Physics-Based Digital Twin Platform for Changing Production Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital twin creation methods rely heavily on Big Data Analytics and Machine Learning, which can lead to misleading results when conditions differ from the data collection interval and require additional data collection for new materials, resulting in inefficiencies and increased costs.
Innovation Solution
A method combining mathematical methods with cloud technology to create digital twins using physics-based scientific codes, allowing for adaptable and efficient digital twin development through APIs, enabling users to develop and deploy digital twin applications easily and quickly, without relying solely on sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Machine Learning algorithms are trained with sensor data collected during a determined period of time, then digital twin can be created, but additional data collection period and additional costs are required when new conditions or materials need to be tested
Solution Approach 1:
The patent pre-generates comprehensive sensor data for various production conditions, speeds, and raw materials through simulation before actual production needs arise. This preliminary data generation eliminates the need for extended data collection periods when testing new conditions, as all possible scenarios are already covered in the pre-generated dataset.
Solution Approach 2:
The patent creates virtual copies of production lines through digital twins that can be used for testing and simulation. These digital copies allow evaluation of new conditions, materials, and parameters without requiring physical experimentation or additional data collection from actual production lines, thereby saving time and resources.
2Productivity
If Machine Learning algorithms are trained with data obtained from production line at specific interval, then digital twin can be created, but misleading results occur when physical conditions are outside the training interval
Solution Approach 1:
The patent generates comprehensive sensor data covering multiple production conditions, speeds, and scenarios through simulation. This universal dataset trained on diverse conditions enables the digital twin to accurately estimate behavior across a wide range of operating parameters, making the system universally applicable rather than limited to specific training intervals.
Solution Approach 2:
The patent systematically varies parameters such as production speed, raw material types, and environmental conditions during data generation to create a robust training dataset. This parameter diversification ensures the digital twin remains reliable when conditions change, as it has been trained on a broad spectrum of possible scenarios rather than a narrow interval.
3Adaptability or versatility
If new type of raw material is requested to be tested with digital twin, then new set of data should be collected and Machine Learning algorithms should be retrained, but this results in stopping the line and additional operations
Solution Approach 1:
The patent uses digital twins as virtual copies to test new raw materials and production scenarios. This allows evaluation of new materials through simulation without requiring physical experimentation on the actual production line, eliminating the need to stop production for testing and maintaining continuous operations.
Solution Approach 2:
The patent pre-generates data for various raw material types through simulation before actual testing is needed. When new materials need to be evaluated, the system can immediately query the pre-generated dataset or perform rapid virtual testing, eliminating production stoppages and additional data collection operations.
4Measurement precision
If Big Data Analytics and Machine Learning methods are used for digital twin creation, then digital twin can be created from sensor data, but additional sensors and infrastructure costs increase
Solution Approach 1:
The patent creates virtual production lines through digital twins that replicate the behavior and characteristics of physical systems. These digital copies enable comprehensive analysis and testing without requiring additional physical sensors or infrastructure, as the virtual models can be configured and tested independently of the physical setup.
Solution Approach 2:
The patent replaces the need for extensive physical sensor infrastructure with computational models and simulations. Instead of deploying additional sensors to capture data for every possible scenario, the system uses mathematical models and virtual experimentation to generate and analyze data, reducing hardware complexity and costs.
Data Source
AI summary
Disclosed is a method to create a digital twin which is a copy of a machine or a production system in a computer environment and usage of the created digital twin in the computer-based calculation. Digital twins which are developed by the method are created by physical laws/equations rather than solely by sensor data. Developed digital twins could be created/served with API's by dividing thereof into small pieces and could be resold through a market place in form of an application.