Neural network controller identification for refining process

Abstract

The article discusses the task of identifying a neural network controller for the installation of rectification of oil refining production. A rectification process research model is used to evaluate the effectiveness of the controller. The control parameters of the rectification process that are used to identify the controller and evaluate its effectiveness are determined. In a numerical study, the possibility of using a neural network controller to control the rectification process is shown. As a basic option for a comparative study, we used a PID-regulator, which is the standard version in production today. The advantage of a neural network controller in controlling processes in the context of the implementation of various target trajectories is shown. The proposed model of a neural network controller can be adapted and used for computer control of the rectification process.

Control Synthesis as Machine Learning Control by Symbolic Regression Methods

“…Typically, machine learning implies the use of neural network technologies [2][3][4][5][6]. At first glance, it seems that a variety of neural network structures can satisfy any control problems.…”