Integer-Free Optimal Scheduling of Smart Appliances

Abstract

With the growing number of smart appliances in the electrical grid, flexible optimization techniques capable of scheduling large scale systems of such devices are needed in order to control them optimally. However, scheduling approaches in the literature predominantly use mixed-integer formulations with fixed time slots. This usually yields a poor scalability due to the computational complexity of the resulting mixed-integer optimization problems. Furthermore this approach can be rather inflexible since it restricts the task executions to specific time intervals. This may become problematic if the (actual) optimal task execution would only span a fraction of one time slot, as it might happen for applications with very short individual tasks.This paper presents an optimization based scheduling approach that creates a continuous schedule, i.e. the task execution is not bound to specific time slots. Furthermore, no integer variables are used in the process. This makes it not only applicable to processes with arbitrarily short execution time lengths, but also has the potential of a reduced computational complexity compared to mixed-integer based approaches. Preliminary case studies demonstrate the advantages of the proposed approach.