Search

• Features are selected from the perspectives of machine learning and field application. • Stepwise FS path combining optimization with machine learning algorithms is proposed. • Best performance using existing features and corresponding feature set are revealed. • Recommendations for feature supplementation to further improve performance are given. • Feature sets are verified to be general and effective by experiments and comparisons. Fault diagnosis (FD) is vital for enhancing chiller efficiency and reliability. Feature selection (FS) is the prerequisite and key to diagnose faults. This paper addresses two intriguing questions from machine learning (ML) and field perspectives. Question-1: Based on commonly installed sensors, what is the best performance that the FD models based on ML algorithms can achieve, and what features are relevant? Question-2: Which features can enhance diagnostic performance? and to what extent? This paper designs a stepwise FS process. First, a field investigation is conducted to gather information on sensors installed in actual chillers. Based on actual field installation, feature calculation cost, and thermodynamic mechanism, three levels of initial feature libraries are created, each containing an increasing number and type of features. An FS method combining an optimization algorithm with an FD model based on ML algorithm is proposed. In the end, the insight into the best diagnostic performance achieved by ML-based models using existing sensors and the corresponding optimal feature subsets is provided, and recommendations for feature supplementation to further improve diagnostic performance are also provided. Compared with other literature-reported feature subsets, the recommended feature subsets show better generality and effectiveness on seven commonly used ML-based models. [ABSTRACT FROM AUTHOR]

The massive collection of data via emerging technologies like the Internet of Things (IoT) requires finding optimal ways to reduce the created features that have a potential impact on the information that can be extracted through the machine learning process. The mining of knowledge related to a concept is done on the basis of the features of data. The process of finding the best combination of features is called feature selection. In this paper we deal with multivariate time-dependent series of data points for energy forecasting in smart buildings. We propose a methodology to transform the time-dependent database into a structure that standard machine learning algorithms can process, and then, apply different types of feature selection methods for regression tasks. We used Weka for the tasks of database transformation, feature selection, regression, statistical test and forecasting. The proposed methodology improves MAE by 59.97% and RMSE by 40.75%, evaluated on training data, and it improves MAE by 42.28% and RMSE by 36.62%, evaluated on test data, on average for 1-step-ahead, 2-step-ahead and 3-step-ahead when compared to not applying any feature selection methodology. [ABSTRACT FROM AUTHOR]

• Developed and applied a feature selection algorithm to unitary equipment at part load conditions. • Feature relevancy analysis of the air and refrigerant side is performed. • Feature selection algorithm verified using Artificial Neural Network. • A subset of 33% of features are identified with the same level of accuracy as a complete dataset. • Developed guidelines for the formulation of future models using algorithms. A variable-speed air-conditioning system provides more control than single-or multi-stage units, resulting in increased energy efficiency and more precise temperature control. Current commercial and residential direct expansion (DX) equipment testing standards require the embedded control placed on variable-speed equipment to be deactivated during performance testing and instead the fan and compressor speeds are fixed. For this reason, the test and field performances differ, creating a performance as well as an energy consumption gap. This paper presents a machine learning-based approach to investigate critical features from the air and refrigerant side influencing the cooling capacity of unitary air conditioning equipment. High-fidelity experimental data obtained from 4 different state-of-the-art high-efficiency units, 1 fixed-speed 35.2(10) kW(tons) commercial and 3 variable-speed residential unitary equipment of 12.3(3.5), 14(4), and 17.6(5) kW(tons) of cooling capacity have been analyzed by utilizing a feature selection methodology, Elastic Net (EN). Influential features with higher relevance scores corresponding to the total cooling capacity are fed into a supervised Artificial Neural Network (ANN), formulated as a predictive model to evaluate the validity of features selected. Results show that the proposed method of feature selection when applied, ANN is able to predict the equipment cooling capacity with a mean absolute percent error of less than 2% for all tested units, thus approving the proposed input features. The findings may be used as recommendations for the development of a semi-empirical model for better cooling capacity and COP predictions. [ABSTRACT FROM AUTHOR]

[Display omitted] • Multimodal occupancy detection framework for residential buildings. • Two variants of Occ-STPN models with feature level fusion and decision level fusion. • Feature extraction and occupancy detection models for various data modalities. • Transferability of a trained occupancy detection model from one building to another. • Occupancy prediction performance evaluation on three open-sourced datasets. For several years now, smart building energy systems have been a research area of intensive activity. In light of the increasing need for sustainable buildings and energy systems, this trend motivates an increasing need for a solution to reduce carbon dioxide emissions and improve energy efficiency. This paper proposes a high-performing and transferable occupancy detection framework that combines sensor data from different data modalities, including time series environmental data (temperature, humidity, and illuminance), image data, and acoustic energy data using ensemble method. To draw out the best prediction performance in each modality, the proposed framework was developed, including various models that were designed to learn the occupancy patterns reflected in the physical data streams. To tackle the time series environmental data, we designed two variants of an occupancy detection spatiotemporal pattern network (Occ-STPN) that performs both feature level and decision level fusion, respectively. We also propose a new metric; the fading memory mean square error (FMMSE), that provides a fair evaluation and penalization of delayed occupancy predictions. Multiple open-sourced datasets, including the Electricity Consumption and Occupancy and the University of California, Irvine's (UCI) building occupancy detection dataset, along with our own real data collected from six different houses, were used to validate the algorithms' performance. The experimental results presented herein break down the performance for each sensing modality, and a detailed analysis of the performance is also discussed. [ABSTRACT FROM AUTHOR]

