Prediction of Detonation Velocity and N−O Composition of High Energy C−H−N−O Explosives by Means of Artificial Neural Networks

Abstract

The possibilities of the application of Data Science Methods in predicting certain macroscopic properties have been examined in energetic compounds. Artificial neural networks, one of the most promising methods of Data Science, has been used for predicting detonation velocity based on a trained set comprising of a large data set containing 104 data points extracted from over 65 explosive compounds and compositions with diverse characteristics and properties. The utility of the method has been demonstrated through validation for over 37 explosive compounds again with diverse characteristics constituting to a data set of 74 data points. The usefulness and versatility of the method is clear as it exhibits similar predictive accuracy on comparison with the similar data derived from two other well‐known empirical models. Such predictive capabilities will be a great tool for engineers and scientists working with high energetic explosives for quick and simple prediction of detonation velocity given the chemical composition and vice versa.

Applying machine learning to balance performance and stability of high energy density materials

“…Figure 4 D presents a distribution of the residuals of the XGBoost predictions, which satisfactorily follows normal distributions. The currently obtained value of r is comparable to a previous ANN prediction of D for 65 explosive compounds and compositions, wherein, r = 0.978 for training set and 0.985 for test set ( Chandrasekaran et al., 2019 ). The currently obtained values of R 2 and RMSE are close to another prediction of 54 nitrogen-rich energetic compounds obtained by least square support vector machine (LS-SVM) method, in which RMSE = 0.17 km·s −1 , r = 0.96 for the training set and RMSE = 0.17 km·s −1 , r = 0.97 for the test set ( Wang et al., 2014 ).…”

“…Till date, the models that have been designed, developed, and employed for ML of HEDMs include multiple linear regression, artificial neural network (ANN), kernel ridge regression (KRR), support vector regression, random forest (RF), k -nearest neighbors, decision tree, least absolute shrinkage, selection operator regression, Gaussian process regression, etc. ( Xu et al., 2012 ; Wang et al., 2012 ; Fathollahi and Sajady, 2018 ; Elton et al., 2018 ; Barnes et al., 2018 ; Kang et al., 2020 ; Zhang et al., 2017 ; Chandrasekaran et al., 2019 ; Nefati et al., 1996 ). The validation metrics derived from these data-driven models brought a high confidence in their use for a reasonably reliable prediction of D , p C-J , Q max , heat of formation, impact sensitivity, decomposition temperature ( T d ), and other critical properties of HEDMs ( Xu et al., 2012 ; Wang et al., 2012 ; Fathollahi and Sajady, 2018 ; Elton et al., 2018 ; Barnes et al., 2018 , Barnes et al, 2020 ; Kang et al., 2020 ; Gupta et al., 2016 ; Chandrasekaran et al., 2019 ; Nefati et al., 1996 ).…”

Machine Learning Enabled Quickly Predicting of Detonation Properties of N‐Containing Molecules for Discovering New Energetic Materials

“…Before this, suitable neural network should be trained on a small database with limited samples and property data. [ 25 ] Herein, various neural networks were developed via training on different small databases with different sample numbers. And then, the obtained neural networks were evaluated for extended prediction of the detonation properties of new N‐containing molecules.…”

“…Detonation velocity was predicted by machine learning based on a trained set comprising of a large dataset containing 104 data points. [ 25 ] Machine learning (ML), materials informatics (MI), and thermochemical data were combined to screen molecular candidates based on high Δ H e values. [ 26 ] Property prediction and molecular screening strategies based on machine learning have high potential on discovering new high‐energy density materials.…”

Thermal Decomposition Mechanism of 1,3,5,7-Tetranitro-1,3,5,7-tetrazocane Accelerated by Nano-Aluminum Hydride (AlH3): ReaxFF-Lg Molecular Dynamics Simulation

“…They can release huge amounts of energy after decomposition, which make them widely used in military and aerospace fields. 1 Nevertheless, with the improvement of the requirement of comprehensive properties of EMs, conventional EMs cannot gradually meet the requirements of high energy density materials. As a result, a growing number of researchers are working to synthesize EMs with better properties.…”