Abstract: As the proportion of distributed generation represented by wind power and photovoltaic power in the distribution network keeps increasing, there are higher requirements for the rationality and reliability of power grid planning. This paper proposes a distributed generation planning method based on k-medoid clustering. The optimal planning location of distributed generation is obtained by taking the optimal line loss of distribution network as the optimization objective,and the grid loss sensitivity factor and node voltage level as clustering characteristic indexes. The optimal planning capacity of the multiple distributed generation is obtained using partial line loss calculation strategy based on branch current variation. Finally,an IEEE 33 bus system is used as the test system. The results show that the proposed method not only has high computational efficiency,but also can give the comprehensive optimal access scheme under multiple characteristic indexes at one time,thus it has strong practicability.