The Method of Constructing a Neural Network Cascade for Simulating Virtual Sensors in the Concept of "Smart" Oil Storage Facility for Agricultural Purposes

Abstract

Technological parameters and technical level of the equipment at an oil storage facility influence motor fuel's quality and its waste during reception, storage and transfer. The use of intelligent systems during the oil storage and handling process enhances quality preservation and reduction of motor fuel waste caused by evaporation, oxidation and hydration while stored in above-ground horizontal steel tanks. Systems managing "smart" oil-storage facilities combine technologies for on-line collection, transmission and storage of information with instant data processing and analysis, and managerial decision-making techniques. A methodological framework, that includes algorithms and a program with sensors to monitor indicators of an automated horizontal oil reservoir, has been developed to control the technological parameters (temperature, pressure, fuel level) of the tanks during storage of light oil products, and to protect fuel against flooding and evaporation. The application of the neural network forecasting technique for fuel waste from evaporation during storage, and processing of the data array, made it possible to calculate with a 98% accuracy rate the gasoline waste during storage in horizontal on-ground tanks with up to 100 m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup> in volume capacity. The application of a neural network enables development of new fuel storage algorithms and calculation of the optimal storage amount to minimise losses. The concept and developed digital intelligent control solutions for oil storage allows combining data in oil management into a single information space, and to control the automated oil storage system with application of neural networks, deep learning and Big Data.