An Autonomous System via Fuzzy Logic for Residential ... - IEEE Xplore

An Autonomous System via Fuzzy Logic for Residential Peak Load Management in Smart Grids Azim Keshtkar Dept. of Mechatronic Systems Engineering Simon Fraser University BC, Canada [email protected]

Siamak Arzanpour Dept. of Mechatronic Systems Engineering Simon Fraser University BC, Canada [email protected]

Abstract—Residential Heating, Ventilation, and Air Conditioning (HVAC) systems can play significant role in the future smart grids in order to balance demand and supply patterns as they are the main electrical load during peak load periods. Programmable thermostats and programmable communicating thermostats are widely used for automatic control of residential HVAC systems with the aim of energy management and providing thermal comfort while users set their daily/weekly schedules and preferences. On the other hand, the programs such as Time-of-Use (TOU) rates, Real-time Pricing (RTP), and Demand Response (DR) are often applied by utilities in order to encourage users to reduce their consumption during peak load periods. However, it is often an inconvenience for residential users to manually modify their schedules and preferences based on the electricity prices that vary over time. Hence, in this paper an autonomous thermostat capable of responding to different parameters such as time-varying prices, while saving energy and maintaining user’s thermal comfort is presented. The developed thermostat is the result of integration of fuzzy logic, wireless sensors, and smart grid initiatives. To implement and validate the approach; a house simulator that represents a smart thermostat is developed in Matlab-GUI. The simulation results demonstrate the overall improvement with respect to energy saving and conservation without jeopardizing occupant’s thermal comfort. Keywords—Autonomous Systems; Smart Grid Initiatives; Fuzzy Logic; HVAC Systems; Smart Thermostats

I.

INTRODUCTION

Residential HVAC systems account for nearly half of the average household energy bill in Canada and the U.S. [1, 2]. Hence, rising electricity prices and their shifting from flat-rate tariffs to time-varying prices such as TOU and RTP will impact the consumers that their energy bills are significantly related to HVAC systems [3]. As mentioned in [4], these devices also caused approximately half of the additional critical peak electrical power usage in California. Therefore, demand-side management strategies in residential buildings particularly for HVAC systems can provide the potentials to benefit both utilities in reducing peak load demand and consumers during high electricity prices [5-8]. Programmable Thermostats (PTs) are widely used by households to control their HVAC systems automatically with the aim of managing energy consumption, saving cost, and providing thermal comfort [9]. However, being limited to a single indoor temperature and lack of communicating with existing smart meters are the major barriers of current PTs [10, 11].

Fazel Keshtkar Dept. of Computer Science Southeast Missori State University MO, USA [email protected]

Nowadays, with the advancement in wireless sensors and communication networks; PTs have been extended into programmable communicating thermostats (PCT) [10]. Currently, PCTs can communicate with smart meters in order to help in the peak load demand using demand response initiatives applied by utilities. The PCTs receive price signals in TOU or RTP programs sent by utilities through smart meters. Then, it automatically applies the specific temperature tolerances (offsets) sat by user in advance to reduce the initialized set points (load demand) based on the electricity rates at different times of a day. However, pre-set tolerances initialized by users often jeopardize residential users’ thermal comfort particularly during hot summers and cold winters [3, 4, 12]. In addition, it is often a hassle and confusing for residential users to manually respond to electricity prices that change with time [13]. Nowadays, the management of the peak load problems is being shifted more towards the demand-side (i.e., users). However, lack of knowledge among residential customers to fully react to smart grid initiatives (i.e., RTP and TOU prices) are the major problem that should be considered in demandside. in addition, most consumers do not have time to modify their devices even based on their interests or in some cases they forget to setback their devices in response to RTP [10]. Hence, the need for in-home energy management systems such as thermostats capable of making autonomous intelligent decisions while providing user comfort and saving energy and cost is necessary [14, 15]. Furthermore, the capability of wireless sensor nodes to detect and measure different variable of interests (i.e., temperature, humidity, occupancy) can improve the limitations of existing energy management systems such as thermostats [16-19]. Authors in [17] developed an optimization-based residential energy management approach using wireless sensors. They attempted to minimize the energy expenses of the residential users in which they schedule home devices to off-peak hours according to the TOU rates. However, their approach cannot handle RTP programs. In addition, their scheduling approach shifts residential loads to off-peak hours. As a result, their proposed approach results in increasing peakto-average ratio (PAR) in demand-side. Moreover, the concepts of integration of intelligent techniques and wireless sensors such as neural networks and fuzzy logic were considered and implemented in different HVAC applications such as control of thermal comfort [20,

