Structured Machine Data Models for Cross-Machine Control Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing manufacturing control systems face challenges in efficiently transferring knowledge and control settings between machines due to lack of standardized data modeling, leading to inefficient machine learning and data analysis across different manufacturing environments.
Innovation Solution
The implementation of a standardized structured data model in a controller that describes machines using predefined categories and labels, enabling a parent-child relationship, and instantiating a machine learning model trained on a similar machine to control operations, allowing for efficient data collection, analysis, and transfer of control settings between machines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a standardized structured data model is implemented across machines, then knowledge and control settings can be efficiently transferred between machines, but device complexity increases due to the need for standardized modeling frameworks
Solution Approach 1:
The patent implements a universal standardized structured data model that can be applied across multiple machines and manufacturing environments. This model uses predefined categories (machine, component, subsystem, measured parameter) and labels that universally describe machine operations, enabling knowledge transfer from one machine to another without requiring machine-specific customizations
Solution Approach 2:
The patent transforms unstructured or semi-structured machine data into a standardized structured format with specific parameters and categories. By changing the data representation parameters to follow a consistent model, the system enables efficient knowledge transfer and machine learning model portability across different machines
2Measurement precision
If machine learning models are trained on individual machines without standardized data models, then models can be highly accurate for specific machines, but training time increases significantly for each new machine
Solution Approach 1:
The patent performs preliminary structuring of machine data according to a standardized model before machine learning training. By pre-organizing data into consistent categories and labels, the system prepares the data in advance to enable faster model training and transfer, reducing the time required when deploying models to new machines
Solution Approach 2:
The patent enables copying of machine learning models and control settings between machines by using the standardized structured data model as a common framework. Once a model is trained on one machine with standardized data, it can be copied and adapted to other machines with similar standardized data structures, eliminating the need to train from scratch
3Productivity
If standardized data models are implemented across manufacturing equipment, then data analysis and reporting efficiency improve, but ease of operation decreases due to stricter data requirements
Solution Approach 1:
The patent segments machine data into distinct standardized categories (machine, component, subsystem, measured parameter) with hierarchical relationships. This segmentation organizes data systematically, making analysis more efficient while providing clear guidance on what data to collect and how to structure it, thereby balancing standardization requirements with operational ease
Data Source
AI summary
Following activation of a first machine, a standardized structured data model is instantiated in a controller of the first machine that describes the first machine according to predefined categories populated with predefined labels that are indicative of measured parameters of the first machine, components of the first machine, and subsystems of the first machine.