EBC-ML Interface for Accurate Virtual Factory Productivity Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing virtual production systems lack the ability to accurately predict productivity and efficiency due to limitations in equipment behavior catalogues that only include equipment properties, necessitating an interface between equipment behavior catalogues and machine learning frameworks to incorporate equipment behaviors and operation conditions.
Innovation Solution
An apparatus and method for interfacing an equipment behavior catalogue (EBC) framework with a machine learning (ML) framework, enabling the integration of ML models by specifying an interface between the EBC and ML frameworks, allowing for the development, evaluation, and execution of ML models to enhance equipment performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If equipment catalogues include only equipment property information, then the catalogue structure is simple, but the ability to accurately predict productivity and efficiency is limited
Solution Approach 1:
The patent combines equipment property information with equipment behavior information (including operation conditions, state transitions, and operation results) into a unified equipment behavior catalogue. This merging allows the system to maintain prediction accuracy while managing complexity through integrated data structures that correlate properties with behavioral patterns.
Solution Approach 2:
The patent introduces an interface layer between the equipment behavior catalogue and machine learning frameworks. This intermediary component enables accurate prediction by facilitating data exchange and model integration without requiring direct modification of the catalogue structure, thus resolving the contradiction between prediction accuracy and structural complexity.
2Productivity
If ML models are integrated into equipment behavior catalogues, then productivity prediction is improved, but system complexity increases
Solution Approach 1:
The patent segments the system into distinct functional modules: the equipment behavior catalogue storage, the interface layer, and the machine learning framework. This segmentation allows productivity prediction to be improved through ML integration while managing system complexity by keeping each module independent and well-defined with clear interfaces between them.
Solution Approach 2:
The patent creates a universal interface that can accommodate multiple ML models and various equipment types within the same framework. This multi-functional interface design enables productivity improvement across different equipment while maintaining consistent system architecture, thus managing complexity through standardization.
3Extent of automation
If equipment behavior data is collected and processed, then ML model training is enabled, but data processing time increases
Solution Approach 1:
The patent implements preliminary actions by pre-processing and structuring equipment behavior data before ML model training begins. Data collection and initial processing occur in advance, with behavior data being organized and stored in the equipment behavior catalogue ready for training. This reduces the actual training time by having data prepared beforehand.
Solution Approach 2:
The system enables self-service through automated data collection and processing mechanisms that operate without manual intervention. The equipment behavior catalogue automatically captures operation conditions, state transitions, and results, and the interface layer automatically prepares data for ML training, reducing processing time through automation.
Data Source
AI summary
The present invention relates to an apparatus and method for interfacing an equipment behavior catalogue (EBC) framework with a machine learning (ML) framework, and the apparatus includes an EBC framework including an EBC storage and a set of equipment instance models, and an ML framework interfaced with the EBC framework and including a collection of tools and libraries that support development and learning of ML models.