基于机器学习和优化AI算法预测壳芯强度OA

Reliable shell core strength prediction is essential for improving the casting process performance and ensuring efficient material utilization in manufacturing processes.Based on the constructed database,and combined with Pearson correlation coefficient and SHAP analysis,this study explores the applicability of eight machine learning techniques to predict the tensile strength of shell cores.The results show that the XGBoost and GBR models achieve excellent fitting performance,with determination coefficient(R2)of 93.5%and 92.1%,respectively,along with comparatively lower mean squared error(MSE),mean absolute error(MAE),and root mean squared error(RMSE).Feature importance analysis using SHAP identifies that the resin and urotropine contents are dominant variables influencing the tensile strength.The XGBoost and GBR models show better predictive stability,as their predictions closely match observed values and exhibit consistent trends.Overall,the findings highlight the effectiveness of machine learning-based modelling for tensile strength estimation of shell cores,offering improved predictive capability relative to traditional analytical approaches.