preprint 2015; arXiv:1509.02971.
– reference: Cho K., Van Merriënboer B., Bahdanau D., Bengio Y. On the properties of neural machine translation: encoder-decoder approaches. arXiv preprint 2014; arXiv:1409.1259.
– volume: 529
  start-page: 484
  year: 2016
  end-page: 489
  ident: bib0009
  article-title: Mastering the game of Go with deep neural networks and tree search
  publication-title: Nature
– volume: 104
  start-page: 86
  year: 2021
  end-page: 100
  ident: bib0018
  article-title: Online reinforcement learning for a continuous space system with experimental validation
  publication-title: J Process Control
– reference: Haarnoja T., Zhou A., Abbeel P., Levine S. Soft actor-critic: off-policy maximum entropy deep reinforcement learning with a stochastic actor. arXiv preprint 2018; arXiv:1801.01290v2.
– volume: 61
  start-page: 8443
  year: 2022
  end-page: 8461
  ident: bib0019
  article-title: Soft actor-critic deep reinforcement learning with hybrid mixed-integer actions for demand responsive scheduling of energy systems
  publication-title: Ind Eng Chem Res
– year: 2014
  ident: bib0036
  article-title: Deterministic policy gradient algorithms
  publication-title: Proceedings of the 31st international conference on machine learning
– start-page: 626
  year: 2006
  ident: bib0041
  article-title: Handbook of natural gas transmission and processing
– volume: 138
  year: 2022
  ident: bib0007
  article-title: Deep learning of complex process data for fault classification based on sparse probabilistic dynamic network
  publication-title: J Taiwan Inst Chem Eng
– volume: 16
  start-page: 2829
  year: 2020
  end-page: 2838
  ident: bib0024
  article-title: Physically consistent soft-sensor development using sequence-to-sequence neural networks
  publication-title: IEEE T Ind Inform
– volume: 133
  year: 2020
  ident: bib0014
  article-title: Reinforcement learning for batch bioprocess optimization
  publication-title: Comp Chem Eng
– volume: 291
  year: 2021
  ident: bib0023
  article-title: Deep reinforcement learning optimization framework for a power generation plant considering performance and environmental issues
  publication-title: J Clean Prod
– volume: 74
  start-page: 303
  year: 2018
  end-page: 311
  ident: bib0022
  article-title: ThermalNet: a deep reinforcement learning-based combustion optimization system for coal-fired boiler
  publication-title: Eng Appl Artif Intell
– volume: 521
  start-page: 436
  year: 2015
  end-page: 444
  ident: bib0003
  article-title: Deep learning
  publication-title: Nature
– volume: 114
  start-page: 111
  year: 2018
  end-page: 121
  ident: bib0002
  article-title: Machine learning: overview of the recent progresses and implications for the process systems engineering field
  publication-title: Comp Chem Eng
– volume: 65
  start-page: e16689
  year: 2019
  ident: bib0011
  article-title: Toward self-driving processes: a deep reinforcement learning approach to control
  publication-title: AIChE J
– volume: 138
  year: 2022
  ident: bib0006
  article-title: Online abnormal interval detection and classification of industrial time series data based on multi-scale deep learning
  publication-title: J Taiwan Inst Chem Eng
– volume: 127
  start-page: 282
  year: 2019
  end-page: 294
  ident: bib0010
  article-title: Reinforcement learning – overview of recent progress and implications for process control
  publication-title: Comp Chem Eng
– volume: 113
  start-page: 287
  year: 2013
  end-page: 299
  ident: bib0038
  article-title: Coke oven gas: availability, properties, purification, and utilization in China
  publication-title: Fuel
– volume: 122
  start-page: 78
  year: 2021
  end-page: 84
  ident: bib0004
  article-title: Multiblock temporal convolution network-based temporal-correlated feature learning for fault diagnosis of multivariate processes
  publication-title: J Taiwan Inst Chem Eng
– volume: 518
  start-page: 529
  year: 2015
  end-page: 533
  ident: bib0008
  article-title: Human-level control through deep reinforcement learning
  publication-title: Nature
– reference: Schulman J., Wolski F., Dhariwal P., Radford A., Klimov O. Proximal policy optimization algorithms. arXiv preprint 2017; arXiv:1707.06347v2.
