Available online at www.sciencedirect.com
ScienceDirect
Procedia Environmental Sciences 00 (2016) 000–000
www.elsevier.com/locate/procedia
Improving Sustainability Concept in Developing Countries
Implementation of Household’s Amenity Maintaining System
Based on Behavior Estimation
Shoki Kawanoa, Tomoya Imanishia, Yasushi Ikedab, Hiroaki Nishia, Eiko Uchiyamab
Graduate School of Science and Technology, Keio University,{ kawano,imanishi}@west.sd.keio.ac.jp, [email protected],
3-14-1 Hiyoshi, Kohoku, Yokohama, Kanagawa 223-8522, Japan
b
Graduate School of Media and Governance, Keio University, [email protected],[email protected],
5322 Endo, Fujisawa, Kanagawa 252-0082, Japan
a
Abstract
Research of residential EMS (Energy Management System) is mainly being promoted with a focus on energy saving initially.
After that it began to focus strikes the balance between the indoor comforts of the occupants and energy saving. Moreover, the
control object is diversified in the home, there is a need for more detailed parameters of individuals in order to satisfy the
comfort of the individual. This paper present a novel method to estimate behaviours of the resident in a house. In addition, it is
to provide comfortable environment at the same time energy saving by using the estimation results.
© 2016 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of IEREK, International experts for Research Enrichment and Knowledge Exchange.
Keywords: EMS; HEMS; behaviours estimation;
1. Introduction
Energy saving is currently one of the main issues in the world. Many countries face various energy challenges due to limited
fossil fuels and to growing electric consumption. In case of Japan, EMS (Energy Management System) attracts more attention
because of the increasing demand for efficient energy management, especially after the Great East Japan Earthquake in 2011.
From the aspect of responsivity and energy saving, levelling peak load through air-conditioning control is widely studied in
residential EMS. Furthermore, home automated system including lightings and windows are considered to improve the overall
comfort of the indoor environment1, 2. Exiting control system of home appliances (e.g. air-conditioner, windows, and lights) apply
motion sensor to detects motion of human body3. Hence, integration of motion detection with location measurement can enhance
the control system.
This paper present a novel method to estimate behaviors of the resident in a house. Indoor information about the location and
the amount of activity, are utilized for the behavior prediction. Moreover, a control system of home appliances is enhanced by the
estimated information. The position of a person is detected by using BLE (Bluetooth Low Energy). The proposed indoor positioning
system exploits trilateration that takes account of three independent measurements from beacons to a smartphone. Besides, activity
information is measured by using a gyro-sensor and a three-axis acceleration sensor mounted in the smartphone.
The presented behavior estimation and control system are applied to the experimental house placed at Keio University in Japan.
Results are validated through a series of questionnaire gathered from residents and the proposed control method proved an
improvement of indoor amenity over the non-control method.
1878-0296 © 2016 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of IEREK, International experts for Research Enrichment and Knowledge Exchange.
2
Author name / Procedia Environmental Sciences 00 (2016) 000–000
This paper comprises seven sections. In Section 2, introduce EMS. In Section 3, explain the Behavior Estimation system. In
Section 4, introduce the experimental house. In Section 5 explain in advance experiment. In Section 6 and 7 describe the
experimental method and results. Finally, conclusions are presented in Section 8.
2. Energy management system
There is an energy management system EMS (Energy Management System) as an effective means for leveling of power demand.
EMS monitors energy consumption of the customer and is a management system that controls electrical equipments. Although
EMS is the primary purpose of reducing the electric charges, it is also the environment-oriented system to achieve optimal
management of the energy demand. For example, a power consumption monitoring systems and demand control system 4, 5. The
demand control system monitors power consumption of the customer. It also for controll electrical devices as needed. We have a
control target mainly air conditioning and refrigeration. If the power consumption likely to exceed a predetermined demand value,
EMS controls the outdoor unit and pump of electrical equipment and reduce the power consumption. In particular, the EMS built
for the household is colled HEMS.
