Advanced Ultrasound techniques and Machine learning algorithms for particle size distribution analyses

Abstract:

This study focuses on the development and application of ultrasound attenuation spectroscopy combined with machine learning algorithms, specifically Random Forest, for measuring particle size distribution in solid-liquid flows. The research utilized the ultrasound attenuation coefficient technique to analyze particle size distribution in a solid-liquid suspension containing ballotini glass beads with a density of 2500 kg/m^3 in a stirred vessel. The experimental setup, ultrasound signal measurement, and data post-processing were thoroughly discussed. By employing machine learning algorithms like Random Forest, the study successfully predicted particle size distribution with an impressive accuracy of 98% based on input from ultrasound measurements. Validation of the ultrasound results was conducted using microscope imaging, demonstrating excellent agreement between the two techniques.