Neural networks for chemical process

Abstract

Neural Networks have been studied and applied to solving several engineering problems increasingly due to their capabilities in providing good mathematical models of processes, in particularly, the applicability on chemical processes such as nonlinear process identification and control. Therefore, this paper presents the neural networks with respect to the implementation on chemical processes including details regarding the development of neural networks on chemical processes step by step from initially the selection of neural networks structure, training, testing and validating. Here, the case study is defined on steel pickling process with demonstration on the implementation of the neural networks mainly in two parts: process models and control. Each part has been concluded focusing on the structure and approach. This study has shown the effectiveness and applicability of the neural networks on the problem solving of chemical processes which give satisfactorily performances with respect to process modeling and control, and their capability in the application to highly nonlinear and complex chemical processes.