Abstract
Feedforward neural networks have been used for kinetic parameters determination and signal filtering in differential scanning calorimetry. The proper learning function was chosen and the network topology was optimised using an empiric procedure. The learning process was achieved using various simulated thermoanalytical curves computed for several thermodynamic and kinetic parameters. Various amounts of simulated noise were added on the power signals. The resilient-propagation algorithm led to the best minimisation of the error computed over all the patterns. Relative errors on the thermodynamic and kinetic parameters were evaluated and compared to those obtained with the usual thermal analysis methods. The results obtained are very promising, and the errors are much lower than with usual methods, especially in the presence of noisy signals. This study shows that simulated thermoanalytical curves produced by Joule effect may be used for the deconvolution of the response of the apparatus, by using artificial neural networks.
Similar content being viewed by others
References
Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators.Neural Networks 1989; 2: 359
Irie B, Miyake S. Capabilities of three layered perceptrons.IEEE Second Int Conf on Neural Networks, San Diego 1988; 1: 641
Sbirrazzuoli N, Brunei D, Elégant L. Different kinetic equations Analysis (Special review).J Therm Anal 1992; 38: 1509–1524
Sbirrazzuoli N, Girault Y, Elégant L. Kinetic investigation of the polymerization of an epoxy resin by DSC and temperature profile determination during cure.Die Angewandte Makromolekulare Chemie 1993; 211: 195–204
Sestak J. Thermophysical Properties of Solids. In Comprehensive Analytical Chemistry, G. Shevla (ed.), 1984; 12: Part D, Elsevier, Prague
Sbirrazzuoli N, Girault Y, Elégant L. Simulations for evaluation of kinetic methods in differential scanning calorimetry. Part 1 — Comparisons between Freeman-Carroll, Ellerstein, Achar-Brindley-Sharp and multiple linear regression methods.Thermochimica Acta 1995; 2280: 1–18
Sbirrazzuoli N. Simulations for evaluation of kinetic methods. Part 2 — Effect of additional noise on single peak methods.Thermochimica Acta 1995; 2611: 1–16
Cesari E, Gravelle PC, Gutenbaum J, Hatt J, Navarro J, Petit JL, Point R, Torra V, Utzig E, Zielenkiewicz W. Recent progress in numerical methods for the determination of thermokinetics.J Therm Anal 1981; 20: 47–59
McAvoy TJ, Su HT, Wang NS, He M, Norwath J, Semerjian H. A comparison of neural networks and partial least squares for deconvoluting fluorescence spectra.Biotech Bioeng 1992; 40: 53–62
De Weijer AP, Lucasius CB, Kateman G, Heuvel HM, Mannee H. Curve fitting using neural computation.Anal Chem 1994; 66: 23–31
Gallant SR, Fraleigh SP, Cramer SM. Deconvolution of overlapping Chromatographic peaks using a cerebellar model arithmetic computer neural network.Chemometrics and Intelligent Laboratory Systems 1993; 18: 41–57
Coenegracht PMJ, Metting HJ, van Loo EM, Snoeijer GJ, Doornbos DA. Peak tracking with a neural network for spectral recognition.J Chromatogr 1993; 631: 145–160
Fechner T. Nonlinear noise filtering with neural networks: comparison with Wiener optimal filtering.Third Int Conf Artificial Neural Networks, London, IEEE, 1993; 143–147
Leung H, Haykin S. Rational function neural network.Neural Computation 1993; 5: 928–938
Ying L, Astola J, Neuvo Y. A new class of nonlinear filters — Neural filters.IEEE Trans Signal Processing 1993; 41(3): 1201–1222
Blanck T, Brown S. Nonlinear multivariate mapping of chemical data using feed-forward neural networks.Analitica Chimica Acta 1993; 65: 3081–3089
Sundgren H, Winquist F, Lukkari I, Lundstrom I. Artificial neural networks and gas sensor arrays: quantification of individual components in a gas mixture.Meas Sci Technol 1991; 2: 464–469
Livingstone DJ, Hesketh G, Clayworth D. Novel method for the display of multivariate data using neural networks.J Mol Graphics 1991; 9: 115–118
Li Z, Cheng Z, Xu L, Li T. Nonlinear fitting by using a neural net algorithm.Anal Chem 1993; 65: 393–396
Becks KH, Block F, Drees J, Langefeld P, Seidel F. B-quark tagging using neural networks and multivariate statistical methods — A comparison of both techniques.Nucl Instr and Meth 1993; A 329: 501–517
Sbirrazzuoli N, Cachet C, Cabrol-Bass D, Forrest TP. Indices for the evaluation of neural network performances as classifiers: application to structural elucidation in infrared spectroscopy.Neural Computing & Applic 1993; 1: 229–239
SNNS: Stuttgart Neural Network Simulator, University of Stuttgart, Institute for Parallel and Distributed High Performance Systems (IPVR), release 3.1
Sbirrazzuoli N, Girault Y, Elégant L. The Malek method in the kinetic study of polymerization by differential scanning calorimetry.Thermochimica Acta 1995; 249: 179–187
Vyazovkin SV, Lesnikovich AI. An approach to the solution of the inverse kinetic problem in the case of complex processes. Part 1. Methods employing a series of thermoanalytical curves.Thermochim Acta 1990; 165: 273–280
Riedmiller M, Braun H. A direct adaptative method for faster backpropagation learning: the RPROP Algorithm.Proc IEEE Int Neural Networks, San Francisco, 1993; 1: 586–591
Bonnet P, Reggia J, Samuelides M. Réseaux neuronaux: une approche connexionniste de l'intelligence artificielle, Teknea. Paris, 1991
Hérault J, Hutten C. Réseaux neuronaux et traitement du signal, Hermès, Paris, 1994
McCulloch WS, Pitts W. A logical calculus of the ideas imminent in nervous activity.Bull Math Biophis 1943; 5: 115
Hopfield JJ. Neural networks and physical systems with emergent collective computational abilities.Proc Nat Acad Sci 1982; 79: 2554
Wythoff B. Backpropagation neural networks: a tutorial, chemometrics and intelligent laboratory Systems 1993; 18: 115–155
Zupan J, Gasteiger J. Neural networks: a new method for solving chemical problems or just a passing phase?Analitica Chimica Acta 1991; 248: 1–30
Riedmiller M. Advanced supervised learning in multilayer perceptron. From Backpropagation to adaptative learning algorithms.Computer Standards and Interfaces 1994; 16(3): 265–278
Shiffman W, Joost M, Werner R. Optimization of the backpropagation algorithm for training multilayer perceptrons.Proc European Symposium on Artificial Neural Networks, ESANN 93, Brussels 1993; 97–104
Ishiwatari H, Kammruzzaman J, Kumagai Y. Novel noise filtering ability of heterogeneous five layered neural network trained identity mapping using BP algorithm.Proc 35th Midwest Symposium on Circuits and Systems, New York, IEEE 1992; 2: 875–878
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Sbirrazzuoli, N., Brunel, D. Computational neural networks for mapping calorimetric data: Application of feed-forward neural networks to kinetic parameters determination and signals filtering. Neural Comput & Applic 5, 20–32 (1997). https://doi.org/10.1007/BF01414100
Issue Date:
DOI: https://doi.org/10.1007/BF01414100