DEEP LEARNING-BASED APPARATUS AND METHOD FOR DETERMINING BREAKDOWN OF POWER TRANSFER DEVICE USING NOISE DATA

摘要

A method for determining a failure of an electric facility using noise data based on deep learning according to the present invention comprises: acquiring noise data for each type of failure of the electric facility; Dividing the noise data into a plurality of unit noise data having a predetermined length; Extracting main component noise data from the unit noise data; Generating STFT images of the plurality of unit noise data by performing a Short Time Fourier Transform (STFT) transformation on the extracted principal component noise data; Generating a deep learning model for generating a deep learning model for determining a failure of the electric equipment by using the STFT images as training data; And determining a failure type classified as the largest number of failure types classified through the deep learning model for the target noise data as a failure type of the target noise data.