Abstract: As an important component of smart grid, more and more kinds of energy sources have been connected into microgrid and it makes various operating states of microgrid, and it also leads to the planning and design of microgrid more complex. To analyze the reasonability of capacity configuration of microgrid, firstly taking reducing operating cost as the objective of dispatch optimization, a multi-type energy source optimal scheduling method based on deep Q-learning was proposed, and then, the optimal operating strategy of microgrid under different planning schemes was solved, and the economic benefits of different schemes were compared. Finally, taking a certain business district in Hefei city for example, the analysis and modeling of this example were conducted, and three energy source schemes, i.e., the combined cooling heating and power system with water source heat pump plan, the electric power alternative plan and the integrated energy plan were designed, and the economy of the microgrid construction under these plans were compared. Besides, a suggestion on the microgrid construction plan for this district was given.