Multi-Source Domain Adaptation Using Ambient Sensor Data DOI Creative Commons
Jawher Dridi, Manar Amayri, Nizar Bouguila

et al.

Applied Artificial Intelligence, Journal Year: 2024, Volume and Issue: 38(1)

Published: Nov. 19, 2024

Smart buildings have gained increasing interest recently by providing several advanced solutions, especially AI-based solutions. Activity recognition and occupancy estimation are among the outcomes of smart that can help provide advantages such as energy management security Previously, domain adaptation (DA) has been widely considered researchers to transfer knowledge from source domains, where we abundant labeled data, a target data is scarce. It tedious time-consuming task label with building applications which why unsupervised DA do in unlabeled domain. Semi-supervised (SSDA) also small amount Most (UDA) SSDA methods one target. However, it possible exploit multiple domains instead single enhance performance Multi-source (MSDA) more difficult than single-source but efficient. In this research, adapt MDSA evaluate them using sensorial datasets.

Language: Английский

DMFF: Deep multimodel feature fusion for building occupancy detection DOI Creative Commons
Kailai Sun

Building and Environment, Journal Year: 2024, Volume and Issue: 253, P. 111355 - 111355

Published: Feb. 27, 2024

The 2022 Global Status Report for Buildings and Construction (Buildings-GSR) indicates that construction activities have returned to pre-pandemic levels in most major economies, alongside more building energy consumption. To achieve the Net Zero emissions target by 2050, particularly post-pandemic era, accurate occupancy information is important enhance efficiency improve comfort. While remarkable progress has been made existing studies, they struggle make full use of multi-sensor data high accuracy. Furthermore, there a expectation multimodel multi-temporary fusion Transformer. In this study, we present Transformer-based multimodal, multi-temporal feature method (DMFF) detection. transfer domain knowledge from artificial intelligence into area, DMFF includes pretrain-finetune pipeline leverages pre-trained visual sound models. Multiple Transformer encoders are employed extract features different modalities. Then, propose self-attention mechanism modality learn relationships among various sensors. Our demonstrates superior performance on real dataset, outperforming machine deep learning methods (e.g., Convolutional Neural Networks, Random Forest, Multilayer Perceptrons). Applied room setting, shows promising potential savings. code demo accessible at https://github.com/kailaisun/multimodel_occupancy.

Language: Английский

Citations

4
A dialectical system framework for building occupant energy behavior DOI
Mei Yang, Hao Yu, Xiaoxiao Xu

et al.

Energy and Buildings, Journal Year: 2025, Volume and Issue: unknown, P. 115649 - 115649

Published: March 1, 2025

Language: Английский

Citations

0
Diagnostic Bayesian network in building energy systems: Current insights, practical challenges, and future trends DOI Creative Commons
Chujie Lu, Ziao Wang, Martín Mosteiro-Romero

et al.

Energy and Buildings, Journal Year: 2025, Volume and Issue: unknown, P. 115845 - 115845

Published: May 1, 2025

Language: Английский

Citations

0
Future technologies for building sector to accelerate energy transition DOI Creative Commons
Fabrizio Ascione, Sandro Nižetić,

Fuqiang Wang

et al.

Energy and Buildings, Journal Year: 2024, Volume and Issue: unknown, P. 115044 - 115044

Published: Nov. 1, 2024

Language: Английский

Citations

3
High-accuracy occupancy counting at crowded entrances for smart buildings DOI
Kailai Sun, Xinwei Wang, Tian Xing

et al.

Energy and Buildings, Journal Year: 2024, Volume and Issue: 319, P. 114509 - 114509

Published: Sept. 1, 2024

Language: Английский

Citations

2
An experimental comparative study of energy saving based on occupancy-centric control in smart buildings DOI
Irfan Qaisar, Wei Liang, Kailai Sun

et al.

Building and Environment, Journal Year: 2024, Volume and Issue: unknown, P. 112322 - 112322

Published: Nov. 1, 2024

Language: Английский

Citations

1
A Systematic Empirical Evaluation of Machine Learning Algorithm on Energy Prediction DOI Creative Commons
Ibtehaj Ul Haque, Waqar Khan Usafzai,

Yusra Khan

et al.

Pakistan Journal of Engineering Technology & Science, Journal Year: 2024, Volume and Issue: 12(1), P. 117 - 124

Published: July 29, 2024

The energy crisis has alerted all scientists and researchers in every field. evident increase the usage of electronic items causes high electricity consumption. Especially residential households high-rise apartments, buildings show electrical loads. This study is designed to forecast need using AI prediction models find most efficient model predict future loads any household considering its surroundings, like weather, air pressure, room temperature others. a comprehensive comparative that provides comparison between regression models. finding by optimum model, we can achieve best results predicting load.

Language: Английский

Citations

0
A Review of Bayesian Network for Fault Detection and Diagnosis: Practical Applications in Building Energy Systems DOI
Chujie Lu, Ziao Wang, Martín Mosteiro-Romero

et al.

Published: Jan. 1, 2024

Language: Английский

Citations

0
Multi-Source Domain Adaptation Using Ambient Sensor Data DOI Creative Commons
Jawher Dridi, Manar Amayri, Nizar Bouguila

et al.

Applied Artificial Intelligence, Journal Year: 2024, Volume and Issue: 38(1)

Published: Nov. 19, 2024

Smart buildings have gained increasing interest recently by providing several advanced solutions, especially AI-based solutions. Activity recognition and occupancy estimation are among the outcomes of smart that can help provide advantages such as energy management security Previously, domain adaptation (DA) has been widely considered researchers to transfer knowledge from source domains, where we abundant labeled data, a target data is scarce. It tedious time-consuming task label with building applications which why unsupervised DA do in unlabeled domain. Semi-supervised (SSDA) also small amount Most (UDA) SSDA methods one target. However, it possible exploit multiple domains instead single enhance performance Multi-source (MSDA) more difficult than single-source but efficient. In this research, adapt MDSA evaluate them using sensorial datasets.

Language: Английский

Citations

0