– volume: 149
  year: 2021
  ident: bib0021
  article-title: Actor-critic reinforcement learning to estimate the optimal operating conditions of the hydrocracking process
  publication-title: Comp Chem Eng
– year: 2018
  ident: bib0017
  article-title: Reinforcement learning: an introduction
– volume: 67
  start-page: e17306
  year: 2021
  ident: bib0025
  article-title: Using process data to generate an optimal control policy via apprenticeship and reinforcement learning
  publication-title: AIChE J
– reference: Cho K., Van Merriënboer B., Gulcehre C., Bahdanau D., Bougares F., Schwenk H., Bengio Y. Learning phrase representations using RNN encoder-decoder for statistical machine translation. arXiv preprint 2014; arXiv:1406.1078.
– volume: 101
  start-page: 99
  year: 1998
  end-page: 134
  ident: bib0026
  article-title: Planning and acting in partially observable stochastic domains
  publication-title: Artif Intell
– volume: 60
  start-page: 3092
  year: 2014
  end-page: 3100
  ident: bib0001
  article-title: Process data analytics in the era of big data
  publication-title: AIChE J
– volume: 142
  year: 2020
  ident: bib0015
  article-title: Machine-learning-based simulation and fed-batch control of cyanobacterial-phycocyanin production in Plectonema by artificial neural network and deep reinforcement learning
  publication-title: Comp Chem Eng
– year: 2018
  ident: bib0035
  article-title: Addressing function approximation error in actor-critic methods
  publication-title: Proceedings of the 35th international conference on machine learning
– volume: 132
  year: 2022
  ident: bib0005
  article-title: Supervised convolutional autoencoder-based fault-relevant feature learning for fault diagnosis in industrial processes
  publication-title: J Taiwan Inst Chem Eng
– volume: 75
  start-page: 40
  year: 2019
  ident: 10.1016/j.jtice.2023.104779_bib0013
  article-title: Continuous control of a polymerization system with deep reinforcement learning
  publication-title: J Process Control
  doi: 10.1016/j.jprocont.2018.11.004
– ident: 10.1016/j.jtice.2023.104779_bib0032
  doi: 10.3115/v1/W14-4012
– volume: 142
  year: 2020
  ident: 10.1016/j.jtice.2023.104779_bib0015
  article-title: Machine-learning-based simulation and fed-batch control of cyanobacterial-phycocyanin production in Plectonema by artificial neural network and deep reinforcement learning
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2020.107016
– volume: 67
  start-page: e17306
  year: 2021
  ident: 10.1016/j.jtice.2023.104779_bib0025
  article-title: Using process data to generate an optimal control policy via apprenticeship and reinforcement learning
  publication-title: AIChE J
  doi: 10.1002/aic.17306
– start-page: 626
  year: 2006
  ident: 10.1016/j.jtice.2023.104779_bib0041
– volume: 9
  start-page: 1735
  year: 1997
  ident: 10.1016/j.jtice.2023.104779_bib0028
  article-title: Long short-term memory
  publication-title: Neural Comput
  doi: 10.1162/neco.1997.9.8.1735
– volume: 133
  year: 2020
  ident: 10.1016/j.jtice.2023.104779_bib0014
  article-title: Reinforcement learning for batch bioprocess optimization
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2019.106649
– ident: 10.1016/j.jtice.2023.104779_bib0037
– volume: 113
  start-page: 287
  year: 2013
  ident: 10.1016/j.jtice.2023.104779_bib0038
  article-title: Coke oven gas: availability, properties, purification, and utilization in China
  publication-title: Fuel
  doi: 10.1016/j.fuel.2013.05.070
– volume: 529
  start-page: 484
  year: 2016
  ident: 10.1016/j.jtice.2023.104779_bib0009
  article-title: Mastering the game of Go with deep neural networks and tree search
  publication-title: Nature
  doi: 10.1038/nature16961
– volume: 101
  start-page: 99
  year: 1998
  ident: 10.1016/j.jtice.2023.104779_bib0026
  article-title: Planning and acting in partially observable stochastic domains
  publication-title: Artif Intell
  doi: 10.1016/S0004-3702(98)00023-X
– ident: 10.1016/j.jtice.2023.104779_bib0029
  doi: 10.3115/v1/D14-1179
– ident: 10.1016/j.jtice.2023.104779_bib0033
– ident: 10.1016/j.jtice.2023.104779_bib0031
– volume: 66
  start-page: 244
  year: 2014
  ident: 10.1016/j.jtice.2023.104779_bib0040
  article-title: Model-based optimization of sulfur recovery units
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2014.01.019
– volume: 132
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0005
  article-title: Supervised convolutional autoencoder-based fault-relevant feature learning for fault diagnosis in industrial processes
  publication-title: J Taiwan Inst Chem Eng
  doi: 10.1016/j.jtice.2021.104200
– volume: 122
  start-page: 78
  year: 2021
  ident: 10.1016/j.jtice.2023.104779_bib0004
  article-title: Multiblock temporal convolution network-based temporal-correlated feature learning for fault diagnosis of multivariate processes
  publication-title: J Taiwan Inst Chem Eng
  doi: 10.1016/j.jtice.2021.04.062
– volume: 138
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0007
  article-title: Deep learning of complex process data for fault classification based on sparse probabilistic dynamic network
  publication-title: J Taiwan Inst Chem Eng
  doi: 10.1016/j.jtice.2022.104498
– volume: 518
