Boiler　system　has　the　characteristics　of　high　complexity,　strong　simultaneity,　multiple　measuring　points,　multiple　faults,　and　the　traditional　fault　diagnosis　method　can　not　meet　the　requirements.　A　fault　diagnosis　model　based　on　fuzzy　neural　network　is　built　by　the　combination　of　fuzzy　logic　technology　and　the　improved　BP　neural　network　algorithm.　The　model　is　used　to　fault　diagnosis　and　it　can　better　solve　the　ambiguity,　simultaneity　and　correlation　of　the　boiler　fault.　Through　analyzing　the　fault　example　of　high　temperature　superheater　damage,　and　the　diagnosis　results　of　the　application　of　the　model　are　verified　that　they　are　consistent　with　the　actual　operation　conditions.