Abstract: This study introduces an innovative method utilizing wide-area measurement data to detect and identify abnormal events in power systems. The proposed approach compresses phasor measurement unit(PMU) data using an unequal distance compression algorithm and employs local outlier factor(LOF) technology for event detection. Additionally, a generator synchronization identification method based on fuzzy equivalence relation(FER) clustering is introduced, with ten synchronization indicators describing generator set characteristics. Simulation analysis of the WECC179 node system confirms that the proposed method effectively detects abnormal power system events, identifies synchronous generator groups, and accurately locates faults, offering crucial support for stability analysis and control.