Geothermal power plants, harnessing the Earth’s internal heat, are an essential catalyst for the transition to clean energy. However, the reliable operation of these plants requires effective maintenance strategies. This paper explores predictive maintenance in geothermal power plants, focusing on the application of Autoencoders and XGBoost. Autoencoders, mainly used for unsupervised learning, are apt for detecting anomalies in normal equipment behavior. XGBoost is an efficient ensemble model suitable for predicting the remaining useful life of critical components. Both algorithms can be used to make interpretable models that can help pinpoint the root cause of equipment failure. This paper reviews research applying these strategies to analogous systems, emphasizing the need for solutions tailored to geothermal contexts. It also highlights challenges, including limited data and literature, that underscore the need for collaboration between geothermal experts and computer scientists to unlock the full potential of predictive maintenance in geothermal power plants.

Press the Back button in your browser, or search again.

Copyright 2024, Stanford Geothermal Program: Readers who download papers from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the original publisher.