Gas　disaster　is　one　of　the　most　serious　disasters　in　coal　mine　safety　production.　Therefore,　it　is　of　great　significance　to　strengthen　the　coal　mine　gas　disasters　forecasting　warning　technology　research　for　improving　the　ability　of　prevention　and　control　gas　disaster　in　coal　mine　and　promoting　the　development　of　digital　mine　in　our　country.　The　biggest　characteristics　of　using　grey　prediction　model　GM　(1,1)　is　that　the　algorithm　is　quite　simple,　and　also,　when　building　the　model,　less　data　can　be　used.　It　is　convenient　for　modelling　and　operation,　but　the　effect　of　forecast　of　the　grey　prediction　model　for　systems　with　volatility　is　not　very　ideal,　and　the　prediction　accuracy　will　reduce　gradually　along　with　the　extrapolation　of　time.　BP　neural　network　has　a　good　performance　for　prediction　of　nonlinear　system,　but　when　the　network　was　trained,　it　often　requires　large　amounts　of　data.　This　paper　is　based　on　the　grey　prediction　model,　using　advantages　of　grey　prediction　that　the　model　algorithm　is　simple　and　the　procedure　of　model　building　needs　less　data,　and　the　BP　neural　network　that　the　performance　of　grey　forecast　model　for　nonlinear　system　prediction　is　good.　We　revise　the　grey　prediction　model　through　BP　neural　network　and　finally　we　build　an　improved　gas　concentration　prediction　model　based　on　grey　theory　and　BP　neural　network,　and　carry　out　a　specific　computer　simulation.　Results　show　that　the　model　effectively　improved　the　precision　of　gas　prediction.