No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Data-Driven Models for Building Occupancy Estimation
Abstract
The availability of accurate occupancy information from different spaces in a building allows for significant reduction in the energy consumption of heating, ventilation, air conditioning, and lighting systems. This paper investigates the application of particle filters and time series neural networks to inferring the number of occupants of individual rooms from time series data collected by a set of occupancy-indicative sensors. Our approach is purely data driven and does not require developing customized and complex physics-based models to predict the occupancy level of the many rooms in a building. We evaluate the efficacy of the proposed methods on two data sets, one contains measurements of dedicated sensors while the other one contains measurements of HVAC sensors that are commonly available in commercial buildings. Our results indicate that time series neural networks are superior in this application, estimating the number of occupants with a root-mean-squared error of 0.3 and 0.8 in the two data sets with a maximum of 7 and 67 occupants, respectively.
... In addition to latter widely researched OB modeling approaches, several studies explored the potential of fuzzy logic modeling [50], neural networks [65,31,76,77], and K-nearest neighbor (K-NN) [65,31] and k -means clustering [45]. Peng et al. [65,31] modeled HVAC control using k-NN and neural networks. ...
... Neural networks for OB modeling are also researched by Golesten et al. [76] and Arkadian et al. [77]. ...
... Peng et al.[65,31,114] developed a framework for occupant-centric HVAC control that is evaluated for energy savings in an university building in Singapore. Park et al.[101] evaluated the smart lighting control in multiple oces in Austin, while the Golesten et al.[76] included an occupancy estimation framework in two dierent building control testbeds. As a consequence, from a very limited number of existing studies with eld implementation, there are little normative guidelines and policies on including the occupants in BPS schemes as well as standardized evaluation protocols and implementation procedures. ...
... This data can be collected by sensors that commonly exist in buildings (passive infrared, carbon-dioxide, water and electricity meters, etc.) or acquired through the Building Management System (BMS). These techniques fall into two categories: the ones that are suitable for detecting the binary occupancy state of a space (i.e., occupied and vacant) [3,18,31,34,46], and those that can determine the exact number of occupants [7,14,26,32]. We collectively refer to both as occupancy detection techniques. ...
... Data Set E includes data from a residential building in KIT's Energy Smart Home Lab [29]. It was previously used for occupancy detection in [32]. Three types of sensors were installed in this building, monitoring the Volatile Organic Compounds (VOC) concentration, the number of Bluetooth Low Energy (BLE) key fobs in the range of a BLE receiver, and the number of connected network devices (NW) every 10 seconds. ...
... It is a variant of Sequential Monte Carlo (SMC) where particles (or samples) are used to estimate the underlying hidden state sequence. It has been previously used for binary occupancy detection [19,41] and for occupancy count determination [32], given measurements of multiple sensors. ...
Recent years have witnessed a steady increase in the number of occupancy detection algorithms and people counting systems designed for residential and commercial buildings, yet comparing the accuracy of existing solutions has been impossible to date due to the lack of publicly available test data sets, open-source implementation of the state-of-the-art algorithms, and consensus on the evaluation metrics. This paper addresses this problem by presenting the design and implementation of an open-source toolkit for occupancy detection. ODToolkit is capable of importing and converting sensor data acquired from various buildings into a common data format, provides implementation of a broad suite of data-driven occupancy detection techniques, and calculates a set of evaluation metrics for each experiment. We present several case studies to show how this toolkit facilitates the development of new occupancy detection algorithms. In particular, we extend this toolkit by implementing novel domain-adaptive occupancy detection algorithms and compare them with the benchmark supervised learning algorithms on multiple data sets. Furthermore, we investigate what sensing modalities and precision are needed to achieve a desired level of accuracy for occupancy estimation through sensor fusion. ODToolkit code and documentation are available at https://odtoolkit.github.io/.
... Ref. [7] tries the Naive Bayes Classifier to predict if there is any occupancy in the room as well as the occupancy count up to three. Other efforts [8][9][10][11][12] focus on counting any number of occupants in the rooms. These works utilize the data from different sources such as CO 2 , VOC (Volatile Organic Compounds), room temperature, air flow, humidity, reheat etc. sensors and count up to any number of occupants in the room leveraging the Particle Filters and time series neural network [8], general time series analysis algorithms [9] , Seasonal Decomposition [10], J48 [11], KNN, RF, MLP, LDA [12,13], and regression model [14]. ...
... Other efforts [8][9][10][11][12] focus on counting any number of occupants in the rooms. These works utilize the data from different sources such as CO 2 , VOC (Volatile Organic Compounds), room temperature, air flow, humidity, reheat etc. sensors and count up to any number of occupants in the room leveraging the Particle Filters and time series neural network [8], general time series analysis algorithms [9] , Seasonal Decomposition [10], J48 [11], KNN, RF, MLP, LDA [12,13], and regression model [14]. Some works [15,16] suggested that only temperature data is not sufficient for predicting accurate occupancy pattern but they drive their work in supervised approach. ...
