Aiming　at　the　difficulties　of　vibration　testing　in　low-speed　and　heavy-duty,　acoustic　emission　testing　is　proposed　in　this　paper,　and　independent　component　analysis　method　is　used　in　acoustic　emission　signals　processing.　The　traditional　independent　component　analysis　method　is　restricted　to　non-gaussian,　stationary　and　mutually　independent　source　signals,　and　the　number　of　observations　is　assumed　to　be　more　than　that　of　sources,　these　assumptions　in　the　mechanical　fault　diagnosis　are　difficult　to　achieved,　so　fault　diagnosis　methods　for　low-speed　and　heavy-duty　based　on　acoustic　emission　testing　and　EMD-ICA　is　proposed.　The　proposed　method　is　that　the　acoustic　emission　signals　are　decomposed　into　a　series　of　intrinsic　mode　functions　(IMF)　by　the　empirical　mode　decomposition　(EMD)　method,　and　then　the　obvious　impact　and　periodicity　IMFs　and　the　original　signals　are　composed　into　new　signals,　which　was　separated　independent　signals　to　find　the　fault　information.　The　analysis　results　of　project　show　that　the　proposed　method　is　very　effective.