Because the traditional self-organized criticality (SOC) research point of power system and its physical model—the sand pile model and cannot effectively explain the new characteristics of cascading accidents under the existing development trend of the power grid, we propose the concept of community self-organized criticality and its physical model—the overlapping sand pile model indicates that the grid cascading accident is essentially a sub-area reaching the SOC state. Firstly, the energy function is applied and combined with the direct method modeling method to quantify the energy correlation between the grid lines, and the weighted undirected complex network model is constructed. Then, based on the model, the improved the local greedy optimization algorithm is used to realize the self-organized criticality sub-area identification of the power grid. Finally, the simulation results verify the effectiveness of the above model and algorithm. In the execution environment of a complex network model with line as the basic structural unit and energy function quantization weight, the local greedy optimization algorithm can effectively mine the SOC sub-area in the power grid. Then, we can provide new ideas for grid cascading accident safety analysis.