Intelligent Buildings of the Future: Cyberaware, Deep Learning Powered, and Human Interacting

Manic, Milos; Amarasinghe, Kasun; Rodríguez-Andina, Juan J.; Rieger, Craig

doi:10.1109/mie.2016.2615575

Cited by 94 publications

(50 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where A z is the area of z th zone. The unnormalized posterior probability of given contaminant source location can be estimated by using (5,24)…”

Section: Bayesian Source Localization Using Dmgpesmentioning

confidence: 99%

“…However, the computational cost of these simplified models is still prohibitive for online applications. Moreover, these models use model order reduction which results in a loss of spatial information.Alternatively, recent studies have explored various machine learning algorithms to develop computationally efficient emulators to infer the indoor events [24]. A similar approach is adopted in this paper, where the contaminant fate and transport model is replaced by a statistical emulator in the Bayesian framework.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Deep Gaussian Process-Based Bayesian Inference for Contaminant Source Localization

2018

View full text Add to dashboard Cite

This paper proposes a Bayesian framework for localization of multiple sources in the event of accidental hazardous contaminant release. The framework assimilates sensor measurements of the contaminant concentration with an integrated multizone computational fluid dynamics (multizone-CFD) based contaminant fate and transport model. To ensure online tractability, the framework uses deep Gaussian process (DGP) based emulator of the multizone-CFD model. To effectively represent the transient response of the multizone-CFD model, the DGP emulator is reformulated using a matrix-variate Gaussian process prior. The resultant deep matrix-variate Gaussian process emulator (DMGPE) is used to define the likelihood of the Bayesian framework, while Markov Chain Monte Carlo approach is used to sample from the posterior distribution. The proposed method is evaluated for single and multiple contaminant sources localization tasks modeled by CONTAM simulator in a single-story building of 30 zones, demonstrating that proposed approach accurately perform inference on locations of contaminant sources. Moreover, the DMGP emulator outperforms both GP and DGP emulator with fewer number of hyperparameters.arXiv:1806.08069v1 [stat.AP] 21 Jun 2018 approaches do not provide full uncertainty analysis of contaminant source location and characteristics. Primarily due to their ability to fully quantify the uncertainties, the Bayesian framework for contaminant source localization and characterization is gaining prominence in various studies [16][17][18][19]. However, barring few notable exceptions of conjugacy like linear models with Gaussian priors, estimation of the Bayesian posterior distribution is analytically intractable. Thus, the Bayesian framework is often implemented by approximating the posterior distribution using sampling methods like Markov Chain Monte Carlo (MCMC) [20].Implementation of the MCMC based Bayesian inference framework for contaminant source localization and characterization is challenging due to: 1) for acceptable accuracy, MCMC requires 10 4 − 10 7 samples [16,19,21,22]; 2) transient nature of the contaminant dispersion phenomena [19,23]. Note that each MCMC sample requires simulating the contaminant fate and transport model, rendering the Bayesian framework computationally intractable for real-time contaminant source localization and characterization [16]. To resolve the first challenge, many state of the art approaches relies on simplification of the contaminant fate and transport model [18]. However, the computational cost of these simplified models is still prohibitive for online applications. Moreover, these models use model order reduction which results in a loss of spatial information.Alternatively, recent studies have explored various machine learning algorithms to develop computationally efficient emulators to infer the indoor events [24]. A similar approach is adopted in this paper, where the contaminant fate and transport model is replaced by a statistical emulator in the Bayesian framework. Notwithstanding t...

show abstract

“…where A z is the area of z th zone. The unnormalized posterior probability of given contaminant source location can be estimated by using (5,24)…”

Section: Bayesian Source Localization Using Dmgpesmentioning

confidence: 99%

mentioning

confidence: 99%

Deep Gaussian Process-Based Bayesian Inference for Contaminant Source Localization

2018

View full text Add to dashboard Cite

show abstract

“…In practice, they are mostly combined into local small power grids defined as microgrids [1][2][3]. Moreover, some RES are installed near buildings or households and locally connected to energy supply networks, providing not only electrical but thermal energy as well [4,5]. It is necessary to highlight that buildings are one of the most energy-intensive sectors.…”

Section: Introductionmentioning

confidence: 99%

A New Concept of Active Demand Side Management for Energy Efficient Prosumer Microgrids with Smart Building Technologies

Ożadowicz

2017

Energies

View full text Add to dashboard Cite

Abstract:Energy efficient prosumer microgrids (PMGs) with active and flexible demand side management (DSM) mechanisms are considered to be crucial elements of future smart grids. Due to an increasing share of renewable energy and the growing power demand, appropriate tools to manage not only the loads but also small generation units, heating and cooling systems, storage units and electric vehicles should be provided for them. Therefore, this paper proposes an innovative approach to both physical and logical organization of an active DSM system for future building-integrated prosumer microgrids (BIPMGs), based on standard building automation and control systems (BACS) as well as Internet of Things (IoT) paradigm. New event-triggered control functions with developed universal, logical interfaces for open BACS and IoT network nodes are presented and their implementation in smart metering as well as fully integrated energy management mechanisms is analyzed. Finally, potential energy efficiency improvements with proposed BACS functions are discussed, based on BACS efficiency requirements defined in the EN 15232 standard.

show abstract

“…Therefore, the real smart home systems should have the likeness of an artificial intelligence (AI) that can independently make decisions and regulate the activities of all systems of a building or a house depending on the constantly changing conditions and needs of users. Methods of AI and machine learning (ML) can be effectively used for such tasks as building energy management and energy efficiency [3], improving the security of both systems and users, making systems adaptable to users (it will be most useful for people with disabilities).…”

Section: Introductionmentioning

confidence: 99%

“…Basically, the work in the field of application of ANN in systems of smart homes or intelligent buildings is centered around predicting the energy consumption of buildings [3][4][5] and the management of various house systems by predicting user actions based on accumulated data on their previous activities (for example, switching on and off lighting devices in [6]). …”

Section: Introductionmentioning

confidence: 99%

Adaptive system of intelligent environment based on neural networks

Nikolaev¹

2017

Proceedings of the IV International Research Conference "Information Technologies in Science, Management, Social Sphere and Med

View full text Add to dashboard Cite

Abstract-This paper describes the concept of adaptive system of intelligent environment for smart homes and intelligent buildings. This intelligent system is based on one of the methods of artificial intelligence and machine learning -artificial neural networks. The key feature of the proposed system is the recognition of current human activity and the switching of the intelligent environment system in the most optimal mode for performing this activity. These operations are performed by neural networks.Keywords-intelligent environment; smart home; intelligent building; artificial intelligence; machine learning, artificial neural networks; deep learning; convolutional neural networks.

show abstract

Intelligent Buildings of the Future: Cyberaware, Deep Learning Powered, and Human Interacting

Cited by 94 publications

References 29 publications

Deep Gaussian Process-Based Bayesian Inference for Contaminant Source Localization

Deep Gaussian Process-Based Bayesian Inference for Contaminant Source Localization

A New Concept of Active Demand Side Management for Energy Efficient Prosumer Microgrids with Smart Building Technologies

Adaptive system of intelligent environment based on neural networks

Contact Info

Product

Resources

About