Calibration Curve Creation Using Independent Component Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calibrating target components from test objects require multiple observation data items and are prone to accuracy degradation due to noise and variability in test object composition or structure, especially when dealing with unevenness or outliers in the data.
Innovation Solution
A calibration curve creation method that uses a computer to acquire and process observation data by separating it into independent components, estimating mixing coefficients, and constructing a regression equation using β divergence and robust regression to improve accuracy, even with a single observation data item, by performing preprocessing steps like normalization and whitening.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple observation data items are used for calibration, then measurement precision may be improved through averaging, but the complexity of the calibration process increases and time consumption increases
Solution Approach 1:
The patent performs preliminary independent component analysis and preprocessing on a single observation data item to extract pure component spectra and mixing coefficients before calibration. This preliminary extraction of calibration parameters from one data item eliminates the need for multiple repeated measurements and averaging, thereby reducing calibration time while maintaining accuracy through the robustness of the ICA-based method
Solution Approach 2:
The patent extracts the essential calibration information (pure component spectra and mixing coefficients) from a single observation data item through independent component analysis. By taking out only the necessary calibration parameters rather than requiring multiple full datasets, the method achieves accurate calibration without the time cost of collecting multiple observation data items
2Adaptability or versatility
If independent component analysis is performed on observation data containing noise, then component separation may be achieved, but measurement precision degrades due to noise interference
Solution Approach 1:
The patent applies preprocessing steps (normalization and whitening) before independent component analysis to cushion against the harmful effects of noise. Normalization standardizes the data scale and whitening transforms the data to have unit variance and zero mean, creating a预处理 layer that protects the ICA algorithm from noise interference and improves both noise resistance and calibration accuracy
Solution Approach 2:
The patent converts the presence of noise into a benefit by using robust regression methods that are specifically designed to handle outliers and noisy data. The robust regression transforms the harmful noise into a manageable factor that the algorithm can accommodate, thereby maintaining or even improving calibration accuracy in noisy environments
3Measurement precision
If robust regression method is used to reduce outlier impact, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent performs preliminary preprocessing (normalization and whitening) and independent component analysis before applying robust regression. By preparing the data in advance through ICA to separate pure components and estimate mixing coefficients, the robust regression operates on already-structured data, which reduces the computational burden and complexity compared to applying robust regression directly to raw noisy data
Data Source
AI summary
A calibration curve creating method includes: (a) acquiring observation data of a plurality of samples of a test object; (b) acquiring content of a target component of each sample; (c) estimating a plurality of independent components when the observation data of each sample is separated into the plurality of independent components, and acquiring a mixing coefficient corresponding to the target component for each sample; and (d) acquiring a regression equation of a calibration curve. (c) includes acquiring an independent component matrix by performing a first preprocessing including normalization of the observation data, a second preprocessing including whitening, and an independent component analysis process in this order. β divergence is used as an independence index of the independent component analysis process, and a robust regression method is used in (d).