Drive Analytics AI Model Using Physics-Informed Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing monitoring and analytics systems for drive systems and apparatuses face challenges in acquiring and utilizing detailed and reliable data, often relying on high data volumes that are difficult to obtain, leading to inefficient and unreliable predictive maintenance and anomaly detection due to the lack of consideration for physics-based domain knowledge.
Innovation Solution
A method for obtaining a domain-informed ML/AI model that integrates physics-based knowledge with data-driven approaches, using physics-informed neural networks (PINNs) to train models on-premise, leveraging historical and simulated data to ensure compliance with physical laws, enabling real-time, physics-consistent analysis and predictive capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high amounts of detailed and reliable data are acquired from deployed drive systems for training analytics algorithms or ML models, then the accuracy of behavior determination is improved, but the difficulty of data provision and acquisition increases
Solution Approach 1:
The patent applies preliminary action by generating synthetic training data through physics-based simulations before deploying the ML model to actual drive systems. This pre-generation of training data using domain knowledge and physics models eliminates the need to collect large volumes of real operational data, thereby improving measurement precision while reducing the difficulty of data acquisition
Solution Approach 2:
The patent introduces physics-based domain knowledge and simulation models as intermediaries between theoretical physics principles and actual drive system data. These intermediaries generate synthetic training data that bridges the gap between limited real-world data and the extensive training data required for accurate ML models, thereby improving accuracy while avoiding the challenges of direct data collection
2Reliability
If analytics-based algorithms and ML-based algorithms are developed for monitoring and predictive maintenance, then the capability for early fault detection is improved, but the consumption of high amounts of data increases
Solution Approach 1:
The patent applies self-service by enabling the ML model to generate its own training data through physics-based simulations and domain knowledge rather than relying on external data collection from deployed systems. This self-generated training data approach maintains high reliability for early fault detection while eliminating the need to consume large quantities of external operational data
Solution Approach 2:
The patent performs preliminary generation of training data through physics-based simulations before the ML model deployment. This pre-computation of synthetic training data using domain knowledge reduces the need for continuous consumption of real operational data, thereby maintaining reliability while reducing data quantity requirements
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for obtaining a domain-informed machine learning/artificial intelligence, ML/AI, model for drive analytics. The method comprises obtaining first data indicative of a set of data points, wherein each data point is associated with a behavior of a drive apparatus and/or drive system. The method further comprises obtaining second data indicative of domain knowledge comprising physics knowledge associated with a behavior of the drive apparatus and/or drive system and/or with an environment of the drive apparatus and/or drive system. The method further comprises training a machine learning/artificial intelligence, ML/AI, model by jointly utilizing the first data and the second data to obtain the domain-informed ML/AI model for drive analytics.