Abstract: In allusion to the fact that the current gridding partition method system for integrated energy system is not yet mature and in which the complementarity of load characteristics is not taken into account, a grid partition method system for integrated energy system, in which the load characteristics was considered, was proposed. Firstly, using the improved k-means clustering algorithm the initial division of energy grid was performed, i.e., on the basis of traditional k-means clustering algorithm the dividing range of gridding number and the position of initial clustering center were determined. Secondly, a method to revise energy gridding, in which the load characteristics was considered, was proposed, and after the accomplish of initial division of energy gridding the load characteristics was utilized to modify the gridding to reduce the inner peak valley rate of load in each energy gridding and the peak load of total grid. Finally, by means of fuzzy ideal decision-making the optimal number of energy gridding division was chosen, and the horizontal comparison of the number of divided number of optimal gridding was conducted. Comparison results show that comparing with the grid division method without considering the load characteristics, using the proposed method the peak-to-valley rates of electrical, heating and cooling loads can be reduced by 7.1%, 0.5% and 4.5% respectively, and the overall peak load of electrical, heating and cooling can be reduced by 7.9%, 11.5% and 3.9% respectively.