Power distribution network fault cause detection method based on short-time Fourier transform and convolutional neural network

Abstract

The invention discloses a power distribution network fault cause detection method based on short-time Fourier transform and a convolutional neural network, and the method comprises the steps: firstly obtaining fault recording data of a to-be-detected power distribution network, and intercepting one-cycle fault data after a fault occurs; performing feature extraction on the fault data through short-time Fourier transform to obtain a time-frequency feature map, and expanding the data by adopting a data enhancement method; constructing a convolutional neural network CNN classification model, and training the constructed convolutional neural network CNN classification model by using the time-frequency feature map after data expansion; and constructing a test data set, and testing the trained convolutional neural network CNN classification model to realize accurate identification and classification of fault causes. The method can quickly and accurately identify and classify fault causes of an actual power distribution network, and realizes real-time monitoring of dynamic behaviors of a power system.

Description

technical field [0001] The invention relates to the technical field of power systems, in particular to a method for detecting fault causes of distribution networks based on short-time Fourier transform and convolutional neural network. Background technique [0002] With the continuous development of the power system, the distribution network, as an important part of the user-oriented, has the characteristics of large scale, wide line, long line, high failure rate, etc. In this situation, it is easily affected by various faults, such as: equipment failure, tree lap fault, foreign object lap fault, etc. When a fault occurs in the distribution network, the traditional method of finding the cause of the fault is manual line inspection. This method not only It consumes a lot of manpower and material resources, and takes a long time, which can easily lead to the expansion of the accident. At the same time, there are problems of poor identification effect and easy misjudgment and m...

AI Technical Summary

Problems solved by technology

Most of the existing technologies identify and classify fault causes based on waveform similarity, but in the actual distribution network system, the waveforms of faults caused by different reasons have a high similarity, and the three-phase current and three-phase voltage When the waveform is used as a feature, the characteristic information of the fault phase is more obvious, which will cover up the characteristic information of the fault cause, seriously affecting the identification and classification of the fault cause

Images