Classification of Frequency Disturbance Event in Power Systems Considering Optimal PMU Placement

Power system frequency disturbance events are caused by various generation and transmission events, including generator tripping, load disconnection, line tripping, etc. Accurately determining the frequency disturbance events is of great significance for the subsequent reasonable suppression measures and improvement of grid stability. In this paper, we take advantage of the far-reaching development of deep learning to establish a convolutional neural network model applied to the IEEE 39-bus system, in which frequency, voltage, rate of change of frequency, and relative angle shift are converted into images as inputs to the model, to detect the events accurately and directly; furthermore, we propose a greedy algorithm embedded in the optimized arrangement of PMUs trained by a deep learning model, which is based on the optimally ranked PMU node feature data as inputs in the context of partial observability. PMU node feature data as input to obtain the most suitable PMU placement location under partial observability, and ultimately the FDEs classification is achieved accurately, quickly, and economically.