Predicting the rate of forest fire spread toward any directions based on a CNN model considering the correlations of input variables

Modeling forest fire spread rate is a complex problem, and the existing models are unable to accurately predict the rate of fires spreading towards any directions. In this paper, a convolutional neural network (CNN)-based model is designed to predict the spread rate of forest fires spreading in any directions and using the spread direction as one of the model’s inputs. Several outdoor burning experiments were designed and conducted in order to obtain a dataset on which the model can be trained and validated. Correlation analysis was performed on the variables, and their positions are arranged in a fourth-order matrix according to the strength of their correlations to reflect the correlations in space for feature extraction by the CNN. A deep neural network (DNN)-based model is also designed for comparison to demonstrate the advantages of considering the correlation between variables. The comparison with the improved Wang’s model proves that the model proposed in this paper has higher prediction accuracy compared with the traditional model. The validation experiments were carried out in terms of fire spread rate or fire line’s position. The proposed spread model can provide the technical support for managing the forest fires.