Machine learning estimation method for sensitivity and mechanical properties of energetic substances and relationship between the sensitivity and the mechanical properties of the energetic substances

Abstract

The invention belongs to the technical field of compound performance evaluation, and discloses a machine learning estimation method for sensitivity and mechanical properties of energetic substances and a relationship between the sensitivity and the mechanical properties of the energetic substances. The estimation method comprises the following steps: constructing a quantitative structure-activityrelationship model of impact sensitivity and bulk modulus of seven nitro energetic substances based on an artificial neural network and a method for determining independent screening and sparse operators by taking a molecular descriptor and molecular structure information calculated by E-Dragon as characteristics; and determining the relationship between the impact sensitivity and the mechanical property of the nitro energetic substance and the quantitative relationship between the impact sensitivity and the mechanical property of the nitro energetic substances respectively with the molecularstructure by utilizing the constructed quantitative structure-activity relationship model of the impact sensitivity and the bulk modulus of the nitro energetic substances. According to the method, seven QSPR models of the impact sensitivity and the bulk modulus of the nitro nitro energetic substances are established on the basis of molecular descriptors calculated by EDragon and several common molecular structure information, so that the process of experimental research on energetic materials is shortened, and design and comprehensive evaluation of novel energetic compounds are facilitated.

Description

technical field [0001] The invention belongs to the technical field of compound performance evaluation, and in particular relates to a machine learning estimation method for energetic material sensitivity, mechanical properties and their relationship. Background technique [0002] At present, energetic materials are a class of compounds or mixtures containing explosive groups or oxidants and combustibles, which can independently undergo chemical reactions and output energy. They are important components of military explosives, propellants and rocket propellant formulations. Energetic materials are widely used in national defense technology industry, aerospace industry and civilian fields. The study of such compounds is not only of great academic significance, but also has great application value. However, due to the long cycle, high cost, high risk, and many influencing factors of the experiment, the reproducibility of the results is low, and the performance data of unsynthe...

AI Technical Summary

Problems solved by technology

The study of the mechanical properties of energetic materials is of great significance for guiding the formulation of energetic materials and the design of structural parts, as well as for the safety assessment and life prediction of energetic materials. However, there is no research on the mechanical properties and structures of energetic materials. Quantitative Structure-Activity Relationship Studies

[0006] 2. There is still room for improvement in the accuracy of the sensitivity QSPR prediction model

In 2012, state-of-the-art 10 characterized by 10 molecular descriptors was studied on the same data set using ANN and MLR methods, and the R 2 They are 0.8658 and 0.7222 respectively, and the prediction accuracy has been greatly improved, but more work is still needed to explore and build a model with higher accuracy, stronger generalization ability and more representative

[0007] 3. There are still relatively few studies on the correlation between different properties such as sensitivity and mechanical properties

[0010] (1) At present, there is no research on the quantitative structure-activity relationship between the mechanical properties of energetic materials and their structures, and the relationship between sensitivity and mechanical properties;

[0011] (2) The prediction accuracy of the existing sensitivity prediction model still has room for improvement, the generalization ability is insufficient, the representativeness is not strong, and the subsequent interpretation of the model is insufficient;

[0012] (3) The calculation process of the mechanical properties of the material is complex and time-consuming, which is only suitable for the calculation of a small number of specific molecules, and cannot calculate a large number of samples in a short time, and has the disadvantage that it cannot be applied to unsynthesized materials in experiments

[0014] (1) Difficulty in obtaining data: It is difficult to obtain experimental data on the mechanical properties of a large number of monomeric energetic materials, and the acquisition of calculation data also requires a large amount of computing resources and high cost;

[0015] (2) The available methods are limited: Although deep learning and other methods more powerful than machine learning have been developed better and better, due to the limitation of the existing data volume, deep learning and other methods cannot be applied to them. However, common machine learning methods often have the problem of insufficient follow-up interpretation.

Images