Tabu search for the optimization of household

energy consumption
Duy Long Ha Student Member IEEE, Stephane Ploix, Eric Zamai, Mireille Jacomino
Laboratoire d’Automatique de Grenoble, INPG, UJF, CNRS UMR5528,
BP46, F-38402 Saint Martin d’Hères Cedex, France
duy-long.ha, stephane.ploix,eric.zamai,[email protected]

Abstract— This paper focuses on Demand-Side load Manage- Energy management can be formulated as a scheduling
ment applied to residential sector. A home automation system problem where energy is considered as a resource shared by
controlling household energy is proposed. It is decomposed into appliances, and periods of energy consumption are considered
three layers: anticipation, reactive and device layers. This paper
deals with anticipation layer that allocates energy in taking as tasks. Generally speaking, these approaches coordinate con-
into account predicted events. It consists in computing both sumption activities in scheduling all tasks as soon as possible
the starting times of some services and to determine set points in order to reduce the overall consumption while satisfying
of others while satisfying the maximal power constraint. A maximum energy resource constraint. These approaches do
Constraint Satisfaction Problem formulation has been proposed. not manage the differences between predictions and effec-
Because the complexity is NP-Hard, a tabu search is used to solve
the problem. It maximizes user comfort and minimizes energy tive values. [5] proposes a solution based on one-day user
cost. An application example is presented. consumption predictions. A parallel and distributed genetic
algorithm optimizes the consumption of buildings in order to
I. I NTRODUCTION adjust the consumption of appliances to energy provider needs.
In [6], an adaptation of the static Resource Constraint Project
Demand-Side load Management(DSM) [1] is a set of meth-
Scheduling Problems (RCPSP) is presented to improve the
ods that coordinate the activities of energy consumers and
management of electric heating systems. This approach is able
energy providers in order to best fit energy production ca-
to coordinate the electric heaters while satisfying a maximum
pabilities to consumer needs. Thanks to DSM, energy demand
power resource constraint. Nevertheless, the problem requires
peaks, which on the one hand, have negative environmental
precise predictive models and, moreover, it is NP-hard. [7]
impacts and on the other hand, increase energy production
presents a new three-layer household energy control system
costs [2], can be reduced. In residential sector, the development
capable both to satisfy the maximum available electrical power
of Home Automation (HA) systems make it possible for
constraint and to maximize user satisfaction criteria. This
energy consumers to be involved in DSM in adapting their
approach carries out more reactivity for fitting the energy
consumption to production needs [3].
provider needs. Rooms equipped with electric heaters are used
[2] presents basic kinds of DSM control:
to illustrate the capability of the control mechanism for using
• Direct control that shifts power requests by directly natural thermal accumulation to adjust power consumption in
interrupting the high power consuming appliances. real time.
• Local control that consists in setting up a policy that This paper proposes a control algorithm, which consists in
encourages consumption at off-peak periods in reducing finding the global solution for the household energy man-
energy costs. agement problem (HEMP). In order to fit the housing con-
However, these kinds of control are not very reactive and does sumption to the available energy, the home automation system
not take into account user comfort. controls the equipment in housing in determining starting time
A home automation system [4] basically consists of appli- of some services and also in controlling the temperature set
ances linked via a communication network allowing appliances point of HVAC systems. Because this problem is NP-hard, a
to communicate one each other. These home automation metaheuristic Tabu Search is proposed in this paper to solve
systems can carry out a new load management mechanism the HEMP.
which is called distributed control [3]. This DSM control
allows energy providers to charge user for the actual energy II. C ONTROL MECHANISMS
production cost in a very precise way. It also allows users In classic predictive scheduling like job shop or RCPSP,
to adjust their power consumption according to energy price the data of scheduling procedure are: the resource and the
variation. In the peak period, the domestic customer would duration of a task and, the earliest and the latest starting time.
be able to decide whether to wait and save money or to use All these data have to be well known before the procedure of
appliances even so. This strategy is more reactive than the scheduling starts. However, the main issue in HA scheduling
basic DSM control but more complex to control when comfort problems is the presence of uncertainties of predictions: solar
has to be taken into account. radiation, outdoor temperature, starting times and durations
of services requested by inhabitants. Uncertainties are so im- III. P ROBLEM MODELING
portant in predictions that even robust scheduling approaches This paper focuses on the anticipative layer in tackling the
are not very efficient. Uncertain events are indeed often more prediction mechanism and its combinational issue. The antici-
important than events that can be predicted. In order to solve pation layer makes plans for energy allocations, which consist
this issue a three layer architecture is proposed: an device in determining both the starting dates of the timed services,
layer, a reactive layer and an anticipative layer. This structure and the set points of the permanent services mainly composed
of control improves the adaptability of the system and reduces of HVAC systems of which the set points corresponds to
the solution research space of the energy allocation plan. temperatures. The following notations have been adopted:
A. Device layer • SRVi denotes a service and SRV S the set of all the