2.1. Home energy management system
The purpose of HEMS (Home Energy Management System) the EMS specializing in home is mainly in the monitoring of the
indoor environment and energy usage. Furthermore HEMS achieve energy saving controls the equipment. In general, HEMS is
composed of a sensor, controller and home appliances are controlled. Local information network in which these devices are
connected is referred to as HAN (Home Area Network). Studies have been conducted on a wide range of devices as the control
object to the home appliances, air conditioners and EV in HAN 6.
Housing such as equipment has been introduced in Japan is called Smart House, demonstrated experiments has been conducted
in various places. For example, Sekisui House was demonstration of smart house with solar panels and electric cars in Yokohama,
Kanagawa Prefecture7. In addition, Mitsubishi Electric built the smart house that enables device control of a variety of housing
facilities in Kamakura, Kanagawa Prefecture2010-20148. Moreover, Tokyo Gas build a housing complex rather than the House
and collect the data of the hot water system used solar heat, and fuel cell system9.
Additionally, Toyota home and Misawa home sell a smart home that implements the HEMS, Several electrical manufacturer
also sells HEMS10, 11.
3. Behavior estimation system
Initially, research of HEMS is mainly being promoted with a focus on energy saving. Then it began to focus strikes the balance
between the indoor comforts of the occupants and energy saving. Subsequently, it is found that the comfort is that there is a case
in which thermal comfort is impaired by individual difference and temperature variation of indoor thermal sensation. Moreover,
the control object is diversified in the home, there is a need for more detailed parameters of individuals in order to satisfy the
comfort of the individual. For example the control of the lighting requires the position information of the person, the brightness
should be changed by the actions. Therefore, it estimates the behavior information in order to control device and energy saving.
Advances in sensing technology, human behavior estimation techniques are beginning to progress. Conventional methods shall
be installed cameras and expensive sensors in advance and user should wear a special device. Therefore, privacy issues in the case
of the camera, the cost of the problem in the case of sensors and the convenience of the problem can be mentioned in the case to
be attached to the user. There are studies behavior estimation using a smartphone, but not yet the outcome of satisfaction in terms
of precision. Consider the accuracy is improved if acquires position information utilizing the room characteristics.
In the proposal method, we use a smartphone and estimate Location information and amount of activity. It is estimated the
relative positional relationship between a smartphone and beacon by using a beacon that transmits Bluetooth signals (BLE) is set
to the living room12. In addition, we estimate the behavior of the occupants by combining the activity data and the result of
smartphone location. Finally, it will be the provision of domestic comfortable environment by HEMS tailored to each of the action.
3.1. Location information
We build a location estimation system using BLE and smartphones to obtain the location information of the resident. BLE is
one of the extended specifications for short-range wireless communication technology, it has become possible to communicate
with a very low power energy as compared to previous Bluetooth.
Author name / Procedia Environmental Sciences 00 (2016) 000–000
3
In the proposed method, smart phone is used as a receiver. MyBeacon® Pro MB004 Ac is used as originating terminal.
MyBeacon is inexpensive and it is easy to set up for battery driven by a compact and lightweight.
Fig.1 MyBeacon pro MB004 AC
Fig.2 smart phone (iPhone)
We use the iPhone as a beacon receiver. By utilizing the radio wave intensity from the beacon to estimate the relative distance
between the beacons. By using this feature, to estimate whether the resident with iPhone is present in any area of the indoor.
Measurement interval is set to every second, iOS application is implemented in Objective-C, and the server side was implemented
in PHP and JavaScript. Detailed beacon placement method and the measurement method in the experiment environment, we will
be described in detail later.
3.2. Activity information
We build a behavior estimation system to know the position and behavior of the residents in order to solve the problem such as
the temperature gradient in the thermal sensation and in the same room with individual differences. It is performed using a smart
phone with a built-in gyro sensor and three-axis accelerometer to measure the residents Activity in order to estimate action13. In
this research, the Mets value is used as an indicator of the amount of activity.
While it is iPhone to use as a device that receives the transmission of beacons, also used as a device for measuring the amount
of activity at the same time in this study. The iPhone has a built-in gyro sensor and three-axis acceleration sensor, it is possible to
obtain the value of the sensor from the application. Based on the obtained value, the application to estimate the amount of activity
(Mets value) is implemented in Objective-C. Version of the OS is using iOS7.1.
