In order to accurately perceive the electricity consumption status of multiple loads in industrial parks, a method of identifying core-loads-clusters of industrial parks considering electricity consumption similarity is proposed. Firstly, the technical principle of core-loads-clusters identification is analyzed and abstracted into a load clustering problem considering electricity consumption similarity. After that, a cluster space of power perception samples is constructed based on power perception elements, and a series of core load clusters are identified by using electricity consumption similarity as clustering basis. Finally, based on the field data of electric tool factory, a series core-loads-clusters are identified and the refined power perception process of this factory based on core-loads-clusters is simulated to verify the effectiveness of this method.