The device layer is composed by devices together with their services to be achieved. SRV S can be partitioned into
existing control systems generally embedded into equipments permanent services SRV SP and timed services SRV ST .
• DEV (SRVi ) = DEVj denotes the device that achieves
by manufacturers. It is responsible of adjusting device controls
in order to reach given set points in spite of perturbations. This the service SRVi with ∀DEVj , DEVj ∈ DEV S
th
• ∆k denotes the k anticipative period following the
layer gathers two kinds of services:
current time
• the permanent services, such as HVAC systems, which
are linked with a one-to-one relation with device A. Timed services on loads
• the timed services, such as cooking or washing, which are
A timed service SRVi ∈ SRV ST is modelled as an on/off
bounded in time. Contrary to permanent services, several service. After starting, it remains on until the service is ended.
timed services can occur on the same device but not at Its consumption during a period ∆k is then modeled by:
the same time. Ei,k ∈ {0, ∆ × Pi } where Pi characterizes the average power
The interest of this layer is to render devices more abstract consumption of the service. The durations of timed services
for other layers: continuous phenomena and fast dynamics are are modeled by number di ∈ N∗ of anticipative periods. Let
hidden at this level. EST (SRVi ) and LST (SRVi ) be respectively the earliest and
B. Reactive layer the latest starting times for the service coming from user
comfort expectations. The starting time requested by user is
Objective of the reactive layer is to manage the real-time
denoted RST (SRVi ). The variable si represents the computed
adjustements of energy allocation. This layer is responsible of
starting time of timed service SRVi .
decision making in case of violation of predefined constraints
dealing either with energy or with comfort [8]. The control B. Permanent services on loads
actions may be either to enable or disable controllers of the Permanent services are generally characterized by a con-
device layer. trolled physical variable. In housing, the permanent services
C. Anticipative layer mainly deals with HVAC and water heating systems. Let’s
The anticipative layer is responsible of managing predicted focus on HVAC systems. Anticipation requires a relevant
events dealing with electric sources and loads in order to thermal air environment model to predict the consumption of
avoid as much as possible the use of reactive layer. The HVAC system. [9] and [10] have proposed precise models of
prediction procedure forecasts several information about future a room. Nevertheless, given the importance of uncertainties in
user requests but also about the future available energetic predictions, for example, outdoor temperature, thermal mod-
ressources and about the price fluctuation of energy. This eling parameter, which may cover several hours, the simple
layer has slower dynamics and includes predictive models with thermal dynamic models presented in [11], [12] has been
learning mechanisms 1 . This layer also contains an anticipa- preferred:
tive control mechanism that schedules energy production and dTi (t) 1
consumption several hours in advance. This layer adjusts set Ci = Pi (t) − (Ti (t) − T out (t)) (1)
