Arc Fault Detection Using Wavelet Decomposition and SVM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current arc fault detection systems in photovoltaic (PV) systems, particularly Arc Fault Circuit Interrupters (AFCI), are ineffective in accurately detecting series arc faults due to assumptions about unique spectral signatures and frequency patterns, leading to false detections and reduced effectiveness in preventing system damage.
Innovation Solution
A system comprising a controller trained with a classification model to distinguish between arc events and non-arc events, which receives signals from PV components, extracts features, and classifies them using wavelet decomposition and support vector machines (SVM) to accurately detect arc faults in real-time, mitigating the limitations of existing spectral analysis methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spectral analysis methods are used for arc fault detection, then detection capability is improved, but false detections increase due to assumptions about unique spectral signatures
Solution Approach 1:
The patent transforms the detection approach by changing from spectral domain analysis to time-domain wavelet analysis. Instead of relying on frequency spectrum characteristics that assume unique spectral signatures, the system uses wavelet decomposition to extract time-localized features that capture the transient nature of arc faults without making restrictive spectral assumptions, thereby reducing false detections while maintaining detection accuracy
Solution Approach 2:
The patent replaces the traditional spectral analysis mechanism with a wavelet-based time-frequency analysis mechanism. This substitution allows the system to analyze arc fault signals in the time domain with multi-resolution capabilities, capturing both transient and steady-state characteristics without being constrained by spectral signature assumptions, thus improving reliability by reducing false positives
2Productivity
If wavelet decomposition is used to transform signals, then data set size is reduced for real-time classification, but computational complexity increases
Solution Approach 1:
The patent extracts only the most relevant features from the wavelet decomposed signals for classification, rather than processing the entire transformed data set. By selecting specific wavelet coefficients and energy parameters that are most indicative of arc faults, the system reduces the dimensionality of the data passed to the classifier, thereby maintaining real-time processing capability while managing computational complexity
Solution Approach 2:
The patent segments the signal processing task into distinct stages: wavelet decomposition, feature extraction, and classification. This segmentation allows each stage to be optimized independently, with the wavelet decomposition providing a structured framework for feature extraction that reduces the overall computational burden compared to analyzing raw signals or full spectral data
Data Source
AI summary
In some examples, a system comprises a first component; a second component configured to receive signals from the first component via one or more wires; and a controller. In at least some examples, the controller is coupled to the one or more wires and is trained with a classification model to distinguish between signals indicating arc events and signals not indicating arc events. In at least some example, the controller is further configured to: receive the signals; extract features that are at least partially related to the received signals; classify the extracted features using the classification model; determine an occurrence of the arc event based on the classification; and provide an output signal indicating an arc event.