Process modeling strategy combining analytical and data based techniques - I. NN identification of reaction rates with known kinetics coefficients

This work deals with the fusion of the data-based and analytical submodels in the process engineering. In contrast to the traditional way of process reaction rates identification by an exhaustive and/or expensive search for the most appropriate parameterized structure, a neural network (NN) based procedure is developed here to identify the reaction rates in the framework of a first principles process model. Since the reaction rates are not measured variables a particular network training structure and algorithm are developed to make possible the supervised NN learning. Our contribution is focused on the general modeling of a class of nonlinear systems representing several industrial processes including crystallization and precipitation, polymerization reactors, distillation columns, biochemical fermentation and biological systems. The proposed algorithm is further applied for estimation of the precipitation rate of calcium phosphate and compared with alternative solutions.