Industrial Digital Twin Pretraining With Physics-Informed Models

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Industrial systems face challenges in developing data-driven models that accurately capture process physics due to limited real-time sensor measurements, high computational costs of physics-based models, and the complexity of hybrid models, leading to suboptimal operations and predictive inaccuracies.

Innovation Solution

A method and system for developing physics-informed data-driven models using a combination of sensor and experimental data, preprocessing, physics-based mathematical formulations, and iterative training to minimize both data-based and physics-based errors, enabling real-time monitoring and predictive analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If physics-based models are used to capture process physics, then model precision and physical consistency are improved, but computational cost and simulation time increase significantly

Engineering Contradiction:
Improvemodel precisionVSAvoidsimulation time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent pre-computes and stores system matrices (mass matrix, stiffness matrix, damping matrix) and their derivatives during an offline phase. These pre-computed matrices are then reused during real-time simulation, avoiding repeated computationally expensive matrix operations and enabling fast evaluation of physics-based models while maintaining accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent separates the model evaluation process into two distinct phases: an offline phase for pre-computing and storing system matrices and parameters, and an online phase for real-time simulation using the pre-computed data. This segmentation allows complex physics-based computations to be performed beforehand, enabling fast real-time predictions without sacrificing model fidelity.

Inventive Principle:
Principle #1Segmentation

2Productivity

If data-driven models are used to reduce computational cost, then real-time monitoring capability is improved, but ability to capture complete process physics deteriorates

Engineering Contradiction:
Improvereal-time monitoring capabilityVSAvoidprocess physics capture
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent merges data-driven models with physics-based models to create a hybrid approach. The data-driven component captures complex nonlinear relationships from historical data, while the physics-based component ensures physical consistency and captures fundamental process physics. The combined model achieves both real-time performance and accurate physics representation.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent transforms the physics-based model into a parameterized form where system matrices and parameters are pre-computed offline. This parameterization allows the model to be evaluated efficiently during real-time operations while maintaining the physical relationships embedded in the original physics-based formulation.

Inventive Principle:
Principle #35Parameter changes

3Manufacturing precision

If extensive experimental data is collected to train data-driven models, then model accuracy is improved, but data acquisition cost and time increase

Engineering Contradiction:
Improvemodel accuracyVSAvoiddata acquisition time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs extensive model training and parameter calibration during an offline phase using available experimental and operational data. This preliminary action allows the model to learn from historical data without delaying real-time applications. The trained model can then be deployed for real-time predictions without requiring additional data collection.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent enables the model to self-calibrate and adapt to changing operating conditions by continuously learning from operational data while maintaining physical constraints. This reduces the need for extensive manual data collection and experimentation, as the model can autonomously improve its accuracy using readily available operational data.

Inventive Principle:
Principle #25Self-service

4Reliability

If hybrid models combining physics-based and data-driven approaches are used, then model reliability is improved, but model complexity increases

Engineering Contradiction:
Improvemodel reliabilityVSAvoidmodel complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent structures the hybrid model into distinct modular components: data-driven models for capturing nonlinear relationships, physics-based models for ensuring physical consistency, and integration mechanisms for combining them. This segmentation makes the complex hybrid model more manageable, interpretable, and easier to implement while maintaining high reliability.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4632504A1Method and system for developing pretrained models for industrial digital twins
Publication Date: 2025.10.15 TATA CONSULTANCY SERVICES LTD
  • EP4632504A1 patent drawingFigure 1
  • EP4632504A1 patent drawingFigure 2
  • EP4632504A1 patent drawingFigure 3A

AI summary

Industrial systems, and equipment are constrained by limited real-time sensor measurements, and monitoring of key performance indicators is difficult which may lead to sub optimal operations of systems. Embodiments of the present disclosure provide a method and system for developing pretrained models for industrial digital twins. A data associated with entity is preprocessed to obtain preprocessed data. The data associated with the entity is mapped with known parameter or unknown parameter of the entity. A n data-driven model is developed based on the identified parameters to select top m models. Top k unknown parameter is selected based on highest value of a parameter attribution score for selected top m model. Estimated value is determined for the top k unknown parameter to develop and iteratively train a physics informed data driven model. A physics-based error and a data-based error must be less than predefined physical discrepancy threshold and data discrepancy threshold.