Scholarlink

<p>Corner accuracy in WEDM is significantly affected by machining parameters, including wire vibration, wire lag, and excessive discharge energy. This study presents an optimization methodology to minimize the corner geometric deviation (CE) in WEDM of In(690) using a cost-effective and chemically stable re-circulating molybdenum wire electrode to enhance machining accuracy. Two corner angles (30° and 60°) alongside machining responses, including cutting-speed (Cs), surface-roughness (SR), Kerf-width (KW). The influence of machining characteristics is assessed by varying the spark-on-time (S<sub>on</sub>), Servo-sensitivity (Sc), wire-speed (Ws), Discharge-current (Dc), and Gap-Voltage (Vg). The main-effect plots and analysis of variance (ANOVA) were employed to determine the statistical significance of these parameters of the WEDMed results. Furthermore, three ML algorithms, i.e., AdaBoost Regression (ABR), Random Forest Regression (RFR), and Bayesian Ridge Regression (BRR), were used to model the machining responses, with their accuracy compared based on R<sup>2</sup>, RMSE, MSE, and MAE metrics. ABR demonstrated superior predictive performance. A hybrid approach integrating Grey Relation Analysis (GRA) and ABR, termed Gray-Machine Learning Reasoning Grade (GMRG), is developed to create a high-accuracy process model addressing multiple process aspects. A mathematical model based on GMRG is formulated to solve the multi-objective optimization problem using an Improved Gray Wolf Optimizer (IGWO). For computational performance comparison, Teaching-Learning-Based Optimization (TLBO) and Social Group Optimization (SGO) were also applied. Results indicated that IGWO achieved faster convergence than its counterparts. The optimum parameter setting is determined as S<sub>on</sub>=30µs, Sc = 3, Ws = 75&#xa0;mm/min, Dc = 8&#xa0;A, and Vg = 90&#xa0;V. The proposed optimization method demonstrates a 14.60% improvement in machining outcomes when compared to the GMRG-based result.</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

Enhancing corner accuracy and machining quality in wire-EDM of In(690) using recirculating molybdenum wire: a hybrid machine learning and metaheuristic optimization approach

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Anshuman Kumar,
  • Chandramani Upadhyay,
  • Siva Nagaraju Dusanapudi,
  • R. L. Krupakaran,
  • Rajagopalan Karthikeyan,
  • Kamsali Venkat

摘要

Corner accuracy in WEDM is significantly affected by machining parameters, including wire vibration, wire lag, and excessive discharge energy. This study presents an optimization methodology to minimize the corner geometric deviation (CE) in WEDM of In(690) using a cost-effective and chemically stable re-circulating molybdenum wire electrode to enhance machining accuracy. Two corner angles (30° and 60°) alongside machining responses, including cutting-speed (Cs), surface-roughness (SR), Kerf-width (KW). The influence of machining characteristics is assessed by varying the spark-on-time (Son), Servo-sensitivity (Sc), wire-speed (Ws), Discharge-current (Dc), and Gap-Voltage (Vg). The main-effect plots and analysis of variance (ANOVA) were employed to determine the statistical significance of these parameters of the WEDMed results. Furthermore, three ML algorithms, i.e., AdaBoost Regression (ABR), Random Forest Regression (RFR), and Bayesian Ridge Regression (BRR), were used to model the machining responses, with their accuracy compared based on R2, RMSE, MSE, and MAE metrics. ABR demonstrated superior predictive performance. A hybrid approach integrating Grey Relation Analysis (GRA) and ABR, termed Gray-Machine Learning Reasoning Grade (GMRG), is developed to create a high-accuracy process model addressing multiple process aspects. A mathematical model based on GMRG is formulated to solve the multi-objective optimization problem using an Improved Gray Wolf Optimizer (IGWO). For computational performance comparison, Teaching-Learning-Based Optimization (TLBO) and Social Group Optimization (SGO) were also applied. Results indicated that IGWO achieved faster convergence than its counterparts. The optimum parameter setting is determined as Son=30µs, Sc = 3, Ws = 75 mm/min, Dc = 8 A, and Vg = 90 V. The proposed optimization method demonstrates a 14.60% improvement in machining outcomes when compared to the GMRG-based result.