Scholarlink

<p>Laser-assisted turning (LAT) offers an effective route to improve the machinability of fused silica-based ceramic composites by promoting thermally assisted material removal and reducing brittle fracture. This study investigates the effects of cutting speed, feed rate, depth of cut, and laser power on surface roughness, cutting temperature, vibration, and cutting force using a Box–Behnken design (27 runs). Quadratic response surface models and ANOVA were developed to quantify factor effects and interactions. In parallel, an artificial neural network (ANN) was trained and validated in MATLAB to capture nonlinear response behavior; the resulting ANN-predicted responses were imported into JMP Pro 18 to examine prediction trends and perform desirability-based multi-response optimization. ANN performance was evaluated using RMSE (in original units) and R<sup>2</sup> for the training, validation, and test subsets. The optimal LAT condition (43&#xa0;m/min, 0.075&#xa0;mm/rev, 0.20&#xa0;mm, 500 W) yielded an overall desirability of 0.496 and produced a stable machining regime characterized by reduced force and vibration and improved surface finish within a favorable thermal window. Surface integrity was quantified through heat-affected zone and microcrack assessments, confirming reduced damage under optimized conditions, and SEM-based tool wear analysis of the PCD insert indicated mild, largely uniform flank wear without catastrophic edge chipping or failure. Overall, the proposed DOE–RSM–ANN framework provides a data-driven basis for selecting LAT parameters that jointly improve surface quality, process stability, and tool condition for fused silica-based ceramic composites.</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

Machine Learning Guided Optimization of Laser-Assisted Turning of Fused-Silica Composites

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Rajat Jain,
  • P. S. C. Bose

摘要

Laser-assisted turning (LAT) offers an effective route to improve the machinability of fused silica-based ceramic composites by promoting thermally assisted material removal and reducing brittle fracture. This study investigates the effects of cutting speed, feed rate, depth of cut, and laser power on surface roughness, cutting temperature, vibration, and cutting force using a Box–Behnken design (27 runs). Quadratic response surface models and ANOVA were developed to quantify factor effects and interactions. In parallel, an artificial neural network (ANN) was trained and validated in MATLAB to capture nonlinear response behavior; the resulting ANN-predicted responses were imported into JMP Pro 18 to examine prediction trends and perform desirability-based multi-response optimization. ANN performance was evaluated using RMSE (in original units) and R2 for the training, validation, and test subsets. The optimal LAT condition (43 m/min, 0.075 mm/rev, 0.20 mm, 500 W) yielded an overall desirability of 0.496 and produced a stable machining regime characterized by reduced force and vibration and improved surface finish within a favorable thermal window. Surface integrity was quantified through heat-affected zone and microcrack assessments, confirming reduced damage under optimized conditions, and SEM-based tool wear analysis of the PCD insert indicated mild, largely uniform flank wear without catastrophic edge chipping or failure. Overall, the proposed DOE–RSM–ANN framework provides a data-driven basis for selecting LAT parameters that jointly improve surface quality, process stability, and tool condition for fused silica-based ceramic composites.