The smart substation system configuration description (SCD) has many nesting levels and a large amount of information data, which is time-consuming and laborious to find and check. The existing SCD file comparisons are mostly carried out between different versions of the same substation, and it is difficult to find the SCD file configuration error. It is proposed to construct a sample data matrix using SCD file description information, and cluster the IED according to substation interval and IED type. The IED comparison analysis based on the clustering results facilitates the discovery of IED configuration differences and configuration errors. The example shows that Markov random walk combined with DBSCAN can well identify IED interval and IED type. The IED configuration comparison in the same SCD file can clearly show the similar interval, similar IED and dual-configuration IED configuration difference, which reduces calibration workload and improve the operation reliability of intelligent station.