Research on the recognition method of secondary equipment misoperation for new power system

In order to reduce the probability of power accidents caused by relay protection equipment misoperation, this paper proposes a new type of secondary equipment misoperation identification method. Firstly, high-definition images of the operation state of secondary equipment are obtained through acquisition equipment, and kernel principal component analysis, phase correlation algorithm, and improved scale-invariant feature transformation are used to complete the denoising, fusion and feature extraction of the images. Secondly, according to the feature extraction results, convolutional neural network is used to realize misoperation state recognition. The experimental results show that the false detection rate of this method is less than 1.8%, while the leakage rate is less than 0.2%, which can comprehensively monitor the misoperation of different types of secondary equipment.