Quantum-inspired neural network-based small current ground fault detection method and system

CN121935799BActive Publication Date: 2026-06-09NANJING GUODIAN NANZI POWER GRID AUTOMATION CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
NANJING GUODIAN NANZI POWER GRID AUTOMATION CO LTD
Filing Date
2026-03-26
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing technologies for detecting low-current grounding faults suffer from problems such as simulation-real-domain discrepancies, easy loss of transient features, and weak ability of traditional models to capture sudden changes, resulting in low accuracy in fault location, especially in high-resistance grounding scenarios where identification is difficult.

Method used

A quantum-inspired neural network-based approach is adopted. Real-time monitoring data is processed through a denoising network, and the quantum-inspired neural network is used to capture transient mapping features. By combining multi-layer quantum-inspired neurons and residual connections, fault line selection and type identification can be achieved.

Benefits of technology

It can recover fault signals with high fidelity in complex noise environments, significantly improving the accuracy of fault line selection, especially maintaining high identification accuracy in high-resistance grounding scenarios, thus improving the precision of fault detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121935799B_ABST
    Figure CN121935799B_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for detecting low-current grounding faults based on quantum-inspired neural networks. The method includes: establishing a distribution network model using power system simulation software; generating a fault simulation signal through the distribution network model; acquiring the actual fault signal of the distribution network; establishing a mapping relationship between the actual fault signal and the fault simulation signal to eliminate timing and phase deviations and obtain a calibration dataset; training a deep neural network using the calibration dataset to obtain a trained denoising network; adding actual fault labels to the fault simulation signal to obtain training samples; training the quantum-inspired neural network using the training samples to obtain a fault identification network; and inputting real-time monitoring data of the distribution network into the preset denoising network and fault identification network to obtain fault line selection and type identification results. This invention can keenly capture the energy burst behavior during the transient process of a fault, thereby achieving high-precision fault line selection and fault identification in a real grounding environment.
Need to check novelty before this filing date? Find Prior Art