The　arc　fault　is　one　of　most　important　reasons　causing　electrical　fire　accidents.　The　morphological　wavelet　is　introduced　into　the　researching　about　arc　fault　in　this　paper　to　solve　this　problem.　The　analysis　module　of　morphological　wavelet　is　developed　by　LABVIEW.　Combined　with　the　new　testing　system,　the　normal　and　fault　arc　waveforms　of　some　residential　loads　are　sampled　and　analyzed　by　the　module.　Also　the　characters　of　the　arc　fault　which　occurs　in　different　positions　with　the　mixed　loads　are　analyzed.　It　makes　a　new　exploration　for　setting　up　an　intelligent　universal　diagnostic　model　about　arc　fault.