Model Predictive Control for room temperature control in buildings is an effective approach to energy management in buildings. However, the development and maintenance of physical models may be a bottleneck for widespread real life application. Data Predictive Control is an attempt to address this problem by learning the behaviour of the building from historical data and thus reducing the modelling effort. Here, we present an application of a Data Predictive Control approach, based on Random Forests with affine functions and convex optimization, to control the room temperature in a real life apartment. When compared to a conventional hysteresis controller, the applied approach saves 24.9 % of cooling energy while reducing the integral of comfort constraint violations by 72.0 % in a six-day experiment.

To reduce the heating and cooling energy demand of buildings and districts novel control strategies are constantly being developed that require information on the future demand of the controlled entity. Demand forecasting is commonly done with deterministic white box models or fitted grey-box models, however, recently more and more data and machine learning based approaches are being developed. All approaches have weaknesses: white-box models require major modelling effort, grey-box approaches are limited by their model or parameter complexity and machine learning is dependent on hyperparameters, some of which are randomly chosen, and therefore considered unreliable. Here we develop a forecasting approach based on Artificial Neural Networks (ANN) and introduce error correction methods based on online learning and the learned autocorrelation of the forecasting error. We compare the approach to other regression based and grey-box methods in a real case study of a small-scale district energy system with mixed use and unknown lower-level control. We show that the proposed method outperforms the other forecasting methods in terms of average error and coefficient of determination. We further demonstrate that in our case study the error correction methods significantly reduce variance in ANN performance created by randomly initialized parameters in the networks.

No abstract

Cómo citar este artículo/Citation: Richner, P., Heer, Ph., Largo, R., Marchesi, E., Zimmermann, M. (2017). NEST -A platform for the acceleration of innovation in buildings. Informes de la Construcción, 69(548) ABSTRACTThe speed and quality of innovation in the construction sector has to be substantially increased in order to meet the pressing challenges associated with the building stock. For this purpose, the NEST project was started in Switzerland. NEST is a flexible and open research and technology transfer platform for partners from academia and industry where new solutions can be implemented and validated in a real life environment. NEST consists of a backbone, the static part, and research units which serve as office or living space where people live and work. Each unit is addressing specific research topics such as timber construction or digital fabrication and bears numerous innovation objects which are subject to continuous development and evaluation. NEST is a vertical neighbourhood of units, which are connected to a water hub and an energy hub. Once the research questions in a unit are answered and new products have been developed, the unit is deconstructed and replaced by a new unit addressing new topics.Keywords: open innovation, living lab, energy hub, water hub. RESUMEN La calidad en la innovación dentro del sector de la construcción debe ser rápidamente adaptada para cumplir con los inmediatos desafíos relacionados con la mejora de las edificaciones actuales. NEST es una plataforma de investigación y transferencia tecnológica flexible y abierta a universidades e industrias donde nuevas soluciones pueden ser implementadas y validadas en un entorno real. NEST consiste en una estructura fija con una serie módulos individuales que son utilizados como oficinas o apartamentos

In this paper, we consider the problem of controller tuning for an operating unit in a building energy system. As an illustrative plant example we focus on a heat pump. Since the plant is in use, the tuning method is supposed to not intervene with its operation. Moreover, the tuning procedure is supposed to be online, model-free, based only on historical data and needs to provide safety guarantees of the plant in operation. In this regard, we formulate the problem as a black-box optimization and adopt safe Bayesian optimization approaches for controller parameter tuning. These approaches are relatively new to the control community and not intensively studied in control applications. Meanwhile, the underlying systems are often expensive and performing relevant experiments is time consuming. Therefore, a crucial step prior to implementation in reality is validating the methods in simulation to verify their applicability. Toward this end, we derive a physical-based model for the heat pump and identify the unknown parameters using gray-box identification methods. Given the simulation model, we tune the controller parameters in simulation for optimal performance while considering safety constraints of the system.