We consider the problem of inferring the mobility map, which is the distribution of the building occupants at each timestamp, from the temperatures of the rooms. We also want to explore the effects of noise in the temperature measurement, room layout, etc. in the reconstruction of the movement of people within the building. Our proposed algorithm tackles down the aforementioned challenges leveraging a parameter learner, the modified Least Square Estimator. In the absence of a complete data set with mobility map, room and ambient temperatures, and HVAC data in the public domain, we simulate a physics-based thermal model of the rooms in a building and evaluate the performance of our inference algorithm on this simulated data. We find an upper bound of the noise standard deviation (<= 1F) in the input temperature data of our model. Within this bound, our algorithm can reconstruct the mobility map with a reasonable reconstruction error. Our work can be used in a wide range of applications, for example, ensuring the physical security of office buildings, elderly and infant monitoring, building resources management, emergency building evacuation, and vulnerability assessment of HVAC data. Our work brings together multiple research areas, Thermal Modeling and Parameter Estimation, towards achieving a common goal of inferring the distribution of people within a large office building.
... The work by Chen et al. [24] combines CNNs and bi-directional LSTM to predict different classes of occupancy. The data in [31] incorporate HVAC sensory data with indoor climate sensory data to infer the number of occupants using particle filtering and neural networks. In a similar manner, Li et al. [23] tackle DCV by proactively predicting CO 2 concentrations. ...
Residential and industrial buildings are significant consumers of energy, which can be reduced by controlling their respective Heating, Ventilation, and Air Conditioning (HVAC) systems. Demand-based Ventilation (DCV) determines the operational times of ventilation systems that depend on indoor air quality (IAQ) conditions, including CO2 concentration changes, and the occupants’ comfort requirements. The prediction of CO2 concentration changes can act as a proxy estimator of occupancy changes and provide feedback about the utility of current ventilation controls. This paper proposes a Hierarchical Model for CO2 Variation Predictions (HMCOVP) to accurately predict these variations. The proposed framework addresses two concerns in state-of-the-art implementations. First, the hierarchical structure enables fine-tuning of the produced models, facilitating their transferability to different spatial settings. Second, the formulation incorporates time dependencies, defining the relationship between different IAQ factors. Toward that goal, the HMCOVP decouples the variation prediction into two complementary steps. The first step transforms lagged versions of environmental features into image representations to predict the variations’ direction. The second step combines the first step’s result with environment-specific historical data to predict CO2 variations. Through the HMCOVP, these predictions, which outperformed state-of-the-art approaches, help the ventilation systems in their decision-making processes, reducing energy consumption and carbon-based emissions.
... Combined with machine learning, one can use CO 2 data to estimate accurately the number of people within a closed building space [3]. The historical data can then be used to predict the occupancy and adapt the LHVAC system accordingly, thus optimizing the energy performance and shave the consumption peaks [3,21,24,42]. In most of the CO 2 based occupancy estimation, the authors propose either a supervised or semi-supervised domain adaptation method as a way to estimate the number of people within one room. ...
... In their approach to determine occupancy using single passive infrared sensor combined with machine learning techniques Raykov et al. [3] proposed a low-cost occupancy estimation solution that produced a mean absolute error (MAE) of 1, but was tested only in rooms with 14 or less occupants. Golestan et al. [4] developed time series neural networks to estimate the number of room occupants with a RMSE of 0.8 for rooms with maximum 67 occupants. They used a set of occupancy indicative sensors including BLE (Bluetooth Low Energy) beacons. ...
Universities worldwide are experiencing a surge in enrollments, therefore campus estate managers are seeking continuous data on attendance patterns to optimize the usage of classroom space. As a result, there is an increasing trend to measure classroom attendance by employing various sensing technologies, among which pervasive WiFi infrastructure is seen as a low-cost method. In a dense campus environment, the number of connected WiFi users does not well estimate room occupancy since connection counts are polluted by adjoining rooms, outdoor walkways, and network load balancing. This paper develops machine learning-based models, including unsupervised clustering and a combination of classification and regression algorithms, to infer classroom occupancy from WiFi sensing infrastructure. Our contributions are three-fold: (1) We analyze metadata from a dense and dynamic wireless network comprising of thousands of access points (APs) to draw insights into coverage of APs, the behavior of WiFi-connected users, and challenges of estimating room occupancy; (2) We propose a method to automatically map APs to classrooms and evaluate K-means, Expectation-Maximization (EM-GMM) and Hierarchical Clustering (HC) algorithms; and (3) We model classroom occupancy and evaluate varying algorithms, namely Logistic Regression, Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), Linear Regression (LR) and Support Vector Regression (SVR). We achieve 84.6% accuracy in mapping APs to classrooms, while our estimation for room occupancy (with symmetric Mean Absolute Percentage Error (sMAPE) of 13.10%) is comparable to beam counter sensors.