dt Ri
points of devices. The sampling period of the anticipation layer
is denoted ∆. where i refers to a HVAC service SRVi .
Antipative layer is based on the most abstract models. Ci is the heat capacity of SRVi . Ti (t) is the indoor temper-
Because of it follows slower dynamics, inferior layer is ature. Each service is linked to a room equipped with a con-
transparent for anticipation: reactive layer adjusts in real-time trolled electric heater. Pi (t) corresponds both to electric power
the set points coming from the anticipative layer. Because the consumed by the heater and to the heating power provided to
dynamics of the reactive layer is higher, globally speaking, it the room. Ri represents the equivalent total thermal resistance
does not modify much the energy allocation of the anticipative between the room and outdoors. T out stands for outdoor
layer. The controllers of the device layer adjust the device temperature. The thermal incidence of other environments
controls in order to satisfy the set-points coming from highest and of solar radiations are considered as perturbations but,
layers. It is also transparent for higher layers. they could also be taken into account with a more precise
model. The anticipation layer plans the consumption of heating
1 including models dealing with user habits systems according to the available electric power. For the sake
 2) Thermal sensation: Comfort is a subjective feeling,
which is difficult to assess. [13] and [14] have proposed
the ISO7730 thermal comfort standard. The function of
PMV(Predict Mean Vote) is determine following this standard
[14]. In this paper, only indoor temperature is taken into
account. Other elements like outdoor temperature, humidity,
user’s clothes and mean air velocity are assumed to be
constant. In this condition, the ”ideal” temperature or the most
Fig. 1. Unsatisfactoriness service notion
comfortable thermal sensation is about 21°C. Objective of the
control the system HVAC is to maintain the indoor temperature
around this set point temperature. The range of acceptable
of simplicity, the thermal modeling (1) is discretized according
indoor temperature is determined by −1 ≤ P M V ≤ 1,
to the anticipative period ∆. Let Ti be the temperature set
a predicted thermal sensation in anticipative period ∆k is
point for the heater linked to SRVi , which will be computed
P M V (Ti,k ). The criterion of unsatisfactoriness for HVAC
several hours in advance. Pi stands for the average thermal
service SRVi is defined as follow:
power provided by the heater during the period ∆k . PK
| P M V (Ti,k ) |
Ui = k=1 (5)
(− R ∆C ) (− R ∆C ) K
Ti,k+1 = e i i Ti,k + Ri (1 − e i i )Pi,k 3) Energy cost criterion: The DSM control in [15] allows
(− R ∆C )
+ (1 − e i i )Tkout (2) energy providers to charge user for the actual energy pro-
duction cost in real time. Thus this price variation is taken
This model points out that to go from a temperature Ti,k to into account into a HA system. Assuming that during an
Ti,k+1 when outside temperature is Tkout , the average required anticipation period ∆k , the energy cost is ECk . An energy
power during period ∆k is equal to Pi,k . cost of a allocation energy plan EC is given by:
C. Permanent services on sources X
K X
n