21], smart thermostats [22], and using sensorrs for multi-zone control in residential and commercial buildinngs [18, 23, 24]. The application of occupancy sensors to save energy is presented in [22]. They deploy different occcupant detector sensors (i.e., motion sensors, PIR sensors) inn different places of a house in order to estimate the probabiliity of occupant’s states (away, home, or sleep) at home using Hidden Markov Model. There exists a wide range of studies relaated to managing energy consumption based on time-varying priices in residential buildings [6, 15, 25-27]. However, some of thhem [15, 25] only take into account the pricing programs (price-bbased techniques) to save energy using shifting loads duringg high electricity prices. They have not considered the parameterrs such as current house demand (i.e., reducing PAR) or user ppresence at home (i.e., for reducing the set point temperatures off thermostat when the user is not home). In addition, in the m most cases there should be a constant interaction between userrs and devices to benefit from smart grid initiatives i.e., RTP [6, 28]. The main objective of our paper iss proposing an autonomous system using integration of fuzzzy logic, wireless sensors, and smart grid initiatives for reesidential energy management. The proposed approach brinngs forward an autonomous thermostat that can autonoomously control residential HVAC systems while saving eenergy and cost without sacrificing occupant’s thermal comforrt. The developed thermostat is able to handle all electricity pricce programs (i.e., TOU rates, RTP, and flat-rate). By taking into account the house current demand; the proposed autonoomous thermostat sheds the residential load in all conditions evenn during off-peak hours to avoid PAR. We assume that each residential building is equipped with smart meter for reading pricing signals as weell as the current house demand can be measured and reported uusing inexpensive wireless home energy display devices [299] (Zigbee-based communication). Energy display devices aallow residential customers to track their electricity demand annd use in detail at different times of a day and understand their energy bills. Our approach is different from [22, 30]. In [30],, the user has to predefine the price threshold in order to acct differently for different electricity rates. This approach needs constant interaction from users and it does not consideer other important parameters such as user presence/absence at hhome. While our approach considers user presence and outdooor temperature to provide thermal comfort by maintaining the set point temperatures in ASHRAE standard-99 comforrt-zone [31] when the home is occupied. In addition, unlike [32]] in our work we do not focus on predicting or modeling the ellectricity price in RTP programs. In our work, the autonom mous thermostat receives the electricity rates in real-time and acts proactively to shed the residential HVAC load by applying tthe defined fuzzy rule(s) associated to that particular event(s).

II.

DEVELOPMENT OF HOUSE SIMULATOR

One of the main targets of electric generator utilities (EGU) and home appliance manufacturers is embedding capabilities into grid-side and demaand-side devices in order to integrate residential energy manag gement systems into future smart grids. It will help both custom mers and utilities to benefit from smart grid initiatives. Theerefore, smart control of residential end-use electricity demaand such as HVAC systems can play an important role in redu ucing the peak demand and optimizing residential energy con nsumption. These can be performed by designing smart therrmostat for HVAC systems or building smart homes with auto omation systems to control residential loads. Hence, in ord der to consider different scenarios related to energy managem ment and cost in residential HVAC systems, there is a need to t develop a user-friendly simulator to understand how resid dential energy consumption and its associated costs can be influenced by interrelated energy management factors succh as electricity prices, occupancy, house parameters, etc. The proposed simulator shown in Fig. 1 was implemented using MATLAB with Graphical User U Interface (GUI). The user interface of simulator captured d in Fig. 1 shows the main elements and components of it. Thee left top hand side of Fig. 1 shows the indoor (blue) and ou utdoor (green) temperature graphs. The initialized set points (rred) are also shown in this side. The outside weather profiles are read by the simulator t real weather data taken from an excel file, and represent the from the Canada’s National Climatee Archive or can be defined by user. In Fig. 1, the indoor temperature shows the dynamic response of the Home HVAC system m which takes into account factors such as the thermal model of o a house, heating loss, and gain. The left bottom side of Fig. 1 is the hourly consumption graph, which displays the consump ption over the time for one day. The right side shows the ind door/outdoor temperatures, total energy consumed (kWh), tottal cost ($), RTP or TOU rates at different times of simulattion, heat/cool set points at different points of simulation, and th he mode of operation (heat, cool, auto, off). The details related to simulator and the house heating and cooling model in ordeer to compute the dynamic response of indoor temperature baased on house and HVAC parameters were elaborated in our recent works [33, 34]. Technically speaking, the simu ulator assists us to consider and analyze energy consumption and associated costs for

