Manufacturing Snapshot Similarity Analysis for Emerging Problem Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting emerging problems in product manufacturing are inadequate due to their global nature, reliance on manual filtering, high computational costs, sensitivity to data instrumentation, and limitations in handling non-textual data and unknown problem types, leading to false negatives and false positives.
Innovation Solution
A computer-implemented method that retrieves a subset of snapshots from a time-ordered set based on similarity thresholds, determines trends by comparing these snapshots to a baseline, and computes similarity signatures to infer emerging problems without prior knowledge of their occurrence or cause-effect relations, using neural network architectures and metadata analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If trend detection methods are used to detect global changes in events, then the ability to detect overall patterns is improved, but the ability to detect emerging problems representing a small portion of global volume deteriorates
Solution Approach 1:
The patent segments the event detection process by introducing semantic dimensions (product, component, location, time) that divide the global event space into meaningful segments. This allows emerging problems to be detected within specific segments rather than being lost in the global volume, resolving the contradiction between detecting overall patterns and detecting small portions of global volume.
Solution Approach 2:
The patent adds semantic dimensions (product, component, location, time) to transform the detection from a single-dimensional global trend analysis to a multi-dimensional analysis. This dimensional expansion enables the system to identify emerging problems in specific semantic contexts while maintaining awareness of global patterns, thus resolving the contradiction between detecting small portions and global volume.
2Measurement precision
If weak signal detection based on explicit semantic analysis is used to extract event categories, then the ability to detect specific problem types is improved, but the computational cost and complexity of manual filtering increases
Solution Approach 1:
The patent implements self-service by automatically generating and updating semantic models from the event data itself, rather than requiring manual creation and maintenance of filter rules. The system learns semantic patterns autonomously, eliminating the need for manual filtering complexity while maintaining high detection accuracy.
Solution Approach 2:
The patent changes the parameters of semantic analysis from static, pre-defined categories to dynamic, data-driven semantic models. By adapting semantic parameters automatically based on observed events, the system maintains high detection accuracy without the computational burden of manual filter management.
3Quantity of substance
If sentiment analysis is used to detect generic quality signals, then the ability to detect overall quality trends is improved, but the ability to discriminate inherently different non-quality events deteriorates
Solution Approach 1:
The patent segments quality signals by associating them with specific semantic contexts (product, component, location, time). Instead of treating all quality events uniformly, the system divides them into segmented categories, enabling both broad quality trend detection and precise discrimination of different non-quality events within their respective segments.
4Measurement precision
If anomaly detection methods are used to monitor all available data, then the ability to detect isolated occurrences is improved, but the ability to detect recurrent issues with root causes deteriorates
Solution Approach 1:
The patent applies preliminary action by building semantic models and establishing baseline patterns before detecting anomalies. This preparatory structuring of data with semantic context enables the system to distinguish between isolated anomalies and recurrent issues with root causes, improving reliability of root cause detection while maintaining anomaly detection accuracy.
Data Source
AI summary
A computer-implemented method for inferring an emerging problem in product manufacturing. The method comprises obtaining a time-ordered set comprising one or more snapshots of a product and one or more similarity thresholds. The method also comprises obtaining at least one recent snapshot, the at least one recent snapshot being time-ordered after at least one snapshot of the time-ordered set. The method also comprises retrieving a subset of one or more snapshots from the time-ordered set, the one or more snapshots being time-ordered before the at least one recent snapshot and satisfying, with respect to the at least one recent snapshot, a similarity above at least one of the one or more similarity thresholds. The method also comprises determining a trend from the retrieved subset and a baseline, the trend being a time distribution of the snapshots of the retrieved subset with respect to the baseline.