Neural Network Based Explicit MPC for Chemical Reactor Control

TL;DR

This paper demonstrates how deep neural networks can be trained to replicate model predictive control in chemical reactor systems, maintaining constraints while simplifying control implementation.

Contribution

It introduces a neural network approach that mimics MPC behavior, preserving constraints without dependence on weighting matrices, for chemical reactor control.

Findings

Neural network successfully approximates MPC control inputs

Constraints on states and inputs are maintained

Method validated through simulation of a chemical reactor

Abstract

In this paper, we show the implementation of deep neural networks applied in process control. In our approach, we based the training of the neural network on model predictive control. Model predictive control is popular for its ability to be tuned by the weighting matrices and by the fact that it respects the constraints. We present the neural network that can approximate the behavior of the MPC in the way of mimicking the control input trajectory while the constraints on states and control input remain unimpaired of the value of the weighting matrices. This approach is demonstrated in a simulation case study involving a continuous stirred tank reactor, where multi-component chemical reaction takes place.