CNN-based parameter inversion and CAI strength prediction for multi-low-velocity impacts

In practical engineering applications, composite laminates frequently encounter complex multiple low-velocity impact events. The damage coupling caused by the different Angles Between Impact Positions (ABIP) is a key factor in reducing the load-bearing capacity of the laminates. It is worth noting that in real impact events, the delamination damage information of laminates is easier to capture directly. Therefore, it is crucial to predict the damage tolerance of laminates by analyzing their delamination damage images. This paper adopts an integrated finite element model to present an in-depth study on the damage characteristics and Compression-After-Impact (CAI) strength of carbon/glass hybrid laminates subjected to multiple low-velocity impacts at different ABIP. By leveraging the recognition capabilities of Convolutional Neural Networks (CNN) and taking into account the impact of noise, it aims to establish the implicit mapping relationship between delamination damage images and impact parameters, as well as CAI strength. This approach facilitates the inverse inversion of impact parameters for multiple low-velocity impacts of laminates under different ABIP, as well as effective prediction of CAI strength.