Discrimination of Solid-Liquid Mixtures using a Multisensing System in a Peristaltic Mixing Conveyor that Imitates Intestinal Function

Abstract

Continuous mixing and conveying technology for solid-liquid mixtures is required in the manufacturing process of foods and medicines. To achieve this, we develop a peristaltic mixing conveyor that simulates the function of the human intestines. This device can mix and convey food and medicinal contents by inflating a rubber tube using air pressure. Currently, we are working on a system of content condition estimation using measurement data from the pressure and flow rate sensors installed in the device. However, these measurement methods use air supplied to the device as the measurement target, and the compressibility of air limits the conditions of contents that can be estimated. So, the generalizability of the estimation is low. In this study, a thin pressure-sensitive sensor is installed that can measure the mechanical responses of device contents due to mixing by the device. We also construct a multisensing system that combines conventional pressure/flow rate and pressing force measurements. Sensor data acquired when solid-liquid mixtures are fed into the device are applied to machine learning to distinguish the mixing ratios of the mixtures. Results show that the accuracy of mixing ratio discrimination is improved from $96.7 \%$ to $98.9 \%$ when pressure and flow rate data are combined with pressing force data. The results thus confirm the improved accuracy of content identification when pressure/flow rate and pressing force measurements are combined.