... On the basis of many Internet of Things (IoT) scenarios, particularly smart buildings, smart cities, and Human Activity Recognition (HAR) solutions, we need systems for automatically detecting people's presence, estimating their number, and measuring interpersonal distance [1]- [4]. ...
... To make possible higher energy and cost savings without compromising comfort, most control techniques incorporate occupant presence or count information at the building level. This makes sense because estimating the number of occupants in each zone is difficult without having a number of sensors installed there [16,37]. Should this information become available, the thermal and visual discomfort can be calculated for each individual occupant that is present in a given zone. ...
... In their approach to determine occupancy using single passive infrared sensor combined with machine learning techniques Raykov et al. [20] proposed a low-cost occupancy estimation solution that produced a mean absolute error (MAE) of 1, but was tested only in rooms with 14 or less occupants. Golestan et al. [11] developed time series neural networks to estimate the number of room occupants with a RMSE of 0.8 for rooms with maximum 67 occupants. They used a set of occupancy indicative sensors including BLE (Bluetooth Low Energy) beacons. ...
Universities worldwide are experiencing a surge in enrollments, therefore campus estate managers are seeking continuous data on attendance patterns to optimize the usage of classroom space. As a result, there is an increasing trend to measure classrooms attendance by employing various sensing technologies, among which pervasive WiFi infrastructure is seen as a low cost method. In a dense campus environment, the number of connected WiFi users does not well estimate room occupancy since connection counts are polluted by adjoining rooms, outdoor walkways, and network load balancing. In this paper, we develop machine learning based models to infer classroom occupancy from WiFi sensing infrastructure. Our contributions are three-fold: (1) We analyze metadata from a dense and dynamic wireless network comprising of thousands of access points (APs) to draw insights into coverage of APs, behavior of WiFi connected users, and challenges of estimating room occupancy; (2) We propose a method to automatically map APs to classrooms using unsupervised clustering algorithms; and (3) We model classroom occupancy using a combination of classification and regression methods of varying algorithms. We achieve 84.6% accuracy in mapping APs to classrooms while the accuracy of our estimation for room occupancy is comparable to beam counter sensors with a symmetric Mean Absolute Percentage Error (sMAPE) of 13.10%.
... On the other hand, the integration of machine learning algorithms with automatic control systems in commercial and residential buildings proves the concept of "smart buildings" with the aim to save energy, ensure security or improve occupants' comfort [29]. As an application example in indoor human behaviour study, data-driven models have been developed for the estimation of building occupancy which can assist in emergency response flow and supporting decision making mechanism [30], [31]. Another example is a recommendation system proposed in [29] to demonstrate the use of machine learning techniques for intelligent building lighting controls that are capable of meeting the needs of both individual visual comfort and energy efficiency of the building. ...
Occupants’ comfort perception about the indoor environment is closely linked to their health, wellbeing and productivity. Improvement of comfort level in office buildings has significant positive impacts on both employers and employees. Human comfort in indoor environment usually can be assessed in four aspects: thermal comfort, visual comfort, acoustic comfort and respiratory comfort. In this paper, we present a literature review on the previous research contributions towards studying various aspects of human comfort with a special focus on the respective assessment criteria, data collection methods and data analysis approaches employed by former studies. Previous review work has covered the fundamental concepts associated with human comfort. However, their studies mainly focus on thermal comfort and there is limited work that covers other aspects of comfort. Moreover, few of them discuss how the data is obtained, how to extract useful information from the data and how the data is analyzed. To fill up this gap, this paper conducts the survey from the data-driven point of view. Through the survey, we find that sensor technology has been widely used in the data collection for various types of comfort, while so far the machine learning approaches are mainly applied in the area of thermal comfort study. Finally, some potential future research areas are proposed based on the current status of the research work. The established knowledge in this paper would provide useful insights for engineers or researchers who embark on their research in this area.
... Similarly, building occupancy is inferred in [18] leveraging the wireless networking infrastructure, and security and access control systems. More recently, measurements of the room temperature, and damper and valve position, which are parts of the HVAC system, are used to infer the occupancy state of individual rooms [3,19]. These techniques have two major shortcomings. ...
