Unidirectional grounding fault feeder line identification method under sample imbalance

Abstract

The invention provides a single-phase earth fault feeder line identification method under sample imbalance, aiming at the problems of lack and imbalance of single-phase earth fault data of a power distribution network, sufficient and balanced training data information can be provided for a data driving model, so that more accurate single-phase earth fault feeder line identification work is carried out. By utilizing the semi-supervised learning characteristic of the conditional generative adversarial network, through the game adversarial training of a generator and a discriminator, the generator generates a single-phase earth fault sample of which the data distribution characteristic is consistent with the real data distribution characteristic, thereby realizing the balance of fault feeder waveform data and sound feeder waveform data, increasing the basic number of training set samples, and improving the training efficiency. And the identification accuracy of the data-driven fault feeder line identification method is improved.

Description

technical field [0001] The invention relates to a one-way ground fault feeder identification method under unbalanced samples, aims at the lack and imbalance of single-phase ground fault data in distribution networks, and belongs to the technical field of comprehensive processing of distribution network ground faults. Background technique [0002] As the system network closest to the user terminal in the power system, the distribution network plays a vital role in the power system. With the gradual increase of cable feeders, the line-to-ground capacitive current continues to increase, and the system is prone to ground faults. Due to lightning, bird damage, tree growth, equipment failure and other reasons, distribution network faults occur frequently, among which single-phase ground faults account for about 80% of the total faults. When a single-phase ground fault occurs in the distribution network, the fault current characteristics are weakened due to the use of the arc supp...

AI Technical Summary

Problems solved by technology

Most of the data sets used for the identification of single-phase-to-ground fault feeders in distribution networks have a serious problem of sample category imbalance, and when training data-driven recognition models with sample unbalanced datasets, the network tends to learn most class samples features, but the ability to identify minority samples is weak, and the identification accuracy is low, that is, the data-driven identification model tends to simply identify the sample as a healthy feeder when identifying the input feeder waveform, but cannot correctly identify the faulty feeder, resulting in Ground fault situation worsens

Therefore, the existing data-driven fault line selection methods have problems such as training overfitting and unsatisfactory field application effects in the unbalanced small sample scenario, making it difficult to ensure the high reliability of power supply in the distribution network

Images