Sample Analysis System Handling Non-Fixed Variables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analytical techniques face challenges in comparing samples with a non-fixed number of variables to a database, leading to incomplete information and potential misclassification of defects or natural variations in product quality control.
Innovation Solution
A method that classifies variables into three categories: common, specific to the training samples, and specific to the unknown sample, allowing for a comprehensive comparison by calculating summary values and ratios to evaluate similarity or difference, and presenting this information through graphical representations like pie charts or boxplots.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional data processing models are used to compare samples with a fixed number of variables, then the comparison process is simple and straightforward, but the system cannot handle samples analyzed by techniques that generate a non-fixed number of variables (e.g., gas chromatography)
Solution Approach 1:
The patent segments the variables into three distinct categories: common variables (present in both training and test samples), variables specific to training samples, and variables specific to test samples. This segmentation allows the system to handle non-fixed numbers of variables by processing each category appropriately, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The patent creates a universal comparison framework that can handle both fixed and non-fixed numbers of variables through the same data processing model. The model universally processes common variables using traditional methods while also incorporating variables specific to either training or test samples, making the system multi-functional and adaptable to different analytical techniques.
2Loss of information
If the system limits comparison to only common variables between unknown sample and learning base, then the data processing model remains simple, but valuable information from variables specific to each sample is lost
Solution Approach 1:
By segmenting variables into common, training-specific, and test-specific categories, the patent preserves information from all variables rather than limiting comparison to only common ones. The segmented approach allows the system to process each type of variable appropriately, preventing information loss while managing complexity through organized handling of different variable types.
Solution Approach 2:
The patent applies local quality by treating different variable types differently according to their specific characteristics. Common variables are processed using traditional comparison methods, while variables specific to training or test samples receive specialized processing. This localized approach to variable handling preserves information from non-common variables without uniformly increasing complexity across all processing.
3Measurement precision
If the system tries to incorporate all variables (common and non-common) into the comparison, then complete information is utilized, but the data processing model becomes significantly more complex and difficult to implement
Solution Approach 1:
The patent achieves measurement precision by incorporating all variables into the comparison through segmentation. By dividing variables into three categories and processing each category with appropriate methods, the system utilizes complete information from all variables while avoiding the overwhelming complexity that would result from treating all variables uniformly. The segmented structure makes the comprehensive processing manageable and implementable.
4Reliability
If the system updates the identification model by adding unknown samples that resemble learning samples, then the model improves over time, but the process requires automated decision rules and threshold comparisons that increase system complexity
Solution Approach 1:
The patent implements feedback through the automated model update mechanism. Unknown samples that resemble learning samples are added to the learning base, and the identification model is retrained with this expanded data. This feedback loop continuously improves model reliability by incorporating new information, while the automated decision rules and threshold comparisons provide structured control over the update process, managing the complexity through systematic criteria.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6A
AI summary
The method involves measuring values of characteristics of reference samples and a subject sample. An assembly of measured characteristics is marked. The characteristics which are common to the subject sample and the reference samples are identified as a characteristics class. The characteristics specific to the reference samples are identified as another characteristics class. The characteristics specific to the subject sample are identified as a third characteristics class. An indication of the class assigned to the identified characteristics is associated to the measured values. Independent claims are also included for the following: (1) a computer program comprising a set of instructions to perform a method for analyzing a subject sample (2) a subject sample analyzing system comprising a computer program.