Abstract: Ground resistance is a key indicator for evaluating the lightning protection performance of transmission line towers and the effectiveness of grounding devices. Considering the influence of soil resistivity on the qualification standards of ground resistance, this paper proposes an improved K-Means-based assessment method for transmission line tower grounding devices. By denoising the raw data obtained from online monitoring and fitting the qualification thresholds under different soil resistivity conditions using a power function to achieve normalization, followed by K-Means clustering of the normalized data, the method identifies the health status of towers and categorizes them into several types. Finally, the classification results are optimized by integrating the KNN strategy, enabling online judgment of new data and identification of unknown fault types. Empirical results demonstrate that this method can effectively distinguish four types of towers, providing a feasible data-driven decision-making basis for the inspection and operation of transmission line grounding devices.