Scholarlink

<p>Abrasive waterjet (AWJ) machining can cut hard materials with high thickness and is extensively used in the manufacturing sector. There are many quality features (kerf width, depth of smooth zone, roughness, depth of cut, and grit embedment) in the AWJ cutting process that are conflicting in nature. Therefore, an attempt is made to model the combined effect of all the quality features termed the multi-parameter evaluation index (MPEI). The multiple quality features are combined to a single value (MPEI) by TOPSIS. Further, nonlinear multiple regression (NLMR) and three intelligent methods, namely artificial neural network (ANN), adaptive neurofuzzy inference system (ANFIS), and fuzzy logic (FL), are applied for modeling of MPEI by tuning their hyperparameters. In NLMR, a quadratic model with interaction effect gave a minimum RMSE of 0.17. In ANN, type of activation functions, number of hidden layers, and their neurons are tuned to obtain the minimum error. The architecture 4–2–4–1 with logsig function gave a minimum RMSE of 0.13. For the application of fuzzy logic and ANFIS, their membership functions were tuned, giving a minimum RMSE of 0.08 with pmf and 0.05 with gauss2mf. The overall results showed that ANFIS gave better results, comparatively.</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

Statistical and Artificial Intelligence-Based Modeling of Multi-Parameter Evaluation Index (MPEI) in Abrasive Waterjet Cutting Process

  • 期刊论文

  • 发表日期：

    2025年1月7日

查看全文
  • Paramjit Thakur,
  • Maahi Khemchandani,
  • Manjusha Deshmukh

摘要

Abrasive waterjet (AWJ) machining can cut hard materials with high thickness and is extensively used in the manufacturing sector. There are many quality features (kerf width, depth of smooth zone, roughness, depth of cut, and grit embedment) in the AWJ cutting process that are conflicting in nature. Therefore, an attempt is made to model the combined effect of all the quality features termed the multi-parameter evaluation index (MPEI). The multiple quality features are combined to a single value (MPEI) by TOPSIS. Further, nonlinear multiple regression (NLMR) and three intelligent methods, namely artificial neural network (ANN), adaptive neurofuzzy inference system (ANFIS), and fuzzy logic (FL), are applied for modeling of MPEI by tuning their hyperparameters. In NLMR, a quadratic model with interaction effect gave a minimum RMSE of 0.17. In ANN, type of activation functions, number of hidden layers, and their neurons are tuned to obtain the minimum error. The architecture 4–2–4–1 with logsig function gave a minimum RMSE of 0.13. For the application of fuzzy logic and ANFIS, their membership functions were tuned, giving a minimum RMSE of 0.08 with pmf and 0.05 with gauss2mf. The overall results showed that ANFIS gave better results, comparatively.