Fluctuation characteristics of wind power is one of the most important reasons that affects wind power accommodation capacity of power gird. Mastering fluctuation characteristics of wind power and constructing an accommodation capacity evaluating model based on the characteristics are of great significance for researches on large-scale wind power integration. This paper proposed a novel method to estimate wind power accommodation capacity based on multifractal theory. Firstly, the multifractal parameters are used to describe the fluctuation characteristics of wind power, and the correlation between the fluctuation characteristics and the wind power accommodation capacity is analyzed; then, a swinging door algorithm is applied for identifying wind power fluctuation process and these processes are clustered based on fluctuation parameter and average Euclidean distance between wind power and load. According to the results of clustering, a functional model for evaluating wind power accommodation capacity is proposed based on the singularity parameters of fluctuation process; finally, the validity and rationality of proposed method are verified with a district power grid in Jiangxi and the influence of parameters in model is analyzed.