Need for Optimisation Techniques to Select Neural Network Algorithms for Process Modelling of Reduction Cell

While there exists a broad range of neural networks for a particular task, different neural network architectures are selected depending upon the nature of application in industry. The range of applications covers anything from performance estimation and pattern recognition to process modelling and control. The network selection can be carried out based on economic considerations, such as cost associated with neural network computation time and obtaining data for required model variables. While each of the selected models can be a possible solution, depending upon the performance criteria, they all can be ranked from most suitable to least suitable for a particular application. In this paper, appraisal of neural networks for three industrial applications, involving process modelling of reduction cells for aluminium production, is discussed. Regression analysis techniques and six neural network models are assessed for their performance, using specific assessment criteria. It is shown that there is no single model that is most appropriate for each of the assessment criteria considered in each instance, hence, the decision of which neural network model is most suitable for a specific application is complex, particularly as the assessment criteria are not fundamentally of equal significance. It is shown that optimisation techniques are necessary to select an appropriate model for an application.

[1]  James L. McClelland,et al.  Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations , 1986 .

[2]  R. Palmer,et al.  Introduction to the theory of neural computation , 1994, The advanced book program.

[3]  Karri,et al.  Combined Kohonen and RBF Networks to Predict Electrolyte Additives in Hall-Heroult Cell , 2000 .

[4]  T. Kohonen.  Self-organized formation of topographically correct feature maps , 1982 .

[5]  T. Kohonen.  Self-Organized Formation of Correct Feature Maps , 1982 .

[6]  Donald F. Specht,et al.  A general regression neural network , 1991, IEEE Trans. Neural Networks.

[7]  F. Frost,et al.  Determining the influence of input parameters on BP neural network output error using sensitivity analysis , 1999, Proceedings Third International Conference on Computational Intelligence and Multimedia Applications. ICCIMA'99 (Cat. No.PR00300).

[8]  Charles W. Butler,et al.  Naturally intelligent systems , 1990 .

[9]  David Lowe.  Radial basis function networks and statistics , 2000 .

[10]  H. J. Arnold.  Introduction to the Practice of Statistics , 1990 .

[11]  Bart Kosko,et al.  Neural networks and fuzzy systems: a dynamical systems approach to machine intelligence , 1991 .

[12]  Teuvo Kohonen,et al.  Self-Organization and Associative Memory , 1988 .

[13]  Teuvo Kohonen,et al.  Self-organization and associative memory: 3rd edition , 1989 .

[14]  Timothy Masters,et al.  Advanced algorithms for neural networks: a C++ sourcebook , 1995 .

[15]  D. Broomhead,et al.  Radial Basis Functions, Multi-Variable Functional Interpolation and Adaptive Networks , 1988 .

[16]  Jacek M. Zurada,et al.  Introduction to artificial neural systems , 1992 .

[17]  Karri,et al.  Intelligent Control of Aluminium Reduction Cells using Backpropagation Neural Networks , 2000 .

[18]  Teuvo Kohonen,et al.  An introduction to neural computing , 1988, Neural Networks.

[19]  V Karri,et al.  Effect of altering the Gaussian function receptive field width in RBF neural networks on aluminium fluoride prediction in industrial reduction cells , 1999, ICONIP'99. ANZIIS'99 & ANNES'99 & ACNN'99. 6th International Conference on Neural Information Processing. Proceedings (Cat. No.99EX378).

[20]  Maureen Caudill,et al.  Understanding Neural Networks; Computer Explorations , 1992 .

[21]  Tarun Khanna,et al.  Foundations of neural networks , 1990 .

[22]  V Karri.  RBF neural network for thrust and torque predictions in drilling operations , 1999, Proceedings Third International Conference on Computational Intelligence and Multimedia Applications. ICCIMA'99 (Cat. No.PR00300).

[23]  S. H. Huang,et al.  Artificial neural networks in manufacturing: concepts, applications, and perspectives , 1994 .

[24]  F. Frost,et al.  Optimum back propagation network conditions with respect to computation time and output accuracy , 1999, Proceedings Third International Conference on Computational Intelligence and Multimedia Applications. ICCIMA'99 (Cat. No.PR00300).

[25]  T. Kohonen.  Adaptive, associative, and self-organizing functions in neural computing. , 1987, Applied optics.

[26]  F. Frost,et al.  Performance comparison of BP and GRNN models of the neural network paradigm using a practical industrial application , 1999, ICONIP'99. ANZIIS'99 & ANNES'99 & ACNN'99. 6th International Conference on Neural Information Processing. Proceedings (Cat. No.99EX378).