Physics-Constrained Machine Learning
Mass, energy, and thermodynamics constrained neural networks for steady-state and dynamic chemical systems.
Postdoctoral Research Associate at University of Wisconsin-Madison
Angan Mukherjee
I build uncertainty-aware, physics-constrained, and topology-driven data analytics and machine learning tools for process systems, digital twins, optimization, and decision making.
Current role
Postdoctoral Research Associate, UW-Madison
Research field
Process Systems Engineering
Doctorate
Ph.D., Chemical Engineering, WVU '24
Recognition
Gold Medalist, Jadavpur University '19
About
Researcher at the intersection of chemical engineering, AI/ML, and optimization.
I am a postdoctoral Research Associate in the Department of Chemical and Biological Engineering at the University of Wisconsin-Madison, advised by Prof. Victor M. Zavala. My research focuses on developing scalable paradigms for uncertainty-aware data-driven modeling, optimization, and decision making, including digital twins. I also develop topology-based data analytics methods for process monitoring, control, and visualization of biomolecules (proteins), polymers, and supercooled liquids.
During my Ph.D. at West Virginia University (2019-2024), advised by Prof. Debangsu Bhattacharyya, I developed sparse AI/ML tools for efficient data-driven modeling of complex process systems; mass, energy, and thermodynamics constrained ML models and training algorithms for transient chemical processes under uncertainty; open-source software for physics-constrained ML in chemical engineering, and hybrid first-principles artificial intelligence models for clean energy systems.
Research
Methods for reliable models, interpretable dynamics, and better decisions.
Topological Data Analysis
Topology-based feature extraction, event detection, and process monitoring for high-dimensional dynamic data.
Uncertainty Quantification
Bayesian and variational approaches for scalable parameter estimation, model calibration, closed-loop experimental design, and decision making.
Hybrid First-Principles AI
Affordable hybrid models for clean energy systems, health monitoring, and dynamic optimization under uncertainty.
Experience
Academic appointments and training.
Postdoctoral Research Associate
University of Wisconsin-Madison, Department of Chemical and Biological Engineering
Advisor: Prof. Victor M. Zavala
Ph.D. in Chemical Engineering
West Virginia University
Minor: Process Controls and Statistics. GPA: 4.00/4.00.
B.E./B.S. in Chemical Engineering
Jadavpur University, Kolkata, India
First rank in a class of 86 students. GPA: 9.23/10.00.
Publications
Peer-reviewed journal and conference publications.
VIBES -- A Two-Stage Scalable Bayesian Uncertainty Quantification Framework: Application to a Biomass Valorization Process
Das, P., Mukherjee, A., and Bhattacharyya, D. 2026. arXiv preprint arxiv:2607.06743.
PreprintTopological Data Analysis for High-Dimensional Dynamic Process Monitoring
Mukherjee, A., Soderstrom, T., Kurtz, M., and Zavala, V. 2026. arXiv preprint arxiv:2606.20443.
PreprintComputing Optimal Programs for Programmable Catalysts using Bayesian Optimization
Flory, J., Mukherjee, A., Dauenhauer, P., and Zavala, V. 2026. ChemRxiv.
PreprintPhysics-Constrained Machine Learning for Chemical Engineering
Mukherjee, A. and Zavala, V. 2026. Current Opinion in Chemical Engineering, 51: 101228.
DOIDevelopment of a Hybrid First Principles - Machine Learning Adaptive Modeling Framework for Health Monitoring of Power Plant Boiler Superheaters
Saini, V., Purdy, D., Mukherjee, A., Adeyemo, S., Bhattacharyya, D., Parker, J., Lolla, T., and Boohaker, C. 2026. Fuel, 406: 136795.
DOIA Digital Twin Simulator of a Pastillation Process with Applications to Automatic Control Based on Computer Vision
Gonzalez, L., Pulsipher, J., Jiang, S., Mukherjee, A., Soderstrom, T., and Zavala, V. 2025. Computers & Chemical Engineering, 201: 109205.
DOIDevelopment of Mass, Energy, and Thermodynamics Constrained Steady-State and Dynamic Neural Networks for Interconnected Chemical Systems
Mukherjee, A. and Bhattacharyya, D. 2025. Chemical Engineering Science, 309: 121506.
DOIMass-Constrained Hybrid Gaussian Radial Basis Neural Networks: Development, Training, and Applications to Modeling Nonlinear Dynamic Noisy Chemical Processes
Mukherjee, A., Gupta, D., and Bhattacharyya, D. 2025. Computers & Chemical Engineering, 197: 109080.
DOIDevelopment of Hybrid First Principles - Artificial Intelligence Models for Transient Modeling of Power Plant Superheaters under Load-Following Operation
Mukherjee, A., Saini, V., Adeyemo, S., Bhattacharyya, D., Purdy, D., Parker, J., and Boohaker, C. 2025. Applied Thermal Engineering, 262: 124795.
