PNIPU: Algorithm for optimizing the process of oil treatment based on neural networks and analytical models

Main article: APCS - typical structure

2022: Improved Automated Process Control for Oil Treatment

Perm Polytech reported on October 13, 2022 that University scientists have found a way to increase the profit of enterprises in the preparation of oil.  

Researchers from Perm Polytechnic have improved automated control of the technological process of preparing "black gold" in order to reduce production costs and improve the quality of the finished product. To do this, they used an optimization algorithm based on neural networks and analytical models. Preparation of oil to marketable quality is a process that consists of many stages. In the process, it is important to ensure optimal process parameters in order to increase the efficiency of equipment use. The result that enterprises are striving for is an increase in profits.

The mathematical foundations of algorithms make it possible to implement them on domestic software and computer complexes of automated process control systems, and can also replace them in foreign computer modeling systems.

During field operation, the composition and properties of the oil emulsion entering the oil treatment unit change. The systems of automatic emulsion flow control and laboratory monitoring of oil water cut, as well as the software and hardware complex for controlling this process allow you to quickly track the parameters of the process mode. The quality of the finished "black gold" and the profit of the enterprise depend on the composition of the equipment and the mode of its operation, said one of the researchers, senior lecturer at the Department of Equipment and Automation of Chemical Industries of the Perm Polytechnic, Tatyana Karanevskaya.

Scientists of the Perm Polytechnic proposed an algorithm for optimizing processes for its implementation in the control system of the oil processing plant. It allows you to determine the most effective values ​ ​ of the parameters of the technological mode of operation of the equipment and the flow rate of the oil emulsion, which ensures maximum profit in the sale of finished products. The algorithm is based on analytical models of technological processes, the principle of Bellman optimality for multi-stage production and artificial neural networks. As a result, optimal operating modes of the plant are determined depending on the composition and properties of the oil emulsion. The efficiency of solving the problem is ensured by applying the principle of optimality of multi-stage processes. 

The scientists determined the control parameters for the main processes: separation, dehydration and heating of the oil emulsion. They also developed analytical models of technological processes and prepared training samples for neural networks. Their use makes it possible to determine the optimal values ​ ​ of the parameters of the technological regime, which ensure the necessary quality of "black gold" and obtain maximum profit from the existing oil processing plant, notes the project manager, professor of the Department of Equipment and Automation of Chemical Industries of the Perm Polytechnic, Doctor of Technical Sciences, Alexander Shumikhin.

The researchers confirmed the performance and accuracy of neural network models. They also evaluated the efficiency of the plant at the optimal values ​ ​ of process parameters and the permissible quality of oil. Compared to the existing operating mode of the equipment, the implementation of the optimal technological mode will reduce the cost of oil processing by 15%. The use of the optimality principle and the neural network approach reduces the cost of time and computational resources to optimize processes. According to scientists, this method of optimizing the technological process can be introduced into the operation of an automated operational and control system in the field of oil processing.