Orientation Recognition of Voltage Sag Source Based on Multi-Dimensional Feature Matrix and Convolution Neural Network

The most effective way to solve and improve voltage sag is to remove the disturbance source of voltage sag from the source by location method. For different grid structures and fault types, the existing methods usually extract a single feature based on the monitoring data to realize the location of sag source. Due to the effectiveness and applicability of the method features can not be guaranteed, this paper presents a method for locating voltage sag source based on multi-dimensional feature matrix. Firstly, the characteristics of voltage sag events caused by faults are analyzed. Then, a multi-dimensional feature matrix for the location of voltage sag source is proposed in this paper. By using convolution neural network to establish the mapping relationship between the feature matrix and the voltage sag position, the orientation identification of voltage sag source based on multipoint monitoring data can be realized. Finally, the voltage sag event caused by short circuit fault is simulated in IEEE14 node model, and the effectiveness of the proposed method is verified by simulation data. The simulation results show that this method has higher accuracy than the traditional method, and this method can be applied to different grid structures and different types of faults.