Anticipation has to satisfy a maximum available power con- EC = ECk Pi,k (6)
k=1 i=1
straint: during each anticipative period, a maximum available
power cannot be exceeded. Providing power is a permanent 4) Global criterion: The HEMP is a multi-objectives opti-
service in an one-to-one relation with the source device mization, HA system try to minimize the total unsatisfactori-
that provides power. This available power may depend on ness services
XJ
the sources and on contracts with energy providers. For the M in( Uj ) (7)
sake of simplicity, all the devices supporting the services on j=1
loads are assumed to be purely resistive. Therefore, the total
power consumption corresponds to the summation of power and the same time minimize the total energy cost
consumptions: M in(EC) (8)
X n
Pi,k ≤ P̂k , ∀k (3) Global criterion is a compromise between comfort and cost.
i=1 An aggregation criteria is not suitable because weighting
The supplied power has to be equivalent to the consumed heterogeneous criteria is difficult for inhabitants. Modes can
power. be defined. For instance, the energy cost 8 may be fixed and be
considered as a constraint. For example, consider an economic
D. Criteria mode where inhabitants accept to decrease the comfort level
1) Starting date of timed services: When the energy source in order to reduce the energy bill. HA system can also be set
is sufficient for all the timed services requesting for energy, the to a comfortable mode, which amounts to search a solution
starting time si is equal to the requested starting time RSTi . based only on the comfort criteria. Secondly, HA system can
Nevertheless, an issue appears when the available energy is not search for several solutions and proposes a pareto of them
sufficient for all. Some services must be delayed or executed to user. Then user can choose the solution that best fits his
sooner, it means the actual starting time si 6= RSTi , this expectations. The comfort criterion 7 is defined by thresholds
effect decreases the user comfort. To represent the level of and treated as a constraint. HA system then select the best
user’s uncomfort, a criterion of unsatisfactoriness Ui ∈ [0, 1] solutions based on the energy cost criteria 8.
is defined as follows:
E. Problem complexity
Ui = 0 if si = RST (SRVi )
The HEMP aims at solving two problems. First problem is
Ui = RST(s(SRV
i −EST (SRVi ))
i )−EST (SRVi )
if si < RST (SRVi ) (4) to minimize the total weighted delay of energy allocation Ei .
(LST (SRVi )−si )
Ui = LST (SRVi )−RST (SRVi ) if si > RST (SRVi ) Thus, the well-known problem of minimizing total weighted
delay of job in a single machine is known to be a NP-hard
problem [16]. The second problem in the HEMP deals with 2) Device capacity constraint: Some devices supporting
the management of permanent services such as HVAC. This timed services, such as washing machine or oven, are shared
problem is also NP-Hard in weak sense [7]. The complexity between several services. These services must be executed
of the HEMP is NP-Hard or also NP-Complete. Therfore, an sequentially. Consider two services SRVi and SRVj that
heuristic should be chosen to solved the problem. share the same device: DEV (SRVi ) = DEV (SRVj ). If the
requested starting times RST (SRVi ) < RST (SRVj ), the
IV. C ONSTRAINT S ATISFACTION FORMULATION following constraint arises:
sj ≥ si + di (12)
Many problems in operational research such as graph col-
oring, n queens, scheduling, car sequencing can be formulated 3) Thermal capacities constraint: To reach the temperature
as constraint satisfaction problems (CSP) [17]. A CSP formu- Ti,k+1 from Ti,k , the average required power is Pi,k =
lation consists in a given set of variables V , a given set of f (Ti,k , Ti,k+1 ) where f is defined by (2). The thermal power
domains di ∈ D corresponding to variables of V and a given must be positive and less or equal than the maximal power of
set of constraints C that must be satisfied. The objective of a the heater SRVi denoted P̄i . If this value is negative, it means
CSP is to find values for all variables vi ∈ V such as vi ∈ di that the set point temperature Ti,k ≥ Ti,k+1 and that the set
and that satisfy all the constraints in C. point Ti,k+1 is too low to reach Ti,k+1 . Conversely, if this
amount is greater than Pi , the set point temperature Ti,k+1 is
A. Variables and Domain too high to reach Ti,k .

