Abstract: Using BP neural network prediction model to predict the fire risk of polymer. Through network simulation and fitting of test samples, the model outputs Integrated Fire Hazard Index, IFHI and comprehensive evaluation was made on the thermal hazard and smoke risk index. The results of prediction of IFHI by BP neural network are similar to that of analytic hierarchy process (AHP), and the fire risk is generally in line with the trend of fire risk index. The results of model prediction and evaluation have good consistency and high coincidence with actual data.

Key words: polymer, fire risk, BP neural network