Abstract

Robot-based painting industries optimize operations and enhance product quality by leveraging insights from real and virtual studies, encompassing trajectory patterns, paint film qualities, and machine learning for fault identification. Automation of fault identification procedures is the novel aspect of the study that helps to reduce human error and maintain consistent quality standards in manufacturing. This in-depth investigation examines the analysis of paint paths for robot painting with a focus on three distinctive movement patterns: linear, circular, and zigzag. The investigation includes assessments of smoothness for each route, along with morphological evaluations using Scanning Electron Microscope (SEM) pictures. The surface quality is assessed methodically using Taguchi L9 orthogonal testing, while Analysis of Variance (ANOVA) is utilised to identify the key factors that contribute to variations in paint qualities. In order to enhance quality control, machine learning is included to automate the classification and identification of flaws, utilising sophisticated picture analysis techniques. It is essential to incorporate virtual-environment experiments to ensure the accuracy and applicability of the results in real-world situations. This technique reveals crucial observations on the temporal difference between virtual and real surroundings, providing significant information for enhancing the painting process to better match the actual operational parameters. In addition, the analysis determines that the best combination of roughness is A3B3C2 using the Taguchi method, which results in an outstanding finish with a roughness value of 0.0347 µm. Verifying the efficacy of cutting-edge technology in honing painting techniques and improving end product quality, the machine learning model demonstrates a remarkable 94% accuracy in real-time flaw detection.

中文翻译：

---

通过田口方法以及模糊逻辑与机器学习的集成来优化 IRB1410 工业机器人喷漆流程

摘要

基于机器人的涂装行业利用真实和虚拟研究的见解，包括轨迹模式、漆膜质量和用于故障识别的机器学习，优化运营并提高产品质量。故障识别程序的自动化是该研究的新颖之处，有助于减少人为错误并保持制造过程中一致的质量标准。这项深入的调查研究了机器人绘画的绘画路径分析，重点关注三种独特的运动模式：线性、圆形和锯齿形。该调查包括对每条路线的平滑度进行评估，以及使用扫描电子显微镜 (SEM) 图片进行形态评估。使用田口 L9 正交测试系统地评估表面质量，同时利用方差分析 (ANOVA) 来确定导致油漆质量变化的关键因素。为了加强质量控制，机器学习利用复杂的图像分析技术来自动分类和识别缺陷。必须结合虚拟环境实验来确保结果在现实世界中的准确性和适用性。该技术揭示了对虚拟和真实环境之间时间差异的重要观察，为增强绘画过程以更好地匹配实际操作参数提供了重要信息。此外，分析确定使用田口方法的最佳粗糙度组合是 A3B3C2，这会产生粗糙度值为 0.0347 µm 的出色光洁度。机器学习模型验证了尖端技术在珩磨涂装技术和提高最终产品质量方面的功效，在实时缺陷检测中表现出高达 94% 的准确率。