A method and system for detecting arc faults in distribution boxes
By synchronously collecting current and high-frequency electromagnetic signals in the distribution box and using the deep forest algorithm for feature extraction and fusion, the problems of high false alarm rate and high cost of fault arc detection are solved, and high-accuracy and low-cost fault arc detection is achieved.
CN122131096APending Publication Date: 2026-06-02ZHEJIANG MAIFENG POWER EQUIP CO LTD
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHEJIANG MAIFENG POWER EQUIP CO LTD
- Filing Date
- 2026-03-04
- Publication Date
- 2026-06-02
AI Technical Summary
Technical Problem
Existing fault arc detection technologies have high false alarm rates, are insensitive to low-energy arcs, and multi-mode solutions are complex and costly, making them difficult to deploy effectively in distribution boxes.
Method used
The system synchronously acquires current signals and high-frequency electromagnetic signals from the distribution box, extracts and fuses features using a machine learning model built with a deep forest algorithm, and combines this with an adaptive dynamic threshold to detect fault arcs.
Benefits of technology
It achieves high accuracy and low cost in fault arc detection, reduces false alarm rate, adapts to different electromagnetic environments, and has a compact structure that is easy to deploy.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122131096A_ABST
Abstract
This invention discloses a method and system for detecting fault arcs in a distribution box, comprising: simultaneously acquiring the current signal of the target line inside the distribution box and the 30MHz-300MHz high-frequency electromagnetic signal radiated from the internal space; extracting the peak-to-peak ratio and pulse density of the electromagnetic signal, and the gradient abrupt change, zero-rest phenomenon enhancement, and high-frequency noise energy ratio of the current signal; fusing the two types of features into a joint feature vector, inputting it into a fault arc classification model pre-trained based on a deep forest algorithm; and combining an adaptive dynamic threshold to determine whether a fault arc has occurred. This invention creatively employs feature-level collaborative analysis of high-frequency electromagnetic radiation and current waveforms, achieving dual capture and cross-validation of the essential characteristics of fault arcs. It has the advantages of fast response, high reliability, strong environmental adaptability, and ease of deployment in embedded environments within distribution boxes, effectively solving the problems of high false alarm rate and insensitivity to low-energy arcs in traditional methods.
Need to check novelty before this filing date? Find Prior Art