基于XGBoost模型的多缺陷点阵结构性能参数预测方法OA北大核心CSTPCD

To address the problem that the diversity and randomness of defects in additively manufactured metal lattice structures make it difficult to accurately predict their mechanical properties,a model for predicting the yield stress of lattice structures based on the parameter optimization XGBoost model was proposed.Based on CT test data,a threshold judgement criterion was proposed for the rapid detection and classification of multiple defects.Focusing on the problems of strut diameter size and angular deviation defects,the number of defects dominated by oversized struts,undersized struts,and waviness struts were used as input features of the model,respectively.The number of layers was used as another input feature,considering the randomness of defect locations.Based on the measured data,a simulation model of the defect-containing lattice structure was established,and its yield stress was simulated.Using the parameter optimization method to optimize the prediction model hyperparameters and combining the results of the yield stress simulation,the prediction of the yield stress of the structural member is achieved with an R2 of 0.81 and an RMSE of 18.87.A comparison between the prediction results and the actual sample test results was provided in the experimental section,and the deviation is 7.6％,which shows that the prediction model can accurately predict the yield stress of the structure.