Different glycan assay formats will be tested for integration into microfluidic platform.
Data-driven models will be derived by applying MLPCA to the datasets to determine the data-driven required number of pseudo-bioreactions. In a second step, a decision on the kinetic structure (grey-box vs black-box) will be taken with respect to the data content. Eventually, an adapted identification procedure will be applied and validated with the datasets.
A general observability mapping will be determined based on the previously obtained models from D8.1 and D8.2. Necessary on-line measurements will be selected and software sensors will be derived from the models and these measurements. First off-line validations of the resulting software sensors will be achieved and used as basis for further on-line validations using closed-loop controllers.
The algorithms of outlier detection, noise reduction and smoothing are implemented and validated. Furthermore preprocessing procedures like deviation-, distance- and density-based approaches are implemented in the evaluation program.
Searching for OpenAIRE data...
Author(s): Sebastian Giehring
Published in: Animal Cell Biotechnology - Methods and Protocols, Issue 2095, 2020, Page(s) 271-284