Excursion Screening Models for False-Positive Filtering in Manufacturing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing manufacturing systems face challenges in accurately detecting and classifying excursions, which can lead to inefficiencies and inconsistencies in product quality due to irrelevant excursions being flagged.
Innovation Solution
Implementing an excursion screening model that uses pattern matching and supervised learning techniques to classify excursions detected by excursion detection models, allowing for the filtering out of irrelevant excursions and improving the accuracy of anomaly detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional excursion detection methods are used, then excursions can be detected, but false positives increase and accuracy decreases
Solution Approach 1:
The system segments the excursion detection process into two independent stages: (1) excursion detection models that identify potential excursions from sensor data, and (2) an excursion screening model that classifies detected excursions as relevant or irrelevant. This segmentation allows each model to specialize in its specific task, improving overall detection accuracy while reducing false positives through the filtering capability of the screening model.
Solution Approach 2:
The excursion screening model acts as an intermediary between the excursion detection models and the final decision-making process. It receives detected excursions as input, applies supervised learning classification, and outputs filtered results that distinguish relevant excursions from false positives. This intermediary layer refines the raw detection output before it reaches users or control systems.
2Reliability
If multiple excursion detection models are used, then detection coverage improves, but computational complexity increases
Solution Approach 1:
The system divides the computational workload into two distinct segments: multiple specialized excursion detection models that focus on identifying potential excursions with high sensitivity, and a single excursion screening model that handles the classification and filtering of detected excursions. This segmentation allows the detection phase to maintain high coverage while the screening phase manages computational complexity by consolidating the classification task.
Solution Approach 2:
The system extracts the classification function from the detection process itself and places it in a separate screening model. This extraction allows the detection models to remain computationally efficient while delegating the more complex classification task to a dedicated screening component that processes only the excursions already detected, rather than performing both detection and classification in a single complex model.
3Productivity
If traditional monitoring is used, then all excursions are flagged, but irrelevant excursions consume resources
Solution Approach 1:
The excursion screening model serves as an intermediary filter between raw excursion detections and resource-intensive follow-up actions. It classifies excursions as relevant or irrelevant using supervised learning, allowing the system to allocate computational resources and attention only to genuinely relevant excursions while automatically filtering out irrelevant ones, thus improving resource efficiency without losing important information.
Solution Approach 2:
The system applies different quality standards to different excursions based on their classification. Relevant excursions receive thorough analysis and follow-up actions, while irrelevant excursions are quickly filtered out. This local differentiation of processing quality ensures that resources are concentrated on important cases while maintaining high overall efficiency through rapid dismissal of unimportant cases.
Data Source
AI summary
A method includes identifying at least one excursion corresponding to sensor data obtained from a manufacturing system, initiating an excursion screening process to screen the at least one excursion using an excursion screening model, wherein the excursion screening model is trained to classify the at least one excursion, and causing a screening output resulting from the screening process to be displayed on a user device.