Incorporating prior knowledge into artificial neural networks - an industrial case study

A case study is presented in which an artificial neural network (ANN) model is developed for use in the model-based optimisation of an industrial batch process. The lack of proper (experimental) training data was compensated for by the introduction of prior knowledge about the process into the modelling procedure. This prior knowledge was used for defining the structure of the model and for determining a sufficiently large augmented training data set. The approach is evaluated and the results compared with an ANN model obtained from a different data set in the classic manner, a semi-empirical model, and a model developed using the principal component regression approach.

[1]  Paul Geladi,et al.  Principal Component Analysis , 1987, Comprehensive Chemometrics.

[2]  Ken-ichi Funahashi,et al.  On the approximate realization of continuous mappings by neural networks , 1989, Neural Networks.

[3]  Jack L. Meador,et al.  Encoding a priori information in feedforward networks , 1991, Neural Networks.

[4]  G. J. Gray.  An Investigation of Open Loop and Inverse Simulation as Nonlinear Model Validation Tools for Helicopter Flight Mechanics , 1998 .

[5]  Atsushi Aoyama,et al.  Nonlinear modelling of LPG %C5 content of catalytic reformer debutaniser column , 1997 .

[6]  George Cybenko,et al.  Approximation by superpositions of a sigmoidal function , 1992, Math. Control. Signals Syst..

[7]  Dezhao Chen,et al.  An improved differential evolution algorithm in training and encoding prior knowledge into feedforward networks with application in chemistry , 2002 .

[8]  D. J. Murraysmith,et al.  Methods for the external validation of contiuous system simulation models:a review , 1998 .

[9]  Stanko Strmčnik,et al.  Experimental testing of flexible recipe control based on a hybrid model , 1999 .

[10]  Giuseppe Storti,et al.  Kinetics of titanium dioxide precipitation by thermal hydrolysis , 1986 .

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

[12]  B. S. Thandaveswara,et al.  A non-linear rainfall–runoff model using an artificial neural network , 1999 .

[13]  Ion-Olimpius Stamatescu,et al.  A practical approach to recipe improvement and optimization in the batch processing industry , 1996 .

[14]  De-zhao Chen,et al.  Prior-knowledge-based Feedforward Network Simulation of True Boiling Point Curve of Crude Oil , 2001, Comput. Chem..

[15]  Babu Joseph,et al.  Predictive control of quality in batch polymerization using hybrid ANN models , 1996 .

[16]  Tor Arne Johansen,et al.  Identification of non-linear systems using empirical data and prior knowledge - an optimization approach , 1996, Autom..

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

[18]  John E. Rijnsdorp,et al.  Integrated Process Control and Automation , 1991 .

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

[20]  Hans Vangheluwe,et al.  The Process of Model Building and Simulation of Ill-Defined Systems: Application to Wastewater Treatment , 1997 .

[21]  H. Hotelling.  Analysis of a complex of statistical variables into principal components. , 1933 .