3.2.1. METs (Metabolic equivalents)
METs is the index, represented by several times as compared to resting energy consumption during physical activity. It is used
as an indicator of the strength of the physical activity required to primarily estimate the physical activity level. METs is as it follows,
Table 1 METs value of each action14
4. Experiment environment
In this research, we constructed behavior estimation system in smart house is called "Co-Evolving-House" which was built in
the Keio University Shonan Fujisawa Campus. Total floor area is 78.79 m2. “Co-Evolving-House” is a new experimental house
4
Author name / Procedia Environmental Sciences 00 (2016) 000–000
that industry, government and academia has created in cooperation 15. This house actively incorporate renewable energy from
natural to realize saving. In addition, we aim to develop a house which can be smart controlled renewable energy by using the new
information technologies.
In addition to general home appliances such as air conditioners and lighting equipment, housing is equipped electric blinds, PV
and hot and cold water circulation air conditioning system. These facilities run by the sensing value by the sensor network.
Furthermore, these facilities is centrally managed by web application.
4.1. Control system
In this experiment, we controllable of various devices through an external network using a built-terminal. Major equipment
(Controlled Objects) are electric blinds, hot and cold water circulation air conditioning system, lighting and air conditioning. These
devices can be controlled from an external network is connected to a Linux PC that is called raspberry pi. In addition, these devices
can be easily controlled through the WEB interface.
4.2. Sensor network
We use wireless sensor (SW-4XXX series) manufactured by Seiko Solutions for accurate monitoring of the indoor environment.
This series are using the 920MHz band which is excellent in radio wave propagation characteristics. Since frequency channels is
the maximum 46 channels and 100 units of the wireless device per channel can be connected, we can construct a flexible sensor
networks.
The various sensors (temperature, humidity, wind speed, motion and CO2) was installed in order to get the environmental
parameters. The installation location is shown below as Fig3.
Obtaining the power consumption further by using a commercially available HEMS terminal. We used Perl in communication
between the terminal and the server HEMS.
Fig. 3 Sensor arrangement in “Co-Evolving-House”
Author name / Procedia Environmental Sciences 00 (2016) 000–000
5
4.3. Web application
We were building an application, such as can be operated in a batch equipment that comes with experimental house. So as not
to accept commands from other residents to implement login system user ID and password.
(I) State selection of the control target
The user can set the status of the control target device in the house.
(II) Visualization of environmental information
We assume that the presence of a user wishing to detailed information about the environment, such as CO2 and temperature
and humidity and constructed visualization interface. It is possible to browse to select the past data. In addition, it can also be
displayed by selecting the position of the sensor.
We used PHP to certify users and to retrieve information from database.
Finally, it shows an overview of the experimental house and a system configuration diagram.
Fig.4 The appearance of the house
Fig.5 System configuration
5. Preliminary experiment
There is a need for more detailed parameters of individuals in order to satisfy the comfort of the individual. Therefore, it is need
for us to estimates the behavior information in order to control device and energy saving. In this section, we consider how the
system can be an action estimated in experimental house through the preliminary experiment.
5.1. The beacon arrangement in “Co-Evolving-House”
To determine the deployment plan of beacon was performed preliminary experiments on distance and radio wave strength
(RSSI value). We show the relationship is obtained between distance and radio wave strength of the beacon and the receiver as
shown Fig6. The x-axis indicates the distance between the smartphone and the beacon and the Y-axis indicates radio field strength.
This graph shows, the error of the actual distance and estimated distance increases, when there is a distance of more than 2m
between the beacon and the smartphone. When more than 2m graph becomes flat, there is a possibility that vary greatly by the
difference between the estimated distances.
In this study, beacons and smartphones are arranged in a grid shape so as not to perform the estimation with a distance more
than 2m. In addition, it was set up to support the value of the coordinates (x, y) as shown Fig7.
In addition, it is considered to be defined to some extent the behavior in the home by location attribute of the room. For example,
the kitchen area is likely to have taken action, such as cooking and cleaning. On the contrary, we considered less likely to have
taken action, such as exercise or study. Therefore, the position estimation of the accuracy required to perform an action estimate is
considered that it is necessary to determine the attributes of the area. This time, we divided this house into the following six areas
of the room that is entrance, living room, dining room, bed room, kitchen and bath room as shown in Fig7. Residents is estimated
by there or position estimation of the program in any of the six areas.
