人工神经网络法预测炸药爆速的研究

Predicting the Detonating Velocity of Explosives by Artificial Neural Network

以分子连接性指数作为炸药分子的结构描述符,利用BP人工神经网络算法,通过对40种炸药的训练建立了炸药分子结构与爆速之间的定量模型,并对另外14种炸药进行了爆速预测。结果表明,该模型较好地反映了炸药分子结构与爆速之间的关系,具有较高的预报精度。该方法为新型炸药分子设计时正确估算其爆速提供了一条新的途径。

This paper discussed the quantitative relationship between the detonation velocity and the structure of explosives. Molecular connecting indices(MCIs) are used to represent the structure. Based on the back propagation algorithm, a quantitative model was established after a training process to a train set containing 40 explosives was completed. With the model a forecasting test was made to a predict set of 14 explosives which didn't belong to the train set. The results showed that the yield model reflected the complex relationship between the structure and the detonation velocity, and had high predicting accuracy. This bring forward a novel method for estimating the detonation velocity when designing new explosives.