Pleased to announce the publication of a new software paper in the journal SoftwareX: “ReAutoPrognostics: Reliability-driven AutoML pipelines for machinery degradation prognostics.” co-authored by Stefanos Kontos, Alexandros Bousdekis, Katerina Lepenioti and Gregoris Mentzas.

Predictive maintenance is critical for minimizing downtime, but building reliable ML pipelines for machinery degradation is often complex and highly manual. To solve this, we developed ReAutoPrognostics, an open-source, Python-based software framework designed to automate and standardize this process.

Key features of ReAutoPrognostics: ✅ Integrates Weibull-based degradation modeling and time-domain feature extraction. ✅ Leverages multiple Automated Machine Learning (AutoML) frameworks into a single workflow. ✅Automatically transforms multivariate sensor data into predictive models—no manual hyperparameter tuning required. ✅ Enables reproducible benchmarking for accuracy and computational cost.

By bridging reliability engineering with advanced AutoML, we hope this framework helps researchers and practitioners build more accurate, scalable predictive maintenance solutions.

🔗 Read the paper here: https://www.sciencedirect.com/science/article/pii/S235271102600261X