predictive control

In school and office buildings, the ventilation system has a large contribution to the total energy use. A control strategy that adjusts the operation to the actual demand can significantly reduce the energy use. This is important in rooms with a highly fluctuating occupancy profile, such as classrooms and open offices. However, a standard rule-based control (RBC) strategy is reactive, making the installation 'lag behind' in relation to the demand. As a result, a good indoor climate is not always guaranteed and the actual energy saving potential is lower than predicted.

The transformation of the building energy sector to a highly efficient, clean, decentralised and intelligent system requires innovative technologies like microscale trigeneration and thermally activated building structures (TABS) to pave the way ahead. The combination of such technologies however presents a scientific and engineering challenge. Scientific challenge in terms of developing optimal thermo-electric load management strategies based on overall energy system analysis and an engineering challenge in terms of implementing these strategies through process planning and control.

This study tested the feasibility of employing artificial neural network (ANN)-based predictive and adaptive control logics to improve thermal comfort and energy efficiency through a decrease in over- and under-shooting of control variables. Three control logics were developed: (1) conventional temperature/humidity control logic, (2) ANN-based temperature/humidity control logic, and (3) ANN-based Predicted Mean Vote (PMV) control logic.

For the heating of buildings occupied on a discontinuous basis, intermittent heating control devices are used. This article presents one which incorporates advanced automatic control techniques (predictive temperature control and adaptation of the internal model). The results obtained are compared with those achieved using standard control devices. They are validated on the installation used to determine the initial settings and on slightly different installations in order to compare their robustness with respect to the various characteristics of the heating loop and of the building.