可解释AI驱动的再生木纹石复合材料性能优化OA

The design of recycled wood-grain stone(RWGS)composites is constrained by complex nonlinear interactions among composition,processing parameters,and performance.This study employs explainable artificial intelligence(XAI)to develop a high-fidelity predictive model based on a dataset of 150 formulations and 750 specimens,achieving an average accuracy of 95.3％.SHAP analysis quantifies the dominant role of binder content in flexural strength(32.5％contribution)and wood/stone mass ratio in density(45.7％contribution).The study reveals an inverted U-shaped relationship between hot-pressing temperature and flexural strength,identifying an optimal temperature window of 145～155℃,and uncovers a strong synergistic interaction between binder content and temperature(SHAP interaction value up to 3.5).A new formulation optimized based on these insights achieves an 18.8％improvement in flexural strength compared to traditional empirical designs.This work establishes an XAI-guided framework for material design,offering a novel paradigm for the efficient development of sustainable high-performance composites.