Abstract
Integrating physical knowledge and machine learning is a cost-efficient solution to modelling complex biochemical processes when the underlying mechanisms are not fully understood. However, hybrid model structure identification is still time-consuming for new processes, ruiring iteration over different hypotheses to explain the observed process dynamics while minimizing over-parameterization. Unfortunately, conventional approaches to automatic model structure identification do not always converge for highly nonlinear models and cannot estimate time-varying model parameters. To address this and accelerate the design of new biochemical processes, a Reinforcement Learning (RL) based framework recently reformulated synchronous hybrid model structure-parameter identification into a process optimal control problem. To further investigate other possible solutions, in this study, a novel Physics Informed Neural Network (PINN) based framework was proposed for the first time to infer time-varying kinetic parameters. This framework first combines possible kinetic structures from phenomenological knowledge, then simultaneously identifies the most likely hybrid model structure and time-varying parameter trajectories. To demonstrate the performance of the PINN based framework, several in-silico case studies were conducted using a known ground truth bioprocess. We thoroughly examined the advantages and limitations of the framework, elucidating its potential for high-fidelity hybrid model construction in biochemical engineering research.
Ilya Orson Sandoval
Knowledge-driven Autonomous Systems - Neural ODEs and Reinforcement Learning
I am a PhD candidate at Imperial College London, where my research focuses on the intersection of reinforcement learning, differentiable programming and nonlinear optimal control. Curiosity driven, usually by applied mathematics and computer simulations with applications over multiple fields! Previously, I worked in data science and software engineering within the energy and food industries in Mexico. I have a background in theoretical and computational physics.
Dr. Ehecatl Antonio del Rio Chanona
Principal Investigator of OptiML
Antonio del Rio Chanona is the head of the Optimisation and Machine Learning for Process Systems Engineering group based in thee Department of Chemical Engineering, as well as the Centre for Process Systems Engineering at Imperial College London. His work is at the forefront of integrating advanced computer algorithms from optimization, machine learning, and reinforcement learning into engineering systems, with a particular focus on bioprocess control, optimization, and scale-up. Dr. del Rio Chanona earned his PhD from the Department of Chemical Engineering and Biotechnology at the University of Cambridge, where his outstanding research earned him the prestigious Danckwerts-Pergamon award for the best PhD dissertation of 2017. He completed his undergraduate studies at the National Autonomous University of Mexico (UNAM), which laid the foundation for his expertise in engineering.