<p>Electrical contact interfaces, integral to devices such as connectors, switches, and conductive slip rings, play a crucial role in determining current-carrying capabilities. This study adopts spatial frequency analysis to assess the current-carrying performance of interfaces based on their morphological characteristics. Initially, rough surfaces with various parameters, including the cutoff frequency (<i>N</i>), the spectral index (<i>β</i>), and the gain factor (<i>A</i>), were generated using spatial frequency methods, followed by current-carrying simulations. The findings reveal that the characteristic parameters derived from spatial frequency methods exhibit a significantly stronger correlation with current-carrying performance than traditional roughness metrics. To enhance the practical applicability of this research, a physical-informed machine learning approach for rough surface features extraction was developed, demonstrating high predictive parameter identification accuracy for the three parameters. Overall, this paper introduces a tribo-informatics approach for analyzing and extracting rough surface features that considers the curvature characteristics of contact points. This approach holds significant potential for evaluating the processing state of current-carrying interfaces, selecting high-quality surfaces, and forecasting current-carrying performances.</p>