To improve the energy utilization of distributed energy storage systems (DESS) based on modular multilevel converters (MMC), it is crucial to address the state of charge (SOC) balance of energy sub-modules (ESM). To address the limitations of existing SOC balance control strategies, a fast SOC balance control strategy based on internal and external hierarchical control is proposed. The outer layer aims to address the SOC differences between upper and lower arms or phase legs. To achieve this, this paper proposes improving the model predictive control (MPC) of MMC, in conjunction with adaptive equalization coefficients, to rapidly adjust power imbalances. The inner layer introduces an adaptive virtual resistance approach. Based on the SOC of ESM, the dominant ESM is determined, and the virtual resistance of each unit is adjusted adaptively to generate a corresponding voltage gradient. Combined with the MMC sorting algorithm, the ESM are allocated power according to their respective SOC, thereby achieving fast SOC balance and improving the energy utilization of DESS. A simulation model is built in Matlab/Simulink to demonstrate the effectiveness and feasibility of the proposed control strategy.