Bioreactor Batch Regulation via NIR Spectra and PCA
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In bioreactors, existing methods for controlling batch processes are inefficient due to the inability to directly measure product quality during the process, leading to significant waste production from variations in microorganism behavior and electrophysiological states, as they rely on empirical values and provide only starting points for corrective interventions without specifying how to proceed.
Innovation Solution
The method involves determining a process fingerprint through near-infrared spectrometry and Principal Component Analysis (PCA), comparing it to a nominal trajectory, and making automatic adjustments using conventional control devices, with direction changes based on trial adjustments and numerical optimization, such as the Hooke-Jeeves method, to maintain the process within an acceptable corridor.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of substance
If conventional closed-loop or open-loop controllers with predetermined empirical values are used to regulate environmental conditions in a bioreactor, then the process can be controlled with simple equipment, but product quality cannot be measured directly during the batch process leading to significant waste production
Solution Approach 1:
The patent implements a feedback mechanism where the actual process trajectory is continuously determined through NIR spectrometry and PCA, compared to the nominal trajectory, and this deviation information is fed back to automatically adjust process parameters. This closed-loop feedback enables online regulation that responds to actual process conditions rather than relying on predetermined empirical values, thereby reducing waste while maintaining automated control.
Solution Approach 2:
The patent replaces conventional mechanical measurement and control systems with optical-based NIR spectrometry and computational PCA analysis. This substitution enables non-invasive, continuous monitoring of the bioprocess trajectory without physical sampling or disruption, allowing for real-time detection of deviations and automated corrective actions that reduce waste while maintaining simple equipment architecture.
2Ease of operation
If multivariate control charts with MSPC are used to monitor batch processes and provide starting points for corrective interventions, then online process assessment becomes possible, but the details of how corrective intervention should proceed are not specified requiring manual intervention
Solution Approach 1:
The patent implements a self-service control system where the multivariate monitoring system automatically generates and executes corrective interventions without requiring manual operator input. The system determines the actual trajectory, compares it to the nominal trajectory, calculates deviations, and automatically adjusts process parameters based on predefined control logic. This self-service capability maintains ease of operation through automated decision-making while managing complexity through algorithmic rather than manual control.
Solution Approach 2:
The patent employs preliminary action by pre-establishing the nominal trajectory from reference batches and pre-programming the control logic that determines how to respond to deviations. When the actual trajectory deviates from the nominal trajectory, the pre-programmed control rules automatically dictate the corrective action to be taken. This preliminary preparation of control strategies enables automated responses without requiring complex real-time human decision-making, balancing ease of operation with manageable system complexity.
3Loss of information
If PCA is used to compress high-dimensional NIR spectra into a low-dimensional principle component region, then data volume is reduced while retaining information content, but the system still only provides online information about process status without automatic regulation capability
Solution Approach 1:
The patent closes the automation loop by implementing feedback control based on the PCA-compressed trajectory information. The actual trajectory determined through PCA is continuously compared to the nominal trajectory, and the deviation information feeds back to automatically adjust process parameters. This feedback mechanism transforms the PCA system from a passive monitoring tool that retains information efficiently into an active automatic regulation system that uses the compressed information to drive corrective actions, thereby achieving both information retention and automation.
Solution Approach 2:
The patent uses the PCA-transformed principle component scores as an intermediary representation that bridges the gap between complex high-dimensional spectral data and simple automatic control decisions. The PCA compression creates a low-dimensional trajectory space that serves as an intermediary layer, allowing the control system to make automated regulation decisions based on simplified yet information-rich representations of the process state, thereby enabling automatic regulation without losing critical process information.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables online, automatic regulation of batch processes, reducing waste by optimizing environmental conditions in real-time, learning from successful interventions and adapting to minimize deviations, thereby improving product quality and reducing unnecessary batch terminations.
Implementation Method 1
a spectrometry observation of the current batch process in the near infrared range is undertaken and by application of principle component analysis to the recorded spectra
Data Source
AI summary
The invention relates to the online regulation of a batch process in a bioreactor. According to the invention, spectra of the actual charge in the bioreactor are recorded at successive points during the running batch process. A measuring vector is produced, for each spectrum, in the low-dimensional main constituent region by the main constituent analysis of the high-dimensional spectra. The deviation is calculated between the measuring vector and a corresponding vector of a nominal trajectory consisting of measuring vectors of a reference charge, which are determined in an earlier batch process, and at least one adjustment operation for the batch process is determined and carried out according to said deviation. The direction of the adjustment operation is maintained when the deviation determined from one point to the next decreases, and modified when the deviation determined from one point to the next remains the same or increases.