Search

This paper addresses the design and implementation of a model predictive control framework for temperature control in buildings zones via direct control of their thermal energy inputs. Comfort-centric approach in ensured by selecting building thermal zones to be equal to the physical building rooms. The framework integrates different identification and estimation technologies, machine learning and model predictive control to assure systematic handling of non-modelled disturbances and offset-free control. It is envisioned as the lowest level in the hierarchical decomposition of building subsystems responsible for comfort and shaping the overall thermal energy consumption in building zones. The paper shows how it is deployed on a full scale occupied skyscraper building. To enable optimization of the whole building behaviour a special focus is put on developing the possibility for interaction and coordination with other building subsystems or energy distribution grids. This ensures the scalability of the approach, computational relaxation, technology independency, cost-effective implementation and enables upscaling towards the smart grid and smart city concepts where buildings play decisive roles. [Display omitted] • Direct control of thermal energy per zone. • Enabled interaction with other building subsystems. • Integral part for upscaling towards smart grid and smart city concepts. • Deployment and verification on a scale of the whole skyscraper building. • Modular service built on top of the existing building automation infrastructure. [ABSTRACT FROM AUTHOR]

• We decompose the building energy consumption into a global part and a local part through domain knowledge analysis. • A hybrid prediction model, combining static model and time series model, is proposed. • Proposed hybrid prediction model outperformed other machine learning models for a large margin. • The types of buildings have an influence on the model performance. The task of building energy prediction (BEP) is essential to several emerging research domains, including energy management, control optimization and fault detection. An accuracy prediction model contributes significantly to the improvement of building energy efficiency and flexibility. However, most existing prediction models are either based on single time series model or static model, making them perform poorly for long-term predictions. To solve this problem, this paper proposes a hybrid prediction model, which combines the deep ensemble (DE) model and autoregressive (AR) model together. Its motivation comes from the domain knowledge analysis of building energy consumption, where the energy consumption is decomposed into a global part and a local part. The deep ensemble model is adopted to predict its global part, and the AR model is employed to predict its local part. Comprehensive data experiments are conducted based on 50 real buildings across five building types to validate the model performance. The prediction horizon is the 24-h ahead sliding window prediction for one-year energy consumption. The results indicate that the hybrid prediction model outperforms the LSTM, DE-only, and ARIMA-only model, where its relative improvements of CV-RMSE are 28.7%, 35.98% and 18.47%, respectively. The experimental results also reveal that the types of buildings, i.e., office, lodging, will affect the model performance, which is attributed to their different user-behaviors. Based on the experimental results, it is demonstrated that the integration of static model and time series model is robust to the varying of building types and prediction steps, which should be the preferred choice of BEP task. [ABSTRACT FROM AUTHOR]

The data-driven method is a promising way to predict the energy consumption of buildings, however suffering from the data shortage problem in various scenarios. Even though transfer learning can improve the few-shot prediction performance by utilizing other buildings' data, the centralized approach poses potential privacy risks. To tackle this issue, the paper proposes a privacy-preserving knowledge sharing framework to facilitate the few-shot building energy prediction based on federated learning. First, a private data aggregation scheme is established to encrypt the sensitive data with shared random masks and guarantee the privacy of the data preprocessing and model optimization. Then, to alleviate the intrinsic data heterogeneity, a dynamical clustering federated learning algorithm is proposed to implement the intra-cluster and inter-cluster knowledge sharing along with the iterative clustering process for participating buildings. Finally, the network-based transfer learning approach is incorporated into the distributed framework to establish the customized model based on trained cluster models and further boost the prediction performance for each building. Extensive experiments on the Building Data Genome Project 2 (BDGP2) dataset indicate that the federated approach witnesses a desirable prediction performance while preserving the privacy of building occupants. • Propose a privacy-preserving knowledge sharing method for building energy prediction. • Establish a private data aggregation scheme for the sensitive information. • Develop a dynamical clustering federated learning algorithm for knowledge sharing. • Integrate transfer learning into the framework to build customized prediction models. [ABSTRACT FROM AUTHOR]

