The　rapid　development　of　distributed　edge　intelligence　in　the　Internet　of　things　(IoT)　scenarios　has　resulted　in　massive　edge　devices　continuously　generating　data,　leading　to　great　pressure　on　traditional　centralized　trustworthy　processing　security　protection　systems.　With　the　continuous　enhancement　of　edge　device　computing　capabilities,　it　is　now　possible　to　achieve　the　goal　of　trusted　decision　while　generating　edge　data.　Therefore,　research　on　intelligent　trusted　decision　methods　for　data　production　is　an　urgent　issue　to　be　addressed.　To　tackle　this　problem,　an　edge　intelligent　trusted　decision　mechanism　is　designed.　In　our　trusted　decision　mechanism,　a　unified　mapping　model　based　on　multidimensional　attributes　of　edge　devices　is　first　constructed.　Based　on　the　descriptive　model,　three　important　trusted　decision　components　(TDCs)　are　included:　static,　dynamic,　and　comprehensive　trusted　decision　component　(TDC).　In　static　TDC,　unidirectional　and　bidirectional　trusted　decision　functions　of　edge　devices　are　defined,　respectively.　In　dynamic　TDC,　dynamic　direct　and　indirect　recommendation　trusted　decision　are　included.　The　dynamic　comprehensive　trusted　decision　is　calculated　by　effectively　weighting　them.　In　comprehensive　TDC,　a　particle　swarm　optimization　algorithm　is　introduced　to　adaptively　adjust　the　weighting　factors　of　static　and　dynamic　TDC,　and　the　comprehensive　trusted　decision　is　calculated.　To　verify　the　performance　of　our　proposed　trusted　decision　mechanism,　three　broad　experiments　are　carried　out.　Static　and　dynamic　TDCs　are　verified　on　randomly　generated　data　sets,　while　comprehensive　TDC　is　verified　on　a　real　data　set.　The　simulation　results　show　that　our　static　and　dynamic　TDCs　can　timely　and　accurately　detect　malicious　devices.　Moreover,　the　comprehensive　TDC　based　on　the　real　data　set　performs　well　on　different　performance　measurement　indicators.