The rest of paper is structured as follows: in Section II, the ‘house simulator’ developed in MATLAB is bbriefly explained. The input/output parameters of the system and their fuzzy membership functions are presented in Sectionn III. Fuzzy logic decision-making is discussed in Section IV. S Simulation results are elaborated in Section V and the paper is concluded in Section VI. Fig. 1: Simulator Interrface at Work

various parameters that are directly relaated to energy management in residential buildings. More importantly, the m” to help us in simulator will be used as an “expert system development of existing thermostats (i.e., PCT) and implementation of the advanced intelligentt techniques for future “Smart Thermostats”, such as our Suupervised Fuzzy Logic Learning (SFLL) approach proposed inn this paper. The simulator can assist to investigate the impaact of smart grid initiatives such as TOU rates and RTP on eneergy conservation of residential HVAC systems. To meet abovementioned objectives, we realized that the simulator needs some necessary improvementts with respect to the previous versions presented in [33, 34]. Thherefore, few new concepts such as smart grid initiatives (TOU U, RTP programs, and DR programs), several controls (such ass different offset tolerances), fuzzy logic demand responsivenness, pre-heating and pre-cooling, etc. were augmented into sim mulator for future purposes. These improvements and upgrades were mandatory for verification of different scenarios and brinnging forward an Autonomous Smart Thermostat. III.

INPUT-OUTPUT VARIABLES OF THE SYSSTEM, WIRELESS SENSORS, AND FUZZY LOGIC RULE-BASED

In order to design the autonomous thermoostat; we need to specify the input and output variables of the system. For this purpose, a synergy of fuzzy logic, wireless seensors, and smart grid initiatives is used. There exist many param meters that could be taken into account as system inputs for eneergy management in residential HVAC systems. However, we prefer the parameters that directly relate to energy managgement as well as occupant’s thermal comfort as system inpuuts. In addition, measuring only these variables make the impleementation of the system easier [34]. The parameters are cuurrent “Outdoor Temperature” (To), “Occupant Presence” (Po), current House Electricity “Electricity Price” (Pe), and current “Hourly H Demand” (Dh). The only output of system is Set Point (Sp) temperature that is autonomously adjusted by applying speccific fuzzy rules based on information received from aavailable inputs. Furthermore, the capabilities of sensor nodes too sense, measure, or detect different variables of interest suchh as temperature, humidity, and occupancy can significantlly enhance the capability of existing in-home energy managem ment system such as Thermostats to save energy and cost. Thhey can enable a better management of energy by using the sensors’ data to control HVAC system(s) more effectively. Moreover, the inputs are totally interrelatedd and fuzzy logic can be the best choice in order to compromisse between them. For example, by choosing appropriate fuzzy rrules we can save energy and cost, while providing thermal com mfort. In addition, the main advantage of fuzzy logic controllers compared to conventional controllers is based on thee fact that no mathematical modeling is required for controlller design. In this paper, the geometric pattern of triangle is used to define membership functions of input and output variables. A membership function assigns a truth value bettween 0 and 1 to each point in the fuzzy set’s domain. Input annd output sets are connected through a set of ‘IF-THEN’ rules iin order to obtain

