Artificial neural network-based prediction of jet characteristics from shaped charge parameters

Abstract

To improve the predictive accuracy of shaped charge jet (SCJ) characteristics and optimize the design of charge configurations, an artificial neural network (ANN) model was developed. Shaped charge parameters of charge radius, charge height, liner angle, liner thickness, and standoff distance were selected as the input nodes and SCJ characteristics of effective jet mass, axial head velocity, axial tail velocity, total axial kinetic energy, effective axial kinetic energy, jet length, and effective jet length were output nodes. The ANN model exhibits high predictive accuracy with relative errors generally confined within ±5%. Additionally, based on the trained ANN model, the impact of charge parameters on the characteristics of SCJ was analyzed. The trained ANN model provides not only reliable predictions but also significant insights for the optimization of shaped charge design, facilitating the development of more efficient and effective shaped charges.