Surrogate modeling

• IDAES Surrogates API
• RIPE: Reaction Identification and Parameter Estimation
• HELMET: HELMholtz Energy Thermodynamics
• Sampling
• Visualizing Surrogate Model Results

IDAES contains several surrogate modeling tools, including the IDAES Surrogates API which enables integrating ALAMO, PySMO or Keras surrogate models into IDAES flowsheets.

ALAMOpy, RIPE, and HELMET are data driven machine learning (ddm-learning) tools which leverage ALAMO, a regression application for generating equation-based surrogates. RIPE and HELMET are regression tools for the development of property models for kinetics and thermodynamics of a system. The provided tools include both ALAMOpy and RIPE that can access ALAMO and other solvers through the Python API, and an interface for accessing ALAMO solvers (via an external license) within IDAES flowsheets.

Python-based Surrogate Modeling Objects (PySMO) is a framework for general-purpose surrogate modeling techniques, integrated within the Pyomo mathematical optimization framework (on which IDAES is also based). The provided tools include an interface for accessing PySMO methods (via IDAES-internal scripts) within IDAES flowsheets.

Keras is a deep learning framework that integrates with TensorFlow’s structure for building and training artificial neural networks, and minimizes the number of user actions required to construct accurate networks. OMLT (Optimization and Machine Learning Toolkit) provides an interface to formulate machine learning models and import Keras or ONNX models as Pyomo blocks. The provided tools include an interface for accessing the Keras module (via the publicly available Python package) within IDAES flowsheets.