This　paper　introduces　a　method　of　control　for　a　new　type　of　household　central　air　conditioning　system　which　is　designed　for　a　new　building　mode　with　multilayer　envelope.　According　to　system＇s　characteristics　such　as　multi-variable,　large　hysteresis　and　non-linear,　this　paper　establishes　suitable　system　architecture　and　proposes　a　control　method　based　on　fuzzy　neural　network.　With　fuzzy　neural　network＇s　forward　propagation　and　feedback,　the　system　completes　fuzzy　reasoning　and　self-learning　process.　So　it　can　quickly　and　accurately　analyze　the　measure　data　in　order　to　achieve　real-time　control　effects.　At　last　experimental　results　show　that　the　control　method　is　feasible　and　has　a　good　dynamic　and　steady　-　state　performance.　©　2013　TCCT,　CAA.