A　method　that　effectively　de-noising　vibration　signal　and　extracting　fault　feature　is　necessary　when　diagnose　the　switching　synchronism　fault　of　low　voltage　circuit　breaker(LVCB).　A　vibration　signal　analysis　method　based　on　Hilbert-Huang　transform(HHT)　is　put　forward,　in　which　intrinsic　mode　function(IMF)　components　are　extracted　by　empirical　mode　decomposition(EMD)　to　reflect　local　characteristic　of　LVCB　vibration　signal,　and　top　five　of　IMF　components　energy　not　only　is　representation　of　vibration　feature,　but　also　has　de-noising　work.　By　analyzing　to　LVCB　vibration　signal　in　time-domain,　kurtosis　and　mean　square　value　are　used　as　auxiliary　mechanic　characteristic　index.　With　feature　vector　of　energy　ratio　of　front　five　IMF　components,　kurtosis　and　mean　square　value,　the　neural　network　based　on　particle　swarm　optimization(PSO)　and　radial　basis　function(RBF)　is　expounded　to　model　fault　recognition　of　asynchronous　switching　for　LVCB.　By　results　of　experiment　and　simulation,　it　is　effective　to　analysis　switching　synchronism　with　intelligent　technology　of　comprehensive　such　as　time-domain　analyze,　EMD　and　PSO-RBF　neural　network,　which　is　a　new　diagnosis　method　for　circuit　breaker　in　special　for　three　phases　switching　synchronism　vibration　of　LVCB.　©,　2014,　Chinese　Machine　Press.　All　right　reserved.