DOIAll-Nonlinear Static-Dynamic Neural Networks versus Bayesian Machine Learning for Data-Driven Modelling of Chemical Processes
Mukherjee, A., Adeyemo, S., and Bhattacharyya, D. 2025. The Canadian Journal of Chemical Engineering, 103(3): 1139-1154.
DOIDevelopment of Steady-State and Dynamic Mass and Energy Constrained Neural Networks for Distributed Chemical Systems Using Noisy Transient Data
Mukherjee, A. and Bhattacharyya, D. 2024. Industrial & Engineering Chemistry Research, 63(32): 14211-14239.
DOIOn the Development of Steady-State and Dynamic Mass-Constrained Neural Networks Using Noisy Transient Data
Mukherjee, A. and Bhattacharyya, D. 2024. Computers & Chemical Engineering, 187: 108722.
DOIHybrid Series/Parallel All-Nonlinear Dynamic-Static Neural Networks: Development, Training, and Application to Chemical Processes
Mukherjee, A. and Bhattacharyya, D. 2023. Industrial & Engineering Chemistry Research, 62(7): 3221-3237.
DOIRecent Advances in Real-Time Process Monitoring and Event Detection
Mukherjee, A., Soderstrom, T., Kurtz, M., and Zavala, V. 2026. Systems & Control Transactions (Proceedings of 2027 FOCAPO-CPC Conference), 10-14 January, 2027, Tucson, AZ.
Development of a Scalable Bayesian Uncertainty Quantification Framework for Biomass Valorization
Das, P., Mukherjee, A., and Bhattacharyya, D. 2026. Systems & Control Transactions (Proceedings of 2027 FOCAPO-CPC Conference), 10-14 January, 2027, Tucson, AZ.
Development of Steady-State and Dynamic Mass-Energy Constrained Neural Networks using Noisy Transient Data
Mukherjee, A. and Bhattacharyya, D. 2024. Systems & Control Transactions (Proceedings of 2024 FOCAPD Conference), 3: 330-337.
DOIFor updates, please visit Google Scholar.
Paper-2-Podcast
AI-generated podcasts of my publications.
For AI-generated podcasts of my publications, feel free to visit my Spotify channel PSE-ing Podcasts.
Presentations
Oral and poster presentations.
Oral Presentations
Uncertainty Quantification in Physics-Constrained Machine Learning
Mukherjee, A. and Zavala, V. Texas-Wisconsin-California Control Consortium, Spring 2026 Meeting, 23-24 February, 2026, Austin, TX.
Data-Driven Dynamic Modeling and Uncertainty Quantification Using Variational Inference: Applications to Microkinetic Modeling
Mukherjee, A., Thompson, J., and Zavala, V. 2025 AIChE Annual Meeting, 2-6 November, 2025, Boston, MA.
AbstractEvent Detection in Multivariate Time-Series Data Using Topology
Mukherjee, A., Soderstrom, T., Kurtz, M., and Zavala, V. 2025 AIChE Annual Meeting, 2-6 November, 2025, Boston, MA.
AbstractDevelopment of a Framework for Automatic Discovery of Optimal Hybrid First Principles - Machine Learning Models
Mukherjee, A., Giridhar, N., and Bhattacharyya, D. 2025 AIChE Annual Meeting, 2-6 November, 2025, Boston, MA.
AbstractApproaches to Physics-Constrained Machine Learning
Mukherjee, A. and Zavala, V. Texas-Wisconsin-California Control Consortium, Fall 2025 Meeting, 8-9 September, 2025, Madison, WI.
Development of Steady-State and Dynamic Mass-Energy-Thermodynamics Constrained Neural Network Models for Interconnected Systems Using Noisy Transient Data
Mukherjee, A. and Bhattacharyya, D. 2024 AIChE Annual Meeting, 27-31 October, 2024, San Diego, CA.
AbstractDevelopment of Steady-State and Dynamic Mass-Energy Constrained Neural Network Models Using Noisy Temporal Data for Dynamic Optimization of Distributed Chemical Systems
Mukherjee, A. and Bhattacharyya, D. 2024 AIChE Annual Meeting, 27-31 October, 2024, San Diego, CA.
AbstractHybrid Gaussian Radial Basis Neural Networks: Development, Training, and Applications to Modeling Nonlinear Dynamic Noisy Chemical Processes
Mukherjee, A. and Bhattacharyya, D. 2024 AIChE Annual Meeting, 27-31 October, 2024, San Diego, CA.
AbstractDevelopment of Hybrid First Principles - Artificial Intelligence Models: Applications to an Industrial Steam Superheater System
Mukherjee, A., Saini, V., Adeyemo, S., and Bhattacharyya, D. 2023 AIChE Annual Meeting, 5-10 November, 2023, Orlando, FL.