Fine-grained occupancy information is essential to improve human experience and operational efficiency of buildings, yet it is quite challenging to obtain this information due to the lack of special-purpose sensors for occupancy monitoring, and insufficient training data for developing accurate data-driven models. This paper addresses this challenge by (a) utilizing recurrent neural network models to uncover latent occupancy patterns in individual rooms from trend data available through the building management system, and (b) applying a domain adaptation technique to transfer existing occupancy models trained in a controlled environment (i.e., the source domain) to another environment (i.e., the target domain) where labelled data is sparse or non-existent. We adjust the model parameters based on the apparent differences between the two environments and apply the adapted model to estimate the number of occupants in the target domain. Our results from two test commercial buildings in two continents indicate that the adapted model yields only slightly lower accuracy than a model that is originally built on the target domain given a large amount of labelled data. Furthermore, we study how much labelled data is required from the target domain for the semi-supervised domain adaptation technique to achieve promising results.
Nowadays occupancy monitoring solutions are an intrinsic component of modern buildings to track and optimize the utilization of spaces and resources. Such solutions form a ground to the decision support of buildings management and rely on a large set of sensors that are commonly part of the LHVAC (lighting, heating, ventilation, and air-condition) systems. However, with the increasing privacy concerns it is becoming challenging to deploy seeing/hearing sensors. In this paper we propose an accurate occupancy estimator using machine learning (ML) and blind sensors to assess the actual occupancy state and predict future states. We also design a smart control system, relying on the occupancy module, to actuate the ventilation system in an energy efficient manner while maintaining the safety properties dictated by standards. In order to improve the prediction accuracy of ML models, we use the DBSCAN outlier to filter the input data and Z-Score to normalize such data. The system is fully implemented and tested on a data set of 8000 samples. The experimental results show that the prediction accuracy of our system is 96.6% with a considerable impact of the DBSCAN outlier. Furthermore, our smart control system enables an improvement of the energy performance and provides an efficient peak shaving to the energy consumption of the ventilation system.
There is increasing interest in deploying building-scale, general-purpose, and high-fidelity sensing to drive emerging smart building applications. However, the real-world deployment of such systems is challenging due to the lack of system and architectural support. Most existing sensing systems are purpose-built, consisting of hardware that senses a limited set of environmental facets, typically at low fidelity and for short-term deployment. Furthermore, prior systems with high-fidelity sensing and machine learning fail to scale effectively and have fewer primitives, if any, for privacy and security. For these reasons, IoT deployments in buildings are generally short-lived or done as a proof of concept. We present the design of Mites, a scalable end-to-end hardware-software system for supporting and managing distributed general-purpose sensors in buildings. Our design includes robust primitives for privacy and security, essential features for scalable data management, as well as machine learning to support diverse applications in buildings. We deployed our Mites system and 314 Mites devices in Tata Consultancy Services (TCS) Hall at Carnegie Mellon University (CMU), a fully occupied, five-story university building. We present a set of comprehensive evaluations of our system using a series of microbenchmarks and end-to-end evaluations to show how we achieved our stated design goals. We include five proof-of-concept applications to demonstrate the extensibility of the Mites system to support compelling IoT applications. Finally, we discuss the real-world challenges we faced and the lessons we learned over the five-year journey of our stack's iterative design, development, and deployment.
The 5G and beyond advancements will impact significantly the evolution of many vertical industries such as the Surveillance and Safety sector. To this end, the requirements posed by its associated services can be satisfied to a certain degree by 5G network slicing and by control and user plane separation -allowing for exploitation of edge computing-, while aspects related to network deployment resilience and end-to-end service orchestration of infrastructure, network and applications across edge and cloud domains are still under investigation. The 5G-COMPLETE project focuses on delivering Beyond 5G technology innovations in the edge and orchestration domains (among others), that can support highly demanding vertical services/applications and relevant use cases such as Advanced Surveillance; along with holistic 5G deployment paradigms and experimentation deployments for testing and evaluation. This paper discusses the service and technical requirements and provides an overview of the proposed technologies and deployment solutions.
Even though occupancy inference is of utmost importance for numerous real-time and real-life applications a widely-accepted approach to predict occupancy does not exist. In this paper, an assessment of widely-recommended approaches and data processing for occupancy is overviewed. Furthermore, the correlation and meta-analysis between various sensor features like motion sensing, temperature, humidity, and energy consumption were tested. Random Forest classifier a widely-applied artificial model for occupancy inference prediction is evaluated in 4 different real-life data sets including various features. The results of both a univariate and multivariate model are examined. Random Forest classifier results during an experimental phase are presented to reveal the best model. The outcomes of the current research indicate that even in similar spaces data analysis and correlation have different results while the multivariate model is more accurate than the bivariate model.
5G changes the landscape of mobile networks profoundly, with an evolved architecture supporting unprecedented capacity, spectral efficiency, and increased flexibility. The MARSAL project targets the development and evaluation of a complete framework for the management and orchestration of network resources in 5G and beyond by utilizing a converged optical-wireless network infrastructure in the access and fronthaul/midhaul segments. In this paper, we present a conceptual view of the MARSAL architecture, as well as a wide range of experimentation scenarios.
