Mine ventilation resistance is decided to coefficient of frictional resistance. and it is very vital to airflow distribution of using air place, With the special conditions of roadways of I-beam supporting in the article, the coefficient of frictional resistance is calculated through an improved BP artificial neural network model, and the article analyses the difference between BP neural network calculation and actual measure. Suggest to apply the BP artificial neural network on pattern recognition of coefficient of frictional resistance. The main advantages of it include simple structure, efficient convergence, precise forecasting, etc. The precision in calculation of model is higher, it has a certain practical value.