NEURAL NETWORK PREDICTING OF STIFFNESS AND STRENGTH IN PLANE OF STITCHED COMPOSITES

Models estimating stiffness and strength of stitched composites were established based on BP algorithm of neural network, and the longitudinal stiffness and strength networks were trained according to experimental data. The nonlinear relationship between input parameters and output parameters was determined and the analysis software of stitched laminates was completed. The longitudinal stiffness and strength of stitched laminates with new stitching parameters and new equivalent unstitched laminate performance parameters were predicted using the software. The results of the prediction and the test data are in good agreement. A very effective approach to estimate the stiffness and strength of stitched composites has been found.