DEVELOPMENT OF A NEURAL NETWORK FOR MAINTAINING THE QUALITY OF DICHLOROETHANE IN THE MATLAB ENVIRONMENT AND SIMULATION OF A VIRTUAL ANALYZER IN MATLAB SIMULINK

Muravyova Elena Alexandrovna, Maskov Gazinur Gaisarovich

The article presents the development of a neural network (HC) for quality control of dichloroethane in the MATLAB environment, as well as the simulation of a virtual analyzer in MATLAB Simulink. The architecture chosen is a Feed-forward backpropagation network, trained using the Levenberg—Marquardt algorithm. The HC includes a hidden layer with ten sigmoid neurons and an output layer with two linear neurons. The input parameters are the consumption of hydrogen chloride, ethylene, and oxygen, and the output parameters are the concentrations of dichloroethane and CO₂. The following metrics were used to evaluate the effectiveness of training: mean square error (MSE), error distribution histogram and linear regression. The simulation results in Simulink confirmed the adequacy of the NS operation, which allows it to be integrated into the process control system.

Keywords:neural networks, MATLAB, quality control, error back propagation, modeling, virtual analyzer, regression analysis