[Display omitted] • Most of the reviewed studies (62%) focused on group-based comfort models. • ML models, especially PCMs outperformed conventional models (PMV and adaptive). • ML-based controls can improve IAQ while reducing energy consumption and CO 2 levels. • The most frequently used algorithms were SVM, ANN and Ensemble Learning. This paper provides a systematic review on the application of Machine Learning (ML) in thermal comfort studies to highlight the latest methods and findings and provide an agenda for future studies. Reviewed studies were investigated to highlight ML applications, parameters, methods, performance and challenges. The results show that 62% of reviewed studies focused on developing group-based comfort models, while 35% focused on personal comfort models (PCMs) which account for individual differences and present high prediction accuracy. ML models could outperform PMV and adaptive models with up to 35.9% and 31% higher accuracy and PCMs could outperform PMV models with up to 74% higher accuracy. Applying ML-based control schemas reduced thermal comfort-related energy consumption in buildings up to 58.5%, while improving indoor quality up to 90% and reducing CO 2 levels up to 24%. Using physiological parameters improved the prediction accuracy of PCMs up to 97%. Future studies are recommended to further investigate PCMs, determine the optimum sample size and consider both fitting and error metrics for model evaluation. This study introduces data collection, thermal comfort indices, time scale, sample size, feature selection, model selection, and real world application as the remaining challenges in the application of ML in thermal comfort studies. [ABSTRACT FROM AUTHOR]

• A framework of building energy modeling for data predictive control is developed. • The framework integrates active learning strategy with feature selection technique. • The framework is demonstrated in a virtual testbed (EnergyPlus and BCVTB). • The case study shows improvement in energy modeling accuracy and DPC performance. Three gaps impede the development of cost-effective and accurate data-driven building energy modeling/models (DBEM) for energy forecasting and predictive control strategies. Gap 1: data bias is common in building operation data, but this topic is hardly studied in DBEM; Gap 2: high data dimensionality is common in DBEM, but a systematic and scalable methodology is lacking to solve the problem; Gap 3: the interactions between data bias and high dimensionality have not been systematically studied for DBEM and predictive control in buildings. In this paper, to address the three gaps mentioned above, we develop a framework that integrates active learning and feature selection for DBEM used for whole building data predictive control (or DPC, which is a branch of model predictive control). The framework provides a systematic methodology and automatic workflow that starts with raw data from building automation systems to the establishment of data-driven energy models for DPC controllers. The developed strategies and framework are evaluated in a virtual testbed based on EnergyPlus and BCVTB. Improved performance and reduced computational complexity are observed from the DBEM built with the developed framework, as well as the DPC controller based on that DBEM, indicating the effectiveness of the developed framework. [ABSTRACT FROM AUTHOR]

• Most effective temporal features from smart meter data for building characterization are found. • State-of-the-art feature selection methods and a proposed customized approach are utilized. • Importance-in-prediction based features that improve building classification are determined. • Number of utilized features is substantially reduced while the accuracy is also enhanced. The present paper aims at determining the most influential features to be extracted from smart meter data to facilitate machine learning-based classification of non-residential buildings. Smart meter-driven remote estimation of the chosen characteristics (the buildings' performance class, use type, and operation group) is significantly helpful in buildings' commissioning, benchmarking, and diagnostics applications. As the first step, state-of-the-art feature selection methods and a proposed customized approach are utilized for determining the most influential parameters in the pool of temporal features, proposed in a previous study. Next, importance-in-prediction based features, generated from an hour-ahead load prediction pipeline, that can improve the classification accuracy are proposed and added as additional input parameters. Finally, interpretations about some of the most influential features for different classification targets are provided. The obtained results demonstrate that, while aiming at estimating the buildings' use type, through performing feature selection and adding importance-in-prediction based features, the number of utilized features is reduced from 290 (initial pool of features proposed in a previous study) to 29, while also increasing the accuracy from 71% to 74%. Similarly, number of employed features for estimating the performance class is decreased from 224 to 17 and the achieved accuracy is improved from 56% to 62%. Finally, using only 6 selected features, compared to 287 features in the initial set, the obtained accuracy for the classification of operation group is increased from 98% to 100%. It is thus demonstrated that the proposed methodology, through selecting and utilizing notably fewer features, results in a notable simplification of the feature extraction procedures, improves the achieved accuracy, and facilitates providing interpretations about the reason behind the influence of some of the most important features. [ABSTRACT FROM AUTHOR]