5. Using Data-Driven Models¶

In addition to the physics-based modeling functionalities described so far, ProgPy also includes a framework for implementing data-driven models.

A data-driven model is a model where the behavior is learned from data. In ProgPy, data-driven models derive from the parent class progpy.data_models.DataModel. A common example of a data-driven model is one that uses neural networks (e.g., progpy.data_models.LSTMStateTransitionModel).

Some data-driven methodologies we will be exploring in this section include Long Short-Term Memory (LSTM), Dynamic Mode Decomposition (DMD), and Polynomial Chaos Expansion (PCE). The data-driven architecture also includes surrogate models, which can be used to create models that approximate the original/higher-fidelity models, generally resulting in a less accurate model that is more computationally efficient.

For more information, refer to the DataModel documentation.

Before we get started, make sure to install the data-driven dependencies using the following command:

pip install 'progpy[datadriven]'

Table of Contents¶

• General Use
  • Building a New Model from Data
  • Surrogate Models
• Long Short-Term Memory (LSTM)
• Dynamic Mode Decomposition (DMD)
• Polynomial Chaos Expansion (PCE)
• Extending
• Conclusion

General Use¶

Building a New Model from Data¶

A version of this notebook will be added in release v1.9

Surrogate Models¶

A version of this notebook will be added in release v1.9

Long Short-Term Memory (LSTM)¶

A version of this notebook will be added in release v1.9

Dynamic Mode Decomposition (DMD)¶

A version of this notebook will be added in release v1.9

Polynomial Chaos Expansion (PCE)¶

A version of this notebook will be added in release v1.9

Extending¶

A version of this notebook will be added in release v1.9

Conclusion¶

In this section, we were able to look at various ways to use data-driven models. The next section 06 Combining Models examines how prognostics models can be combined.