"Synthetic IBMS Alarm Event Dataset with Ground Truth for Alarm Correlation and Root Cause Analysis"

"This dataset provides a synthetically generated Building Management System (BMS) alarm dataset designed for research in alarm correlation, root cause analysis (RCA), and alarm flood detection in smart building environments. The dataset simulates realistic alarm behavior from a large-scale integrated BMS comprising HVAC, power distribution, fire detection and alarm suppression systems (FDASS), CCTV and access control, pumping, and diagnostic subsystems. It contains 179,466 alarm records spanning 180 days, with 15,426 annotated root cause events across 24 distinct fault propagation scenarios and 20 event types. Each alarm is annotated with ground truth labels including root event identification, correlation group membership, propagation delays, and alarm flood flags. The dataset addresses the scarcity of publicly available labeled BMS alarm data for benchmarking alarm correlation and RCA algorithms."