Department of Electrical Engineering, Graduate School at Shenzhen, Tsinghua University, Shenzhen 518055, China.
Shenzhen Power Supply Co. Ltd., Shenzhen 518038, China.
Sensors (Basel). 2018 Aug 10;18(8):2623. doi: 10.3390/s18082623.
The fast detection of classical contaminants and their distribution on high-voltage transmission line insulators is essential for ensuring the safe operation of the power grid. The analysis of existing insulator contamination has traditionally relied on taking samples during a power cut, taking the samples back to the lab and then testing them with elemental analysis equipment, especially for sugars, bird droppings, and heavy metal particulates, which cannot be analysed by the equivalent salt deposit density (ESDD) or non-soluble deposit density (NSDD) methods. In this study, a novel method called laser-induced breakdown spectroscopy (LIBS) offering the advantages of no sample preparation, being nearly nondestructive and having a fast speed was applied for the analysis of metal contamination. Several LIBS parameters (laser energy and delay time) were optimized to obtain better resolution of the spectral data. The limit of detection (LOD) of the observed elements was obtained using a calibration curve. Compared to calibration curves, multivariate analysis methods including principal component analysis (PCA), k-means and partial least squares regression (PLSR) showed their superiority in analyzing metal contamination in insulators. Then, the elemental distribution of natural pollution was predicted using LIBS to fully capture information about the bulk elements (Na, Ni, Cu, Mn, Ca, etc.) of entire areas with PLSR. The results showed that LIBS could be a promising method for accurate direct online quantification of metal contamination in insulators.
快速检测高压输电线路绝缘子上的经典污染物及其分布对于确保电网的安全运行至关重要。传统上，对现有绝缘子污染的分析依赖于在停电期间取样，将样本带回实验室，然后用元素分析设备进行测试，特别是对于糖、鸟粪和重金属颗粒等物质，等效盐密（ESDD）或不溶性沉积物密度（NSDD）方法无法对其进行分析。在这项研究中，一种称为激光诱导击穿光谱（LIBS）的新方法被应用于金属污染分析，该方法具有无需样品制备、几乎无损和快速的优点。优化了几个 LIBS 参数（激光能量和延迟时间）以获得更好的光谱数据分辨率。使用校准曲线获得了观察到的元素的检出限（LOD）。与校准曲线相比，多元分析方法（包括主成分分析（PCA）、k-均值和偏最小二乘回归（PLSR））在分析绝缘子中的金属污染方面显示出了优越性。然后，使用 LIBS 预测自然污染的元素分布，通过 PLSR 充分捕捉整个区域的大量元素（Na、Ni、Cu、Mn、Ca 等）的信息。结果表明，LIBS 可能是一种用于准确在线直接定量绝缘子中金属污染的有前途的方法。
