PREDICTION OF RESIDUAL STRENGTH OF LAMINATED COMPOSITES SUBJECTED TO LOW VELOCITY IMPACT

摘要

Barely visible impact damage in composite structures is difficult to detect. The predominant failure mechanism is delamination, which is easily detected by C-scan. Using pixel data from C-scan image, coupled with acoustic emission amplitude distribution data from compression after impact testing, and applying it to a back propagation neural network, correlations on ultimate strength can be made with great accuracy. This paper demonstrates the ability to predict the ultimate compressive strengths of composite structures using this approach.