Simulator-Informed Neural Networks

Accelerate Physics-based Models with Industrial Machine Learning

What are the goals and aims of this project?

How are SimINN Surrogate Models created?

How can Surrogate Models be used?

Comments, Questions, and Additional Resources  

Leverage Speed.

Machine learning surrogates of physics-based models are accurate, flexible, and fast.

Complete typical engineering workflows in a fraction of the time.

Constrained optimization, the workhorse of modern engineering, is robustified and accelerated.  

Embrace Uncertainty.

Run parameter estimation algorithms in real-time.

Generate confidence bounds to enable effective decision-making.

Computationally intensive algorithms become feasible with surrogate models.

Capture Performance.

Unlock total system performance with optimization.

Accurate performance predictions reduce unexpected events.

Optimize even when model parameters are only partially known.