Fig. 2: Conceptual of fuzzy au utonomous thermostat

the corresponding output(s). The stteps executed by the fuzzy system are: Fuzzification of input variables, Rule evaluation, a Defuzzification. Fig. 2 Aggregation of the rule outputs, and shows the conceptual design of the system. s The membership functions and reasons that why these variables are chosen as system inputs i are summarized as follows: i is important because Outdoor temperature as one of inputs in cold countries such as Canada, demand d of load significantly depends on the temperature of thee day. Normally, when the temperature is very cold/hot; the deemand of electricity is high. In this paper, we assume that the outdoor o temperature can be taken from outdoor wireless temperrature sensor. An excel file is used to emulate wireless temperature sensor. The outdoor temperature is collected hourly. Fig. 3 (a) shows the mperature. membership functions of outdoor tem It is essential to know whether or o not the occupant is in the controlled space (rooms or house)). This is because that the control system must act differently when w a person is present for providing thermal comfort. It is asssumed that the occupancy sensors give a real output value of 0 or 255. As depicted in Fig. 3 (b), this interval is fuzzified into two different values of the bsent’. linguistic variable ‘Present’ and ‘Ab Furthermore, with the advan ncements in smart grid technologies, there is a significant opportunity to realize both her through smart meters. energy efficiency and DR togeth designing and producing thermo ostats capable of reading electricity prices through communiccating with deployed smart meters; enables them to proactivelly respond to time-varying prices without user interaction. In this paper, in order to b the thermostat and emulate two-way communication between smart meter; an excel file contains TOU rates or RTP is used. Fig. 3 (c) shows the membership fu unctions of electricity price read from smart meter. “Hourly house electricity demaand” as another important input of system can help the autono omous thermostat to reduce the load demand when other homee appliances are operating. The role of this variable as system input is being more crucial particularly during the times that ussers shift their loads to offpeak hours. In many cases, shifting loads to off-peak times will g these times the thermostat increase the PAR. Therefore, during autonomously sheds the residential HVAC load by reducing n in Fig. 3 (d), it is assumed the set point temperatures. As shown that for example in Canada, the average of highest hourly

(a)

(b)

(c)

(d) Fiig. 3: Membership functions of input parameters

demand kW for a residential building is equall to 1.80 kW, the medium 1.20 kW, and the lowest is 0.60 kW [11, 5]. In addition, the only output of the system m is the set point temperatures (Sp). They are autonomously addjusted based on designed fuzzy logic algorithm through evaaluating available inputs received from wireless sensors as well as the electricity prices taken from smart meter. The trianguular membership functions of the output are shown in Fig. 4. As shown in Fig. 4, the system output (set ppoints) is divided into 9 different zones. For the first zone (SP1) that in the most cases is tuned when the home is unoccupied; tthe zone covers 3 °C. It is often recommended to keep the teemperature of an empty house between 14-16 °C [31]. Other zonnes cover 2 °C of the total range (16-24 °C). Each of these set point zones overlaps adjacent set point zones or membersship functions by half. The spacing of set point zones is hencee 1 °C. The used spaces are chosen to achieve precise indoorr temperature. In addition, the selected zones and spaces cann give us such a freedom to change the set point temperatures to tune the fuzzy rules in order to save energy and cost. IV.

FUZZY LOGIC DECISION-MAKING ALGORITHM

In building automation in which a HVA AC system is the primary load in consuming energy; the aaim of a global controller is to maintain the indoor temperatture between the desired intervals. In our case, the objectivve of the design thermostat is to autonomously maintain inddoor temperature within the ASHRAE comfort-zone based on information taken from WSN, electricity prices, and current houuse load demand without any interaction from its users. F Furthermore, the thermostat should save energy without saacrificing user’s thermal comfort. ASHRAE recommends that relative humidity (RH) is maintained below 60%. The RH shouuld be bigger than 30% as well. However, influence of the hum midity is not great

