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

Abstract

Among the greatest challenges in soft sensor development for bioprocesses are variable process lengths, multiple process phases, and erroneous model inputs due to sensor faults. This review article describes these three challenges and critically discusses the corresponding solution approaches from a data scientist’s perspective. This main part of the article is preceded by an overview of the status quo in the development and application of soft sensors. The scope of this article is mainly the upstream part of bioprocesses, although the solution approaches are in most cases also applicable to the downstream part. Variable process lengths are accounted for by data synchronization techniques such as indicator variables, curve registration, and dynamic time warping. Multiple process phases are partitioned by trajectory or correlation-based phase detection, enabling phase-adaptive modeling. Sensor faults are detected by symptom signals, pattern recognition, or by changing contributions of the corresponding sensor to a process model. According to the current state of the literature, tolerance to sensor faults remains the greatest challenge in soft sensor development, especially in the presence of variable process lengths and multiple process phases.

Online 2D Fluorescence Monitoring in Microtiter Plates Allows Prediction of Cultivation Parameters and Considerable Reduction in Sampling Efforts for Parallel Cultivations of Hansenula polymorpha

“…With increasing LVs, the independent (spectral) and dependent (offline) data of the calibration dataset are fitted more closely to one another as the amount of included covariance increases. However, PLS models with a high number of LVs are prone to overfitting, leading to the inclusion of non-representative signal variations, such as spectral noise [ 66 , 67 , 68 ]. The optimal number of LVs is usually identified using external validation datasets or internal cross-validation methods.…”

Generalizability of Soft Sensors for Bioprocesses through Similarity Analysis and Phase-Dependent Recalibration

“…Individual data sets with significantly differing correlations can be removed. Another optimization possibility is the addition of a synchronization method [ 23 ] to prepare data sets with varying lengths for MSPC-based selection because previous data sets that indicate similar temporal profiles of the online variables to the current process are chosen for automatic recalibration. However, this neglects the fact that data sets may be adequate for recalibration despite their temporal variances.…”

Industrial data science – a review of machine learning applications for chemical and process industries

“…21). 80,82,282 DTW can also be used to classify anomalous batches and to identify correlating parameters (Fig. 22).…”