It　is　a　comprehensive　problem　to　accurately　set　the　ink　key　opening　of　every　ink　zone　for　a　specific　press　according　to　the　color　coverage　of　the　graphical　content　which　will　be　printed　on　the　sheet.　In　this　paper,　a　new-type　neural　network　is　proposed　to　preset　the　key　opening　of　the　ink　zones,　which　combines　the　fuzzy　ART　neural　network　and　BP　neural　network,　and　is　with　the　capability　of　half-incremental　learning　and　good　generalization.　The　actual　qualified　printed　sheets　are　used　as　training　samples　to　train　fuzzy　ART-BP　network,　so　that　the　nonlinear　mapping　relationship　among　graphic　digital　information,　printing　conditions,　and　ink　key　control　parameters　is　established.　The　experimental　results　indicate　that　the　fuzzy　ART-BP　network　algorithm　could　effectively　shorten　adjusting　time　of　the　ink　key　setting　and　improve　printing　efficiency.　©　2013　Springer-Verlag.