NIR Model Training With Virtual Reference Values for Faster Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing NIR models require analytical reference values, which are costly, time-consuming, and resource-intensive, making it difficult to efficiently interpret near-infrared spectra.
Innovation Solution
A method using model-based reference values trained with historical metadata and NIR spectral data to reduce the reliance on analytical reference values, improving the efficiency and accuracy of NIR model interpretation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If analytical methods are used to obtain reference values, then measurement precision is improved, but loss of time and loss of substance increase
Solution Approach 1:
The patent creates a virtual copy of the analytical measurement process through a reference model that predicts reference values without requiring physical sampling and laboratory analysis. The reference model learns from historical data where analytical values are known, enabling it to generate accurate reference values for new samples without actual analytical measurement, thus eliminating the time-consuming physical analysis process while maintaining measurement precision.
Solution Approach 2:
The system performs preliminary training of the reference model using historical data where analytical reference values are already known. This preliminary action allows the model to be pre-configured with the relationships between NIR spectral data and accurate reference values, so that when new samples are measured, the model can immediately provide accurate reference values without requiring time-consuming analytical measurement processes.
2Measurement precision
If analytical methods are used to obtain reference values, then measurement precision is improved, but cost increases
Solution Approach 1:
The patent replaces expensive physical analytical measurements with a computational model that generates virtual reference values. The reference model is trained on historical data from actual analytical measurements, allowing it to reproduce accurate reference values without requiring repeated expensive laboratory analyses. This copying approach maintains measurement precision while eliminating the high costs associated with repeated analytical testing.
Solution Approach 2:
The system enables self-service by allowing the reference model to generate its own reference values without requiring external analytical laboratory services. Once trained, the model autonomously predicts accurate reference values for new samples, eliminating the need to pay for external analytical testing while maintaining the same level of measurement precision.
3Measurement precision
If analytical methods are used to obtain reference values, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent simplifies the overall system by replacing the complex physical analytical measurement process with a simpler computational model. Instead of requiring physical sampling, laboratory equipment, and complex analysis procedures, the system uses a reference model that processes NIR spectral data through computational algorithms, significantly reducing device and process complexity while maintaining reference value accuracy.
Solution Approach 2:
The system substitutes the mechanical and physical analytical measurement process with an information-processing approach. Rather than requiring physical sampling, transportation, and laboratory analysis, the reference model uses computational mechanisms to generate reference values from NIR spectral data, replacing complex mechanical systems with simpler computational processes.
4Loss of time
If model-based reference values are used, then loss of time is reduced, but measurement precision may deteriorate
Solution Approach 1:
The system incorporates feedback mechanisms where the reference model continuously learns from historical data and compares its predictions against actual analytical values. This feedback loop allows the model to refine its accuracy over time, ensuring that the time-saving benefits are achieved without sacrificing measurement precision. The model adapts to new data and improves its predictive accuracy through ongoing training and validation.
Solution Approach 2:
The reference model performs preliminary training using extensive historical data where accurate analytical reference values are already known. This preliminary action ensures that the model is pre-calibrated with the correct relationships between NIR spectral features and accurate reference values, so that when the model makes predictions for new samples, it does so with high accuracy, eliminating the trade-off between speed and precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach reduces the need for costly and time-consuming analytical methods, lowering resource consumption and accelerating the interpretation of NIR spectra while maintaining accuracy and reliability.
Implementation Method 1
receiving NIR spectral data
Data Source
AI summary
Disclosed herein is a computer-implemented method for measuring a target value with a NIR model including:(a) receiving NIR spectral data,(b) determining at least one target value with the trained NIR model based on the spectral data, where the NIR model was trained based on NIR training data including at least one model-based reference value determined with a reference model and historical NIR spectral data, and(c) providing the at least one target value.