AbstractHybrid Series and Parallel All-Nonlinear Dynamic-Static Neural Networks: Development, Training, and Application to Chemical Processes
Mukherjee, A. and Bhattacharyya, D. 15th AIChE Midwest Regional Conference, 11-12 April, 2023, Chicago, IL.
ProgramDevelopment of Mass and Energy Constrained Neural Networks
Mukherjee, A. and Bhattacharyya, D. 2022 AIChE Annual Meeting, 13-18 November, 2022, Phoenix, AZ.
AbstractData-Driven Modeling of Complex Nonlinear Systems Using Hybrid Series and Parallel Nonlinear Static-Nonlinear Dynamic Neural Networks
Mukherjee, A. and Bhattacharyya, D. AIChE Advanced Manufacturing and Processing Conference, 1-3 June, 2022, Bethesda, MD.
AbstractModeling Complex Nonlinear Systems Using Concatenated Static-Dynamic Neural Networks
Mukherjee, A. and Bhattacharyya, D. 2021 AIChE Annual Meeting, 7-19 November, 2021, Boston, MA.
AbstractDevelopment of a Hybrid First Principles - Artificial Intelligence Approach for Dynamic Modeling of Complex Systems
Mukherjee, A. and Bhattacharyya, D. 2020 Virtual AIChE Annual Meeting, 16-20 November, 2020.
AbstractPoster Presentations
Scalable Parameter Estimation and Uncertainty Quantification in Physics-Constrained Machine Learning
Mukherjee, A. and Zavala, V. Hougen PSE Symposium, 11-12 May, 2026, Madison, WI.
Data-Driven Modeling and Uncertainty Quantification using Variational Inference
Mukherjee, A. and Zavala, V. Texas-Wisconsin-California Control Consortium, Spring 2026 Meeting, 23-24 February, 2026, Austin, TX.
Data-Driven Dynamic Modeling and Uncertainty Quantification using Variational Inference
Mukherjee, A. and Zavala, V. Optimal Control & Decision Making under Uncertainty for Digital Twins, Institute for Mathematical & Statistical Innovation, 27-31 October, 2025, Chicago, IL.
AbstractTopological Data Analysis for Multivariate Process Monitoring
Mukherjee, A., Soderstrom, T., Kurtz, M., and Zavala, V. Texas-Wisconsin-California Control Consortium, Fall 2025 Meeting, 8-9 September, 2025, Madison, WI.
Detection of Relaxation Events in Supercooled Liquids using Topological Data Analysis
Mukherjee, A., Kung, P., Voyles, P., and Zavala, V. Midwest Thermodynamics and Statistical Mechanics Conference, 1-3 June, 2025, Madison, WI.
ProgramPhysics-Constrained Machine Learning
Mukherjee, A. and Bhattacharyya, D. Texas-Wisconsin-California Control Consortium, Fall 2024 Meeting, 23-24 September, 2024, Madison, WI.
Development of Steady-State and Dynamic Mass-Energy Constrained Neural Networks using Noisy Transient Data
Mukherjee, A. and Bhattacharyya, D. Foundations of Computer-Aided Process Design Conference, 14-18 July, 2024, Breckenridge, CO.
AbstractHybrid Series/Parallel All-Nonlinear Dynamic-Static Stochastic Neural Networks: Development, Training and Application to Chemical Processes
Mukherjee, A. and Bhattacharyya, D. 2023 AIChE Annual Meeting, 5-10 November, 2023, Orlando, FL.
AbstractDevelopment of Algorithms for Mass and Energy Constrained Dynamic Neural Network Models
Mukherjee, A. and Bhattacharyya, D. 2023 AIChE Annual Meeting, 5-10 November, 2023, Orlando, FL.
AbstractNew Data-Driven Modeling Paradigms in Systems Engineering Using Novel Neural Network Structures
Mukherjee, A. 2023 AIChE Annual Meeting, 5-10 November, 2023, Orlando, FL.
AbstractDevelopment of Algorithms for Mass-Constrained Dynamic Neural Networks
Mukherjee, A. and Bhattacharyya, D. Foundations of Process/Product Analytics and Machine Learning Conference, July 30-August 3, 2023, UC Davis, CA.
AbstractData-Driven Modeling of Complex Nonlinear Systems Using Hybrid Series and Parallel Nonlinear Static-Dynamic Stochastic Neural Networks
Mukherjee, A. and Bhattacharyya, D. 2022 AIChE Annual Meeting, 13-18 November, 2022, Phoenix, AZ.
AbstractOpen-Source Software
Codes for data-driven process modeling.
Skills
Computational tools and analytical strengths.
Analytical
Data-driven modeling, AI/ML, dynamic optimization, linear and optimal control, process monitoring, data reconciliation, neural networks, system identification, Bayesian optimization, uncertainty quantification, design of experiments, and digital twins.
Software
MATLAB, Python, Pyomo, TensorFlow, PyTorch, Julia, R, Aspen Custom Modeler, C, C++, and MS Office.
Contact