In　order　to　meet　the　requirements　of　accurate　measurement　and　real-time　monitoring　of　human　internal　temperature,　this　paper　proposes　to　use　multi-band　microwave　radiometer　to　measure　human　internal　temperature.　According　to　the　nonlinear　correspondence　between　the　skin　depth　of　human　skin　and　the　working　frequency　of　microwave　temperature　measurement,　three　frequency　bands　are　designed　to　measure　the　actual　temperature　distribution　of　human　multilayer　skin　tissue,　and　then　a　three-layer　microwave　radiation　forward　transfer　model　is　established.　This　paper　also　studies　the　inversion　method　of　the　model.　For　the　problem　that　BP　neural　network　is　easy　to　fall　into　local　optimization　and　slow　convergence　speed,　this　paper　uses　particle　swarm　optimization　(PSO)　and　Levenberg　Marquardt　(LM)　to　optimize　it.　The　internal　temperature　distribution　of　the　human　body　obtained　by　PSO-LM-BP　is　close　to　the　theoretical　values,　which　verifies　the　correctness　of　the　inversion　method.　©　2021　IEEE.