31/08/26 Mon 15:37pm
Hybrid Modeling Techinque
A hybrid model combines a physics based structure with a data driven ML component where physics provides skeleton structure and the data fills the gaps.
Note
The goal of hybrid modelling is to account for the details that physics based modeling misses ( like air resistance, and other minor details that are assumed)
Architecture
1.Physics + ML Residual Correction
2.Physics Informed Machine Learning
3.System Identification
Iteratively solving for the coefficients of the physics equation of the real world asset
Analogy
Think of it as trying to understand the workings of a black box. You can provide inputs to it and receive outputs, you have to figure out the inner workings by providing different inputs and checking their outputs.