Prediction of Explosion Heat of Aluminized Explosive Based on Artificial Neural Network

In this study, a three-layer artificial neural network(ANN) model was constructed to predict the explosion heat (Q) of aluminized explosive. Elemental composition was employed as input descriptors and explosion heat was used as output. The dataset of 24 aluminized explosives was randomly divided into a training set (17) and a prediction set (7). After optimized by adjusting various parameters, the optimal condition of the neural network was obtained. Simulated with the final optimum neural network, calculated explosion heat shows good agreement with experimental values. It is shown here ANN is able to produce accurate predictions of the explosion heat of aluminized explosive.