Identifying important state variables for a blind controller

In literature many different blind controller have been proposed and it has been shown in most cases, that they can provide energy savings and better comfort. In the early approaches they where based only on one state variable, but over time more variables have been taken into account and the structure got more sophisticated. With an increasing number of state variables the performance may rise but also the cost and complexity. In this paper we present a systematic way to identify the important state variables for a blind controller by comparing the performance of the same controller with different sets of state variables. Our work is based on an adaptable hierarchical fuzzy controller which gets optimized by a multi-objective genetic algorithm in terms of energy consumption and thermal comfort, and is then tested in a dynamic simulation environment. For making this more realistic, we include stochastic processes which try to mimic the state of occupancy and the behavior of occupants.