(606b) Model Predictive Control and Estimation – Towards Decision Making in the Cloud

Technological advancements in communication and computation have led to the widespread appearance and use of cloud computing and data management, with significant impact on many applications, spanning from logistics, marketing, energy networks, to traffic control and monitoring.

Even so that the use of cloud technologies for process control, monitoring and operation is still in its infancy, they bear large potential to improve process performance, individualize control loops, and shade light on process phenomena which are not well understood. As will be outlined in the talk, the combination of model predictive control, model based estimation techniques, as well as learning and adaptation strategies are ideally suited to fully exploit the potential of cloud computing. On one side the combination of estimation, learning, and control allows to identify and adapt towards unknown and changing process parameters and disturbances, while maintaining satisfaction of constraints. It enables inclusion of preview information, such as disturbances and set point changes, and allows to cope with the often unavoidable, inherently asynchronous operation due to network communication. Additionally, cloud supported estimation and control strategies enable to exploit formerly unseen computational power, without the need to significantly modify existing local control hardware. Cloud based approaches allow to service and monitor many processes at the same time, while facilitating central maintenance of the underlying algorithms. Model and data mining based analysis of the collected data from the repeated operation of one process, or multiple similar processes, furthermore allows to shade light on phenomena not well understood, or to identify reasons for process failure or degradation.

We outline the potential of cloud based model predictive control and estimation considering two example processes – the operation of batch bioreactors, as well as age aware charging and monitoring strategies of Li-Ion batteries.