Noise Level Estimation in Chromatograms Using Segmented Waveform Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the noise component in chromatograms and similar measurement data are inaccurate due to fluctuating noise factors, making it difficult to separate noise from peak components effectively.
Innovation Solution
A time-frequency analysis method that segments waveform data based on positive and negative value changes, excludes segments exceeding a reference value, and calculates noise levels statistically to accurately estimate noise magnitude, while normalizing and correcting for frequency and environmental factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If white noise approximation is used to estimate noise component, then the estimation process is simple, but the noise level estimation accuracy is insufficient
Solution Approach 1:
The patent segments the chromatogram into multiple sections and performs noise estimation separately for each section. This allows the method to adapt to local noise characteristics while maintaining a systematic approach, resolving the contradiction between simplicity and accuracy by dividing the complex estimation task into manageable segments that can be processed independently.
Solution Approach 2:
The patent changes the estimation parameters by using local standard deviation and skewness calculations instead of global white noise assumptions. By adapting the noise estimation parameters to local chromatogram characteristics, the method achieves higher accuracy without significantly increasing computational complexity.
2Measurement precision
If noise component is estimated from a portion of chromatogram considered to include no peak, then the estimation can be performed, but it is difficult to identify the portion with absolutely no peak component
Solution Approach 1:
The patent applies local quality by estimating noise characteristics separately for different sections of the chromatogram rather than assuming uniform noise properties throughout. Each section's noise is estimated based on its local characteristics, allowing the method to work effectively even when peak-free portions are difficult to identify globally.
Solution Approach 2:
The patent employs dynamic noise estimation that adapts to local conditions in each chromatogram section. By calculating local standard deviation and skewness for each segment, the method dynamically adjusts to varying noise characteristics without requiring explicit identification of peak-free regions.
3Ease of operation
If the noise component is approximated by white noise, then the calculation is straightforward, but the fluctuating noise factors cannot be accounted for
Solution Approach 1:
The patent performs preliminary segmentation of the chromatogram into multiple sections before conducting noise estimation. This preliminary action allows the subsequent noise calculation to account for local variations in noise characteristics, improving reliability while keeping the overall process straightforward through systematic preparation.
Solution Approach 2:
The patent replaces the simple white noise model with a more sophisticated statistical approach using local standard deviation and skewness calculations. This substitution maintains computational straightforwardness while significantly improving the ability to account for noise fluctuations through statistical characterization.
Data Source
AI summary
A method includes: performing a time-frequency analysis on measurement data to obtain waveform data representing a temporal change in the intensity of each of a plurality of frequency components; dividing the waveform data of each of a plurality of predetermined frequencies into a plurality of segments so that each section where positive values successively occur and each section where negative values successively occur in a time-axis direction are defined as one segment; calculating the area of each of the segments to obtain segment values; creating, for the waveform data of each of the predetermined frequency components, a selected segment group by excluding a segment whose segment value exceeds a predetermined reference value from the segments in the waveform data; and determining a noise level of each of the predetermined frequency components based on the average value of the segment values of the segments included in the selected segment group.