Adaptive FDC Limits Using Hyperplane Classification Boundaries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fault detection and classification (FDC) systems in manufacturing equipment are inefficient, requiring extensive manpower for monitoring, using obsolete data, and failing to account for equipment drift and interactions between sensor statistics, leading to inaccurate fault detection and increased downtime.
Innovation Solution
A method involving machine learning models that process current and historical sensor data to identify features and use hyperplane limits for adaptive fault detection and classification, automatically adjusting limits based on equipment changes and interactions between sensors, reducing false positives and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional FDC systems use fixed thresholds for fault detection, then the system structure is simple, but the measurement precision and reliability deteriorate due to equipment drift and sensor interactions
Solution Approach 1:
The patent implements dynamic FDC limits that automatically adapt to equipment drift and changing operating conditions. Instead of fixed thresholds, the system continuously updates limits based on historical data and machine learning models, allowing the detection criteria to evolve with the equipment while maintaining simple operation for users.
Solution Approach 2:
The system performs self-adjustment by automatically learning from historical sensor data and product outcome data. The machine learning models continuously improve the FDC limits without requiring manual intervention, enabling the system to self-optimize detection accuracy while accounting for equipment drift and sensor interactions.
2Productivity
If manual monitoring is used for fault detection, then the device complexity is low, but the productivity and reliability worsen due to extensive manpower requirements and human error
Solution Approach 1:
The patent replaces manual monitoring with automated machine learning-based detection. The system uses algorithms to analyze sensor data and determine fault conditions, substituting human operators with computational processes that provide consistent, scalable, and error-free monitoring while handling complex multivariate analysis automatically.
Solution Approach 2:
The system introduces machine learning models as intermediaries between raw sensor data and fault detection decisions. These models process and interpret complex sensor interactions, translating raw data into actionable fault predictions while shielding users from the underlying complexity of multivariate sensor analysis.
3Adaptability or versatility
If obsolete data is used for FDC limits, then the adaptability is low, but the reliability worsens due to failure to account for equipment drift
Solution Approach 1:
The system performs preliminary actions by continuously collecting and analyzing historical sensor data and product outcomes to proactively update FDC limits. The machine learning models learn from past equipment behavior and product results, preparing the system to detect future faults more accurately while adapting to gradual equipment drift before it causes failures.
Solution Approach 2:
The system implements feedback loops where product outcome data and sensor data continuously inform FDC limit adjustments. The machine learning models receive feedback from actual product results and use this information to refine detection criteria, ensuring the system adapts to equipment changes while maintaining high reliability through continuous learning and adjustment.
Data Source
AI summary
A method includes receiving, from sensors, current trace data including current sensor values associated with producing products. The method further includes processing the current trace data to identify features of the current trace data and providing the features of the current trace data as input to a trained machine learning model that uses a hyperplane limit for product classification. The method further includes obtaining, from the trained machine learning model, outputs indicative of predictive data associated with the hyperplane limit and processing the predictive data and the hyperplane limit to determine: first products associated with a first product classification and second products associated with a second product classification based exclusively on the subset of the plurality of features; and third products associated with the first product classification or the second product classification based on an additional feature not within the subset.