(561e) Modelling of Multivariable Chemical Processes Using a Nonlinear Autoregressive Model with Exogenous Input

Process development and continuous request for productivity led to an increasing complexity of industrial units. In order to operate a successful plant or process, continuous improvement must be made in the areas of safety, quality and reliability. Improvement in these areas will lead to cost reductions which help make the plant a viable operation in a competitive market. The intrinsic highly nonlinear behavior in the industrial process, especially when a chemical reaction is used, poses a major problem for the formulation of good predictions and the design of reliable control systems. The main contribution of this paper is to obtain a reliable model of a process dynamic behavior. The use of this model is used in order to reflect the normal behavior of the process and allows distinguishing it from an abnormal one. The simplicity of the developed neural model under all regimes (i.e. steady-state and unsteady state), used in this case is realized by means of a NARX (Nonlinear Auto-Regressive with eXogenous input) model and by an experimental design. The proposed neural model is implemented by training a Multi-Layer Perceptron Artificial Neural Network (MLP-ANN) with input-output experimental data. An analysis of the inputs number, hidden neurons and their influence on the behavior of the neural predictor is carried out. The performance of the proposed model has been tested on a real plant as a distillation column under vacuum conditions. Validation statistical criteria are used to validate this experimental data. Satisfactory agreement between identified and experimental data is found and results show that the neural model predicts better the evolution of the process dynamics.