Industrial Intelligence A structured map of industrial AIA structured map of industrial AI
LinkedIn
Technology
Status
Fit
Effect
Hover any cell to preview
Engineering & Simulation · Engineering
ML for Engineering / Physics-informed & Hybrid ML
Physics-informed engineering models
Scaling Adjacent medium effect
Core capability
Because the model is constrained by physics as well as data, it can produce results that are more trustworthy in engineering contexts than purely data-driven models alone.
How it works
The model is trained not only to match historical data but also to respect physical laws, which helps reduce unrealistic outputs and makes the results more trustworthy in engineering applications.
Application here
Models with built-in physical constraints deliver better predictions when there is not enough training data to rely on machine learning alone.
Business impact
This is particularly useful when engineering data is scarce or expensive, because physical knowledge helps compensate for limited data.
Limitations
It requires strong domain expertise to encode the right physics. Wrong assumptions can make the model worse rather than better.
In production
This is already valuable when pure machine learning is too unreliable but full first-principles simulation is too slow or difficult to operationalize.
Research
The frontier is toward systems that can recover or refine the hidden rules of a process from measurement data, making advanced modeling possible even when the full physics is not already written down.
Examples
Liquid‑carryover detection in a refinery centrifugal compressor using physics‑informed analytics
Description: In a hydrogen compressor train at a refinery, an analytics platform combined thermodynamic performance models and ML to detect liquid carry‑over from small deviations in polytropic head, efficiency, and power balance; the model flagged the anomaly before physical failure and inspection confirmed liquid ingress from an upstream reciprocating compressor.


Physics‑informed machine learning for gas‑turbine performance degradation monitoring
Description: A physics‑informed ML framework for heavy‑duty gas turbines integrates thermodynamic balance equations and component maps with an LSTM network to predict performance degradation; the model uses operational data and ISO‑condition physics to estimate efficiency loss and guide predictive maintenance with higher robustness than purely data‑driven models.



Physics‑informed hybrid optimization for robotic welding (PHOENIX framework)
Description: In robotic welding, a physics‑informed hybrid optimization framework (PHOENIX) embeds heat‑transfer and metallurgy constraints into a neural‑network controller, enabling real‑time prediction of welding instability with up to 98% accuracy at 50 ms; the approach reduces reliance on large labeled datasets while maintaining stability margins.
https://www.youtube.com/watch?v=VoY9EXdETa4https://papers.phmsociety.org/index.php/phmap/article/download/3723/2188https://www.nature.com/articles/s41467-025-60164-y
Sources
NVIDIA PhysicsNeMo — ; Raissi et al., Physics-Informed Neural Networks —
https://developer.nvidia.com/physicsnemohttps://doi.org/10.1016/j.jcp.2018.10.045