Abstract: Aiming at the problems of difficult identifications of configuration modification, ambiguous definition of associated devices and unclear test boundaries in the retrofit and expansion of smart substations, resulting in large power-outage scope and low commissioning efficiency, this paper proposes a test boundary determination method based on the hash algorithm and spectral clustering. Firstly, the algorithm of "two-layer vectorized modeling + sub-cyclic redundancy check (sub-CRC)" is adopted to extract core configuration information of intelligent electronic devices (IEDs) from the substation configuration description (SCD) file. Hash vectors and sub-CRC check codes are calculated to realize automatic identification of device addition, deletion and modification as well as accurate module-level localization. Secondly, the spectral clustering algorithm is introduced to construct an association-domain division model. By integrating influence-transmission intensity and topological characteristics, four classification criteria for association domains and dynamic calibration rules are defined to determine the minimum test boundary. Verified by a retrofit-and-expansion project of a 220 kV smart substation, the proposed method can significantly narrow the test scope and mitigate the power-outage impacts, providing technical support for the safe and efficient implementation of retrofit and expansion projects of smart substations.