for the people with very light or sed dentary activities. ASHRAE comfort-zone (19–23 °C) for cold regions r and (18–22 °C) for cool regions is commonly accepteed in research and practice during heating operation. The low wer values of two defined intervals above considerably help us u in saving energy and cost during high electricity demands and prices. The rule-based layer in any fuzzzy logic structure represents the knowledge of the outside world.. It specifies how to react to input signals as well. To do so, the thermostat t keeps evaluating the available inputs and makes decision d about the system output (adjusting set point) accorrding to the defined rules. Therefore, the proposed SFLL hass to constantly be tuned to balance the new set point in order to o save energy and cost. For this purpose, the system invokes the t fuzzy rules to compute the amount of Sp. Simultaneously, the t fuzzy rules has to work as a smart agent to determine the best indoor temperature in w the home is occupied. order to provide thermal comfort when All these objectives are performed by b applying the defined and tuned fuzzy rules that some listed d in Table I. In this Table, ‘VC’ represents ‘Very Cold’, ‘H’, ‘M’, and ‘L’ stand for ‘High’, ‘Medium’, and ‘Low’ respectively. ‘P’ shows ‘Present’ and A indicates ‘Absent’. This base includes rules which are structured as follows: ‘ and Po is ‘P’ and Dh is If current To is ‘VC’ and Pe is ‘H’ ‘H’ Then Sp is ‘Sp6’. Furthermore, the aggregated of fuzzy sets must be he control variable. This is transformed into crisp values for th the goal of the Defuzzification Interface. I We propose the Mamdani technique for Defuzzificcation of output because it provides a natural framework to in nclude expert knowledge in the form of linguistic rules which h is very important in our case. The Center of Gravity (CO OG) approach described in equation 1 is used for Defuzzzification, where μA(x) is

TABLE II. SCENARIOS FOR VERIFICATION N OF THE PERFORMANCE OF SFLL

TABLE I. SOME OF FUZZY RULES #Rule 1 2 3 4 5 6 7 8 9 10 11 12

To

VC VC VC VC Cold Cold Cold Cold Cool Cool Natural Natural

Inputs Pe Dh L H H H H L L M H H H L

Po

L H H L H H L H H L H H

P P A P P P A P P P P P

Output Sp SP8 SP7 SP2 SP8 SP6 SP7 SP1 SP7 SP5 SP6 SP3 SP4

membership function of fuzzy set A and m iis the number of rules applied to the controller. Therefore, the Defuzzified value of the output (Sp) specifies the value of set point for that specific time. Finally, the control signal is seent to actuate an on/off relay which results in turning on/ooff a residential HVAC system. ∑

.

∑

V.

(1)

SIMULATION RESULTSS

The proposed autonomous system uusing SFLL is embedded and implemented in the simulatorr. We verify the thermostat equipped with proposed autonom mous system with respect to changing in electricity prices,, load demand, occupancy, and outdoor temperature. Within the verifications; providing thermal comfort and energy ssavings will be evaluated as well. In order to verify the propposed algorithm, the input parameters are set at different vaalues to emulate different scenarios that might occur. To do so, we plan ten different scenarios during a day as depictedd in Table II to observe the thermostat’s response accordinglyy. All inputs data listed in Table II were chosen in such a way to be within the for winter season defined fuzzy ranges. The electricity rates fo are taken from Hydro one, Ontario, Canada. We suggest two different modes namely ‘Economy Mode’ and ‘Comfort Mode’ for the developed autonomous thermostat. The user can choose one for tthe operation of HVAC system. In Comfort Mode we have a sm mall shift in load reduction and there is a large load reductioon for Economy Mode. Maximizing comfort implies minimizinng load reduction benefits, and vice versa. In Economy Mode we have chosen the ASHRAE comfort-zone between (18-222 °C), and for Comfort Mode the range between (19-23 °C) iis preferred. Fig. 5 shows the response of therm mostat where it autonomously adjusts the set points based on information listed in Table II. In this case the Comfort Mode has been Comfort Mode the chosen as user preferred mode. Although in C thermostat saves energy and cost, energy cconsumption and cost can be managed better when Economy M Mode is preferred. In Fig. 5, the response of thermostat to changging in electricity price can be observed from 7:00 AM to 11:00 AM. At this interval (see Table II scenarios 4 and 5) thee price increases from 7.2 cents (low) to 12.9 cents (high) andd the thermostat autonomously reduces the set point to save cosst and energy.

