Generalization Capability of Mixture Estimation Model for Peristaltic Continuous Mixing Conveyor

Abstract

We propose herein a method for estimating the mixing state of the contents of a peristaltic continuous mixing conveyor simulating the intestine, developed for mixing and conveying powders and liquids. This study serves to improve a previously proposed method for estimating the mixing state using a logistic regression model with the pressure and flow rate sensors installed in the device as inputs. Moreover, the estimation accuracy of the proposed method is better than that of the previous method. The generalizability of the proposed method is evaluated for four conditions in which the feeding order of the contents, powder, and liquid are changed. The feeding order is as follows: powder first, liquid first, and powder and liquid alternately. As a result, a highly accurate estimation of mixing is achieved under the condition wherein the powder component is in the unit adjacent to the lid, but not under the condition wherein the liquid component is fed first. It is speculated that this is because the movement of the powder component inside the device is more easily reflected by the pressure and flow rate sensors installed in the device than in the liquid component.

Keywords

• Soft robotics
• robot sensing systems
• machine learning
• predictive models
• data acquisition
• product safety