Detecting the presence of persons and estimating their quantity in an indoor environment has grown in importance recently. For example, the information if a room is unoccupied can be used for automatically switching off the light, air conditioning, and ventilation, thereby saving significant amounts of energy in public buildings. Most existing solutions rely on dedicated hardware installations, which involve presence sensors, video cameras, and carbon dioxide sensors. Unfortunately, such approaches are costly, are subject to privacy concerns, have high computational requirements, and lack ubiquitousness. The work presented in this article addresses these limitations by proposing a low-cost occupancy detection system. Our approach builds upon detecting variations in Bluetooth Low Energy (BLE) signals related to the presence of humans. The effectiveness of this approach is evaluated by performing comprehensive tests on five different datasets. We apply several pattern recognition models and compare our methodology with systems building upon IEEE 802.11 (WiFi). On average, in multifarious environments, we can correctly classify the occupancy with an accuracy of 97.97%. When estimating the number of people in a room, on average, the estimated number of subjects differs from the actual one by 0.32 persons. We conclude that our system’s performance is comparable to that of existing ones based on WiFi, while significantly reducing cost and installation effort. Hence, our approach makes occupancy detection practical for real-world deployments.
Detecting the presence and estimating the number of subjects in an indoor environment has grown in importance recently. For example, the information if a room is unoccupied can be used for automatically switching off the light, air conditioning, and ventilation, thereby saving significant amounts of energy in public buildings. Most existing solutions rely on dedicated hardware installations, which involve presence sensors, video cameras, and carbon dioxide sensors. Unfortunately, such approaches are costly, subject to privacy concerns, have high computational requirements, and lack ubiquitousness. The work presented in this article addresses these limitations by proposing a low-cost system for occupancy detection. Our approach builds upon detecting variations in Bluetooth Low Energy (BLE) signals related to the presence of humans. The effectiveness of this approach is evaluated by performing comprehensive tests on 5 different datasets. We apply different pattern recognition models and compare our methodology with systems building upon IEEE 802.11 (WiFi). On average, in different environments, we can correctly classify the occupancy with an accuracy of 97.97\%. When estimating the number of people in a room, on average, the estimated number of subjects differs from the actual one by 0.32 persons. We conclude that the performance of our system is comparable to existing ones based on WiFi, while leading to a significantly reduced cost and installation effort. Hence, our approach makes occupancy detection practical for real-world deployments.
A review of the literature pertaining to building occupancy detection, counting, and tracking – three areas under the umbrella of building occupancy estimation – is presented with a focus on mathematical approaches and corresponding metrics. The idea is to provide the reader with a background on the hardware and techniques used for occupancy inference, subsequent to data collection, with emphasis placed on the algorithmic characterization of occupancy estimation. The various approaches employed by researchers to tackle the problem are surveyed and summarized, including: data collection, cleaning processes, algorithm utilization and categorization, as well as data structuring and organization. The scope of prediction and performance metrics are used to establish a benchmarking system through a comprehensive summary of (indoor) occupancy estimation, presented in the context of the mathematical tools utilized.
Human occupancy counting is crucial for both space utilisation and building energy optimisation. In the current article, we present a semi-supervised domain adaptation method for carbon dioxide-Human Occupancy Counter (DA-HOC), a robust way to estimate the number of people within in one room by using data from a carbon dioxide sensor. In our previous work, the proposed Seasonal Decomposition for Human Occupancy Counting (SD-HOC) model can accurately predict the number of individuals when the training and labelled data are adequately available. DA-HOC is able to predict the number of occupancy with minimal training data, as little as one-day data. DA-HOC accurately predicts indoor human occupancy for a large room using a model trained from a small room and adapted to the larger room. We evaluate DA-HOC with two baseline methods-support vector regression technique and SD-HOC model. The results demonstrate that DA-HOC's performance is better by 12.29% in comparison to SVR and 10.14% in comparison to SD-HOC. CCS CONCEPTS • Human-centered computing → Ubiquitous and mobile computing design and evaluation methods;
The inter-temporal consumption flexibility of commercial buildings can be harnessed to improve the energy efficiency of buildings, or to provide ancillary service to the power grid. To do so, a predictive model of the building's thermal dynamics is required. In this paper, we identify a physics-based model of a multi-purpose commercial building including its heating, ventilation and air conditioning system during regular operation. We present our empirical results and show that large uncertainties in internal heat gains, due to occupancy and equipment, present several challenges in utilizing the building model for long-term prediction. In addition, we show that by learning these uncertain loads online and dynamically updating the building model, prediction accuracy is improved significantly.
