Neural network approximation of iron oxide reduction process

被引:23
作者
Wiltowski, T [1 ]
Piotrowski, K
Lorethova, H
Stonawski, L
Mondal, K
Lalvani, SB
机构
[1] So Illinois Univ, Coal Res Ctr, Carbondale, IL 62901 USA
[2] So Illinois Univ, Dept Mech Engn & Energy Proc, Carbondale, IL 62901 USA
[3] Silesian Tech Univ, Dept Chem & Proc Engn, Gliwice, Poland
关键词
artificial neural network (ANN); feed-forward multilayer network; iron oxides reduction; isothermal solid-state reaction kinetics; backpropagation error algorithm;
D O I
10.1016/j.cep.2004.08.007
中图分类号
TE [石油、天然气工业]; TK [能源与动力工程];
学科分类号
0807 ; 0820 ;
摘要
The kinetics of Fe2O3 to FeO reduction process was investigated using the thermogravimetric data. The authors' previous experimental results indicated that initially the reduction of hematite is a surface controlled process, however once a thin layer of lower oxidation state iron oxides (magnetite, wustite) is formed on the surface, it changes to diffusion control. In order to analyze the time-behavior of Fe2O3 reduction under various process conditions, artificial neural network (ANN) was tested for modeling of this complex reaction pathways. The data used included the reduction of hematite in various temperatures by CO, H-2 and a mixture of CO and H-2. The ANN model proved its applicability and capability to mimic some extreme (minimum) of reaction rate within specific temperature range, when the classical Arrhenius equation is of limited use. (c) 2004 Elsevier B.V. All rights reserved.
引用
收藏
页码:775 / 783
页数:9
相关论文
共 51 条
[1]   An artificial intelligence treatment of devolatilization for pulverized coal and biomass in co-fired flames [J].
Abbas, T ;
Awais, MM ;
Lockwood, FC .
COMBUSTION AND FLAME, 2003, 132 (03) :305-318
[2]   Use of neural network for modeling of non-linear process integration technology in chemical engineering [J].
Abilov, A ;
Zeybek, Z .
CHEMICAL ENGINEERING AND PROCESSING-PROCESS INTENSIFICATION, 2000, 39 (05) :449-458
[3]   Neural network and hybrid model: a discussion about different modeling techniques to predict pulping degree with industrial data [J].
Aguiar, HC ;
Maciel, R .
CHEMICAL ENGINEERING SCIENCE, 2001, 56 (02) :565-570
[4]   Application of artificial neural networks in modeling limestone-SO2 reaction [J].
Bandyopadhyay, JK ;
Annamalai, S ;
Gauri, KL .
AICHE JOURNAL, 1996, 42 (08) :2295-2302
[5]   A nonlinear observer based on hybrid modelling of chemical reactors [J].
Baratti, R ;
Servida, A .
CHEMICAL ENGINEERING COMMUNICATIONS, 2000, 179 :219-231
[6]   Control of nonlinear chemical processes using neural models and feedback linearization [J].
Braake, HABT ;
van Can, EJL ;
Scherpen, JMA ;
Verbruggen, HB .
COMPUTERS & CHEMICAL ENGINEERING, 1998, 22 (7-8) :1113-1127
[7]   Product and process development using artificial neural-network model and information analysis [J].
Chen, JH ;
Wong, DSH ;
Jang, SS ;
Yang, SL .
AICHE JOURNAL, 1998, 44 (04) :876-887
[8]   An adaptive optimal control scheme based on hybrid neural modelling [J].
Costa, AC ;
Alves, TLM ;
Henriques, AWS ;
Maciel, R ;
Lima, EL .
COMPUTERS & CHEMICAL ENGINEERING, 1998, 22 :S859-S862
[9]   Adaptive hybrid neural models for process control [J].
Cubillos, FA ;
Lima, EL .
COMPUTERS & CHEMICAL ENGINEERING, 1998, 22 :S989-S992
[10]   Hybrid modelling of biochemical processes: A comparison with the conventional approach [J].
deAzevedo, SF ;
Dahm, B ;
Oliveira, FR .
COMPUTERS & CHEMICAL ENGINEERING, 1997, 21 :S751-S756