Multimodal and multi-view predictive maintenance: A case study in the oil industry

This paper explores comprehensive prognostics and health management in the oil and gas industry. It provides a proactive approach to equipment maintenance by detecting potential problems and predicting faults before they occur. Distillers are crucial oil and gas industry components, requiring regular maintenance and monitoring to maintain optimal performance and prevent unplanned maintenance. Deep learning models have shown promising results for predictive maintenance, and multimodality can bring generalization capabilities to these models. This study proposes a multimodal (and multi-view) approach for predictive maintenance in an oil and gas industry distiller dataset. The goal is to demonstrate that this approach can achieve more liable and generalizable models for predictive maintenance than state-of-the-art neural networks when training on a medium-scale dataset.