CNN-based self-supervised voltage sag source recognition method

Abstract

The present invention discloses a CNN-based self-supervised voltage sag source recognition method. The method comprises the following steps: collecting voltage sag data, pre-processing the voltage sagdata; constructing a convolutional coder and a convolutional decoder based on an autocoder, to establish a CNN self-supervised model, so that the CNN self-supervised model extracts a feature by usinga convolution layer and a pooling layer, and performs classification by using a BP classification network; dividing the pre-processed voltage sag data into a training set and a test set, inputting the training set into the CNN self-supervised model in batches, to train a feature extraction capability and a classification capability of the CNN self-supervised model; and inputting the test set intothe trained CNN self-supervised model, to perform voltage sag source recognition on the test set. According to the method, a voltage sag source can be recognized accurately.

Description

technical field [0001] The invention relates to the field of power quality disturbance source identification, in particular to a CNN-based self-monitoring voltage sag source identification method. Background technique [0002] With the increasing level of industrial equipment, building electrical automation and intelligence, the impact of voltage sag on the production and operation of large industrial and commercial users is becoming more and more significant, especially in semiconductor manufacturing, precision instrument processing, and automobile manufacturing. The electronic equipment industry is very sensitive to voltage sags. When the effective value of the voltage is lower than 90% and the duration reaches 1 to 2 cycles, it will trip and stop. Voltage sag is a common power quality problem. Motor starting, transformer switching, short-circuit faults, etc. will cause voltage sag. Production interruption and delay caused by voltage sag interference are on the rise. The ...

AI Technical Summary

Problems solved by technology

The existing problems in some existing voltage sag source identification methods are as follows: manual setting of features needs to be based on a certain understanding of the data to be extracted, relying on expert experience to select the target features to be extracted, and then using various means to identify Targeted extraction of sag features, but there are a large number of unknown interferences in actual engineering, and the accuracy of feature extraction and identification is affected according to the unchanging expert experience. It can be seen that the traditional scheme of voltage sag source identification scheme has a problem of low accuracy

Images