Reservoir production capacity prediction of Zananor field based on LSTM neural network

Abstract

This paper aims to explore the application of artificial intelligence in the petroleum industry, with a specific focus on oil well production forecasting. The study utilizes the Zananor field as a case study, systematically organizing raw data, categorizing different well instances and production stages in detail, and normalizing the data. An individual long short-term memory (LSTM) neural network model is constructed with monthly oil production data as input to predict the monthly oil production of the experimental oilfield. Furthermore, a multivariate LSTM neural network model is introduced, incorporating different production data as input sets to enhance the accuracy of monthly oil production predictions. A comparative analysis is conducted with particle swarm optimization optimized recurrent neural network results. Finally, gray relational analysis and principal component analysis methods are compared in feature selection. Experimental results demonstrate that the LSTM model is more suitable for the study area, and the multivariate model outperforms the univariate model in terms of prediction accuracy, especially for monthly oil production. Additionally, gray relational analysis exhibits higher accuracy and greater applicability in feature selection compared to principal component analysis. These research findings provide valuable guidance for production forecasting and operational optimization in the petroleum industry.