Identification and Location of Voltage Sag Sources Based on Multi-Label Random Forest

With the continuous construction and development of the smart grid, the complexity of power grid equipment and the regionalization of power consumption patterns have posed many new challenges to the traditional identification method of voltage sag sources based on physical characteristics. To identify and locate voltage sag sources effectively, a novel method based on the multi-label random forest (ML-RF) algorithm is proposed in this paper. Firstly, the simulation model of voltage sag source (VSS) is constructed to obtain bus voltage data in different fault situations. Secondly, the features of all buses voltage data are respectively extracted by using the discrete wavelet transform and then compose system features data, which can assist in differentiating various voltage sag sources and its location. Finally, the parameters of the multi-label random forest (ML-RF) classification model are further analyzed and optimized, then the voltage sag sources are identified based on the system features data. Compared with similar algorithms, the experimental results indicate that the proposed algorithm can accurately identify and locate all kinds of voltage sag sources simulated in the power grid, and has high accuracy and validity. The proposed method can provide assistance for the formulation of voltage sag control scheme and the clarification of accident liability.