6
Author name / Procedia Environmental Sciences 00 (2016) 000–000
Fig. 6 The distance between the beacon and the smartphone
Fig. 7 Area division in “Co-Evolving-House”
5.2. Distance estimation method
In this study, we performed a three-point measurement by receiving a radio wave of the nearest three beacon receiver to estimate
the coordinates of the smartphone. Specifically, we obtain the distance to the smartphone from all beacons that are arranged on a
grid and select the nearest three beacon. Next we triangulate using three beacons that is chosen and obtain the relative position
from each of the beacons. Lastly, smartphone’s coordinate is estimated based on the absolute coordinates of the selected beacon.
Applications for estimating is implemented the objective-C and communication portion of the terminal and the server was using
PHP.
5.3. Activity information
It was carried out preliminary experiments for the purpose of verifying whether the amount of activity estimation can be
performed properly in accordance with the present method. Preliminary experiments. We detected the amount of activity of the
four actions (sleeping, cooling, studying and eating) in preliminary experiments. Also a detection target in four of the action this
experiment. Experimental participants were the detection target behavior for 3 minutes. The results are shown in the Fig.8.
As a result actually found to have remained similar value as the activity amount of size that are published from, it can be said
to have correctly determined. However, since slightly higher than the value which has been published, the threshold values of the
experiment is determined in this experiment.
Author name / Procedia Environmental Sciences 00 (2016) 000–000
7
Fig. 8 METs value of each detection behavior
6. Experiment
The purpose of this paper is to estimate the action by measuring METs and location information. In addition, it is to provide
comfortable environment tailored to act at the same time energy saving. Therefore, we recruited experimental participants and
measured the positional information and activity level and estimated behavior estimation. At the same time, the equipment in the
room was controlled to match the behavior.
We show the comparison experimental conditions and the detail experimental content. Men and women in their 20s participated
in the experiment, this experiment was conducted in December 2014. METs values and area condition for action estimation is set
based on preliminary experiments and reference data. (Table2)
Table 2 Detection actions and detection conditions
Detection behavior
sleep
cook
study
eat
Conditions of METs value
1.30 and less
1.50 and more
1.40 and less
1.40 and less
Area conditions
Bedroom
Kitchen
Dining room
Dining room
The first, we explained how to operate of equipment and indoor environment for experiment participates. Next, experiment
participants to act in accordance with the shift. The contents of shift is the order, such as (I)Study in the dining, (II)Cooking in the
kitchen, (III)Diet in the dining and (IV)Sleep in the bedroom. Time of each action is 30 minutes, across the rest of 3 minutes
between each action. It is defined as a set. (Fig.9)
They acted first set is the absence of control, the second set is the presence of control. Finally we asked them a questionnaire
for assessment of the comfort of the environment.
We will be a detection target a basic behavior of residents as sleep, eat, light work (study) and housework (cook). METs values
are used those measured previously by using the amount of activity estimated application that was developed.
While “study” and “eat” of the detection conditions are the same, which is determined by the results before and after behavior
estimate. In particular, when there is a "cook" to detect behavior within estimated before 40 minutes and they exist in the dining
room with METs values below 1.40, the detection action is "eat", the other is a study.
8
Author name / Procedia Environmental Sciences 00 (2016) 000–000
Fig. 9 Move route by behavior shift
The specific control content that has been carried out to provide a comfortable environment within the room in each action are
as follows.
(I)Behavior estimation result is "studying"
・Provide ventilation spaced window if the carbon dioxide concentration is greater than 600ppm, to close the window where it
falls below.
