Decoupling Neural Network Model Predictive Control. Algorithm Development and Application to Reactive Distillation Process

      Аннотация: Nowadays, effective control of reactive distillation process has become one of the major challenges facing Process Systems Engineers because of the complex and multivariable nature of this process. Good control of this process can only be achieved by considering and trying to eliminate the interactions existing among all the variables involved in the process. This can be accomplished by employing Decoupling Control. Also, taking the complexity as well as the nonlinearity of the process into consideration, a versatile model like Neural Network model is preferred for its control. This work has addressed these issues by developing the algorithms required to implement Decoupling Neural Network Model Predictive Control on multivariable and complex reactive distillation process. The algorithms, including Simulink diagrams and MATLAB codes, provided in this work will go a long way in contributing to efficient, realizable and convenient control of a complex process like reactive distillation both in academic researches and in industries.

Ключевые слова: optimization, Neural Network, Model predictive control, Matlab/Simulink, Esterification, process identification, decoupling, reactive distillation, Hysys