M. Fischetti, E. Carrizosa, R. Haaker, J. M. Morales González
Machine Learning (ML) models are increasingly used in industry to detect anomalies. We want to go a step further: once an anomaly is detected, we aim to identify the optimal control strategy that restores the system to safety and minimizes the changes required to do so. We frame this challenge as a counterfactual problem: given a ML model that classifies the system as either "good" or "faulty," our goal is to determine the minimal adjustment to return it to the "good" state. We leverage a mathematical model that finds the optimal counterfactual solution while respecting system-specific constraints.
We applied this novel methodology to the maintenance of real-world offshore-wind-turbine oil transformers, thanks to our collaboration with Vattenfall. Given the high cost and risks associated with offshore wind turbine maintenance, quickly and efficiently bringing the system back to safety with minimal changes has the potential for substantial operational impact.
Palabras clave: Mathematical models, counterfactual analysis, wind energy, optimal control, operations research, machine learning
Programado
Optimization and Learning in Energy
11 de junio de 2025 15:30
Auditorio 2. Leandre Cristòfol