MACHINE LEARNING FOR PREDICTING OIL MASS FLOW: TRAINING PROCESS AND RESULTS

The Central Oil Processing Facility (COPF) is a critical element of oil and gas infrastructure, ensuring the purification and stabilization of oil before its further transportation or processing. Challenges such as equipment instability, measurement inaccuracies, and high operating costs lead to downtime and financial losses. This study proposes the use of machine learning (ML) methods, such as LSTM and Attention, which enable effective anomaly detection and prediction of potential equipment failures. These approaches enhance system performance and reduce costs. The presented examples demonstrate significant improvements in productivity, cost reduction, and oil quality, while also discussing the prospects of further integration of artificial intelligence into the oil and gas industry.