Based　on　vector　wavelet　kernel　function,　a　method　for　analog　circuit　diagnosis　based　on　genetic　algorithm　(GA)　and　least　squares　wavelet　support　vector　machine　(LSWSVM)　is　proposed.　Using　wavelet　package　as　a　tool　for　extracting　feature,　the　GA-LSWSVM　is　then　applied　to　the　filet　circuit　after　training　by　GA;　the　simulation　results　have　shown　that　the　method　can　enhances　the　accuracy　and　generalization　ability.　©　2010　IEEE.