Insulators are critical components in transmission lines, playing a vital role in supporting and insulating, especially in ultra-high-voltage (UHV) transmission lines. However, the accumulation of industrial dust and other pollutants on the surface of insulators leads to a decline in insulating performance, which can trigger contamination flashover and cause significant damage to the transmission system. Therefore, monitoring the contamination level of transmission line insulators is a key factor in ensuring the safe and reliable operation of the power grid. Laser-induced breakdown spectroscopy (LIBS) is an in situ, rapid elemental analysis technique that offers the advantage of on-site, non-destructive analysis without requiring sample preparation. The remote sensing capability is one of the distinctive strengths of LIBS. In this study, artificial contamination was used as the target for analysis. A novel remote LIBS analyzer was used to conduct remote analysis of the elemental composition and contamination levels of glass insulator surfaces. A novel in-situ, rapid analytical method for determining the contamination level of insulators using remote LIBS was established. In the experimental process, under working conditions of a 2-meter testing distance, a laser energy output of 50 mJ, a laser frequency of 20 Hz, an integration time of 2.0 seconds, and a delay time of 2.0 μs, effective qualitative determination of elements such as Mg, Si, Al, Ca, and Na in artificial contamination was achieved. The quantitative analysis revealed a good linear relationship between the Na intensity in the contamination and the equivalent salt density. This indicates that the remote LIBS analyzer has a strong spectral response to Na in the contamination. Using the characteristic spectral lines of soluble salts, such as Mg and Na, obtained from the LIBS spectra, principal component analysis (PCA) and K-nearest neighbors (KNN) algorithms were employed to cluster and distinguish contamination levels effectively. The KNN classification model achieved an accuracy of 94.4%, a precision of 93.7%, and a recall of 96.4%, demonstrating its high effectiveness in identifying contamination levels. This study demonstrates that remote LIBS can achieve in-situ multi-element analysis of contamination on insulator surfaces. Combined with machine learning, it enables direct recognition of contamination levels. This provides methodological support for the further development of LIBS technology in power industry applications and the development of targeted analysis equipment for future in-situ applications.