Biomass soft sensor for aPichia pastorisfed‐batch process based on phase detection and hybrid modeling

Abstract

A common control strategy for the production of recombinant proteins in Pichia pastoris using the alcohol oxidase 1 (AOX1) promotor is to separate the bioprocess into two main phases: biomass generation on glycerol and protein production via methanol induction. This study reports the establishment of a soft sensor for the prediction of biomass concentration that adapts automatically to these distinct phases. A hybrid approach combining mechanistic (carbon balance) and data‐driven modeling (multiple linear regression) is used for this purpose. The model parameters are dynamically adapted according to the current process phase using a multilevel phase detection algorithm. This algorithm is based on the online data of CO2 in the off‐gas (absolute value and first derivative) and cumulative base feed. The evaluation of the model resulted in a mean relative prediction error of 5.52% and R² of .96 for the entire process. The resulting model was implemented as a soft sensor for the online monitoring of the P. pastoris bioprocess. The soft sensor can be used for quality control and as input to process control systems, for example, for methanol control.

Ensemble‐based adaptive soft sensor for fault‐tolerant biomass monitoring

“…Neverthe- less, the transferability of the submodels to a following fedbatch phase may be limited, especially in the case of the MIR model. As shown in other studies, the biomass in the fed-batch phase could be well modeled using base consumption and off-gas measurements [6] suggesting transferability of Base and CER submodels. The MIR submodel, in contrast, predicts the biomass concentration indirectly via the decrease of nutrients and sugars in the medium.…”

“…In addition, the function of the base as an additional nitrogen source implies a good correlation with biomass concentration. This linear correlation has already been demonstrated by Brunner et al [6]. The calibration of the model was performed by least-squares regression.…”

“…However, several secondary variables are linked to biomass concentration and can be used for prediction in soft sensor models. Existing literature suggests the online measurement of turbidity [ 1 ], infrared [ 2 , 3 ], or fluorescence spectra [ 4 ], as well as O 2 and CO 2 concentration in the exhaust gas [ 5 , 6 ]. Furthermore, actuator information such as the addition of pH correction agents [ 6 ] can be suitable, as well.…”

Challenges in the Development of Soft Sensors for Bioprocesses: A Critical Review

“…One reason is that they are often the only means of determining critical process parameters (CPP) or critical quality attributes (CQA) online at all ( Capito et al, 2015 ; Melcher et al, 2015 ; Sauer et al, 2019 ; Spann et al, 2019 ; Walch et al, 2019 ; Pais et al, 2020 ; Wasalathanthri et al, 2020a ). Making these quantities measurable by means of soft sensors, in turn, allows CPPs or CQAs to be closed-loop controlled ( Birle et al, 2015 ; Matthews et al, 2016 ; Voss et al, 2017 ; Brunner et al, 2020 ; Gomis-Fons et al, 2020 ). This type of control, also called inferential control, plays an important role in the automation of bioprocesses, since by far not all process quantities to be closed-loop controlled can be measured directly ( Rathore et al, 2021 ).…”

“…The pH and pH correction agent (flow and cumulative amount) signals were used to distinguish the phases. Brunner et al (2020) used the off-gas CO 2 signal to detect the consumption of the carbon source and thus the end of the batch phase in a P. pastoris fed-batch bioprocess. To make the detection of this landmark (CO 2 peak) more robust, a threshold for the cumulative amount of pH correction agent was additionally implemented in the phase detection algorithm.…”