Real-Time Identification of Distribution Network Operation Modes Based on t-SNE Dimension Reduction and Birch Clustering

The frequent changes in operation modes of distribution networks with high penetration of renewable energy sources necessitate the development of rapid and accurate methods for operation mode identification, which holds significant practical importance. To address this, a method for online identification of distribution network operation modes based on real-time measurement data from smart meters is proposed. Firstly, the bus voltage and power data acquired by smart meters are preprocessed, and the $t$-distributed stochastic neighbor embedding (t-SNE) algorithm is employed to reduce the dimensionality of the bus voltage and power data for each historical time section. Subsequently, a small number of typical historical sampling moments are selected, and the correspondence between the data at these moments and the distribution network operation modes is manually established. The Birch clustering method is then utilized to automatically establish the correspondence between the data at the remaining historical sampling moments and the distribution network operation modes. Finally, the real-time measurement data are reduced in dimensionality, and the operation mode of the distribution network at the current moment is identified based on the known correspondences. Case study analysis validates that the proposed method can accurately and rapidly identify the operation mode of the active distribution network online using voltage and power measurement data collected by smart meters.