  start-page: 529
  year: 2015
  ident: 10.1016/j.jtice.2023.104779_bib0008
  article-title: Human-level control through deep reinforcement learning
  publication-title: Nature
  doi: 10.1038/nature14236
– volume: 61
  start-page: 8443
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0019
  article-title: Soft actor-critic deep reinforcement learning with hybrid mixed-integer actions for demand responsive scheduling of energy systems
  publication-title: Ind Eng Chem Res
  doi: 10.1021/acs.iecr.1c04984
– volume: 60
  start-page: 3092
  year: 2014
  ident: 10.1016/j.jtice.2023.104779_bib0001
  article-title: Process data analytics in the era of big data
  publication-title: AIChE J
  doi: 10.1002/aic.14523
– volume: 16
  start-page: 2829
  year: 2020
  ident: 10.1016/j.jtice.2023.104779_bib0024
  article-title: Physically consistent soft-sensor development using sequence-to-sequence neural networks
  publication-title: IEEE T Ind Inform
  doi: 10.1109/TII.2019.2952429
– year: 2018
  ident: 10.1016/j.jtice.2023.104779_bib0017
– volume: 14
  start-page: 179
  year: 1990
  ident: 10.1016/j.jtice.2023.104779_bib0027
  article-title: Finding structure in time
  publication-title: Cogn Sci
  doi: 10.1207/s15516709cog1402_1
– volume: 138
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0006
  article-title: Online abnormal interval detection and classification of industrial time series data based on multi-scale deep learning
  publication-title: J Taiwan Inst Chem Eng
  doi: 10.1016/j.jtice.2022.104445
– volume: 114
  start-page: 111
  year: 2018
  ident: 10.1016/j.jtice.2023.104779_bib0002
  article-title: Machine learning: overview of the recent progresses and implications for the process systems engineering field
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2017.10.008
– volume: 130
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0012
  article-title: Robust control and training risk reduction for boiler level control using two-stage training deep deterministic policy gradient
  publication-title: J Taiwan Inst Chem Eng
  doi: 10.1016/j.jtice.2021.06.050
– volume: 65
  start-page: e16689
  year: 2019
  ident: 10.1016/j.jtice.2023.104779_bib0011
  article-title: Toward self-driving processes: a deep reinforcement learning approach to control
  publication-title: AIChE J
  doi: 10.1002/aic.16689
– volume: 74
  start-page: 303
  year: 2018
  ident: 10.1016/j.jtice.2023.104779_bib0022
  article-title: ThermalNet: a deep reinforcement learning-based combustion optimization system for coal-fired boiler
  publication-title: Eng Appl Artif Intell
  doi: 10.1016/j.engappai.2018.07.003
– year: 2018
  ident: 10.1016/j.jtice.2023.104779_bib0035
  article-title: Addressing function approximation error in actor-critic methods
– volume: 521
  start-page: 436
  year: 2015
  ident: 10.1016/j.jtice.2023.104779_bib0003
  article-title: Deep learning
  publication-title: Nature
  doi: 10.1038/nature14539
– volume: 167
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0016
  article-title: Multi-step lookahead Bayesian optimization with active learning using reinforcement learning and its application to data-driven batch-to-batch optimization
  publication-title: Comput Chem Eng
  doi: 10.1016/j.compchemeng.2022.107987
– ident: 10.1016/j.jtice.2023.104779_bib0034
– volume: 291
  year: 2021
  ident: 10.1016/j.jtice.2023.104779_bib0023
  article-title: Deep reinforcement learning optimization framework for a power generation plant considering performance and environmental issues
  publication-title: J Clean Prod
  doi: 10.1016/j.jclepro.2021.125915
– volume: 149
  year: 2021
  ident: 10.1016/j.jtice.2023.104779_bib0021
  article-title: Actor-critic reinforcement learning to estimate the optimal operating conditions of the hydrocracking process
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2021.107280
– volume: 143
  year: 2020
  ident: 10.1016/j.jtice.2023.104779_bib0020
  article-title: Real-time optimization using reinforcement learning
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2020.107077
– ident: 10.1016/j.jtice.2023.104779_bib0030
– volume: 127
  start-page: 282
  year: 2019
  ident: 10.1016/j.jtice.2023.104779_bib0010
  article-title: Reinforcement learning – overview of recent progress and implications for process control
  publication-title: Comp Chem Eng
  doi: 10.1016/j.compchemeng.2019.05.029
– year: 2014
  ident: 10.1016/j.jtice.2023.104779_bib0036
  article-title: Deterministic policy gradient algorithms
– volume: 61
  start-page: 6106
  year: 2022
  ident: 10.1016/j.jtice.2023.104779_bib0039
  article-title: Hydrogen recovery from coke oven gas. Comparative analysis of technical alternatives
  publication-title: Ind Eng Chem Res
  doi: 10.1021/acs.iecr.1c04668
– volume: 104
  start-page: 86
  year: 2021
  ident: 10.1016/j.jtice.2023.104779_bib0018
  article-title: Online reinforcement learning for a continuous space system with experimental validation
  publication-title: J Process Control
  doi: 10.1016/j.jprocont.2021.06.004
SSID ssj0000070225
Score 2.3684473
Snippet •The DRL-based controller integrating with the process dynamics is developed.•The dynamic feature is extracted from the historical data using the Seq2seq...