・Make the LED lighting in white
・Start the air conditioner in the living
(II) Behavior estimation result is "cooking"
・Make the LED lighting in orange
・Start the air conditioner in the living
(III) Behavior estimation result is "diet"
・Make the LED lighting in orange
・Start the air conditioner in the living
(IV) Behavior estimation result is "sleep"
・Move down the blind of bedroom
・Start the air conditioning bedroom side only, I cut the living room side air conditioners
We summarized control as follows Table3
Table 3 Detection behavior and control content
Behavior
LED lights
Air conditioners
Electric brinds
Ventiration window
sleep
cook
study
eat
off
bulb color
daylight color
bulb color
only bedroom side
moderation
moderation
only libingroom side
close
no control
no control
no control
no control
no control
Depending on the CO2 concentration
no control
The start of the experiment during the initial state of each device is living Room and bedroom air conditioner was running, LED
lighting set on, all blinds were opened and ventilation window was closed. Air conditioning settings, mode heating, air flow is
normal set temperature was 24°. Some parts of the described as "somewhat conservative operation" drove the set temperature as
20°. Comparing the difference in power consumption when it is multiplied with not multiplied by the control. At that time, to
perform a questionnaire as feedback to confirm whether it is secured comfort.
Author name / Procedia Environmental Sciences 00 (2016) 000–000
9
It was conducted a survey in the experiment participants so controlled by the action estimated shall satisfy.
Questionnaire method is as follows,
(I)The ease of behavior (behavioral comfort)
・comfortable environment to act (cooking, study, sleep and diet).
(II) Brightness and color of the lighting
・proper lighting environment for action.
(III) Simplicity of the system operation (Ease of use of the remote control web application that was developed)
・Whether the operation is simple.
(IV)Thermal environment
・Thermal environment (temperature, humidity) is comfortable.
7. Results and discussion
7.1. The accuracy of the behavior estimation
It shows a table that is described the percentage of correct answers whether it could the same action as the schedule. The correct
answer rate of each behavior estimate is as it follows “sleep” is 100%, “cook” is 94.3%, “study is 76.5 and “eat” is 78.3%.This is
a result that if the detection action and the actual behavior were matched was examined every one minute, and records the correct
answer rate of about 80%. Further, if detected continued for several minutes, it was fully consistent with real action.
7.2. The evaluation of the amount of power consumption
Proposed control results were achieved approximately 16% power reduction compared to the case without the proposed control.
This result is obtained by averaging the consumption of 30 minutes of the nine experiment participants. (Fig.10)
Reason for the reduced power, the effect of control of the lighting and the air conditioner based on the behavior estimation result.
When don’t use the proposed control, the experiment participants did not hardly control the lighting and air conditioning conditions
at the start of the experiment, regardless of the action. Since it was prepared an initial state such as not to be frustrated, it is
considered that did not action for environmental control in this experiment.
Among the proposed control was controlled air conditioners and lighting, depending on the action. For example, when it detects
an action studies was a modest air conditioning because not moving. It was reduced the power consumption at the same time make
a comfortable environment. In addition, using the location information of the behavior estimation, it was running only air
conditioning there are experimental participants. We think that used such a control reduce energy consumption.
Next, it shows a figure of a power consumption reduction of each device follow as Fig.11.
Fig. 10 Comparison of the 30 minutes of the integrated
Fig. 11 Comparison of the two hours of accumulated
power consumption
power consumption (per device)
It is understood that it is possible to reduce power consumption in lighting and air conditioning which is controlled.
10
Author name / Procedia Environmental Sciences 00 (2016) 000–000
7.3. Questionnaire result
Questionnaire about comfort for the behavior as follows Fig12.
The results, proposed control was approximately 1% higher rating as compared with the case where no control. (The average
four questions) This means that the proposed system is considered to be able to provide a comfortable environment for the behavior
of residents.
In the proposed control, indoor environment changes automatically depending on the action in situations where residents do
not think the complaints. Opinion was good in that regard. For example, the experiment participants wants to white light during
the study, but it may be cumbersome to manipulate. It had opinion of the convenient change automatically at this time.
Next, we describe the questionnaire evaluation. (Lighting and color of the lighting, simplicity of the system operation and
thermal environment)
Fig. 12 Questionnaire average behavioural comfort
Fig. 13 Evaluation values for the thermal environment, operation and lighting
For thermal environment, as compared with the case of not performing the proposed control and control to obtain an evaluation
unchanged almost. It means that experience of the experimental participants in both cases did not change. Without operating the
air conditioner, if we did not control when observed the experiment participants, both the living room and the bedroom conditioner
was running on the same state as the initial state. On the other hand the proposed control can have the original figure to position
and behavior of residents. Therefore, while maintaining the satisfaction with the thermal environment, it can be said to have been
able to make a high power saving of control.