The problem of household energy management can be 0 ≤ Pi,k ≤ P̄i (13)

formulated as a constraint satisfaction problem. The CSP is
V. TABU S EARCH
formally defined by (V, D, C).
1) Starting times: The starting times si of the timed ser- Tabu Search(TS) is based on a metaheuristic, which has
vices belong to the set V . The domain di of si is defined been originally developed by Glover [18] and [19]. It has been
by: successfully applied to a variety of combinatorial optimization
problem. The basic principle of TS is to pursue Local Search
(LST (SRVi ) − EST (SRVi )) until a local optimum is found and then, to allow non-
di = EST (SRVi ) + n (9)
∆ improving moves. A list called Tabu list T , that records the
recent history of the search, prevents from revisiting solutions
2) Temperature set points: The temperature set points for that have already been considered. The procedure of TS can be
each anticipation period ∆k has also to be included into V . summarized as follows: from an initial solution s0 ∈ S where
In order to reduce the dimension of the search space, the S denotes the search space i.e. the instantiation of variables
domain of acceptable values for indoor temperature Ti,k is V according to domains of D, a neighborhood N (s0 ) of the
discretized into nd domains. In fixing the acceptable limit solution s0 is generated in performing elementary steps from
of the thermal sensation criterion −1 ≤ P M V ≤ 1, two the initial solution. The best solution of the neighborhood is
thresholds characterizing acceptable indoor temperatures can then chosen as a candidate solution. This candidate is pushed
be deduced: T min ≤ Ti,k ≤ T max . The discretized domain di into the tabu list T in order to prevent future reconsideration of
of indoor temperatures satisfies: this solution. The length of the tabu list T , denoted t, is called
tabu tenure. To select a candidate solution from N (sl ), an
T max − T min aspiration criterion is also considered. It consists in allowing
di = {T min + j ; j ∈ {0 . . . nd − 1}} (10)
nd a move, if this move steps to a solution which has a better
objective criterion than the current best-known solution s∗ .
Other discretizations, such as temperature for water heating Even if it is in the tabu list.
service, are managed in the same way.
A. Principle TS implementation
B. Constraints 1) Penalty function of violation constraint: [20] has pro-
posed to use tabu search as a general solver for CSP. Each
1) Maximal energy constraint: Maximal energy constraint
constraint cj ∈ C has a penalty function pcj and an appropriate
models the limited capacity of the electrical source. It may
weight wcj representing the importance of the constraint. The
vary in time. During a anticipation period ∆k , the maximum
penalty function is used to allow the TS to balance between
available energy Ekmax = Pkmax × ∆ where Pkmax stands for
feasible search space and infeasible space. Three kinds of
the maximum power capacity of the source. Obviously, the
constraint, (11), (12) and (13), can be formulated as linear
total power consumption of all the loads cannot exceed this
constraints g(x) ≤ Acj where Acj is a constant. Let sl be a
value:
X solution of the CSP that corresponds to a feasible instantiation
Ei,k ≤ Ekmax (11) of the variables of V . Let Acj + ǫ be the tolerance of Acj .
i pcj (sl ) denotes the value of the penalty function for the
constraint cj for the solution S. pcj (sl ) is a nonnegative B. Control tabu tenure
number defined as follows: The tabu tenure t is very sensitive and must be tuned
g(sl ) − Acj carefully because the performance of TS depended highly on
pcj (sl ) = M ax( , 0) (14)
ǫ it.
The total amount of violation of a constraint cj ∈ C by a 1) The fixed tabu tenure: Recently [16] has proposed a
solution sl is : TS using different tabu tenure values including: small (St),
X medium (Mt) and large (Lt) tabu tenure values. The setting of
p(sl ) = wcj pcj (sl ) (15)
j
t depends on the number of elements in the neighborhoods,
which can be generated at each iteration. In this paper, St ≈
2) Intensification phase: The intensification phase (IP) con-
10% | N (s0 ) |. It means that the maximal number of prohib-
sists in recording a fixed length of consecutive solutions in a
ited elements of the current neighborhood represents 10% of
short-term-memory. This search phase aims at performing a
the total number of generated elements of the neighborhood.
more thorough examination. Let f0 (s); s ∈ S be the objective
In the same way: M t ≈ 25% | N (s0 ) |, Lt ≈ 50% | N (s0 ) |.
