Publications
2021
Zhang, Liang; Wen, Jin; Li, Yanfei; Chen, Jianli; Ye, Yunyang; Fu, Yangyang; Livingood, William
A review of machine learning in building load prediction Journal Article
In: Applied Energy, vol. 285, pp. 116452, 2021, ISSN: 0306-2619.
Abstract | Links | BibTeX | Tags: Building energy forecasting, Building energy system, Building load prediction, Data engineering, Feature engineering, Machine learning
@article{ZHANG2021116452,
title = {A review of machine learning in building load prediction},
author = {Liang Zhang and Jin Wen and Yanfei Li and Jianli Chen and Yunyang Ye and Yangyang Fu and William Livingood},
url = {https://www.sciencedirect.com/science/article/pii/S0306261921000209},
doi = {https://doi.org/10.1016/j.apenergy.2021.116452},
issn = {0306-2619},
year = {2021},
date = {2021-01-01},
journal = {Applied Energy},
volume = {285},
pages = {116452},
abstract = {The surge of machine learning and increasing data accessibility in buildings provide great opportunities for applying machine learning to building energy system modeling and analysis. Building load prediction is one of the most critical components for many building control and analytics activities, as well as grid-interactive and energy efficiency building operation. While a large number of research papers exist on the topic of machine-learning-based building load prediction, a comprehensive review from the perspective of machine learning is missing. In this paper, we review the application of machine learning techniques in building load prediction under the organization and logic of the machine learning, which is to perform tasks T using Performance measure P and based on learning from Experience E. Firstly, we review the applications of building load prediction model (task T). Then, we review the modeling algorithms that improve machine learning performance and accuracy (performance P). Throughout the papers, we also review the literature from the data perspective for modeling (experience E), including data engineering from the sensor level to data level, pre-processing, feature extraction and selection. Finally, we conclude with a discussion of well-studied and relatively unexplored fields for future research reference. We also identify the gaps in current machine learning application and predict for future trends and development.},
keywords = {Building energy forecasting, Building energy system, Building load prediction, Data engineering, Feature engineering, Machine learning},
pubstate = {published},
tppubtype = {article}
}
The surge of machine learning and increasing data accessibility in buildings provide great opportunities for applying machine learning to building energy system modeling and analysis. Building load prediction is one of the most critical components for many building control and analytics activities, as well as grid-interactive and energy efficiency building operation. While a large number of research papers exist on the topic of machine-learning-based building load prediction, a comprehensive review from the perspective of machine learning is missing. In this paper, we review the application of machine learning techniques in building load prediction under the organization and logic of the machine learning, which is to perform tasks T using Performance measure P and based on learning from Experience E. Firstly, we review the applications of building load prediction model (task T). Then, we review the modeling algorithms that improve machine learning performance and accuracy (performance P). Throughout the papers, we also review the literature from the data perspective for modeling (experience E), including data engineering from the sensor level to data level, pre-processing, feature extraction and selection. Finally, we conclude with a discussion of well-studied and relatively unexplored fields for future research reference. We also identify the gaps in current machine learning application and predict for future trends and development.
Zhang, Liang; Wen, Jin
Active learning strategy for high fidelity short-term data-driven building energy forecasting Journal Article
In: Energy and Buildings, vol. 244, pp. 111026, 2021, ISSN: 0378-7788.
