Prediction of Shock Wave Velocity Induced by a Combined Millisecond and Nanosecond Laser Based on Convolution Neural Network

Abstract

The variation of shock-wave velocity with time induced by a millisecond-nanosecond combined pulse laser (CPL) on silicon is investigated. The convolution neural network (CNN) is used to predict the shock-wave velocity induced by a single ns laser and CPL with a ns laser energy density of 6, 12 and 24 J/cm2, ms laser energy density of 0 and 226.13 J/cm2, and pulse delay of 0, 0.4 and 0.8 ms. The four-layer CNN model was applied, ns laser energy density, ms laser energy density, pulse delay and time were set as the input parameter, while the shock-wave velocity was set as the output parameter. The correlation coefficient (R2), mean absolute error (MAE) and root mean square error (RMSE) of the CNN model on the test data set was 0.9865, 3.54 and 3.01, respectively. This indicated that the CNN model shows a high reliability in the prediction of CPL-induced shock-wave velocity with limited experimental data.