criterion and p0 (s) be the value of the penalty function for
Firstly, the TS with fixed tabu tenure is stopped after 2000 not-
solution s. If p(s) > 0, it means that several constraints are
improving iterations. Next, the best-known solution s∗ is sent
violated. First step of intensification phase is to reach rapidly
to the intensification phase with t = St. The stop condition is
the feasible solution space. Only p(s) is considered as an
1000 not-improving iterations.
objective:
2) Dynamic control tabu tenure: Varying continuously the
minimize q(s) = p(s) subject to s ∈ S (16) tabu tenure results in a balance between intensification and
When a feasible solution is found(p(s) = 0), the second phase diversification phases. On the one hand, a setting t := St
try to improve this solution. The objective function of TS is corresponds to the intensification phase. On the other hand,
only f (sl ) and only steps satisfying p(s) = 0 are allowed: the settings t := M t, t := Lt corresponds to the diversification
phase of TS. Consider that the tabu tenure varies according to
minimize q(s) = f (s) subject to s ∈ S : p(s) = 0 (17) a sequence [St, M t, St], i.e. tabu tenure starts with t := St.
3) Diversification phase: Diversification phase is an al- After 1.5t iterations of TS, the tabu list T is reset and
gorithmic mechanism that tries to alleviate this problem in tabu tenure changes to t := M t. Three strategies of varia-
forcing the search into previously unexplored areas of the tion t are considered: [St, M t, Ht, M t, St], [M t, Ht, M t, St],
search space. This phase may lead TS to infeasible solution [Ht, M t, St]
space: moving to instantiation that violated some constraints VI. R ESULTS
is permitted. The long-term-memories is considered. The ob-
jective function is: An illustrative example based on two permanent services
and two timed services is presented in this section. The two
minimize q(s) = f (s) + p(s) subject to s ∈ S (18) permanent services SV1 and SV2 are heating services in two
4) Initial solution: To start the TS approach, an initial so- rooms. Thermal parameters are C = 0.025kW h/°C, R =
lution s0 must be prepared. Considering that s0 is determined: 40kW ◦ C for both rooms. The average outdoor temperature
the temperature set point of a thermal environment is set to is: T out = 5°C and the initial temperature is T0,0 = 20◦ C for
its optimal value 21°C. The starting date of timed services are the first room and T1,0 = 18°C for the second one. The energy
set to their requested starting date RSD(SRVi ). consumption plan covers a period of 2h. The anticipation
5) Neighborhood structure: A move, denoted period ∆ is fixed to 0.1h. Domain of the temperature set-point
−−−−−−−→
s0 move(a, b) s′0 , is a step from an element s0 of the belongs to [15◦ C, 30◦ C]. The discretization of temperature
search space to s′0 , which is considered as a element of domains is given by: nd = 15. The first timed service SV3
the neighborhood of s0 . Let V and V ′ be the instantiated is defined by: EST (SV3 ) = 0.2h, RST (SV3 ) = 0.5h,
variables corresponding respectively to s0 and s′0 . First, V ′ LST (SV3 ) = 0.8h, P (SV3 ) = 1kW . The second one is
−−−−−−−→ defined by: SV4 : EST (SV4 ) = 0.6h, RST (SV4 ) = 0.8h,
is initialized with V . Then, the move(a, b) is determined by
two parameters a and b [20]. They represent the assignment: LST (SV4 ) = 1.2h, P (SV4 ) = 1kW . The total power
va′ ∈ V ′ va′ := (da )b where (da )b stands for the bth element constraint is limited to 2kW . The consumption of the best
of the domain da . Therefore, the size of the neighborhood solution found by TS is illustrated in figure 1. Available power
N (s0 ) is calculated as follows: doesn’t allow the simultaneous run of two timed services. SV3
X is on time but SV4 has been delayed by 0.2h. In figure 2, the
| N (s0 ) |= (| Di | −1) (19) temperature set points of SV1 and SV2 have been increased
i before the run of SV3 in order to accumulate energy into the
∗
The best element s′ 0 of the neighborhood becomes the room. The user feels warm during several period but he doesn’t
candidate
−−−−−−− solution
−→ ∗ for the next iteration of TS. The feel too cold afterwards.
s0 move∗ (a, b) s′ 0 is pushed into the tabu list. At the next 20 random cases of study have been used to test the
iteration, only the moves that are not in the tabu list, are performance of the different strategies of TS implementation.
allowed except for the one that satisfies the aspiration criterion. Two scores have been computed: the number of times that
 St Mt Ht St,Mt,Ht,Mt,St Mt,Ht,Mt,St Ht,Mt,St 25

