Unsupervised machine learning for detection of faulty beam position monitors

Unsupervised learning includes anomaly detection techniques that are suitable for the detection of unusual events such as instrumentation faults in particle accelerators. In this work we present the application of a decision trees-based algorithm to faulty BPMs detection at the LHC. This method is fully integrated into optics measurements at LHC and has been successfully used during commissioning and machine developments (MD) for different optics settings in 2018.