A methodology based on Hidden Markov Models for occupancy detection and a case study in a low energy residential building

Candanedo, Luis M.; Feldheim, Véronique; Deramaix, Dominique

doi:10.1016/j.enbuild.2017.05.031

Cited by 96 publications

(31 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For instance, previous research used machine learning techniques to predict occupancy [109][110][111] and occupants' interactions with buildings [112][113][114][115] and energy use. 116 These techniques are particularly beneficial in modelling with large datasets regarding occupant-building interactions.…”

Section: Unresolved Modelling Issues and Future Requirementsmentioning

confidence: 99%

Critical review and illustrative examples of office occupant modelling formalisms

D’Oca

Gunay

Gilani

et al. 2019

Building Services Engineering Research and Technology

View full text Add to dashboard Cite

It is widely understood that occupants can have a significant impact on building performance. Accordingly, the field has benefited extensively from research efforts in the past decade. However, the methods and terminology involved in modelling occupants in buildings remains fragmented across a large number of studies. This fragmentation represents a major obstacle to those who intend to join in this research endeavor as well as for the convergence and standardization of methods. To address this issue, this paper investigates occupant modelling methods for the key domains of electric lighting, blinds, operable windows, thermostats, plug loads, and occupancy. In the reviewed literature, five broad categories of occupant model formalisms were identified: schedules, Bernoulli models, discrete-time Markov models, discrete-event Markov models, and survival models. Illustrative examples were provided from two independent datasets to demonstrate the strengths and weaknesses of these model forms. It was shown that Markov models are suitable to represent occupants' adaptive behaviors, while survival models are suitable to represent occupancy, non-adaptive behaviors, and infrequently executed adaptive behaviors, such as the blinds opening behavior. Practical application: The engineering application of the occupant modelling formalisms that are critically reviewed in this paper is that these models are highly beneficial for incorporating occupants' presence and behaviors into building design and control. Building design can be improved significantly regarding energy use and occupant comfort when the most suitable occupant models are implemented in simulation-aided building design process. Ultimately, like for any modelling domain, the most suitable model is dependent on the modelling objective (e.g. optimizing passive design, equipment sizing), building type and size, occupant-related domain (e.g. occupancy, window-opening behavior), and climate zones. Furthermore, there is great potential in improving occupant comfort and energy savings of existing buildings when occupants' presence and interactions with buildings' systems and components are predicted accurately using occupant models.

show abstract

Section: Unresolved Modelling Issues and Future Requirementsmentioning

confidence: 99%

Critical review and illustrative examples of office occupant modelling formalisms

D’Oca

Gunay

Gilani

et al. 2019

Building Services Engineering Research and Technology

View full text Add to dashboard Cite

show abstract

“…The applied approach provided 90.27% average accuracy for five input features and 70.46% average accuracy for two input features. In [2], an occupancy detection methodology based on HMM was used to infer the daily and hourly average occupancy schedules. The HMM based on the first order difference of CO2 data at 5 min time average achieved the best accuracy (90.24%).…”

Section: Related Workmentioning

confidence: 99%

“…The reason for that uncertainty may be acquisition errors, or incomplete knowledge. Some of the mentioned methods dealt with the uncertainty of decision based on probability theory [36] using HMM such as [2, 4, 5, 12, 18, 22, 23] or based on fuzzy set theory [37] such as [16, 29]. Probability is the likelihood of whether an event will occur using historical data.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Sensor-based occupancy detection using neutrosophic features fusion

Fayed

Abu-Elkheir

El-Daydamony

et al. 2019

Heliyon

View full text Add to dashboard Cite

Occupancy detection using ambient sensors has many benefits such as saving energy and money, enhancing security monitoring systems, and maintaining the privacy. However, sensors data suffers from uncertainty and unreliability due to acquisition errors or incomplete knowledge. This paper presents a new heterogeneous sensors data fusion method for binary occupancy detection which detects whether the place is occupied or not. This method is based on using neutrosophic sets and sensors data correlations. By using neutrosophic sets, uncertain data can be handled. Using sensors data fusion, on the other hand, increases the reliability by depending on more than one sensor data. Accordingly, the results of experiments applied using Random Forest (RF), Linear Discriminant Analysis (LDA), and FUzzy GEnetic (FUGE) algorithms prove the new method to enhance detection accuracy.

show abstract

“…The authors in [9] use of HMM for occupancy modelling based on observations from smart electricity meters. However, in most cases, the observations contain environmental variables such as the CO2 concentration ( [10][11][12]). An auto-regressive Hidden Markov Model, which is an extension of the above-mentioned HMM, is applied in the work of [13].…”

Section: Related Workmentioning

confidence: 99%

CO₂-based grey-box model to estimate airflow rate and room occupancy

Wolf

Alonso²,

Calı̀

et al. 2019

E3S Web Conf.

View full text Add to dashboard Cite

In the existing building stock, heating, cooling and ventilation often run on fixed schedules assuming maximal occupancy. However, fitting the control of the HVAC system to the building’s real demand offers large potential for energy savings over the status quo. Building occupants’ presence as well as mechanically supplied and infiltrated airflow rates provide information that enables to define tailored strategies for demand-controlled ventilation. Hence, real-time estimations of these quantities are a valuable input to demand-controlled built environments. In this work, the use of stochastic differential equations (SDE) to estimate the room occupancy, infiltration air-rate and ventilation air-rate is investigated. In particular, a grey-box model based on a carbon dioxide (CO2) mass balance equation is presented. The model combines knowledge about the physical system with statistical, data-driven parameter estimation. Furthermore, the proposed model contains uncertainty parameters. This is in contrast to purely deterministic models based on ordinary differential equations, where uncertainty is usually disregarded. The suggested model has been tested in a naturally ventilated and in a mechanically ventilated environment; the performance in these two cases has been compared. We show that the ability to address measurement errors and non-homogeneous conditions in the room air implies that the suggested SDE-based grey-box approach is suitable in the context of demand-controlled ventilation.

show abstract

A methodology based on Hidden Markov Models for occupancy detection and a case study in a low energy residential building

Cited by 96 publications

References 19 publications

Critical review and illustrative examples of office occupant modelling formalisms

Critical review and illustrative examples of office occupant modelling formalisms

Sensor-based occupancy detection using neutrosophic features fusion

CO₂-based grey-box model to estimate airflow rate and room occupancy

Contact Info

Product

Resources

About

A methodology based on Hidden Markov Models for occupancy detection and a case study in a low energy residential building

Cited by 96 publications

References 19 publications

Critical review and illustrative examples of office occupant modelling formalisms

Critical review and illustrative examples of office occupant modelling formalisms

Sensor-based occupancy detection using neutrosophic features fusion

CO2-based grey-box model to estimate airflow rate and room occupancy

Contact Info

Product

Resources

About

CO₂-based grey-box model to estimate airflow rate and room occupancy