Related Experiment Video
Updated: Jul 19, 2025
07:26
In Situ Microscopy for Real-time Determination of Single-cell Morphology in Bioprocesses
Published on: December 5, 2019
8.0K
In-line particle size measurement during granule fluidization using convolutional neural network-aided process
Orsolya Péterfi1, Lajos Madarász1, Máté Ficzere1
1Department of Organic Chemistry and Technology, Faculty of Chemical Technology and Biotechnology, Budapest University of Technology and Economics, Műegyetem rkp. 3., H-1111 Budapest, Hungary.
Summary
This study introduces a machine learning vision system for real-time particle size analysis of fluidized granules. The method accurately monitors particle size distribution (PSD) in-line, offering a new process analytical technology (PAT) tool.
Area of Science:
- Chemical Engineering
- Materials Science
- Computer Science
Background:
- Accurate monitoring of particle size distribution (PSD) is crucial in pharmaceutical and chemical manufacturing.
- Traditional offline methods for PSD analysis are time-consuming and do not allow for real-time process adjustments.
- Developing in-line process analytical technology (PAT) tools is essential for efficient process control.
Purpose of the Study:
- To develop and evaluate a machine learning-based image analysis method for real-time monitoring of particle size distribution (PSD) in fluidized granules.
- To assess the feasibility of using machine vision as an in-line PAT tool for granule characterization.
- To compare the performance of the developed method against established offline PSD measurement techniques.
Main Methods:
- A direct imaging system comprising a fiber-optic endoscope, light source, and high-speed camera was utilized.
- A custom 3D-printed device simulated particle movement in a fluidized-bed granulator.
- Convolutional neural network (CNN)-based software was employed for image analysis and granule detection.
- Volumetric PSDs were determined for granule mixtures (100-2000 μm) and compared with dynamic image analysis and laser diffraction.
Main Results:
- The machine learning software successfully detected in-focus granules amidst dense particle flow.
- Real-time PSD measurements showed similar trends to offline reference methods (dynamic image analysis and laser diffraction).
- The system demonstrated effective particle size analysis within the 100-2000 μm range.
Conclusions:
- The developed machine vision system is feasible for real-time, in-line particle size analysis of fluidized granules.
- This approach offers a valuable PAT tool for enhancing process understanding and control in granulation processes.
- Machine learning-based image analysis provides a robust alternative to traditional offline PSD measurement techniques.
