Hybrid neural networks models for a membrane reactor

Hybrid Neural Networks Models for a Membrane Reactor Mohammed Al-Yemni Artificial neural networks (ANN) have become an established discipline and have gained extensive interest within chemical engineering. In recent years, research effort has focused on the use of “hybrid artificial neural networks” (HANN) models that combine both the deterministic and the ANN elements. Several methods have been proposed for combining ANN with first principle relations. In this thesis, a new hybrid scheme, which is similar to that developed by Kasprow for a space-independent and time-dependent fedbatch microbial reactor, was developed for a space-dependent steady-state enzymatic reactor. This scheme combines ANN with mass balances and assumed rate expressions. It was shown that this new hybrid scheme performed significantly better than both blackbox ANN model and the hybrid ANN with only mass balance equations. An enzymatic tubular membrane reactor (TMR) was selected as a case study due to the availability of a reliable deterministic/computational model, which can provide simulated process data as needed, as well as its potential industrial importance. Also, two modeling schemes were developed, a fully 'black box' model (BANN), based on ANN technique only, and a simple hybrid model, combining ANN with mass balances (HANN1). Qualitative and quantitative comparisons of the predicted profiles of the above three modeling schemes indicated that the new hybrid scheme (HANN2) performed better than the other two schemes. As a result of adding biochemical knowledge, in the form of mass balances and simplified rate expressions, the new hybrid scheme allowed the process data to be interpolated and extrapolated more accurately.

[1]  Wei-Kang Yuan,et al.  A hybrid neural network-first principles model for fixed-bed reactor , 1999 .

[2]  Rubens Maciel Filho,et al.  Hybrid neural modeling of bioprocesses using functional link networks , 2002 .

[3]  Aline Carvalho da Costa,et al.  A HYBRID NEURAL MODEL FOR THE OPTIMIZATION OF FED-BATCH FERMENTATIONS , 1999 .

[4]  J. Hale,et al.  Ordinary Differential Equations , 2019, Fundamentals of Numerical Mathematics for Physicists and Engineers.

[5]  Dale E. Seborg,et al.  Model structure determination in neural network models , 2000 .

[6]  J. A. Wilson,et al.  A generalised approach to process state estimation using hybrid artificial neural network/mechanistic models , 1997 .

[7]  W. Fred Ramirez,et al.  Optimization of Fed‐Batch Bioreactors Using Neural Network Parameter Function Models , 1996 .

[8]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[9]  James T. Luxhoj.  Neural Networks in Bioprocessing and Chemical Engineering , 1997 .

[10]  S. Feyo de Azevedo,et al.  Hybrid modelling of biochemical processes: A comparison with the conventional approach , 1997 .

[11]  M.H. Hassoun,et al.  Fundamentals of Artificial Neural Networks , 1996, Proceedings of the IEEE.

[12]  Lyle H. Ungar,et al.  A hybrid neural network‐first principles approach to process modeling , 1992 .

[13]  P. Greenfield,et al.  Enzymatic saccharification of cellulose in membrane reactors , 1980 .

[14]  Mark A. Kramer,et al.  Modeling chemical processes using prior knowledge and neural networks , 1994 .

[15]  S. P. Chitra.  Use neural networks for problem solving , 1993 .

[16]  Thomas J. Mc Avoy,et al.  Use of Neural Nets For Dynamic Modeling and Control of Chemical Process Systems , 1989, 1989 American Control Conference.