Research on the material fire risk index based on the GA-BP neural network

Abstract

To improve the material fire risk index (IFHI) forecast precision and accuracy, we raise a way that the prediction model consists of the genetic algorithm (GA) and BP neural network. There are many merits in this model. For example, the weights and thresholds of BP neural networks can be improved by the advantages that the genetic algorithm can reach the peak of the forecasting accuracy of the whole situation. And we pose MSE, MAE, and MAPE to measure the effectiveness. The result of MSE, MAE, and MAPE is 0.005843, 0.0183, 1.08158. The experimental results show that the IFHI of the polymer can be accurately predicted by GA-BP neural networks.