Abstract:  In order to complete the physical fire test efficiently and quickly and to give full play to the potential of data， this paper focuses on the application of data mining and deep network in the entity fire test of automatic sprinkler system. Combined with the physical fire tests of the automatic sprinkler system， this paper selectes the appropriate algorithm, determines the parameters in the model and sets up the verify model of the data mining eventally. The model can reduces the risk,improves the quality of the test.It can shorten the period of test and formate a more intelligent decision of entity test .

Key words: data mining, entity fire test, deep model