Prediction and evaluation of projectile damage in composite plates using the neural network–cloud model

Abstract

Composite plates are widely used in the aircraft manufacturing industry. The projectile damage of composite plates is affected by complex factors such as material, structure, impact velocity, and impact angle. A reliable method is needed for efficient structural health monitoring. In this paper, a composite plate damage prediction and evaluation model based on the cloud model and neural network is proposed; the five types of experimental characteristics are used as input parameters, and the depth and diameter of the damage area are used as output parameters to train the neural network–cloud model. This method transforms the quantitative data of impact damage of the composite plate into qualitative damage by introducing the cloud model, which makes the damage situation more intuitive. The results show that the accuracy of the prediction model is 97.23%, the accuracy of the evaluation model is 92.41%, and the comprehensive accuracy of the model is 89.85%. The composite damage prediction model has a good prediction performance.