Dynamic Clustering of Transient Signals for Partial Discharge Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing transient signals, such as partial discharges in electrical distribution networks, require excessive computation and are prone to noise, leading to inefficient processing and reduced precision.
Innovation Solution
A method for dynamic clustering of transient signals, where similar transients are grouped into characteristic signatures, reducing the number of signatures to process, thereby decreasing computation time and increasing signal-to-noise ratio, using a computer-implemented system with a processor and memory to execute instructions for clustering and signature analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If distinct processing of each transient signal is performed, then individual signal analysis is achieved, but computation effort is excessively high
Solution Approach 1:
The patent groups similar transient signals into clusters, merging multiple individual signal processing tasks into cluster-level processing. By identifying and combining transients with similar characteristics, the system processes representative signals from each cluster rather than every individual transient, thereby reducing overall computation effort while maintaining analysis precision through cluster-based pattern recognition.
Solution Approach 2:
The patent segments the set of transient signals into distinct clusters based on similarity criteria. This segmentation divides the large volume of individual transient processing into manageable groups, where each cluster is processed separately using representative signals, thus improving processing efficiency without sacrificing the precision of individual signal analysis.
2Loss of information
If each transient signal is processed separately, then detailed signal characteristics are obtained, but processing time increases significantly
Solution Approach 1:
The patent combines multiple transient signals into clusters based on their characteristic similarities, processing representative signals from each cluster instead of every individual transient. This merging approach retains essential signal details through cluster representation while significantly reducing the total number of processing operations, thereby decreasing computation time without substantial loss of signal information.
Solution Approach 2:
The patent creates representative copies or prototypes for each cluster of similar transients. Instead of processing every original transient signal individually, the system uses these representative copies to characterize entire clusters, preserving the essential signal characteristics while reducing processing time through the use of surrogate representations.
3Productivity
If clustering of transient signals is performed, then computation time is reduced, but processing complexity increases
Solution Approach 1:
The patent employs dynamic clustering algorithms that adaptively group transient signals based on their characteristics. The clustering process dynamically adjusts group formations as new signals are processed, enabling the system to handle varying signal patterns efficiently. This dynamic approach balances processing speed with manageable complexity by using adaptive grouping strategies rather than static, pre-defined categories.
Solution Approach 2:
The patent utilizes parameter-based clustering where transients are grouped according to key characteristic parameters. By changing the clustering parameters and similarity criteria, the system can optimize the balance between processing speed and algorithmic complexity, adjusting the level of detail and grouping granularity to match computational resources and signal characteristics.
Data Source
Figure 1~2
Figure 3~4
Figure 5~7
AI summary
A method for clustering of transient signals is provided. The method comprises the steps of acquiring the transient signals as they come, dynamically building up clusters of similar transient signals in a hyperspace based on comparison and clustering rules so that each new one of the transient signal acquired ends up in a cluster with similar transient signals formerly acquired, analyzing the clusters to determine respective signatures defined by the transient signals gathered in the clusters, and processing the signatures to detect a phenomenon connectable to an intrinsic attribute of the transient signals.