Abstract | Links | BibTeX | Tags: Active learning, Block design, Building energy forecasting, Expected error reduction, Model predictive control, Training data quality
@article{ZHANG2021111026,
title = {Active learning strategy for high fidelity short-term data-driven building energy forecasting},
author = {Liang Zhang and Jin Wen},
url = {https://www.sciencedirect.com/science/article/pii/S0378778821003108},
doi = {https://doi.org/10.1016/j.enbuild.2021.111026},
issn = {0378-7788},
year = {2021},
date = {2021-01-01},
journal = {Energy and Buildings},
volume = {244},
pages = {111026},
abstract = {The quality of a data-driven model is heavily dependent on the quality of data. Data from building operation often have data bias problems, which means that the data sample is collected in a way that some members of the intended data population are less likely to be included than others. Data-driven energy forecasting models built on such data hence are biased and could lead to large forecasting errors. Active learning— an effective method to defying data bias—is rarely studied or applied in the area of data-driven building energy forecasting modeling. This paper attempts to fill this gap and explores the application of active learning in data-driven building energy forecasting. The developed strategy in this paper efficiently generate informative training data within a time budget and uses block design to passively consider weather disturbances. The developed active learning strategy is applied and evaluated in both virtual and real-building testbeds against traditional data-driven methods. Via these virtual and real-building evaluation cases, we have demonstrated that the data bias problem typically exists in building operation data is resolved by applying the developed active learning strategy. Building energy forecasting models trained from data generated from the active learning strategy have shown improved performances in both model accuracy and model extendibility perspectives. The effectiveness of the block design module is also validated to effectively consider the impact of weather conditions on active learning design.},
keywords = {Active learning, Block design, Building energy forecasting, Expected error reduction, Model predictive control, Training data quality},
pubstate = {published},
tppubtype = {article}
}
The quality of a data-driven model is heavily dependent on the quality of data. Data from building operation often have data bias problems, which means that the data sample is collected in a way that some members of the intended data population are less likely to be included than others. Data-driven energy forecasting models built on such data hence are biased and could lead to large forecasting errors. Active learning— an effective method to defying data bias—is rarely studied or applied in the area of data-driven building energy forecasting modeling. This paper attempts to fill this gap and explores the application of active learning in data-driven building energy forecasting. The developed strategy in this paper efficiently generate informative training data within a time budget and uses block design to passively consider weather disturbances. The developed active learning strategy is applied and evaluated in both virtual and real-building testbeds against traditional data-driven methods. Via these virtual and real-building evaluation cases, we have demonstrated that the data bias problem typically exists in building operation data is resolved by applying the developed active learning strategy. Building energy forecasting models trained from data generated from the active learning strategy have shown improved performances in both model accuracy and model extendibility perspectives. The effectiveness of the block design module is also validated to effectively consider the impact of weather conditions on active learning design.
2019
Zhang, Liang; Wen, Jin
A systematic feature selection procedure for short-term data-driven building energy forecasting model development Journal Article
In: Energy and Buildings, vol. 183, pp. 428-442, 2019, ISSN: 0378-7788.
Abstract | Links | BibTeX | Tags: Building energy forecasting, Data-driven model, Feature selection
@article{ZHANG2019428,
title = {A systematic feature selection procedure for short-term data-driven building energy forecasting model development},
author = {Liang Zhang and Jin Wen},
url = {https://www.sciencedirect.com/science/article/pii/S0378778818321625},
doi = {https://doi.org/10.1016/j.enbuild.2018.11.010},
issn = {0378-7788},
year = {2019},
date = {2019-01-01},
journal = {Energy and Buildings},
volume = {183},
pages = {428-442},
abstract = {An accurate building energy forecasting model is the key for real-time model based control of building energy systems and building-grid integration. Data-driven models, though have lower engineering cost during their development process, often suffer from poor model generalization caused by high data dimensionality. Feature selection, a process of selecting a subset of relevant features, can defy high dimensionality, increase model interpretability, and enhance model generalization. In building energy modeling research, features are often selected based on domain knowledge. There lacks a comprehensive methodology to guide a systematic feature selection procedure when developing building energy forecasting models. In this research, a systematic feature selection procedure for developing a building energy forecasting model is proposed which attempts to integrate statistical analysis, building physics and engineering experiences. The proposed procedure includes three steps, i.e., (Step 1) feature pre-processing based on domain knowledge, (Step 2) feature removal through filter methods to remove irrelevant and redundant variables, and (Step 3) feature grouping through wrapper method to search for the best feature set. Two case studies are presented here using both simulated and real building data. The simulated building data are generated from a medium-size office building (a DOE reference building) simulation model. The real building data are obtained from a medium-size campus building in Philadelphia, PA. In both cases, the energy forecasting models that are developed using proposed systematic feature selection procedure is compared with models using other feature selection techniques. Results show that the models developed using proposed procedure have better accuracy and generalization.},
keywords = {Building energy forecasting, Data-driven model, Feature selection},
pubstate = {published},
tppubtype = {article}
}
An accurate building energy forecasting model is the key for real-time model based control of building energy systems and building-grid integration. Data-driven models, though have lower engineering cost during their development process, often suffer from poor model generalization caused by high data dimensionality. Feature selection, a process of selecting a subset of relevant features, can defy high dimensionality, increase model interpretability, and enhance model generalization. In building energy modeling research, features are often selected based on domain knowledge. There lacks a comprehensive methodology to guide a systematic feature selection procedure when developing building energy forecasting models. In this research, a systematic feature selection procedure for developing a building energy forecasting model is proposed which attempts to integrate statistical analysis, building physics and engineering experiences. The proposed procedure includes three steps, i.e., (Step 1) feature pre-processing based on domain knowledge, (Step 2) feature removal through filter methods to remove irrelevant and redundant variables, and (Step 3) feature grouping through wrapper method to search for the best feature set. Two case studies are presented here using both simulated and real building data. The simulated building data are generated from a medium-size office building (a DOE reference building) simulation model. The real building data are obtained from a medium-size campus building in Philadelphia, PA. In both cases, the energy forecasting models that are developed using proposed systematic feature selection procedure is compared with models using other feature selection techniques. Results show that the models developed using proposed procedure have better accuracy and generalization.
