Machine learning approaches for density evolution model of TATB-based PBX

摘要：

　　Polymer-bonded explosive (PBX) is a kind of highly-particle-filled composite materials, comprised of explosive crystals and a polymeric binder.In this study, a novel approach for density evolution of PBX molding powder is introduced, which is strongly needed for a realistic simulation of PBX behavior by employing machine learning approaches instead of the traditional densification equation.Support vector machine (SVM) is adapted to model realizations of the density evolution function at different temperatures (117℃, 130 ℃, 143 ℃) and maximum load (135MPa, 150MPa, 165MPa) with a set of parameters.After the successful fitting of the experiment data, the algorithm is employed to model the density evolution model of PBX molding powder, starting from error analysis.Comparing with the experiment data, the maximum mean absolute percentage error (MAPE) is only 1.2481%, and the maximum MAPE is only 0.0241 MPa, which provides a new approach for the study of the mechanical properties of similar materials.

关键词： Polymer-bonded explosive Support vector machine Density evolution

作者: Xiao-chang DUAN Yuan-ge ZHANG Yong TIAN Wei TANG

作者单位: Institute of Chemical Materials, China Academy of Engineering Physics, Mianyang, Sichuan, China

会议类型: 国际会议

会议名称: 2018 International Conference on Defence Technology (2018国际防务技术会议)

会议地点: 北京

会议语种:英文

页码: 274-281

在线出版日期: 2018-10-21（万方平台首次上网日期，不代表论文的发表时间）