Based on decentralized autonomy and centralized coordination, a hierarchical coordinated optimization strategy for multi-microgrid (MMG) system is proposed in this paper. In the strategy, a middle agent (MA) is introduced as a coordinator in the upper level. The internal trading price is formulated by MA based on the power surplus and shortage information uploaded by microgrids (MGs) and fed back to MGs immediately. Each MG, seen as an autonomous unit in the lower level, receives the internal trading price and minimizes cost by optimizing its micro-sources’ generation and exchange power with external grid. Moreover, the equilibrium of MMG’s coordinated operation is acquired by multiple interactions between MGs and MA. Considering that there may be no optimal equilibrium because the pricing model is a non-optimization problem, a relative optimal equilibrium solution set is firstly formed and then the appropriate solution is selected within a certain evaluation criteria in the set. Finally, the proposed model is simulated in a MMG system with three MGs, as well as the rationality and effectiveness are verified in the results. This work is supported by National Natural Science Foundation of China (No. 51677020 and No. 51877181), Soft Science Research Project of Chengdu (No. 2017-RK00-00054-ZF), Major Cultivation Project of Sichuan Education Department (No. 18CZ0018), and the Innovation Fund of Postgraduate of Xihua University (No. ycjj2018074).