The questionnaire about simplicity of system operation proposed controls as compared with the case where it is not the control
was approximately 1% higher rating on average. System operation took using the WEB interface that we have built. The system
has been easy to operate. However, the experimental participants had never to almost operation. In result of the interview, most of
the experimental participants did not do the operation because there was no complaint against the environment.
The results of the lighting, the proposed method has become particularly high evaluation as compared to the case where it did
not control. We think that lighting environment is easily felt to visual change.
8. Conclusion
In this study, we have a combination of the amount of activity information and indoor location information, and estimated
behavior of resident. It proposed a HEMS to provide a comfortable environment for the action in the room on the basis of the
estimation result. We constructed system in "Co-Evolving-House" which was built in the Keio University Shonan Fujisawa
Campus. Proposal HEMS, the visualization of environmental information and power information through the WEB application
and an environmental control function based on indoor location estimation and amount of activity measurement and behavior
estimation of resident.
As a result of experiments proposed control in HEMS environment, while maintaining or improving the satisfaction with the
environment in the room by controlling the various home appliances conforming to resident information. At the same time, It was
reduced the power consumption.
From the above results, the proposed EMS concluded that it is valid with respect to the provision of indoor comfort environment
were obtained.
As future issues to enhance the comfort of the residents, we will use a machine learning to consider the preferences of the
residents of the environment, and consider the temperature gradient of the part that there is a resident.
Author name / Procedia Environmental Sciences 00 (2016) 000–000
11
References
1. K. Kuzuhara and H. Nishi (2013) “Accurate indoor condition control based on PMV prediction in BEMS environments”, Industrial Electronics Society, IECON
2013 - 39th Annual Conference of the IEEE, p. 8142 – 8147.
2. Qela B and Mouftah H.T (2014) “Peak Load Curtailment in a Smart Grid Via Fuzzy System Approach”, IEEE Trans Smart Grid, vol.5, no 2, p.761-768
3. H.Tanaka, Y. Murakami, M Terano (2012) “An empirical study on reduction of electric power consumption for lighting, ventilation fan and Air-Conditioner
with motion sensor control: Part1-Effect of energy conservation on application of the control to laboratory in campus”, Architectural Institute of Japan, p.12811282
4. D. P. Chassin (2010) “What can the smart grid do for you? And what you can do for the smart grid?”, The Electricity Journal, vol.23, p.57-63.
5. M. Naphade, G. Banavar, C. Harrison, J. Paraszczak and R. Morris (2011) “Smarter Cities and Their Innovation Challenges”, IEEE Computer Society, vol.44,
no.6, p.32-39.
6. Y. Koike, T. Ishigure and E. Nihei (1995) “High-bandwidth graded-index”, JOURNAL OF LIGHTWAVE TECHNOLOGY, vol.13, no.7, p.1475-1489.
7. T. Akimoto (2013) “Smart House "kankankyo"”, Solar energy, vol.39, no. 2, p.27-30.
8. N. Kushiro, Y. Ito (2012) “Overview of Ofuna Smart House and its Key Technologies”, Mitubishidenkigiho, vol.86, no.2, p122-125.
9. N. Enomoto (2012) “Isogo Smart House Demonstration Experiment”, Clean energy, vol.21, no.10, p.48-51.
10. “Toyota home’s smart house” http://www.toyotahome.co.jp/smarthouse/
11. R Matsubara (2014) “Development of HEMS corresponding distribution panelboards “smart Cosmo””, Electrical equipment Society, vol.34, no.10, p.746-747.
12. A. Uehara, iBeacon Handbook, 2014.
13. M. Ito and H. Nishi (2013) “Real-time METs estimation for effective indoor amenity control in HEMS”, IEEE International Symposium, Industrial Electronics
(ISIE).
14. “Compendium Physical Activities”, Arizona State University Healthy Lifestyles Research Center,
https://sites.google.com/site/compendiumofphysicalactivities/.
15. “Co-Evolving-House” http://web.sfc.keio.ac.jp/~s12884ry/beta/?p=769..