Uneven batch data alignment with application to the control of batch end-product quality

Area of Science:

• Chemical Engineering
• Process Control
• Data Science

Background:

• Batch processes exhibit variable trajectories due to batch-to-batch differences, complicating end-product quality assessment.
• Aligning time-series data and quality measurements is crucial for effective statistical process monitoring and control.
• Existing methods like indicator variables or dynamic time warping have limitations in synchronizing uneven batch data.

Purpose of the Study:

• To develop a novel approach for aligning uneven batch process data.
• To enable accurate statistical process monitoring and control of batch end-product quality.
• To extend shorter trajectories and predict future batch quality using identified models.

Main Methods:

• Identification of short-window Principal Component Analysis (PCA) and Partial Least Squares (PLS) models.
• Application of identified PCA-PLS models to extend shorter batch trajectories.
• Utilizing moving window estimation for aligning batch data to a specified length.
• Demonstration with a simulated fed-batch fermentation process for penicillin production.

Main Results:

• Successfully aligned uneven batch trajectories from a simulated penicillin fermentation process.
• Demonstrated the capability to predict batch end-product quality using the proposed alignment method.
• Showcased the effectiveness of short-window PCA-PLS models in handling batch-to-batch variations.

Conclusions:

• The proposed novel approach effectively aligns uneven batch data, enhancing process monitoring and control.
• This method provides a robust solution for synchronizing variable-length trajectories and predicting quality outcomes.
• The technique is applicable to various batch processes, including fermentation, for improved operational efficiency and quality assurance.