Large scale smart meter deployments have resulted in popularization of sensor-based electricity forecasting which relies on historical sensor data to infer future energy consumption. Although those approaches have been very successful, they require significant quantities of historical data, often over extended periods of time, to train machine learning models and achieve accurate predictions. New buildings and buildings with newly installed meters have small historical datasets that are insufficient to create accurate predictions. Transfer learning methods have been proposed as a way to use cross-domain datasets to improve predictions. However, these methods do not consider the effects of seasonality within domains. Consequently, this paper proposes Hephaestus, a novel transfer learning method for cross-building energy forecasting based on time series multi-feature regression with seasonal and trend adjustment. This method enables energy prediction with merged data from similar buildings with different distributions and different seasonal profiles. Thus, it improves energy prediction accuracy for a new building with limited data by using datasets from other similar buildings. Hephaestus works in the pre- and post- processing phases and therefore can be used with any standard machine learning algorithm. The case study presented here demonstrates that the proposed approach can improve energy prediction for a school by 11.2% by using additional data from other schools. [ABSTRACT FROM AUTHOR]

Recent studies emphasized the importance of building energy consumption prediction for improved decision making. Data-driven models are being widely used for building energy consumption prediction. Among these, support vector machines (SVM) gained a lot of popularity due to its capability of handling non-linear problems. This paper presents an SVM-based lighting energy consumption prediction model for office buildings. For this study, an office building in Philadelphia, PA was instrumented and the required lighting energy consumption data to train the model were collected from this building. The developed model predicts daily lighting energy consumption based on two features: daily average sky cover and day type. The results showed that the developed model could be a good baseline model for predicting lighting energy consumption, which could be further extended by taking occupant behavior into account. [ABSTRACT FROM AUTHOR]

In recent years, a large number of building energy consumption prediction models, with various intended uses, have been proposed. The majority of these models have either taken a data-driven or a physical modeling approach, with each approach having its own strengths and limitations. Towards leveraging the strengths and reducing the limitations of each approach for improved prediction performance, this paper presents a hybrid machine-learning approach for occupant-behavior-sensitive energy consumption prediction. The proposed approach is composed of three constituent models: (1) a machine-learning model that learns the impact of outdoor weather conditions from simulation-generated data, (2) a machine-learning model that learns the impact of occupant behavior from real data, and (3) an ensemble model that predicts cooling energy consumption based on the outputs of the first two models. The simulation-generated data were created through simulating a set of reference buildings in EnergyPlus. The real data were collected from an office building in Pennsylvania. The proposed hybrid model was validated on an unseen real dataset. It achieved 0.73 kWh RMSE and 9.07% CV in hourly cooling energy consumption prediction, which indicates that the proposed approach is promising. [ABSTRACT FROM AUTHOR]

Abstract: The field of building energy management, which monitors and analyses the energy use of buildings with the aim to control and reduce energy expenditure, is seeing a rapid evolution. Automated meter reading approaches, harvesting data at hourly or even half-hourly intervals, create a large pool of data which needs analysis. Computer analysis by means of machine learning techniques allows automated processing of this data, invoking expert analysis where anomalies are detected. However, machine learning always requires a historical dataset to train models and develop a benchmark to define what constitutes an anomaly. Computer analysis by means of building performance simulation employs physical principles to predict energy behaviour, and allows the assessment of the behaviour of buildings from a pure modelling background. This paper explores how building simulation approaches can be fused into energy management practice, especially with a view to the production of artificial bespoke benchmarks where historical profiles are not available. A real accommodation block, which is subject to monitoring, is used to gather an estimation of the accuracy of this approach. The findings show that machine learning from simulation models has a high internal accuracy; comparison with actual metering data shows prediction errors in the system (20%) but still achieves a substantial improvement over industry benchmark values. [Copyright &y& Elsevier]