Bioprocess Model Parameter Adjustment for Variance Threshold Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Bioprocesses, such as fermentation, face challenges in accurately predicting and monitoring process parameters due to biological variability and process control variations, leading to unnecessary interventions and discontinuations, as existing models struggle to distinguish between harmless deviations and critical errors.
Innovation Solution
A method that adjusts defined model parameters, like the lag term, to align the process model with actual bioprocess conditions, allowing for real-time evaluation and classification of variances, thereby improving prediction quality and reducing unnecessary interventions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a deterministic process model is used to predict bioprocess parameters, then the prediction accuracy improves, but the model cannot distinguish between biological variability and process errors, leading to unnecessary interventions
Solution Approach 1:
The patent segments the sources of variance into distinct categories: biological variability (inherent to the living system) and process errors (controllability issues). By separating these variance sources, the system can apply different evaluation criteria and response strategies to each type, allowing accurate prediction while maintaining the ability to distinguish between normal fluctuations and actual problems requiring intervention.
Solution Approach 2:
The patent changes the parameters used for evaluation by introducing multiple process parameters monitored over time and comparing them against threshold values. This parameter-based approach allows the system to detect patterns that indicate process errors versus normal biological variability, enhancing the model's ability to distinguish between the two while maintaining prediction accuracy.
2Reliability
If the process model is adjusted to account for biological variability, then false alarms are reduced, but the complexity of the monitoring system increases
Solution Approach 1:
The patent implements a feedback mechanism where measured process parameters are continuously compared with predicted values from the deterministic model, and deviations are evaluated against threshold values. This feedback loop allows the system to learn from actual process behavior and adjust its evaluation criteria, reducing false alarms while maintaining manageable system complexity through automated decision-making rules.
Solution Approach 2:
The patent introduces additional parameters for variance evaluation and threshold comparison, which allows the system to account for biological variability without requiring complete reconfiguration of the underlying deterministic model. This selective parameter addition maintains system complexity at acceptable levels while improving reliability.
3Manufacturing precision
If continuous monitoring of multiple process parameters is implemented, then process control quality improves, but the cost and complexity of the monitoring system increases
Solution Approach 1:
The patent applies partial monitoring by selecting specific process parameters that are most indicative of process health and variability, rather than monitoring all possible parameters continuously. This selective approach maintains high process control quality for critical parameters while avoiding the excessive complexity and cost of comprehensive continuous monitoring of all process variables.
Solution Approach 2:
The patent applies different monitoring and evaluation intensities to different process parameters based on their importance and variability characteristics. Critical parameters receive more intensive monitoring and stricter threshold evaluation, while less critical parameters are monitored with lower intensity. This local differentiation of monitoring quality optimizes the balance between process control quality and system complexity.
Data Source
AI summary
A method of monitoring bioprocesses, wherein a course of a bioprocess is predicted using a process model and values for process parameters are estimated during the bioprocess, where at least one process parameter is selected (step a), for which current measured values are determined during the bioprocess, the respective current measured value of the selected process parameter is compared with the corresponding estimated value for this process parameter estimated by the process model (step b), a variance is then compared with a predetermined threshold value (step c), in a step d), at least one model parameter is then changed when the threshold value is exceeded by the variance, where steps b) to d) are executed until the variance fails to meet the threshold value, and in the case that the threshold value is not met after a predetermined number of repetitions, the method is discontinued and a warning is output.