With　the　continuous　development　of　the　mobile　phone　industry,　used　mobile　phone　(UMP)　recycling　has　become　a　hot　topic.　At　present,　few　experts　research　UMP　recognization　(UMPR)　methods　based　on　UMP　recycling　equipment.　Deep　forest　(DF)　identification　model　of　UMPs　for　intelligent　recycling　equipment　has　been　proposed,　which　has　many　parameters　to　be　manually　adjusted.　Aim　at　the　above　problems,　this　paper　combines　differential　evolution　(DE)　algorithm　and　DF　model　to　propose　a　method　for　identifying　UMPs.　First,　the　candidate　multi-scale　feature　parameters　and　DF　parameters　are　given　out.　Next,　we　use　DE　to　find　the　optimal　parameters　with　DF　accuracy　as　the　evaluation　criterion.　Finally,　we　feed　the　obtained　optimal　parameters　into　the　DF　model　to　construct　the　final　UMPR　model.　Simulation　results　based　on　mobile　phone　pictures　of　the　Ministry　of　Industry　and　Information　Technology　prove　the　effectiveness　of　the　method.