Neural Network Approach to Ranking Factors Affecting Energy Efficiency of Oil Production

Abstract

The article describes the relevance of the topic, considers the features of approaches to factor analysis of the energy efficiency of the artificial oil lift. Comparative analysis of the approaches used in assessing the current and projected level of energy efficiency in relation to the planned values is carried out. An approach to the ranking of energy efficiency factors based on intelligent methods is proposed. An assessment of the modern possibilities of automating the factor analysis of the energy efficiency of artificial oil lift is given. To develop an effective methodology for the analysis and planning of energy efficiency based on a relevant factor model, an approach to ranking the factors influencing the energy efficiency of artificial lift using an artificial neural network is proposed. When solving the problem, the method of factor analysis of specific power consumption, the L-BFGS algorithm. Various sets of factors are considered, they were ranked according to the shares of significance, the procedure for excluding factors were carried out on the basis of paired correlation dependences between them. A correlation matrix is built for the adjusted set of factors. On the basis of an expert analysis of the results obtained, their relevance to the cause-and-effect relationships manifested in the practice of operating a mechanized well stock was assessed. The use of the proposed approach to factor analysis using an artificial neural network will improve the reliability of control over the energy efficiency of artificial oil lift. The developed model can be included in the algorithmic and software and hardware support of a promising automated energy efficiency control system and can be used in decision-making by specialists who plan, monitor and predict energy efficiency indicators and assess the results of the implementation of energy-saving measures.