Occupancy profile is one of the driving factors behind discrepancies between the measured and simulated energy consumption of buildings. The frequencies of occupants leaving their offices and the corresponding durations of absences have significant impact on energy use and the operational controls of buildings. This study used statistical methods to analyze the occupancy status, based on measured lighting-switch data in five-minute intervals, for a total of 200 open-plan (cubicle) offices. Five typical occupancy patterns were identified based on the average daily 24-hour profiles of the presence of occupants in their cubicles. These statistical patterns were represented by a one-square curve, a one-valley curve, a two-valley curve, a variable curve, and a flat curve. The key parameters that define the occupancy model are the average occupancy profile together with probability distributions of absence duration, and the number of times an occupant is absent from the cubicle. The statistical results also reveal that the number of absence occurrences decreases as total daily presence hours decrease, and the duration of absence from the cubicle decreases as the frequency of absence increases. The developed occupancy model captures the stochastic nature of occupants moving in and out of cubicles, and can be used to generate a more realistic occupancy schedule. This is crucial for improving the evaluation of the energy saving potential of occupancy based technologies and controls using building simulations. Finally, to demonstrate the use of the occupancy model, weekday occupant schedules were generated and discussed.
Heating, cooling and ventilation accounts for 35% energy usage in the United States. Currently, most modern buildings still condition rooms assuming maximum occupancy rather than actual usage. As a result, rooms are often over-conditioned needlessly. Thus, in order to achieve efficient conditioning, we require knowledge of occupancy. This paper shows how real time occupancy data from a wireless sensor network can be used to create occupancy models which in turn can be integrated into building conditioning system for usage based demand control conditioning strategies. Using strategies based on sensor network occupancy model predictions, we show that it is possible to achieve 42% annual energy savings while still maintaining American Society of Heating, Refrigerating and Air-Conditioning (ASHRAE) comfort standards.
Buildings are known to be the largest consumers of electricity in the United States, and often times the dominant energy consumer is the HVAC system. Despite this fact, in most buildings the HVAC system is run using primitive static control algorithms based on fixed work schedules causing wasted energy during periods of low occupancy. In this paper we present a novel control architecture that uses occupancy sensing to guide the operation of a building HVAC system. We show how we can enable aggressive duty-cycling of building HVAC systems - that is, turn them ON or OFF - to save energy while meeting building performance requirements using inexpensive sensing and control methods. We have deployed our occupancy sensor network across an entire floor of a university building and our data shows several periods of low occupancy with significant opportunities to save energy over normal HVAC schedules. Furthermore, by interfacing with the building Energy Management System (EMS) directly and using real-time occupancy data collected by our occupancy nodes, we measure electrical energy savings of 9.54% to 15.73% and thermal energy savings of 7.59% to 12.85% for the HVAC system by controlling just one floor of our four floor building.
Heating, ventilation, and air conditioning (HVAC) systems are an important target for efficiency improvements through new equipment and retrofitting because of their large energy footprint. One type of equipment that is common in homes and some offices is an electrical, single-stage heat pump air conditioner (AC). To study this setup, we have built the Berkeley Retrofitted and Inexpensive HVAC Testbed for Energy Efficiency (BRITE) platform. This platform allows us to actuate an AC unit that controls the room temperature of a computer laboratory on the Berkeley campus that is actively used by students, while sensors record room temperature and AC energy consumption. We build a mathematical model of the temperature dynamics of the room, and combining this model with statistical methods allows us to compute the heating load due to occupants and equipment using only a single temperature sensor. Next, we implement a control strategy that uses learning-based model-predictive control (MPC) to learn and compensate for the amount of heating due to occupancy as it varies throughout the day and year. Experiments on BRITE show that our techniques result in a 30%–70% reduction in energy consumption as compared to two-position control, while still maintaining a comfortable room temperature. The energy savings are due to our control scheme compensating for varying occupancy, while considering the transient and steady state electrical consumption of the AC. Our techniques can likely be generalized to other HVAC systems while still maintaining these energy saving features.
Heating, ventilation, and airconditioning (HVAC) are among the major energy demand in the buildings sector globally. Improving the energy efficiency of such systems is a critical objective for mitigating greenhouse gas emissions and transitioning towards renewable sources of energy supply. The interest of this paper is to explore means to increase the efficiency of HVAC systems in accommodating occupants' behavior in real time. For instance, rooms in office buildings are not always occupied by occupants during scheduled HVAC service periods. This offers an opportunity to reduce unnecessary energy demands of HVAC systems associated with oc-cupants' behavior. An in-depth analysis of occupants' stochastic behavior within an office building is conducted in this paper. A demand-driven control strategy is proposed that automatically responds to occupants' energy-related behavior for reducing energy consumption and maintains room temperature for occupants with similar performances as a static cooling. In this control strategy, two types of machine learning methods-unsupervised and supervised learning-are applied to learn occupants' behavior in two learning processes. The occupancy-related information learned by the algorithms is used by a set of specified rules to infer real-time room setpoints for controlling the office's space cooling system. This learning-based approach intends to reduce the need for human intervention in the cooling system's control. The proposed strategy was applied to control the cooling system of the office building under real-world conditions. Eleven case study office spaces were selected, representing three typical office uses: single person offices, multi-person offices, and meeting rooms. The experimental results report between 7% and 52% energy savings as compared to the conventionally-scheduled cooling systems.
