Electromagnetic Noise Separation by Partial-Wave Period Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing blind source separation techniques struggle with noise separation in one-channel measurement when frequency bands of multiple electromagnetic noise sources overlap, making it difficult to identify noise sources accurately.
Innovation Solution
A computer system that processes electromagnetic noise signals by defining partial waves, estimating occurrence periods, classifying them into clusters, and filtering out non-similar partial waves within the same cluster to improve noise separation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If blind source separation technique is used for one-channel measurement, then noise source identification can be performed without preliminary information, but noise separation accuracy deteriorates when frequency bands of multiple electromagnetic noise sources overlap
Solution Approach 1:
The patent segments the continuous electromagnetic noise signal into discrete partial waves based on occurrence periods. By dividing the signal into separate partial waves that can be individually analyzed, the system can handle overlapping frequency bands more effectively. Each partial wave is processed separately through classification and filtering, enabling accurate noise source identification even without preliminary information about the noise sources.
2Device complexity
If conventional noise separation methods are used, then processing is simpler, but the ability to separate overlapping frequency bands deteriorates
Solution Approach 1:
The patent introduces a new dimension for signal analysis by classifying partial waves based on occurrence periods rather than solely relying on frequency domain analysis. This temporal dimension (occurrence period classification) provides an additional degree of freedom for separating noise sources with overlapping frequency bands, improving separation accuracy without requiring excessively complex processing.
Solution Approach 2:
The patent changes the analysis parameter from traditional frequency-only analysis to a combined approach using occurrence periods and frequency characteristics. By estimating and utilizing occurrence periods as a key parameter, the system can distinguish between noise sources even when their frequency bands overlap, thereby improving noise separation accuracy while maintaining reasonable processing complexity.
Data Source
Figure 1
Figure 2
Figure 3(a)~3(b)
AI summary
To improve noise separation in one-channel measurement. Provided is a computer system including one or more processors and one or more memory resources. The one or more processors are configured to execute: obtaining a signal including electromagnetic noise; defining a plurality of partial waves included in the signal; estimating, for at least some partial waves of the plurality of partial waves, an occurrence period at which each partial wave occurs in the signal; classifying at least some partial waves of the plurality of partial waves into a plurality of clusters with respect to a plurality of the occurrence periods; and filtering of removing, among the plurality of partial waves belonging to the same cluster, the partial wave, of the plurality of partial waves, that does not have similarity in feature to another partial wave of the plurality of partial waves, from the cluster.