Batch-Run Quality Indicator Using Weighted Time-Series Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining quality indicators in batch production processes face challenges due to varying time intervals and multi-variate data sources, which can lead to inaccurate similarity detection and delayed quality category assignment, especially when indicating failure.
Innovation Solution
A computer-implemented method that converts multi-variate time-series data from production batch-runs into uni-variate data by multiplying with source-specific factors and summing up values, allowing for more accurate comparison with reference time-series to differentiate conforming or non-conforming batch-runs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multi-variate time-series data from multiple data sources is used for quality assessment, then the comprehensiveness of quality evaluation is improved, but the complexity of data processing and comparison increases
Solution Approach 1:
The patent combines multiple uni-variate time-series data from different data sources into a single converted time-series by applying conversion factors and summing the weighted values. This merging approach maintains comprehensive quality assessment while simplifying the comparison process by reducing multi-dimensional data to a unified metric that can be directly compared against reference values.
Solution Approach 2:
The patent transforms multi-variate time-series data into uni-variate converted time-series data by applying parameter changes through conversion factors. Each data source is assigned a specific conversion factor that reflects its importance, and the transformation process converts complex multi-parameter data into a simplified single-parameter representation that preserves quality information while reducing processing complexity.
2Adaptability or versatility
If time-series data with varying time intervals is used for batch production comparison, then the adaptability to different production durations is improved, but the accuracy of similarity detection deteriorates
Solution Approach 1:
The patent applies dynamic scaling through conversion factors that can be adjusted based on the specific characteristics of each batch run. The conversion factors allow the system to adapt to varying production durations and time intervals dynamically, while the mathematical transformation maintains the relative relationships between data points, preserving detection accuracy despite temporal variations.
Solution Approach 2:
The patent changes the temporal parameter representation by applying conversion factors that normalize time-series data from different production durations. This parameter transformation allows batch-runs of varying lengths to be compared on a common scale, maintaining similarity detection accuracy while accommodating production flexibility.
3Measurement precision
If conversion factors are applied to weight different data sources, then the precision of quality category differentiation is improved, but the complexity of data conversion increases
Solution Approach 1:
The patent applies parameter changes through conversion factors that weight different data sources according to their importance for quality assessment. This mathematical transformation enhances the precision of quality category differentiation by emphasizing critical parameters while de-emphasizing less important ones, achieving higher measurement precision through systematic parameter adjustment.
Solution Approach 2:
The patent creates a simplified copy of the original multi-variate data through the converted time-series representation. Instead of processing the full complexity of multi-source data, the system creates a weighted summary copy that retains the essential quality information while reducing conversion complexity through a unified mathematical formula.
Data Source
AI summary
To determine a quality indicator of production batch-run of a production process, a computer compares time-series with multi-source data from a reference batch-run and time-series with multi-source data from the production batch-run. Before comparing, the computer converts multi-variate time-series to uni-variate time-series, by first multiplying data values of source-specific uni-variate time-series with source-specific factors from a conversion factor vector and second summing up the multiplied data values according to discrete time points. The source-specific factors of the conversion factor vector are obtained earlier by processing reference data, including the determination of characteristic portions of the time-series, converting, aligning by time-warping and evaluating displacement in time between characteristic portions before alignment and after alignment.