High voltage (HV) circuit breakers are essential for the safe functioning of power systems，as they control the opening and closing of electrical circuits.However

concealed faults in HV circuit breakers are difficult to detect

and fault diagnosis accuracy often falls short.This study proposes a mechanical fault diagnosis method for high voltage circuit breakers based on the staged searching firefly algorithm combined with an extreme learning machine (SS FA ELM).The method optimizes the random weight matrix of the extreme learning machine using the staged searching firefly algorithm

improving the ELMs ability to nonlinearly model the time domain features of coil current in HV circuit breakers.Experimental results based on fault data from the LW30 252 HV circuit breaker show that the proposed method achieves a diagnostic accuracy of 96.67%.It effectively identifies typical mechanical faults

such as iron core jamming

excessive iron core travel

and operating mechanism jamming

significantly enhancing fault diagnosis accuracy.This approach provides reliable technical support for ensuring the stable operation of power systems.