2016
Li, Xiwang; Wen, Jin
System identification and data fusion for on-line adaptive energy forecasting in virtual and real commercial buildings Journal Article
In: Energy and Buildings, vol. 129, pp. 227-237, 2016, ISSN: 0378-7788.
Abstract | Links | BibTeX | Tags: Building energy forecasting, Data fusion, On-line estimation, Real field implementation, System identification
@article{LI2016227,
title = {System identification and data fusion for on-line adaptive energy forecasting in virtual and real commercial buildings},
author = {Xiwang Li and Jin Wen},
url = {https://www.sciencedirect.com/science/article/pii/S0378778816306922},
doi = {https://doi.org/10.1016/j.enbuild.2016.08.014},
issn = {0378-7788},
year = {2016},
date = {2016-01-01},
journal = {Energy and Buildings},
volume = {129},
pages = {227-237},
abstract = {Accurate, computationally efficient, and cost-effective energy forecasting models are essential for model based control. Existing studies in model based control have mostly been focusing on developing energy forecasting models using simplified physics based or data driven models. However, creating and identification the simplified physics model are often challenging, which requires expert knowledge for model simplification and significant engineering efforts for model training. In addition, the accuracy and robustness of data driven models are always bounded by the training data. To this end, developing high fidelity energy forecasting models with less engineering effort and good performance is still an urgent task. Although the previous studies from the authors have shown great promises in a system identification model and outperformed other data-driven and grey box models, they still have large errors at the special operation situations. Therefore, this paper investigates a novel methodology to develop energy estimation models for on-line building control and optimization using an integrated system identification and data fusion approach. The data fusion approach is able to adapt the forecasting model under the special operation situations based on the real measurements. An eigensystem realization algorithm based model reformation method is developed to convert the system identification models into state space models. Kalman filter based data fusion techniques are then implemented on the state space models to improve the model accuracy and robustness. The developed methodology are evaluated using data from a virtual building (simulated) and a real small size commercial building. Three different data fusion intervals: 15, 30, and 60min, have been tested. The overall building energy estimation accuracy from this proposed methodology can reach to above 95% in the virtual building and around 90% in the real building. The results also show that the shorter data fusion interval used, the higher accuracy can be achieved.},
keywords = {Building energy forecasting, Data fusion, On-line estimation, Real field implementation, System identification},
pubstate = {published},
tppubtype = {article}
}
Accurate, computationally efficient, and cost-effective energy forecasting models are essential for model based control. Existing studies in model based control have mostly been focusing on developing energy forecasting models using simplified physics based or data driven models. However, creating and identification the simplified physics model are often challenging, which requires expert knowledge for model simplification and significant engineering efforts for model training. In addition, the accuracy and robustness of data driven models are always bounded by the training data. To this end, developing high fidelity energy forecasting models with less engineering effort and good performance is still an urgent task. Although the previous studies from the authors have shown great promises in a system identification model and outperformed other data-driven and grey box models, they still have large errors at the special operation situations. Therefore, this paper investigates a novel methodology to develop energy estimation models for on-line building control and optimization using an integrated system identification and data fusion approach. The data fusion approach is able to adapt the forecasting model under the special operation situations based on the real measurements. An eigensystem realization algorithm based model reformation method is developed to convert the system identification models into state space models. Kalman filter based data fusion techniques are then implemented on the state space models to improve the model accuracy and robustness. The developed methodology are evaluated using data from a virtual building (simulated) and a real small size commercial building. Three different data fusion intervals: 15, 30, and 60min, have been tested. The overall building energy estimation accuracy from this proposed methodology can reach to above 95% in the virtual building and around 90% in the real building. The results also show that the shorter data fusion interval used, the higher accuracy can be achieved.