In order to meet the requirements of information privacy among different operators of regional micro-grid as well as to meet the computing challenges brought by large-scale micro-grid merging into clusters in the future, Dantzig-Wolfe decomposition (DWD) is proposed to solve the optimal scheduling problem of regional multi-energy complementary micro-grid clusters in a decentralized manner. Meanwhile, the other three kinds of distributed decomposition algorithms are compared and analyzed. For the bus-type micro-grid cluster, this paper verifies the effectiveness of the proposed DWD decomposition algorithm, and the proposed algorithm can converge to the optimal result in less iteration in winter scenarios. Unlike the other three kinds of distributed decomposition algorithms, the number of iterations of the DWD algorithm varies a little with the increase of the number of micro-grid, which is very suitable for the scenario where large-scale micro-grid merged into clusters in the future.