Active sensing in froth flotation

Abstract

The idea of active sensing is to embed sensor systems with intelligence to require less human interaction. Accurate but limited main measurement systems are complemented with broadband auxiliary measurements that gather data and alert the main measurement to focus on certain area. This is similar with the way that our eyesight works in context of gathering data from our surroundings. The purpose of this study is to introduce and test a control architecture that could improve the operation of froth flotation process. An active sensing architectures on linear quadratic Gaussian control is developed and tested in a simulation environment based on plant data for froth flotation with X-ray fluorescence and visible and near-infrared measurements. The architecture is tested in cases where external disturbances or auxiliary measurement model drifting go unnoticed by the main measurement. In both scenarios, the anomalies are successfully corrected by the active sensing architecture.