Ballistic Performance Evaluation of Multi-layered Armors using Neural Network Algorithm

摘要

For a design of multi-layered armors, extensive full scale and sub-scale penetration tests are required. These test experiments are time consuming, expensive, and highly dangerous. However, the applications of numerical and analytical methods are yet very limited because of the poor understanding of penetration mechanisms. In this paper, we developed an object oriented neural network algorithm for new armor design. The neural network algorithm is easy to use and predicted the penetration depths within maximum 5% error for most of the new metallic and ceramic armors based on the pre-existing penetration database. Among different combinations of material properties, density was the best material property for the evaluation of penetration depth of ceramics.