Impact Energy Release Characteristics of Si/W/PTFE Reactive Materials and Predictive Modeling of Energy Release Efficiency via Machine Learning

Five kinds of typical Si/W/PTFE reactive materials (RMs) are prepared to study their energy release characteristics under high-velocity impact by using shock equations of state (EOS) for porous mixtures and chemical reaction kinetics equations. The energy release characteristic curves of the Si/W/PTFE RMs under the impact velocities ranging from 600 to 1300 m/s are obtained using an improved hemispherical quasi-sealed reaction chamber. The datasets of the impact-induced energy release characteristics of Si/W/PTFE RMs are established. Furthermore, the energy release efficiencies of Si/W/PTFE RMs under impact are predicted using two machine learning models: support vector regression (SVR) and random forest regression (RFR). The datasets for Si/W/PTFE RMs with W contents of 0%, 25%, 50% and 75% are used as test sets, respectively. The mean absolute errors (MAEs) of SVR model are 1.17, 3.10, 3.72, and 1.95, respectively, and the MAEs of RFR model are 1.85, 3.05, 2.86 and 1.24, respectively. The predicted results of SVR and RFR models show a high degree of agreement with the experimental data. Additionally, the SVR model demonstrates superior generalization ability when the shock temperature exceeds 1200 K.