Scenario

Time of Day

To

Pe

Dh

Po

Sp (°C)

1 2 3 4 5 6 7 8 9 10

0:00 to 2:00 2:00 to 6:00 6:00 to 7:00 7:00 to 8:00 8:00 to 11:00 11:00 to 14:00 14:00 to 17:00 17:00 to 19:00 19:00 to 21:00 21:00 to 24:00

2 -5 0 0 -3 8 10 4 -6 14

7.2 7.2 7.2 12.9 12.9 10.9 10.9 12.9 7.2 7.2

0.5 0.4 0.7 1.1 1.6 1 0.8 2 1.8 0.6

P P P P P A A P P P

22.5 23 22.8 19.2 19 15.5 15.5 18.3 20 17.6

Furthermore, in order to obserrve that how the set point temperatures are set using SFLL L; a comparison between Comfort Mode and Economy Mode based on information listed c be observed from Fig.6, in Table II is shown in Fig.6. As it can the reduction in residential HV VAC load demand using Economy Mode is larger particulaarly during high electricity prices (see intervals 7:00-11:00 AM M and 17:00-19:00). In fact, the proposed approach proactively y responds to time-varying prices. In this way, the residential HVAC H system is integrated into smart grid without any interaction from its user. In Economy Mode (see Fig. 6)), the thermostat equipped with SFLL keeps attempting to tunee the set points between 1822 °C. From 21:00 to 24:00 when the outdoor temperature is 14 °C, demand is low (0.6 kW), and the price is low; the n 16 °C while it is 17.6 °C thermostat adjusts the set point on using Comfort Mode (refer to Fig. 6, S10 or Table II, scenario 10). In this case there is no need to keep the house warm when the outdoor temperature is natural. In addition, in order to c house demand (kW) verify the importance of knowing current as one of inputs; the response of thermostat to two similar scenarios when the house demandss are different is considered (see scenarios 2 and 9 in Fig. 6, Eco onomy Mode). In scenario 2, where the demaand is ‘low’ (0.4 kW) the thermostat puts the set point temperrature on 22 °C (S2). While in scenario 9, where the demand is i ‘high’ (1.8 kW); the set point is autonomously set on 20 2 °C (S9). Although the thermostat reduces the set point (S9), that is still in ASHRAE comfort-zone to maintain thermall comfort. In this case, in addition to saving energy; the auttonomous smart thermostat can contribute to reduce the PAR R that is a major problem particularly when the loads are shifteed to off-peak hours.

Fig. 5: Set points adjusted by au utonomous thermostat

VI.

CONCLUSION

In this paper an autonomous system utilizzing a synergy of supervised fuzzy logic learning, wireless sennsors, and smart grid initiatives was discussed. The proposed aapproach brought forward an autonomous thermostat for residential HVAC mfort Mode’ and systems. Two different modes namely ‘Com ‘Economy Mode’ with aim at energy saving without sacrificing thermal comfort were presented. The simulation results demonstrated that the autonomous therrmostat was able to proactively respond to different parameteers such as house electricity demand, outdoor temperature, aand time-varying prices, while saving energy without jeoppradizing user’s thermal comfort. In addition to integrating reesidential HVAC system into smart grid, the proposed methhod was able to contribute to reduce the PAR as well. REFERENCES 1. Office of Energy Efficiency, Energy Efficiency Trrends in Canada, Natural Resource Canada, 2011: Canada. 2. J. J. Conti, et al., Annual energy outlook with projectionns to 2035, Energy Information Administration, U.S., 2010. 3. Faruqui, A. and S. Sergici, Household response to dy dynamic pricing of electricity: a survey of 15 experiments. Journal of Reguulatory Economics, 2010. 38(2): pp. 193-225. 4. McAuliffe, P. and A. Rosenfeld, Response of residenntial customers to critical peak pricing and time-of-use rates during thee summer of 2003. California Energy Commission, 2004. 5. Ontario Home Builders Association,, Reducing electricity consumption in houses, Energy conservation committee report, 2006. 6. Hyungna, Oh., Stratford D., Powsiri Klinkhachorn, SSystem Reliability and Price Responsiveness of Residential Loads, West V Virginia University, Advanced Power & Electricity Research Center 2006. 7. A. Ipakchi, F. Rahimi, Demand response as a market rresource under the smart grid paradigm. IEEE Trans. on Smart Grid, 20100. 1(1): p. 82–88. 8. Strbac, G., Demand side management: Benefits and cchallenges. Energy policy, 2008. 36(12): pp. 4419-4426. 9. Peffer, T., et al., How people use thermostats in hhomes: A review. Building and Environment, 2011. 46(12): pp. 2529-2541. 10. Iain S. Walker, et al., Residential Thermostats: Coomfort Controls in California Homes, 2008. 11. Malinick, T., et al., Destined to disappoint: program mmable thermostat savings are only as good as the assumptions abouut their operating characteristics. ACEEE on Energy Efficiency in Buildings Pacific Grove, CA, 2012. 12. Chanana, S. and M. Arora, Demand Response from m Residential Air Conditioning Load Using PCTs. 13. Allcott, H., Real time pricing and electricity markets. H Harvard University, 2009. 14. Vojdani, A., Smart integration. Power and Energy Magaazine, IEEE, 2008. 6(6): pp. 71-79. 15. Du, P. and N. Lu, Appliance commitment for householdd load scheduling. Smart Grid, IEEE Trans. on, 2011. 2(2): pp. 411-419. 16. J. Yick, B. Mukherjee, and D. Ghosal, “Wireless sensorr network survey”, Computer Networks, Vol. 52, Issue 12, pp. 2292-2330, A August 2008. 17. Erol-Kantarci, M. and H.T. Mouftah, Wireless sensor nnetworks for costefficient residential energy management in the smart grid. Smart Grid, IEEE Trans., 2011. 2(2): pp. 314-325. 18. Cao, X., et al., Building-environment control with wiireless sensor and actuator networks: Centralized versus distributed. Induustrial Electronics, IEEE Trans. on, 2010. 57(11): p. 3596-3605.

