Set-Based Iterative Learning Control for Temperature and Humidity in Buildings

This paper presents the design of a passivity-based iterative learning control (ILC) algorithm for coupled temperature and humidity in buildings. Since buildings are subjected to repeating diurnal patterns of disturbances, ILC algorithms can significantly improve performance. Moreover, since it is a feedforward control scheme, it can be used in conjunction with either model-free or model-based approaches such as the popular model predictive control techniques. However, model-based control is challenging for buildings because of the difficulty in identifying building thermohygrometric models. Furthermore, the control law should be designed in such a way as to address both temperature and humidity set points. We propose a model-free ILC design approach facilitated by the inherent passivity of building thermohygrometric dynamics. We first demonstrate that the building dynamics are strictly output-incremental passive. This property is then exploited to design ILC laws that guarantee convergence in the iteration domain, while being robust to model uncertainty. Since we wish to control both temperature and humidity using only one input - mass flow rate of supply air, convergence to a point is not guaranteed; instead convergence to an ellipse on the temperature-humidity plane is shown. The controller performance is demonstrated through simulation examples.