Fuzzy Inference Systems for Predictive Maintenance in Particle Accelerators

Abstract

Particle accelerators play a pivotal role in scientific research, and their uninterrupted operation is crucial. Predictive maintenance techniques are essential to ensure the reliability and efficiency of these complex systems. In this study, we investigate the application of Fuzzy Inference Systems (FIS) for predictive maintenance in particle accelerators. The FIS offers a nuanced approach to decision-making by considering the interplay of multiple parameters and handling uncertainty inherent in sensor data. We present a case study where Vibration Level and Temperature are monitored in real time, and a FIS is used to calculate maintenance priorities. The results demonstrate that the FIS consistently assigns low maintenance priorities during observed time points, indicating its potential to reduce unnecessary interventions and false alarms. We discuss the advantages of FIS, such as nuanced decision-making and adaptability to complex systems, along with areas for future research and improvement.