Signal Processing via Correlation Filter for Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Synchronous averaging techniques reduce both asynchronous and synchronous waveform components, leading to complex computations and inefficiencies in extracting desired signal components from repetitive waveforms.
Innovation Solution
A signal processing method involving the generation of a processing target signal and a template signal from a source signal, followed by correlation filter processing to enhance periodic signal components while reducing ergodic noise, without the need for envelope processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If synchronous averaging is performed to reduce asynchronous waveform components, then the asynchronous noise is reduced, but the synchronous waveform components that are not aligned with the averaging timing are also reduced, leading to loss of useful signal information
Solution Approach 1:
The patent applies preliminary action by generating a template signal from a reference waveform before performing correlation processing. This template signal serves as a pre-prepared pattern that guides the correlation filter to selectively enhance only the relevant synchronous waveform components, avoiding the information loss inherent in synchronous averaging that blindly suppresses all non-aligned components.
Solution Approach 2:
The patent implements local quality by using correlation filter processing that operates locally on matching patterns between the template signal and the input signal. Instead of globally suppressing all non-synchronous components, the correlation filter selectively enhances only those waveform components that locally match the template pattern, preserving useful signal information while reducing noise.
2Measurement precision
If envelope processing is performed followed by spectral analysis to extract signal components, then the signal extraction capability is improved, but the computational complexity increases due to the need for smoothing operations and filter selection
Solution Approach 1:
The patent extracts the essential signal component directly through correlation filter processing that operates on the time-domain signal. By taking out only the relevant synchronous components through template-matched correlation, the method avoids the complex multi-step process of envelope detection, smoothing filter application, and spectral analysis, significantly reducing computational complexity while maintaining extraction effectiveness.
Solution Approach 2:
The patent substitutes the mechanical signal processing chain (rectification → smoothing filtering → spectral analysis) with a more efficient correlation-based approach. The correlation filter processing replaces multiple mechanical operations with a single mathematical operation that achieves the same signal extraction goal with reduced computational burden.
3Measurement precision
If smoothing filter processing is applied during envelope processing to extract desired signal components, then the signal extraction is improved, but the selection and tuning of proper cut-off frequencies adds complexity and requires additional computational resources
Solution Approach 1:
The patent implements self-service by allowing the correlation filter to automatically adapt to the signal characteristics through template matching. The filter parameters are derived directly from the reference waveform template, eliminating the need for manual cut-off frequency selection and complex filter tuning. The system serves itself by using the inherent structure of the reference signal to configure the optimal filtering behavior.
Data Source
AI summary
A signal processing method includes: a processing target signal generation step of generating a processing target signal that is a time-series signal, based on a source signal that is a time-series signal corresponding to a physical quantity generated in a steady state of a target object; and a correlated signal generation step of performing correlation filter processing between a template signal that is a time-series signal generated based on the source signal, and the processing target signal, and generating a correlated signal that is a time-series signal.