For the phenomenon that the precise experimental result can't be obtained as the process of studying combustible coal's properties is restricted by many kinds of non-linear factors, the combustible coal's firing temperature and its influencing factors were analyzed from non-linear angle, and the intelligent network is carried into training, learning and predicating by use of practical examples. The results were recorded with requirement of testing, and predication of combustible coal's firing temperature was achieved successfully using BP neural network theory.