24

SC1 12/20 10/20 8/20 15/20 17/20 14/20 23

22
SC2 12.8s 32.3s 12.2s 343.7s 274.5s 213.8s 21

T(C)
20

TABLE I 19

18

E XPERIMENTAL COMPUTATION 17

16
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Zone 0
25

24

23
1
22
SV1(kW)

21

T(C)
0.5 20

19

18
0
10 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 17
SV2(kW)

16
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Zone 1
0.5

0
20 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Fig. 3. Predicted temperature in rooms
SV3(kW)

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
2
[2] K. P. Wacks, “Utility load management using home automation,” IEEE
SV4(kW)

1 transaction on Consumer Electronics, vol. 37, pp. 168–174, May 1991.

0 [3] ——, “The impact of home automation on power electronics,” in Applied
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Power Electronics Conference and Exposition, 1993, pp. 3 – 9.
2
Ptotal(kW)

[4] P. Peter and P. Ratko, “Demand side management in private home using
1 lonworks,” 1997.
[5] Y. K. Penya, “Last-generation applied artificial intelligence for energy
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 management in building automation,” in Proceedings of the 5th IFAC
International Conference on Fieldbus Systems and their Application,
July 2003, pp. 79–83.
Fig. 2. Energy consumption plan [6] L. Duy Ha, S. Ploix, E. Zamai, and M. Jacomino, “Control of energy
consumption in home automation by ressource constraint scheduling,”
in The 15th International Conference on Control System and Comput-
erScience, Bucharest, Romania, May 25-27 2005.
the best found solution has been retrieved by the different [7] ——, “A home automation system to improve the household energy
strategies (SC1 ), and the CPU time required to find the control,” in INCOM2006 12th IFAC Symposium on Information Control
solution (SC2 ). All TS strategies have converged after almost Problems in Manufacturing, St Etienne, France, 2006.
[8] S. Abras, S. Ploix, S. Pesty, and M. Jacomino, “A multi-agent home
200 iterations. The dynamic variation of tabu tenure gives automation system for power management,” in The 3rd International
more chance to reach the global optimal solution but it takes Conference on Informatics in Control, Automation and Robotics, Se-
more times to reach the solution. For a given case of study, tubal, Portugal, 2006.
[9] P. Lagonotte, “Analyse de la qualité de modèles nodaux réduits à
there are no TS strategy, which guarantees to reach the optimal l’aide de la méthode des quadripôles,” International Journal of Thermal
region in the search space, but all TS strategies have found a Science, vol. 38, pp. 51–65, 1999.
feasible solution. [10] G. Fraisse, C. Viardo, O. Lafabrie, and G. Achard, “Development of
a simplified and accurate building model based on electrical analogy,”
VII. C ONCLUSION Energy and buildings, vol. 1430, pp. 1–14, 2002.
[11] K. K. Andersen, “Modelling the heat dynamics of a building using
In this paper, an approach that manages power consump- stochastic differential equations,” Energy and Building, 2000.
[12] M. Nathan, “Building thermal performance analysis by using mat-
tion in home automation is presented. Household energy lab/simulink,” in Seventh International IBPSA Conference, Rio de
management consists in setting both starting times of timed Janeiro, Brazil, 2001.
services and set points of permanent services. This problem [13] B. Olesen and K. Parsons, “Introduction to thermal comfort standard
and to the proposed new version of en iso 7730,” Energy and Building,
has been formulated as a CSP with two objectives: cost and vol. 34, pp. 537–548, 2002.
comfort criteria. An adaptation of tabu search has solved [14] V. Candas, “Confort thermique,” Technique de l’ingenieurs, vol.
efficiently the household energy management problem. This BE9085, 2000.
[15] J. Y. Boivin, “Demand side management-the role of the power utility,”
mechanism synchronizes the energy consumption in satisfying Pattern Recognition, vol. 28, no. 10, pp. 1493–1497, 1995.
the maximal power constraint and the user comfort remains at [16] U. Bilge, M. Kurtulan, and F. Kira, “A tabu search algorithm for the
a good level. However, the results show that it is difficult to single machine total weighted tardiness problem,” Euroupean Journal
of Operation Reseach, 2006.
tune a TS strategy for all situations. Different strategies have [17] S. C. Brailsford, C. N. Potts, and B. M. Smith, “Constraint satisfac-
to be performed at the same time. tion problems: Algorithms and applications,” vol. European Journal of
The proposed solution makes it possible for the private Operational Research, no. 119, p. 557581, 1998.
[18] F. Glover, “Tabu-search - part 1,” ORSA Journal of Computing, pp. 190–
households to automatically adjust their consumption in order 206, 1989.
to satisfy power constraints and consequently to participate [19] ——, “Tabu-search - part 2,” ORSA Journal of Computing, pp. 4–32,
into a DSM system. 1990.
[20] N. Koji and I. Toshihide, “A tabu search approach to the constraint
R EFERENCES satisfaction problem as a general problem solver,” European Journal of
Operational Research, vol. 106, pp. 599–623, 1998.
[1] B. G. Thomas, “Load management techniques,” in Southeastcon 2000.
Proceedings of the IEEE, April 2000, pp. 139 – 145.