Application of a class of density peak clustering algorithm in short-term smart grid

The classification of short-term power load data by clustering algorithm can lay a good foundation for the subsequent power load forecasting work and provide a more efficient, safe and reliable direction for the operation of power system. Density peaks clustering algorithm with k-nearest neighbors and weighted similarity can reduce the negative impact of global data on local data, avoid the collateral effect of errors and improve the accuracy of the model. It has better effect on local peak power change and global power forecast. In this paper, the density peaks clustering algorithm with k-nearest neighbors and weighted similarity are compared with DPC clustering method through simulation examples, and the effectiveness of this method is verified.