To content
Department of Biochemical and Chemical Engineering

September 2021

-
in
  • Picture of the Month
The image shows an efficient AI-based method for determining the operating state in stirred extraction column. © AD​/​BCI

September 2021 – Efficient AI-based Online Supervision Method for Determining the Operating State in Stirred Extraction Columns

Recent investigations show the superior monitoring performance of artificial intelligence based process supervision.  Here, an efficient online detection method using AI-based image classification to differentiate between regular operation and flooding is presented. An image showing one of these two states is evaluated in 0.043 seconds/image with a​ validation accuracy of 99.81%. Flooding is indicated by the non-uniform appearance of the liquid-liquid interphase (at the green arrow) and could reliably be found within this image time series.

Moreover, the size of the droplets can also be evaluated within three classes, small, medium and large-sized drops. The obtained results are utilized to develop a smart sensor that can control the column based on the image analysis.