A combined first-principles and data-driven approach to model building

Cozad, Alison; Sahinidis, Nikolaos V.; Miller, David C.

doi:10.1016/j.compchemeng.2014.11.010

Cited by 97 publications

(68 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[17] recently developed the ALAMO approach to machine learning, motivated by the need to obtain simple models from experimental and simulation data. Soon thereafter, the approach was extended to combine data-driven and theory-driven model building [18]. The primary purpose of this paper is to review, evaluate, and illustrate the ALAMO methodology and contrast it to existing techniques in the machine learning literature.…”

Section: Introductionmentioning

confidence: 99%

The ALAMO approach to machine learning

Wilson

Sahinidis

2017

Computers & Chemical Engineering

Self Cite

159

View full text Add to dashboard Cite

ALAMO is a computational methodology for leaning algebraic functions from data. Given a data set, the approach begins by building a low-complexity, linear model composed of explicit non-linear transformations of the independent variables. Linear combinations of these non-linear transformations allow a linear model to better approximate complex behavior observed in real processes. The model is refined, as additional data are obtained in an adaptive fashion through error maximization sampling using derivative-free optimization. Models built using ALAMO can enforce constraints on the response variables to incorporate first-principles knowledge. The ability of ALAMO to generate simple and accurate models for a number of reaction problems is demonstrated. The error maximization sampling is compared with Latin hypercube designs to demonstrate its sampling efficiency. ALAMO's constrained regression methodology is used to further refine concentration models, resulting in models that perform better on validation data and satisfy upper and lower bounds placed on model outputs.

show abstract

Section: Introductionmentioning

confidence: 99%

The ALAMO approach to machine learning

Wilson

Sahinidis

2017

Computers & Chemical Engineering

Self Cite

159

View full text Add to dashboard Cite

show abstract

“…In general, there is no rigorous, all-encompassing analysis of surrogate model selection, sampling strategy, and underlying model; however, several groups actively pursuing various pieces of this puzzle, e.g., Boukouvala et al [10], Nuchitprasittichai and Cremaschi [24], Eason and Cremaschi [25], Sikorski et al [11], Cozad et al [26,27], Wang and Ierapetritou [12], Bhosekar and Ierapetritou [7], Garud et al [8,22]. An overall discussion of current progress in these areas of surrogate modeling is presented by Bartz-Beielstein and Zaefferer [28].…”

Section: Design Of Experiments For Surrogate Modelingmentioning

confidence: 99%

“…This method designs mathematically simple surrogates from a set of basis functions using the least amount of data possible. Cozad et al [27] added constrained regression to the method, which places bounds on the surrogate output, making extrapolation more reliable -an important feature for modeling physical or safety limitations in chemical processes. They exemplified ALAMO on several relevant examples, i.e., for a flash drum, a carbon capture adsorber, and a bound constrained batch reactor.…”

Section: Surrogate-based Optimizationmentioning

confidence: 99%

Overview of Surrogate Modeling in Chemical Process Engineering

McBride

Sundmacher

2019

Chemie Ingenieur Technik

212

108

View full text Add to dashboard Cite

The ability to accurately model and simulate chemical processes has been paramount to the growing success and efficiency in process design and operation. These improvements usually come with increasing complexity of the underlying models leading to substantial computational effort in their use. It may also occur that the structure of the model is sometimes unknown making optimization and study difficult. To circumvent these issues, mathematically simpler models, commonly known as surrogate models, have been designed and used to successfully replace these complex, underlying models with much success. This technique has seen increasing use within the chemical process engineering field and this article summarizes some popular surrogates and their recent use in this area.

show abstract

“…A good alternative is to use mixed-integer optimization and global methods to select among a combination of user-provided potential basis functions, those that provide the best fit taking into account possible extra constraints to guarantee physical coherence. Algebraic modelling environments such as ALAMO (Cozad, Sahinidis and Miller 2014;2015) offer very good support to the fitting task using global MINLP solvers like BARON and adaptive-sampling procedures. In the next section, this methodology is applied to a challenging biotechnological process.…”

Section: Estimation Variables Are Considered As Independentmentioning

confidence: 99%

Developing Grey-Box Dynamic Process Models

et al. 2018

View full text Add to dashboard Cite

This paper presents a methodology for developing grey models of process systems, that is, models that, being based on fundamental principles and laws of nature, combine them with sub-models obtained from experimental data. The method follows two steps: the first one takes advantage of what is known, while the second uses the data and mixed-integer optimization algorithms to identify the structure and parameters of the remaining parts of the model. The method is illustrated in a challenging biotechnological process: the Acetone-Butanol-Ethanol (ABE) fermentation process.

show abstract

A combined first-principles and data-driven approach to model building

Cited by 97 publications

References 19 publications

The ALAMO approach to machine learning

The ALAMO approach to machine learning

Overview of Surrogate Modeling in Chemical Process Engineering

Developing Grey-Box Dynamic Process Models

Contact Info

Product

Resources

About