Predictive Maintenance Techniques for Reducing Equipment Failures through Machinery Fault Prediction

Abstract

In the thriving oil industry, particularly within the operations of the Novus Bolashak oil company in Kazakhstan, equipment efficiency and reliability are pivotal. A significant disruption, like the malfunctioning of gearboxes or other machinery, can result in both costly repairs and considerable losses in production. This paper presents a thorough analysis of advanced predictive maintenance techniques to counteract these potential setbacks. Utilizing Machine Learning (ML) and Deep Learning (DL) methods, we aim to enhance the early detection of equipment faults, with a specific focus on the machinery operated by Tengizchevroil, a main operator of the field and subsidiary of Chevron. Our research employs a range of datasets concerning both gearbox and rotatory machinery. The evaluation of various machine learning models revealed that Random Forest (RF) and Deep Neural Network (DNN) models exhibited remarkable proficiency in anticipating potential faults. In turn, this helps reduce downtime and promotes efficient operations for the Novus Bolashak oil company.