Scholarlink

<p>Electrical Discharge Machining (EDM) is widely employed for the accurate machining of challenging materials such as Inconel 718 alloy. The present work studies the influence of cryogenic treatment of copper electrode, voltage, current and pulse duration on the Material Removal Rate (MRR), Electrode Wear Rate (EWR) and Overcut (OC) in EDM machining of Inconel 718 alloy. This study also investigated the predictive efficacy of various machine learning (ML) Models-Linear Regression (LR), Support Vector Regression (SVR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Adaptive Boosting (AdaBoost) in modeling of MRR and EWR. A comprehensive assessment was conducted utilizing performance metrics such as Root Mean Square Error (RMSE) and Coefficient of Determination (R<sup>2</sup>) and Taylor diagrams, which simultaneously assess standard deviation, correlation coefficient, and centered RMSE. The results exhibited that ensemble models, particularly XGBoost and AdaBoost, achieved enhanced predictive accuracy and generalization performance in both training and testing datasets. XGBoost achieved the highest predictive performance for MRR with an R<sup>2</sup> of 0.9941 and RMSE of 0.0157 on the test set and for EWR, AdaBoost exhibited the best test performance with an R<sup>2</sup> of 0.9912 and RMSE of 0.0020. Taylor diagrams further confirmed the closeness of ensemble models to the ideal prediction zone. These findings are supported by residual analysis and explainable AI tools, SHapley Additive exPlanation (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), which highlighted current and voltage as dominant parameters influencing MRR and EWR. The integration of explainable AI enhances interpretability and validates the alignment of the model with physical EDM principles, enabling the advancement of intelligent decision support systems in advanced machining applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Integrating explainable machine learning models for interpreting edm performance in inconel 718 with cryogenically treated electrodes

  • 期刊论文

  • 发表日期：

    2025年10月20日

查看全文
  • Elumalai Boominathan,
  • Giridharan Krishnan,
  • Praveen Kumar Alagesan,
  • Balamurugan Senthamaraikannan,
  • Chakravarthi Gurijala,
  • Manikandan Durairaj,
  • Krishnaraj Ramaswamy

摘要

Electrical Discharge Machining (EDM) is widely employed for the accurate machining of challenging materials such as Inconel 718 alloy. The present work studies the influence of cryogenic treatment of copper electrode, voltage, current and pulse duration on the Material Removal Rate (MRR), Electrode Wear Rate (EWR) and Overcut (OC) in EDM machining of Inconel 718 alloy. This study also investigated the predictive efficacy of various machine learning (ML) Models-Linear Regression (LR), Support Vector Regression (SVR), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Adaptive Boosting (AdaBoost) in modeling of MRR and EWR. A comprehensive assessment was conducted utilizing performance metrics such as Root Mean Square Error (RMSE) and Coefficient of Determination (R2) and Taylor diagrams, which simultaneously assess standard deviation, correlation coefficient, and centered RMSE. The results exhibited that ensemble models, particularly XGBoost and AdaBoost, achieved enhanced predictive accuracy and generalization performance in both training and testing datasets. XGBoost achieved the highest predictive performance for MRR with an R2 of 0.9941 and RMSE of 0.0157 on the test set and for EWR, AdaBoost exhibited the best test performance with an R2 of 0.9912 and RMSE of 0.0020. Taylor diagrams further confirmed the closeness of ensemble models to the ideal prediction zone. These findings are supported by residual analysis and explainable AI tools, SHapley Additive exPlanation (SHAP) and Local Interpretable Model-agnostic Explanations (LIME), which highlighted current and voltage as dominant parameters influencing MRR and EWR. The integration of explainable AI enhances interpretability and validates the alignment of the model with physical EDM principles, enabling the advancement of intelligent decision support systems in advanced machining applications.