Discrimination of Powder Particle Size and Input Amounts for the Estimation of Mixing Using a Peristaltic Mixing Conveyor Imitating an Intestinal Tube

Continuous mixing and conveying technology for solid-liquid mixtures and highly viscous fluids are required during food and medicine manufacturing. To realize this, we developed a peristaltic mixing and conveying device that mimics the motion of intestines. The device uses air pressure to inflate a rubber tube or other materials to mix and convey the contents. We are working on estimating the state of the contents using pressure and flow sensors and acceleration sensors mounted on the device. In this study, for powder discrimination, a multi-class determination was conducted for powders with different particle sizes and input amounts using the three types of sensor data. We used principal component analysis of time series data for each sensor and decision trees and support vector machine (SVM) as judges. The results indicate that the correct response rate was 97.0
^%
when the three types of sensors were used. Accelerometers have more information than pressure or flow sensors, and using accelerometers increased the percentage of correct responses.