Medium-Voltage Insulator Partial Discharge Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting partial discharges in medium or high voltage electrical devices fail to accurately classify the origin of discharges, making it difficult to identify and correct the underlying issues.
Innovation Solution
A method using automatic learning to classify partial discharges into distinct classes based on statistical quantities derived from sensor measurements of oscillating signals associated with the insulator of electrical conductors, employing a classification model trained on reference samples to determine the class of a new discharge.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical analysis methods are used to detect partial discharges, then detection capability is provided, but classification precision of discharge origin is insufficient
Solution Approach 1:
The patent transforms the detection approach by changing from basic statistical analysis to analyzing multiple specific signal parameters simultaneously (amplitude, duration, frequency characteristics, phase position). This parameter transformation enables the system to distinguish between different discharge origins by evaluating the combined pattern of these parameters rather than relying on simple statistical thresholds.
Solution Approach 2:
The invention adds classification dimensionality by introducing a supervised learning framework that maps multi-dimensional signal characteristics to discrete discharge origin classes. By training a classification model on labeled data containing various discharge types and their origins, the system gains the ability to identify discharge sources in a new dimensional space that combines multiple signal features, thereby resolving the information loss about origin.
2Measurement precision
If automatic learning classification is implemented, then discharge origin identification is improved, but system complexity increases
Solution Approach 1:
The patent employs a copying approach by training a classification model on a dataset that replicates the characteristics of actual discharge signals from various origins. The model learns to recognize patterns by processing copies of training data with known labels, then applies this learned knowledge to classify new, unseen discharge signals. This allows the system to achieve high classification precision without requiring complex real-time analysis infrastructure.
Solution Approach 2:
The invention implements preliminary action by pre-training the classification model offline using a comprehensive dataset of labeled discharge signals. This preliminary training phase prepares the model in advance to handle various discharge scenarios, so that during actual operation, the system can perform rapid classification without complex real-time computations. The heavy processing is done beforehand, simplifying the operational system complexity.
Data Source
AI summary
A method for classifying a partial discharge in an insulator of an electrical conductor of a medium voltage or high voltage electrical device, the method allowing a partial discharge to be classified from between at least one first class and a second class distinct from the first class. The method includes: obtaining a set of samples each corresponding to at least one partial discharge, determining a classification model by automatic learning based on at least one statistical quantity of the samples of the set, acquiring a new sample corresponding to at least one partial discharge, and determining the class of the partial discharge associated with the new sample acquired using the classification model.