Model predictive control is a promising approach to optimize the operation of building systems and provide demand-response functionalities without compromising indoor comfort. The performance of model predictive control relies, among other things, on the quality of weather forecasts and building occupancy predictions. The present study compares the accuracy and computational demand of two occupancy estimation and prediction approaches suitable for building model predictive control: (1) count prediction based on indoor climate modeling and parameter estimation “using common sensors”, (2) count prediction based on data from 3D stereovision camera. The performance of the two approaches was tested in two rooms of a case study building. The results show that the method with dedicated sensors outperforms common sensors. However, if a building is not equipped with dedicated sensors, the present study shows that the common sensor method can be a satisfactory alternative to be used in model predictive control.
This poster presents ongoing work on an occupancy detection system based on data fusion of various low-cost sensors within a real smart building and compares the results of different classification approaches.
Heating, ventilation, and air conditioning (HVAC) systems account for over 50% of the energy consumed by commercial buildings. While "smart" HVAC technologies, such as learning thermostats, are widely available for residential use, commercial buildings typically rely on legacy systems that are difficult to upgrade and require facility managers to manually set HVAC schedules. In this paper, we propose a novel Machine Learning-driven technique to automatically learn custom occupancy-based HVAC schedules for buildings across a large campus. While our technique is compatible with any occupancy sensor, we leverage the existing wireless networking infrastructure that is omnipresent across any modern campus. We analyze building WiFi activity, specifically from smartphones, to infer detailed spatial occupancy patterns in each building, and present an algorithm that learns from these patterns to derive a custom HVAC schedule. Our approach is adaptive and dynamically adjusts its schedules as occupancy patterns change, much like a learning thermostat. To evaluate our techniques, we analyze data from several thousand WiFi access points deployed in 112 office buildings on a university campus. Our analysis reveals significant differences in occupancy patterns across and within buildings, motivating the need for our adaptive learning-based approach. Compared to the current static approach, our results demonstrate that learning HVAC schedules from mobile WiFi activity across the campus can yield a 37% reduction in waste time, a measure of energy savings, and a 3% reduction in miss time, a measure of user comfort.
This paper presents and evaluates a simple methodology based on Hidden Markov models for the problem of unsupervised occupancy detection using the open source program R. The models were created using different environmental parameters such as temperature, humidity, humidity ratio, CO2 and light time series data and were evaluated against ground truth occupancy from a public data set. The accuracies of the models are reported. Also, as a case study, the developed methodology is applied for humidity ratio data calculated from temperature and humidity measured in different rooms (kitchen, living room, office, parents’ room, teenager’s room, laundry room, ironing room and bathroom) in a low energy residential building to infer daily and hourly average occupancy schedules for which there is no ground truth data. The estimated occupancy schedules are commented on by one of the house occupants and discussed. Inferred schedules found with this method could be useful for understanding average occupancy schedules, for detecting regular activities or actions and as an input for residential building energy simulations.
The design of energy-efficient commercial building Heating Ventilation and Air Conditioning (HVAC) systems has been in the forefront of energy conservation efforts over the past few decades. The HVAC systems traditionally run on a static schedule that does not take occupancy into account, wasting a lot of energy in conditioning empty or partially-occupied spaces. This paper investigates the application of non-intrusive techniques to obtain a rough estimate of occupancy from coarse-grained measurements of the sensors that are commonly available through the building management system. Various per-zone schedules can be developed based on this approximate knowledge of occupancy at the level of individual zones. Our experiments in three large commercial buildings confirm that the proposed techniques can uncover the occupancy pattern of the zones, and schedules that incorporate these occupancy patterns can achieve more than 38% reduction in reheat energy consumption while maintaining indoor thermal comfort.
Buildings account for 40% of US primary energy consumption and 72% of electricity. Of this total, 50% of the energy consumed in buildings is used for Heating Ventilation and Air-Conditioning (HVAC) systems. Current HVAC systems only condition based on static schedules; rooms are conditioned regardless of occupancy. By conditioning rooms only when necessary, greater efficiency can be achieved. This paper describes POEM, a complete closed-loop system for optimally controlling HVAC systems in buildings based on actual occupancy levels. POEM is comprised of multiple parts. A wireless network of cameras called OPTNet is developed that functions as an optical turnstile to measure area/zone occupancies. Another wireless sensor network of passive infrared (PIR) sensors called BONet functions alongside OPTNet. This sensed occupancy data from both systems are then fused with an occupancy prediction model using a particle filter in order to determine the most accurate current occupancy in each zone in the building. Finally, the information from occupancy prediction models and current occupancy is combined in order to find the optimal conditioning strategy required to reach target temperatures and minimize ventilation requirements. Based on live tests of the system, we estimate ~30.0% energy saving can be achieved while still maintaining thermal comfort.
