Prediction of peak hail impact force under wind-hail coupling based on artificial intelligence models

Wind-hail disasters frequently inflict substantial damage on sunroom structures. Accurately predicting the resilience of sunroom structures against wind and hail is of critical importance. In this study, a comprehensive series of wind-hail coupling experiments were conducted using a proprietary hail impact simulation integrated device coupled with a high-speed Digital Image Correlation (DIC) system. These experiments were aimed at determining the peak principal strain and displacement of hail impacts on polycarbonate (PC) panel materials used in sun rooms. Following the experimental phase, a correlation analysis between independent and dependent variables was performed. Based on the findings, Back Propagation (BP) and Particle Swarm Optimization–Back Propagation (PSO-BP) neural network models were developed and subsequently translated into mathematical expressions for practical application. The results indicated that the peak hail impact force increases with the diameter and velocity of hail particles as well as with wind speed, but decreases with an increase in the thickness of the PC panels. Additionally, for a given velocity of hail launch, larger particle diameters and thinner PC panels showed a more pronounced influence of wind speed on the peak impact force. In terms of stability and accuracy, both the BP and PSO-BP models demonstrated commendable performance, with the PSO-BP neural network showing enhanced predictive accuracy and generalization capability, thus enabling more precise predictions of the peak impact force of single hail particles under coupled wind-hail conditions.