Search

Abstract: This contribution describes the pattern recognition based data analysis of an existing industrial batch dryer, and the comparison of three artificial intelligence techniques suited to perform classification tasks: neural networks trained using the Levenberg–Marquardt and the Levenberg–Marquardt method with Bayesian regularization, the neuro-fuzzy model based on clustering and grid partition, and the Takagi–Sugeno fuzzy models. The classifiers are used to quantify the dryer batch time and its variation during a certain production period, thus the motivation behind the work is genuine. The presented pattern recognition method implements a supervised learning approach and is based on pressure measurement profiles recorded by the plant data management software—the PI System from OSIsoft. It is found that the neural networks trained with the Bayesian regularization have shown the most robust classification performance with respect to separation threshold selection. Furthermore, it is concluded that the application of artificial intelligence techniques in real chemical manufacturing facilities is feasible and provides useful information for process performance monitoring purposes. The pattern recognition findings presented in this paper are not case specific and can be directly used for the monitoring of a large variety of drying processes since the pressure profile features – vacuum check, pressure decrease, vacuum break – do not depend on the chemicals which are dried. Since the development of the artificial intelligent classifiers is presented in detail and step by step, this work may be interesting as a pattern recognition tutorial for chemical engineers. [Copyright &y& Elsevier]

Abstract: A novel method is presented for the rigorous propagation of uncertainties in initial concentrations and in dosing rates into the errors in the rate constants fitted by multivariate kinetic hard-modelling of spectroscopic data using the Newton–Gauss–Levenberg/Marquardt optimisation algorithm. The method was successfully validated by Monte-Carlo sampling. The impact of the uncertainties in initial concentrations and in the dosing rate was quantified for simulated spectroscopic data based on a second and a formal third order rate law under batch and semi-batch conditions respectively. An important consequence of this study regarding optimum experimental design is the fact that the propagated error in a second order rate constant is minimal under exact stoichiometric conditions or when the reactant with the lowest associated uncertainty in its initial concentration is in a reasonable excess (pseudo first order conditions). As an experimental example, the reaction of benzophenone with phenylhydrazine in THF was investigated repeatedly (17 individual experiments) by UV–vis and mid-IR spectroscopy under the same semi-batch conditions, dosing the catalyst acetic acid. For all experiments and spectroscopic signals, reproducible formal third order rate constants were determined. Applying the proposed method of error propagation to any single experiment, it was possible to predict 80% (UV–vis) and 40% (mid-IR) of the observed standard deviation in the rate constants obtained from all experiments. The largest contribution to this predicted error in the rate constant could be assigned to the dosing rate. The proposed method of error propagation is flexible and can straightforwardly be extended to propagate other possible sources of error. [Copyright &y& Elsevier]