A study was conducted on the lighting operation and workspace occupancy patterns across numerous commercial buildings to better quantify the performance estimates of occupancy sensors across typical space types. By examining how occupants occupy their spaces and manually control their lighting, and comparing these baselines to modeled occupancy sensor control scenarios, energy and dollars savings potentials were investigated.
Accurate occupancy information in commercial buildings can enable several useful applications such as energy management and dynamic seat allocation. Most prior efforts in this space depend on deploying an additional network of deeply coupled sensors to gather occupancy details. This paper presents a novel approach for occupancy detection using only context sources that are commonly available in commercial buildings such as area access badges, Wi-Fi access points, Calendar and Instant Messaging clients. We present models to conduct a situation-centric profiling using such sources and evaluate results of those models. Through a pilot study of a building floor with 5 volunteers for 6 weeks, we demonstrate the potential for detecting occupancies with accuracy as high as 90%.
Probabilistic robotics is a new and growing area in robotics, concerned with perception and control in the face of uncertainty. Building on the field of mathematical statistics, probabilistic robotics endows robots with a new level of robustness in real-world situations.
This book introduces the reader to a wealth of techniques and algorithms in the field. All algorithms are based on a single overarching mathematical foundation. Each chapter provides example implementations in pseudo code, detailed mathematical derivations, discussions from a practitioner's perspective, and extensive lists of exercises and class projects. The book's Web site, http://www.probabilistic-robotics.org, has additional material.
The book is relevant for anyone involved in robotic software development and scientific research. It will also be of interest to applied statisticians and engineers dealing with real-world sensor data.
The rapidly growing world energy use has already raised concerns over supply difficulties, exhaustion of energy resources and heavy environmental impacts (ozone layer depletion, global warming, climate change, etc.). The global contribution from buildings towards energy consumption, both residential and commercial, has steadily increased reaching figures between 20% and 40% in developed countries, and has exceeded the other major sectors: industrial and transportation. Growth in population, increasing demand for building services and comfort levels, together with the rise in time spent inside buildings, assure the upward trend in energy demand will continue in the future. For this reason, energy efficiency in buildings is today a prime objective for energy policy at regional, national and international levels. Among building services, the growth in HVAC systems energy use is particularly significant (50% of building consumption and 20% of total consumption in the USA). This paper analyses available information concerning energy consumption in buildings, and particularly related to HVAC systems. Many questions arise: Is the necessary information available? Which are the main building types? What end uses should be considered in the breakdown? Comparisons between different countries are presented specially for commercial buildings. The case of offices is analysed in deeper detail.
This paper provides an overview over the most common neural network types for time series processing, i.e. pattern recognition and forecasting in spatio-temporal patterns. Emphasis is put on the relationships between neural network models and more classical approaches to time series processing, in particular, forecasting. The paper begins with an introduction of the basics of time series processing, and discusses feedforward as well as recurrent neural networks, with respect to their ability to model non-linear dependencies in spatio-temporal patterns. 1 Introduction The world is always changing. Whatever we observe or measure -- be it a physical value such as temperature or the price of a freely traded good -- is bound to be different at different points in time. Classical pattern recognition, and with it a large part of neural network applications, has mainly been concerned with detecting systematic patterns in an array of measurements which do not change in time (static patterns). ...
Sensing by Proxy: Occupancy Detection Based on Indoor CO2 Concentration
- Ming Jin
- Nikolaos Bekiaris-Liberis
- Kevin Weekly
- Costas Spanos
- Alexandre Bayen
Ming Jin, Nikolaos Bekiaris-Liberis, Kevin Weekly, Costas Spanos, and Alexandre
Bayen. 2015. Sensing by Proxy: Occupancy Detection Based on Indoor CO2 Concentration. In Proc. 9th International Conference on Mobile Ubiquitous Computing,
Systems, Services and Technologies. 1-10.
KIT Energy Smart Home Lab
KIT. Accessed Jan. 2018. KIT Energy Smart Home Lab. Online https://github.
com/aifb/eshl-occupancy/. (Accessed Jan. 2018).
POEM: Powerefficient occupancy-based energy management system
- L Varick
- Stefan Erickson
- Alberto E Achleitner
- Cerpa
Varick L. Erickson, Stefan Achleitner, and Alberto E. Cerpa. 2013. POEM: Powerefficient occupancy-based energy management system. In Proc. 12th International
Conference on Information Processing in Sensor Networks (IPSN). ACM/IEEE, 203-216.
Probabilistic Robotics (Intelligent Robotics and Autonomous Agents)
- Wolfram Sebastian Thrun
- Dieter Burgard
- Fox