Optimization of the operation of an oil refining plant using a neural network forecast of its economic efficiency

The problem of optimal control of an oil refining unit is considered. The proposed approach is based on the construction of a predictive model predicting the economic efficiency of the installation. This model is built by training a recurrent neural network. The effectiveness of the proposed approach is shown by the example of the installation of hydrocracking of tar. Optimization of the forecast economy of the installation according to its control parameters allows us to obtain their optimal values that maximize the predicted economic efficiency. The correctness of the recommendations received was evaluated by experts, as well as by conducting a natural experiment.