Mixed　nitrates　with　relative　low　melting　point　have　widely　been　used　as　thermal　storage　medium　in　the　concentrated　solar　power　stations.　However,　their　maximum　working　temperatures　aren＇t　high　enough　to　integrate　with　more　efficient　power　generation　systems　such　as　the　supercritical　CO2　cycle.　In　the　present　work,　mixed　nitrates　with　elevated　working　temperature　were　obtained　by　adding　KCl/K2CO3.　With　the　help　of　FactSage　software　and　back-propagation　(BP)　algorithm　combined　with　genetic　algorithm　(GA),　the　eutectic　compositions　and　melting　points　of　NaNO3-KNO3-KCl　and　NaNO3-KNO3-K2CO3　were　predicted,　and　the　validity　of　BP　algorithm　combined　with　GA　was　verified.　Based　on　DSC　and　TG　curves,　the　melting　points　of　the　as-prepared　NaNO3-KNO3-KCl　and　NaNO3-KNO3-K2CO3　are　206.12　degrees　C　and　207.24　degrees　C,　respectively,　which　are　in　agreement　with　predicted　values.　Moreover,　their　decomposition　temperatures　were　increased　to　more　than　640　degrees　C.　In　addition,　the　specific　heat　capacity　of　both　mixed　molten　salts　was　measured,　and　it　also　was　calculated　by　using　FactSage　software.　This　work　provides　two　new　ternary　molten　salts　based　on　nitrates　with　higher　working　temperature　and　develops　the　prediction　method　on　phase　diagram　calculation　and　machine　learning.