Forest　fires　will　do　great　harm　to　the　ecological　environment.　At　present,　the　most　important　research　field　is　the　study　of　the　causes　of　forest　fires,　and　the　failure　to　determine　the　causes　will　greatly　hinder　the　prevention,　control　and　safety　management　of　forest　fires.　In　order　to　solve　this　problem,　this　paper　constructs　a　Bayesian　network　to　calculate　the　state　probability　and　sensitivity　after　determining　the　factors　related　to　forest　fire.　Secondly,　on　the　basis　of　constructing　Bayesian　network,　the　Bow-tie　model　is　constructed　according　to　the　consequences　of　forest　fire.　The　constructed　Bow-tie　model　contains　7　influencing　factors　and　5　consequences,　and　26　corresponding　control　measures.　Finally,　through　the　analysis　of　Bayesian　network　and　bow　tie　model,　the　forest　fire　prevention　system　is　constructed.　According　to　the　calculation　results　of　the　Bayesian　network,　the　probability　of　forest　fires　being　at　high　risk　of　occurrence　is　higher　for　three　factors:　agriculture,　forestry　and　water　expenditure,　rural　population　density　and　precipitation.　Changes　in　rural　population　density　and　humidity　have　the　greatest　impact　on　the　occurrence　of　forest　fires.　In　combination　with　the　Bow-tie　model,　human　factors　are　the　most　important　factors　leading　to　the　occurrence　of　forest　fires,　and　the　material,　environmental,　and　management　factors　are　the　secondary　factors.　Based　on　the　forest　control　system,　the　prevention　and　management　measures　are　put　forward.　This　research　can　provide　reference　for　the　control　of　forest　fire　and　reduce　the　occurrence　of　similar　forest　fires.