Abstract

Dynamic production processes in mineral beneficiation have complex flowsheets and non-linear, time-varying behavior. They have also limited measurement capabilities. Hence, virtual models are seen as important tools for assisting in design, planning and operation. The development of operational dimension requires not only a suitable software architecture and virtual model, but also continuous validation or adaptation of the virtual model. In this paper, a framework establishing a digital twin for a flotation process is presented. The model adaptation is treated as a trajectory matching problem and realized with a Differential Evolution algorithm. The results are demonstrating the applicability of the presented approach in simulation environment together with a discussion on additional challenges foreseen in implementations to the real processes.