Fig. 6: Adjusted set points for economy and comfort modes 19. Ota, N.K., The application of wirelesss sensor networks to residential energy efficiency and demand response, 2007. 2 20. Chen, K., Y. Jiao, and E.S. Lee, Fuzzzy adaptive networks in thermal comfort. Applied Mathematics Letters, 2006. 19(5): pp. 420-426. 21. Bermejo, P., et al., Design and simulation of a thermal comfort adaptive system based on fuzzy logic and on-linee learning. Energy and Buildings, 2012. 49: p. 367-379. 22. Lu, J., et al. The smart thermostat: using u occupancy sensors to save energy in homes. in Proceedings off the 8th ACM Conference on Embedded Networked Sensor Systems. 20 010. 23. Ferreira, P., et al., Neural networks bassed predictive control for thermal comfort and energy savings in public buildings. b Energy and Buildings, 2012. p. 238-251. 24. Taylor, K., et al. Sensor/actuator networks n supporting agents for distributed energy management. in Loca al Computer Networks, 2004. 29th Annual IEEE International Conference on. o 25. Kim, T.T. and H.V. Poor, Scheduling g power consumption with price uncertainty. Smart Grid, IEEE Transs. on n, 2011. 2(3): p. 519-527. 26. Koomey, J. and R.E. Brown, The role off building technologies in reducing and controlling peak electricity demand, 2002. Garcia, Optimal residential load 27. Mohsenian-Rad, A.-H. and A. Leon-G control with price prediction in real-timee electricity pricing environments. Smart Grid, IEEE Trans. on, 2010. 1(2): p. 120-133. 28. Qian, L.P., et al., Demand response man nagement via real-time electricity price control in smart grids. Selected Areas in Communications, IEEE Journal on, 2013. 31(7): p. 1268-1280. 29. Texas Instruments: Smart Grid d Tools, available online: http://www.ti.com/lit/an/slaa517d/slaa517d.pdf (2012-2015). 30. Ji Hoon, Y., R. Baldick, and A. Novoseelac, Dynamic Demand Response Controller Based on Real-Time Retail Price for Residential Buildings. Smart Grid, IEEE Transactions on, 2014. 5(1): p. 121-129. 31. ASHRAE Handbook - Fundamentals (I-P edition). 2011: American Society of Heating, Refrigerating and Airr-Conditioning Engineers. 32. Mohsenian-Rad, A.H., et al. Optimal and autonomous incentive-based energy consumption scheduling algorith hm for smart grid. in Innovative Smart Grid Technologies (ISGT), 2010. IEEE. I 33. Keshtkar, A. and Arzanpour, S., Simullation and implementation of an intelligent system for residential energyy management using Xbee-based wireless sensors. IEEE Canadian Jourrnal on Electrical and Computer Engineering, 2014 (In Press). 34. Keshtkar, A., et. al, Smart residentiall load reduction via fuzzy logic, wireless sensors, and smart grid incentivves, Energy and buildings, July 03, 2015 (In press).