A power distribution network fault detection method based on a lightweight neural network

By constructing a lightweight multi-scale neural network architecture, the problems of insufficient generalization ability and high computational complexity in existing power distribution network fault detection methods are solved, achieving efficient extraction and identification of fault features and improving the real-time performance and accuracy of fault detection.

CN122332897APending Publication Date: 2026-07-03NANJING HEXING GRID TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANJING HEXING GRID TECH CO LTD
Filing Date
2026-04-03
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing methods for detecting faults in power distribution networks rely on periodic manual inspections and basic monitoring, which makes it difficult to achieve real-time panoramic perception. Traditional neural network models have insufficient generalization ability when dealing with high-dimensional nonlinear correlations, resulting in high false negative rates, limited recognition accuracy, and high model complexity, making them difficult to deploy in terminal devices with limited computing resources.

Method used

A lightweight multi-scale neural network architecture is constructed, including a PPD module for feature extraction, an MSCat module for fusing multi-scale features, and a DSE module for enhancing cross-scale correlation. By using random channel selection, multi-scale pooling, and upsampling, the number of model parameters and computational complexity are optimized, thereby improving the fault feature extraction capability and recognition accuracy.

Benefits of technology

It significantly improves the diversity and discriminative representation of fault characteristics, enhances the ability to perceive and identify weak faults and complex fault modes, reduces computational burden, and enables efficient deployment on resource-constrained devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122332897A_ABST
    Figure CN122332897A_ABST
Patent Text Reader

Abstract

The application discloses a power distribution network fault detection method based on a lightweight neural network, which comprises data acquisition, data preprocessing, data set division, neural network model construction, model training, model testing and fault detection; through integration of a pyramid lightweight extraction module PPD, a multi-scale feature splicing and fusion module MSCat and a cross-scale dynamic enhancement module DSE, an efficient feature extraction and fusion mechanism is constructed, the perception and recognition ability of multi-scale fault features is effectively improved while the model parameter quantity and the calculation burden are significantly reduced, and thus a solution is provided for realizing accurate, real-time and end-side deployable intelligent fault detection. The application solves the problems of high missing detection rate, limited recognition accuracy, high model complexity and difficulty in deployment in actual power distribution terminals with limited computing resources in the FTU fault detection method.
Need to check novelty before this filing date? Find Prior Art