In　order　to　solve　the　problems　of　excessive　communication　resource　consumption　and　data　security　faced　by　centralized　learning-based　microgrid　energy　management　(MEM),　the　hierarchical　federated　deep　reinforcement　learning　(DRL)　based　energy　management　strategy　is　investigated　in　this　paper.　Firstly,　the　microgrid　component　model　and　its　corresponding　Markov　decision　process　(MDP)　model　of　MEM　under　the　hierarchical　federated　learning　(FL)　framework　is　constructed.　Then,　in　order　to　solve　this　problem,　we　propose　the　hierarchical　federated　dueling　DQN　using　Sine　Cosine　algorithm　(HFS-DDQN)　and　leverage　it　to　design　the　energy　management　strategy.　Simulation　results　show　that　the　proposed　energy　management　strategy　is　able　to　achieve　higher　reward　to　maximize　the　economic　benefit　of　microgrid　under　the　premise　of　protecting　the　privacy　of　the　data　and　at　the　same　time,　it　reduces　the　communication　latency.