工业控制网络与系统研究室知识库

碳纤维复合材料(CFRP)具有质量轻、硬度强等优势，受到工业领域的广泛关注。采用激光清洗碳纤维表面的环氧树脂膜层，有利于提高胶接性。文章结合激光诱导击穿光谱（LIBS）技术，设计激光清洗在线监测系统，实时监测激光清洗的质量。实验所用激光器为激光清洗中大量采用的光纤激光器，效率高、使用灵活，可以在多维空间加工应用。文中首先确定激光清洗速度值，并研究LIBS谱跟随激光单脉冲能量密度的变化规律。实验发现，所获得的LIBS光谱中波长在588.819nm与589.411nm处Na(I)原子谱线，可以表征碳纤维复合材料清洗的效果。在数据分析处理上，采用均值平滑去除背景的方法，处理包络状的光谱连续背景；用DBSCAN密度聚类算法，实现光谱噪声和有效数据的分离；采用皮尔逊系数分析的方法，确定激光清洗的最佳烧蚀次数，为激光清洗实现过程自动优化控制提供判定依据。最后，采用电子显微镜分析了碳纤维表面形貌特征，证实LIBS技术在线监测激光清洗效果的有效性。

Carbon fiber reinforced plastic (CFRP) have the advantages of light weight and strong hardness, and have received widespread attention in the industrial field. Laser cleaning of the epoxy resin film on the surface of the carbon fiber is beneficial to improve the adhesiveness. This article combines laser-induced breakdown spectroscopy (LIBS) technology to design an online laser cleaning monitoring system to monitor the laser cleaning effect in real time. The laser used in the experiment is a large number of fiber laser used in laser cleaning, which has high efficiency and flexible use, and can be used in multi-dimensional space processing. In this paper, the velocity of laser cleaning is determined, and the law of LIBS spectrum following the energy density of laser single pulse is studied. It was found that the Na (I) spectrum lines at 588.819 nm and 589.411 nm in the LIBS spectrum obtained can characterize the cleaning effect of carbon fiber composite materials. In the data analysis and processing, the mean smoothing method is used to remove the background, and the envelope spectrum continuous background is processed; The DBSCAN density clustering algorithm is used to separate the spectral noise from the effective data; Pearson coefficient analysis method is used to determine the optimal ablation times of laser cleaning(3times), which provides a basis for the automatic optimal control of laser cleaning process. Finally, the surface morphology of carbon fiber was analyzed by electron microscope, which confirmed the effectiveness of LIBS technology for online monitoring laser cleaning effect.