Under the "dual carbon goal", with the deepening reform of China's energy market and the gradual formation of the distributed multi-energy market, the integrated power microgrid and its community appear on the user side. The microgrid cluster formed by the interconnection of adjacent microgrids is a multi-energy comprehensive energy system, which is jointly affected by many factors such as multi-energy interaction and coupling, complex game decision relations and uncertainty of energy sources and loads. Therefore, the energy mechanism of this kind of comprehensive energy system is complex, and the management scheme formulation is difficult. On the basis, a multi-agent comprehensive energy microgrid group alliance energy management strategy is proposed based on cooperative game and Gaussian mixture clustering. Through simulation example analysis, it can be seen that the proposed method significantly reduces the operating costs of various microgrids, up to 24 % for each microgrid, thus improving the operating efficiency of the entire microgrid system and maximizing the benefits of the alliance. The influence of higher source load prediction errors on the operating cost of each agent and alliance are less than 3 %.