China Safety Science Journal ›› 2018, Vol. 28 ›› Issue (7): 32-37.doi: 10.16265/j.cnki.issn1003-3033.2018.07.006

• Safety Systematology • Previous Articles     Next Articles

Prediction of detonation velocity of nitro aromatic compoundsbased on quantitative structure-property relationship

HE Pei 1, PAN Yong 1, JIANG Juncheng1, ZHOU Wei2   

  1. 1 College of Safety Science and Engineering,Nanjing Tech University,Nanjing Jiangsu 210009,China;
    2 China Academy of Highway Transportation Sciences,Beijing 100029,China
  • Received:2018-04-18 Revised:2018-06-13 Online:2018-07-28 Published:2020-11-25

Abstract: In order to improve the accuracy of predicting detonation velocities of aromatic nitro compounds,a prediction method based on the QSPR principle was worked out.Molecular descriptors of 21 kinds of aromatic nitro compounds were calculated by using the Dragon software and 4 molecular descriptors relating the most closely to the detonation velocity of aromatic nitro compounds were selected from a large number of descriptors through GA-MLR method.A four-parameter linear theory model was built for predicting the velocities,and the degree of fitting,robustness and predictability were verified by means of the internal authentication method and the external validation method.The results demonstrate that the average absolute error (AAE) for the training set and test set is 0.048 and 0.208 km/s respectively,the root mean square error (RMSE) is 0.058 and 0.302 km/s,and that the effectiveness of the proposed method is great.

Key words: nitro aromatic compounds, detonation velocity, quantitative structure-property relationship (QSPR), genetic algorithm-based multiple linear regression (GA-MLR), prediction

CLC Number: