Interconnected AI Digital Twins for Industrial Anomaly Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Creating a digital twin for industrial assets without high-fidelity physics-based models is challenging due to the time-consuming and error-prone process, especially for older assets or those from various manufacturers, and existing methods fail to automatically generate a stable and accurate digital twin.
Innovation Solution
A system that uses historical system node data to automatically construct interconnected artificial intelligence models, which are trained simultaneously to create a digital twin, and injects synthetic disturbances to simulate abnormal operations, enabling robustness analysis and anomaly detection without relying on pre-existing models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-fidelity physics-based models are used to create digital twins, then model accuracy is improved, but model development time and complexity increase significantly
Solution Approach 1:
The patent replaces traditional physics-based mechanical modeling with machine learning-based data-driven modeling. Instead of using complex physics equations and mechanical system representations, the invention uses neural networks and other ML algorithms to learn system behavior directly from operational data, thereby reducing model development time while maintaining accuracy
Solution Approach 2:
The patent creates virtual copies of physical assets using data-driven approaches. Rather than building detailed physics-based models that require extensive domain knowledge and time, the system creates accurate virtual representations by copying and learning from historical operational data patterns, enabling faster digital twin deployment
2Measurement precision
If high-fidelity physics-based models are used to create digital twins, then model accuracy is improved, but the modeling process becomes more error-prone
Solution Approach 1:
The patent replaces error-prone manual physics-based modeling with automated machine learning processes. The ML algorithms automatically learn system dynamics from data without requiring manual specification of complex physics equations, reducing human error and improving modeling reliability
Solution Approach 2:
The system enables self-service modeling where the digital twin automatically learns and adapts from operational data without requiring continuous expert intervention. The model self-adjusts its parameters and structure based on incoming data, reducing the risk of errors from manual model maintenance
3Productivity
If data-driven approaches are used to create digital twins, then model development time is reduced, but detection accuracy may decrease
Solution Approach 1:
The patent segments the digital twin into multiple specialized ML models, each trained on specific types of operational data and designed to detect particular anomalies. This segmentation allows the system to maintain high detection accuracy across different asset types and failure modes while keeping individual model development time short
Solution Approach 2:
The patent creates universal data-driven models that can be applied across multiple asset types and manufacturing sources. These multi-functional models learn general patterns from diverse data sources, enabling accurate anomaly detection without requiring asset-specific physics models, thus maintaining both speed and accuracy
Data Source
AI summary
In some embodiments, a system node data store may contain historical system node data associated with normal operation of an industrial asset, and a plurality of artificial intelligence model construction platforms may receive historical system node data. Each platform may then automatically construct a data-driven, dynamic artificial intelligence model associated with the industrial asset based on received system node data. The plurality of artificial intelligence models are interconnected and simultaneously trained to create a digital twin of the industrial asset. A synthetic disturbance platform may inject at least one synthetic disturbance into the plurality of artificial intelligence models to create, for each of a plurality of monitoring nodes, a series of synthetic disturbance monitoring node values over time that represent simulated abnormal operation of the industrial asset.