A　mathematical　model　for　patterns　of　the　printing　quality　control　chart　is　established,　and　the　data　of　printing　quality　is　simulated　based　on　Monte　Carlo　method,　Then　the　complexity　of　the　sample　data　is　reduced　by　using　the　method　of　standard　transformation　and　linear　encoding.　A　4-layer　BP　neural　network　model,　as　24-18-16-4,　is　established　through　the　experiments,　and　a　scaled　conjugated　gradient　training　algorithm　is　adopted　to　enhance　the　stability　and　convergence　of　the　network.　The　paper　uses　different　capacity　of　training　samples　in　pattern　recognition　for　control　chart,　and　the　recognition　accuracy　achieves　95.87%.　Results　of　experiments　show　that　this　method　can　improve　the　level　of　quality　control　and　degree　of　automation　for　printing　enterprise.