Prediction of crack quantity of carbon fiber composites based on Sparrow BP Neural Network

Carbon fiber composites was affected by complex alternating load, environment and other factors in the process of service, led to the emergence and development of tiny crack, once the crack reaches a certain degree, composites fractures, serious safety accidents happen, so structural health monitoring of carbon fiber composites is necessary, predicting the quantity of cracks is helpful to monitor the health status of carbon fiber composites better. Common damage prediction methods include Back Propagation Neural Network, Logistic Regression, etc., BP Neural Network has higher accuracy in damage prediction. However, BP Neural Network has the risk of falling into local extreme value. Therefore, using Sparrow BP Neural Network, which is combined by Sparrow Search Algorithm (SSA) with strong global optimization ability and BP Neural Network, to predict the crack quantity of composites. The prediction results by using Sparrow BP Neural Network are closer to the real crack quantity, which can predict the crack quantity of carbon fiber composites more accurately, solve the problem that the internal cracks of the composites are difficult to predict. It helps to promote the widespread application of intelligent algorithms to predict the cracks of composites.