Arc Fault Detection Using Multi-Frequency Envelope Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing arc fault detection devices struggle to accurately differentiate between arc faults and nonlinear loads, often resulting in false tripping due to reliance on limited current and voltage characteristics, requiring expert determination and manual threshold settings.
Innovation Solution
The use of neural networks to analyze low-frequency and high-frequency envelope waveforms, eliminating the need for high-frequency sampling and expert intervention, by converting analog signals to digital and processing them through a neural network-based algorithm for accurate arc fault detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current and voltage characteristics are used to identify arc faults, then detection capability is provided, but false tripping occurs due to inability to differentiate from nonlinear loads
Solution Approach 1:
The patent segments the frequency spectrum into multiple frequency ranges (e.g., first frequency range and second frequency range) and analyzes envelope waveforms at different frequency levels. This multi-frequency segmentation allows the system to capture distinct characteristics of arc faults versus nonlinear loads, improving differentiation precision and reducing false tripping while maintaining reliable detection.
2Measurement precision
If high-frequency sampling is performed to capture arc characteristics, then detection accuracy improves, but system complexity and cost increase
Solution Approach 1:
The patent extracts envelope waveforms from high-frequency signals through frequency range separation and envelope detection, rather than directly sampling high-frequency arc characteristics. This extraction approach captures the essential arc fault information in a simplified form that can be analyzed at lower frequencies, reducing sampling system complexity and cost while maintaining high detection accuracy.
3Adaptability or versatility
If manual threshold setting by experts is used, then detection criteria can be established, but the system requires expert intervention and lacks adaptability
Solution Approach 1:
The patent implements a neural network-based analysis device that learns from training data containing various load conditions and arc fault patterns. The system continuously processes envelope waveform data and provides feedback to the neural network, enabling it to automatically adapt detection thresholds and criteria based on actual operating conditions, eliminating the need for manual expert intervention while maintaining high adaptability.
4Device complexity
If limited arc fault characteristics are used for determination, then detection is simplified, but determination errors increase due to insufficient information
Solution Approach 1:
The patent transitions from analyzing single-frequency or limited-characteristic signals to analyzing envelope waveforms across multiple frequency dimensions. By examining the system from multiple frequency ranges and combining envelope waveform characteristics, the neural network receives multi-dimensional information that significantly improves detection reliability while keeping the processing framework manageable through systematic frequency segmentation.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach achieves higher accuracy and finer granularity in arc fault detection, reducing false determinations and enabling automatic tripping actions, with the neural network model optimized for low-cost MCU implementation and achieving 99.7% detection accuracy.
Implementation Method 1
a first sensor (111) connected in series in said electric power system line and arranged to collect circuit signals of a first frequency range in said electric power system line; a second sensor (112) connected in series in said electric power system line and arranged to collect circuit signals of a second frequency range in said electric power system line
Implementation Method 2
a first analog-to-digital converter is further connected between the first amplifier and the analysis device, and the first analog-to-digital converter converts the amplified circuit signals of the first frequency band from analog signals to digital signals
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention provides an arc fault detection device and method. The device comprises : a first sensor, which collects circuit signals of a first frequency range in an electric power system line; a second sensor, which collects circuit signals of a second frequency range in the electric power system line; a first envelope waveform generator, which receives the circuit signals of the first frequency range that are collected by the first sensor and generates a first envelope waveform; a second envelope waveform generator, which receives the circuit signals of the second frequency range that are collected by the second sensor and generates a second envelope waveform; and an analysis device, which receives the first envelope waveform and the second envelope waveform, and determines, based on the first envelope waveform and the second envelope waveform, whether current in the electric power system line generates an arc fault signal, wherein when an arc signal is generated in the electric power system line, the analysis device drives a switch device to perform tripping. The present invention has high efficiency and does not rely on manpower.