Journal of Building Physics最新文献

Research and development have demonstrated that effective building energy prediction is significant for enhancing energy efficiency and ensuring grid reliability. Many machine learning (ML) models, particularly deep learning (DL) approaches, are widely used for power or peak demand forecasting. However, evaluating prediction models solely based on accuracy is insufficient, as complex models often suffer from low interpretability and high computational costs, making them difficult to implement in real-world applications. This study proposes a multi-perspective evaluation analysis that includes prediction accuracy (both overall and at different power levels), interpretability (global/local perspectives and model structure), and computational efficiency. Three popular DL models-recurrent neural network, gated recurrent unit, long short-term memory, and three tree-based models-random forecast, extreme gradient boosting, and light gradient boosting machine-are analyzed due to their popularity and high prediction accuracy in the field of power demand prediction. The comparison reveals the following: (1) The best-performing prediction model changes under different power demand levels. In scenarios with lower power usage patterns, tree-based models achieve an average CV-RMSE of 13.62%, which is comparable to the 12.17% average CV-RMSE of DL models. (2) Global and local interpretations indicate that past power use and time-related features are the most important. Tree-based models excel at identifying which specific lagged features are more significant. (3) The DL model behavior can be interpreted by visualizing the hidden state at each layer to reveal how the model captures temporal dynamics across different time steps. However, tree-based models are more intuitive to interpret using straightforward decision rules and structures. This study provides guidance for applying ML algorithms to load forecasting, offering multiple perspectives on model selection trade-offs.

Pressurized corridor (PC) ventilation systems are a common method used in existing multi-unit residential buildings (MURBs) to deliver make-up air to individual units, and as a means of controlling inter-zonal odour/contaminant transfer. In PC systems, ventilation air is supplied directly to the common corridor and enters the units via intentional undercuts at the unit entry doors. In practice, the amount of ventilation air supplied to each unit is dependent on the air pressure differential between the two zones, which can be affected by occupant behaviours, such as window and unit exhaust fan operation; wind; or large indoor-outdoor temperature differentials. Accurately characterizing the impact of these variables on building pressure differentials is critical to not only identifying conditions when depressurization events may occur (which would result in a lack of ventilation to dwelling unit and the potential for contaminant movement from units to the corridor), but also understanding how operational changes can improve system operation. This paper will describe the development of an XGBoost regression model for predicting inter-zonal pressure differentials in a contemporary MURB with a PC system. The model was trained and validated using measurements collected as part of a 6-month field study in a 17-storey MURB located in Toronto, Canada, including corridor-to-unit and exterior-to-unit differential pressures, window/door operation, corridor supply air flow rates and interior/exterior temperature and relative humidity. Unit exhaust fan operation was inferred from the unit differential pressure data. This paper addresses feature selection, hyperparameter tuning and accuracy assessment, with a specific emphasis on evaluating the potential for the use of the model as a diagnostic tool and testing environment to evaluate ventilation system performance in multi-unit residential buildings.

The combination of a Solar Assisted Geothermal Heat Pump system (SAGHP) with a multi-zone greenhouse is investigated to take advantage of water flooding in abandoned open pit mines in Canada. The envisioned system includes an Air Handling Unit (AHU), Heat Recovery Ventilation (HRV), daily Thermal Energy Storage (TES), and daily Domestic Hot Water (DHW). The main objective is to satisfy the greenhouse heating, cooling, and dehumidification loads, for the considered application, while minimizing energy consumption. This analysis is conducted using data extracted from a case study of a commercial, multi-zone greenhouse, considering different daily weather conditions throughout a year. To reduce the computation time, a clustering approach based on the K-Means method is applied to obtain a small number of typical weather days. Elbow, Dendrogram, and Silhouette approaches confirmed that it is possible to represent a year as six different Typical Days (TD), which can be further categorized as Heating only (TD1 and TD2), Heating/Cooling (TD3 and TD4), and Cooling only (TD5 and TD6). Dynamic Pinch Approach (DPA) showed a great ability to target the minimum energy consumption and maximize the potential heat recovery for each typical day. The study focuses on energy targeting, with discussion of preliminary design considerations, such as the solar hot water (SHW) system, Thermal Energy Storage (TES), and heat pumping. Results revealed that mine water can significantly improve the energy system efficiency, specifically where heating/cooling or only cooling is dominant (TD3, TD4, TD5, and TD6). For instance, by integrating an AHU with the greenhouse for the TDs where heating/cooling is dominant, 22.5% energy saving is achievable. The incorporation of heat pumping, waste heat recovery, and solar thermal collectors through mixed direct/indirect heat recovery (i.e. via TES) can reduce hot utility usage in the considered application by as much as 40%.

This paper presents a methodology to develop archetype gray-box models and use them in an economic model-based predictive control algorithm to simulate optimal heating load management in response to a newly-introduced static time-of-use tariff for Québec's residential sector, rate Flex-D. The methodology is evaluated through a case study, wherein in situ measurements from a two-storey unoccupied research house of Hydro-Québec are used to develop an 11R6C network with a heuristic zoning-by-floor approach and compute the sequence of optimal electric heating input for the next control horizon. Properly-tuned economic model-based predictive control under rate Flex-D shows potential for an approximately 30% reduction in daily heating cost compared to the reference operation, with a minimal average deviation of indoor air temperature from the reference setpoint. Also, the analysis of the response's sensitivity to weather forecast uncertainties indicates that the most influential uncontrolled input directing the performance of economic model-based predictive control is the structure price signal, rendering the impact of uncertainty in the weather forecast negligible.