Reconstructing shock front of unstable detonations based on multi-layer perceptron

The dynamics of frontal and transverse shocks in gaseous detonation waves is a complex phenomenon bringing many difficulties to both numerical experimental research. Advanced laser-optical visualization structure may provide certain information its reactive front, but the corresponding lead shock needs be reconstructed building complete flow field. Using multi-layer perceptron (MLP) approach, we propose front reconstruction method which can predict evolution wavefront from state front. verified through results one- two-dimensional unstable detonations based on Euler equations with one-step irreversible chemical reaction model. Results show that accuracy proposed depends activation energy mixture, influences prominently cellular instability hence, distortion surface. To select input variables for training evaluate their influence effectiveness method, five groups, one six variables, other four are tested analyzed MLP trained cases different energies, demonstrates inspiring generalization capability. This paper offers universal framework predicting provides novel way interpret waves.