Introducing Hybrid Modeling with Time-Series-Transformers: A Comparative Study of Series and Parallel Approach in Batch Crystallization

Sitapure, Niranjan; Sang-Il Kwon, Joseph

doi:10.1021/acs.iecr.3c02624

Cited by 14 publications

(2 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hybrid modeling advances this concept by combining physical models with data-driven models to capture system dynamics. This method can infer missing (unknown) parameters that theoretical models might not identify, resulting in more robust and reliable models. ,, By leveraging both the strengths of physical models and the flexibility of data-driven approaches, hybrid modeling offers a comprehensive understanding of complex nonlinear systems.…”

Section: Methodsmentioning

confidence: 99%

Machine Learning-Based Adaptive Regression to Identify Nonlinear Dynamics of Biochemical Systems: A Case Study on Bio 2,3-Butanediol Distillation Process

Choi,

Bhadriraju,

Kwon

et al. 2024

ACS Sustainable Chem. Eng.

Self Cite

View full text Add to dashboard Cite

Developing an accurate process model is essential to efficiently operate a process and maximize its economics. While offline data-driven models utilizing historical data generally exhibit satisfactory performance, their effectiveness diminishes in accurately predicting real processes characterized by constant changes and uncertainties over time. Hence, there is a need for an adaptive model that is capable of effectively handling dynamic behavior. In this study, we propose an adaptive data-driven regression model that leverages subset selection techniques and decision thresholds. In addition, a comprehensive analysis was performed to determine the best adaptive regression model, considering case studies with different model parameters and training window sizes, taking into account statistical indicators of model accuracy as well as nonstatistical indicators such as the number of updates, update period, and computation time. The developed adaptive regression model has been successfully demonstrated on a bio 2,3-Butanediol distillation column at GS Caltex, Republic of Korea, suggesting its potential applicability to similar process systems and providing opportunities for future research in process optimization and control.

show abstract

Section: Methodsmentioning

confidence: 99%

Machine Learning-Based Adaptive Regression to Identify Nonlinear Dynamics of Biochemical Systems: A Case Study on Bio 2,3-Butanediol Distillation Process

Choi,

Bhadriraju,

Kwon

et al. 2024

ACS Sustainable Chem. Eng.

Self Cite

View full text Add to dashboard Cite

show abstract

“…To tackle this challenge, a category of models known as hybrid models has emerged, wherein the kinetic parameters of first-principles models are forecasted by a data-driven component. , In these hybrid models, hidden chemical mechanisms can then be characterized by parametric uncertainties. These hybrid models have found applications across a wide array of domains, including bacterial cultivations, chemical reactors, − flowsheet simulators for chemical processes, crystallization, , distillation columns, biopharmaceutical industries, polymerization processes, fermentations, − hydraulic fracturing, intracellular signaling pathways, and many more. In prior research, hybrid modeling has proven effective in addressing issues related to parametric uncertainties within simpler model equations. , However, applying this approach to complex systems described by PDEs presents its own distinct challenges.…”

Section: Introductionmentioning

confidence: 99%

Unveiling Latent Chemical Mechanisms: Hybrid Modeling for Estimating Spatiotemporally Varying Parameters in Moving Boundary Problems

Pahari,

Shah,

Sang-Il Kwon

2024

Ind. Eng. Chem. Res.

Self Cite

View full text Add to dashboard Cite

Hybrid modeling has gained substantial recognition due to its capacity to seamlessly integrate machine learning methodologies while preserving the fundamental physical principles inherent in a model. Although these hybrid models have predominantly been applied to temporal processes governed by ordinary differential equations, the intricacies of various real-world systems, characterized by a diverse array of physical processes, extend far beyond this domain. This study extensively investigates the application of hybrid modeling techniques within the context of a complex biological system regulated by the reaction−diffusion equation, a specific type of partial differential equation. The primary objective is to tackle the challenges associated with latent chemical mechanisms that are concealed from direct observation. The proposed approach introduces a hybrid modeling framework that synergizes neural networks with mathematical methods to estimate parameters exhibiting spatiotemporal variations within a category of problems involving dynamic boundaries. Model training is executed through a back-propagation algorithm, adept at efficiently updating these parameters while ensuring numerical stability. Subsequently, the hybrid model is employed in the context of reaction−diffusion models, yielding results that validate its proficiency in precisely estimating variable diffusivity across both space and time as well as temporal fluctuations in cell proliferation rates and cell carrying density. The mean squared error of the final output predictions by the hybrid model is 9.4 × 10 −6 as compared to that of a simple first-principles model which is 3.12 × 10 −4 .

show abstract

Optimization-based multi-source transfer learning for modeling of nonlinear processes

Xiao,

Vellayappan,

P S

et al. 2024

Chemical Engineering Science

View full text Add to dashboard Cite

Introducing Hybrid Modeling with Time-Series-Transformers: A Comparative Study of Series and Parallel Approach in Batch Crystallization

Cited by 14 publications

References 76 publications

Machine Learning-Based Adaptive Regression to Identify Nonlinear Dynamics of Biochemical Systems: A Case Study on Bio 2,3-Butanediol Distillation Process

Machine Learning-Based Adaptive Regression to Identify Nonlinear Dynamics of Biochemical Systems: A Case Study on Bio 2,3-Butanediol Distillation Process

Unveiling Latent Chemical Mechanisms: Hybrid Modeling for Estimating Spatiotemporally Varying Parameters in Moving Boundary Problems

Optimization-based multi-source transfer learning for modeling of nonlinear processes

Contact Info

Product

Resources

About