Dynamic modeling and optimization of sustainable algal production with uncertainty using multivariate Gaussian processes

Abstract

Dynamic modeling is an important tool to gain better understanding of complex bioprocesses and to determine optimal operating conditions for process control. Currently, two modeling methodologies have been applied to biosystems: kinetic modeling, which necessitates deep mechanistic knowledge, and artificial neural networks (ANN), which in most cases cannot incorporate process uncertainty. The goal of this study is to introduce an alternative modeling strategy, namely Gaussian processes (GP), which incorporates uncertainty but does not require complicated kinetic information. To test the performance of this strategy, GPs were applied to model microalgae growth and lutein production based on existing experimental datasets and compared against the results of previous ANNs. Furthermore, a dynamic optimization under uncertainty is performed, avoiding over-optimistic optimization outside of the model's validity. The results show that GPs possess comparable prediction capabilities to ANNs for long-term dynamic bioprocess modeling, while accounting for model uncertainty. This strongly suggests their potential applications in bioprocess systems engineering.

Hybrid physics‐based and data‐driven modeling for bioprocess online simulation and optimization

“…For instance, autoassociative neural network (AANN) can be used to replace the simple kinetic model for noise filtering (Baughman & Liu, ); current advances in machine learning based dynamic model structure discovery can be implemented at the top‐level to identify the best physical model structure for process prediction and visualization; reinforcement learning can be used as an alternative approach for automatic process optimal control (Petsagkourakis, Sandoval, Bradford, Zhang, & del Rio‐Chanona, ). Other techniques such as Gaussian processes (Bradford et al, ; Tulsyan et al, ) can be also adopted to estimate the uncertainty of model predictions for product quality control and process monitoring. The best combination of these modeling, visualization and optimization strategies should be extensively studied to consolidate efficiency of the current hybrid modeling framework.…”

“…reinforcement learning can be used as an alternative approach for automatic process optimal control (Petsagkourakis, Sandoval, Bradford, Zhang, & del Rio-Chanona, 2019). Other techniques such as Gaussian processes (Bradford et al, 2018;Tulsyan et al, 2018) can be also adopted to estimate the uncertainty of model predictions for product quality control and process monitoring. The best combination of these modeling, visualization and optimization strategies should be extensively studied to consolidate efficiency of the current hybrid modeling framework.…”

“…Then, another k ‐1 set is selected to repeat this procedure until the best model is found. Detailed explanations of these methods and their implementations have been illustrated in recent studies (Bradford et al, ; Del Rio‐Chanona et al, ). Data augmentation and ANN construction were executed in Mathematica 11.…”

“…Moreover, through the use of stochastic optimization, data-driven models have been adopted to optimize long-term fed-batch processes, causing considerable increases in production of different metabolites and yielding highest intracellular contents of bioproducts reported to date (del Rio-Chanona, Manirafasha, Zhang, Yue, & Jing, 2016;Dineshkumar et al, 2015). Since 2018, Gaussian process regression, another technique popularized by the machine learning community, was successfully applied for bioprocess modeling and optimization and online monitoring (Bradford, Schweidtmann, Zhang, Jing, & del Rio-Chanona, 2018;Tulsyan, Garvin, & Ündey, 2018).…”

Review of advanced physical and data‐driven models for dynamic bioprocess simulation: Case study of algae–bacteria consortium wastewater treatment

“…Bradford et al (2018). Unlike ANNs, GPs provide an uncertainty measure representing the prediction uncertainty of the unknown function given the availability of only limited amounts of data.…”