SourceID crossref
elsevier
SourceType Enrichment Source
Index Database
Publisher
StartPage 104779
SubjectTerms Actor-critic networks
Deep reinforcement learning
industrial Claus process
Sequence-to-sequence network
Twin delayed deep deterministic policy gradient algorithm
Title Deep reinforcement learning based controller with dynamic feature extraction for an industrial claus process
URI https://dx.doi.org/10.1016/j.jtice.2023.104779
Volume 146
WOSCitedRecordID wos000951950000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVESC
  databaseName: ScienceDirect Freedom Collection
  customDbUrl:
  eissn: 1876-1089
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0000070225
  issn: 1876-1070
  databaseCode: AIEXJ
  dateStart: 20090101
  isFulltext: true
  titleUrlDefault: https://www.sciencedirect.com
  providerName: Elsevier
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV3db9MwELfKhgQ8TDBADBjyA28hKHHbOHkc-9BAaEJakcpT5NgXlCnKqrQd-w_4tzl_JRmFCZB4iSIrtlPfL-fz9e5-hLwuYpFkTINXSBFOgMehiFUSKhGhuSwBd4XCkE3ws7N0Ps8-jUbffS7MVc2bJr2-zhb_VdTYhsLWqbN_Ie5uUGzAexQ6XlHseP0jwR8BLIIWTEVUaZx_nhria6D3LOXD02torRtWWVb6oART5TNAfd06CnEbYxlUPcGHrMV6GSxsfsFvTFttzM5E9U00N4MRuuoEvgpiZ9B_rNYGTzhD1YcGLy1X9jt89fBLn7N26DJKjnT7-eV66LlggzhB607bSKkxGhjVM-4Nlk3kLQzbLNdQp7at63JjC7DeiAv946Wug8rG-n9sbilrfqqtfa5H1pPhQSzSnFN3yDbj0wzV4_bB--P5h85dp8siMUPh272gr2FlogU3Zvu1nTOwXWYPyY6TDD2wYHlERtDskns-J325Sx4MylI-JrWGEL0BIeohRA2EaA8hqiFEHYSogxDtIURxDCoa2kOIGghRB6En5PPJ8ezwNHSsHKFEc2cVxqkoGH73yURGgjE8IDAmpLar-UQqkUZKAo9kmSpgRQy8KHk8Vaj4x2USpQDjp2SruWzgGaFlqXg0jWJIx2jYsijjoIOSJZ45YiYmao8wv4a5dCXrNXNKnfvYxIvcLHyuFz63C79H3nSdFrZiy-2PJ144uTM6rTGZI6Bu6_j8Xzu-IPf7r-El2Vq1a9gnd-XVqlq2rxzwfgB3GKvZ
linkProvider Elsevier
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Deep+reinforcement+learning+based+controller+with+dynamic+feature+extraction+for+an+industrial+claus+process&rft.jtitle=Journal+of+the+Taiwan+Institute+of+Chemical+Engineers&rft.au=Liu%2C+Jialin&rft.au=Tsai%2C+Bing-Yen&rft.au=Chen%2C+Ding-Sou&rft.date=2023-05-01&rft.pub=Elsevier+B.V&rft.issn=1876-1070&rft.eissn=1876-1089&rft.volume=146&rft_id=info:doi/10.1016%2Fj.jtice.2023.104779&rft.externalDocID=S1876107023001086
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1876-1070&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1876-1070&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1876-1070&client=summon