Distributed algorithms have significant advantages in communication, reliability, flexibility, and other aspects, demonstrating their competitiveness in the economic dispatch of power grids. For this purpose, a multi-period economic dispatch model for microgrids is constructed, which includes transferable loads and purchasing electricity from the large power grid. A dynamic power constraint generation strategy based on the Karush-Kuhn-Tucker （KKT） condition is designed to address the issue that current distributed consensus dispatch algorithms cannot solve the problem of dispatch resource cost functions being non strictly convex. This strategy can dynamically modify the power constraints in the local optimization process, avoiding the centralized participation of transferable loads and purchasing electricity in response, and eliminating the impact of non strictly convex scheduling resource cost functions that prevent convergence of results. Finally, simulation examples verify the correctness and effectiveness of the proposed algorithm.