Fuzzy Control For Improved Buildings Energy Management Systems
Fuzzy Control For Improved Buildings Energy Management Systems
Fuzzy Control For Improved Buildings Energy Management Systems
ABSTRACT: Research on energy efficiency in buildings has proved that the design and even the
sophisticated facilities such as Building Energy Management systems (BEMS), aiming to improve the
indoor environmental conditions while minimising the energy needs, are not sufficient enough due to
users interference. Users are a dynamic part of the building, therefore they should be taken into account
in the control strategy. Latest trends in designing Intelligent Building Energy Management Systems
(IBEMS) integrate a Man Machine Interface that could store the users preferences and adapt the control
strategy accordingly. The objective of the present paper is to present the advantages of the fuzzy
control techniques together with a man machine interface based on a smart card terminal in satisfying
the users preferences. A fuzzy controller is developed and the minimization of energy consumption is
achieved by the use of a suitable cost function for the whole system. The overall control system
including the cost function is modeled and tested using MATLAB/SIMULINK.
KEYWORDS: Building Energy Management System, Man Machine Interface, Fuzzy Controller, Cost
Function
INTRODUCTION
The Energy crisis in the late 70’s combined with the fast development of computers
science has led to the development of the Building Energy Management Systems
(BEMS). The aims of these systems are to monitor and control the environmental
parameters of the buildings and at the same time to minimise the energy consumption
and cost. Since then, BEMS have become commercial tools and are implemented in a
wide range of applications, especially in large office buildings; thus useful experience
is available regarding their benefits and drawbacks.
The objectives of the present paper are to propose the integration of a Man Machine
Interface into BEMS that takes into account the users preferences as a dynamic part of
the system plus to satisfy the criterion of the energy consumption minimization. The
overall control strategy is developed using fuzzy techniques.
CONTROL STRATEGY
The aim of the controller is to maintain users comfort over the zone level. Comfort
always reflects the subjectivity of the users and is the integration of:
(i) Thermal comfort
(ii) Visual comfort
(iii) Indoor air quality
In this specific application the controlled parameter for thermal comfort, is the
Predicted Mean Vote (PMV) introduced by Fanger [1] which depends upon:
(a) The temperature
(b) The relative humidity
(c) The mean radiant temperature
(d) The air velocity
(e) The activity level
(f) The clothing parameter (clo).
The first four parameters are directly measured while the next two are estimated on
the basis of the building type and activities.
The PMV index varies in the range of –3 to +3, while the span of thermal satisfaction
lies between –0.5 to +0.5 with the Percentage of People Dissatisfied (PPD) index,
introduced also by Fanger [1], to be less than 5%.
The controlled parameter selected for estimating visual comfort is the illuminance
level, measured in lux [4].
The indoor air quality is mainly influenced by the concentration of pollutants in the
controlled space. CO2 concentration (measured in ppm) is selected for this
application, as it represents the presence of users as well as various sources of
pollutants in the building.
A fuzzy controller approach [5] has been selected for the control of the above
parameters for the following reasons:
i. The limits of comfort and discomfort are fuzzy and subjective.
ii. Oscillations and overshootings have to be avoided in order to minimise dis-
comfort and energy waste.
The fuzzy controller for each of the controlled parameters (PMV, Illuminance levels,
and CO2 concentration) takes the following inputs:
- The error between the desired and the current value of the parameter at time nT.
- The error at time (n-1)T.
The outputs of the controller are:
(i) Increase / decrease of heating / cooling.
(ii) Increase / decrease of window opening or ventilator.
(iii) Increase / decrease of shading adjusting the daylight contribution.
(iv) Increase / decrease of electric lighting.
Thus the fuzzy controller has six inputs, four outputs and 27 rules.
The membership functions used for this application for each controlled parameter are
of triangular and trapezoidal type [6].
Three membership functions were selected per input for each of the controlled
parameters around the regions zero (Ze), negative (Neg) and positive (Pos).
Five membership functions are used per output (Zero, Small Positive, Small Negative,
Positive, Negative).
The membership functions of the PMV error at time nT (pmve0) and the error at time
(n-1)T (pmve-1) are illustrated in Fig. 2.1.2.a. and 2.1.2.b. The membership functions
for heating / cooling are illustrated in Fig.2.1.2.c.
Figure 2.1.2.a., b Membership functions for the‘pmve0’ and ‘pmve-1’ inputs
The membership functions of the CO2 concentration error at time nT and (n-1)T, as
well as the membership functions for the illuminance levels are of the same type with
those of PMV. The membership functions of the outputs of the PI controller are of the
same type with those for heating/cooling and are illustrated in Fig.2.1.2.c.
Figure 2.1.2.c. Membership functions for heating/cooling outputs of the fuzzy controller
The controller has been modeled using the MATLAB/SIMULINK Software. The
model of the building has been developed by the University of Athens in the
framework of the Joule III Research Programme JOE3 – CT 97 0044.
THE ROLE OF THE COST FUNCTION
Having determined the acceptable limits of the parameters f1, f2, and f3, random
generators are used to generate values within the acceptable limits. The objective is
the determination of the optimum limits of the cost function as a whole. The weights
are estimated using PCA and the resulting values are:
w1 = 0.4462 w2 = 0.4432 w3 = 0.4541
w4 = 0.4396 w5 = 0.4527
The optimum cost function lies between 0 and 2 as shown in Fig.2.2.a. The optimum
energy use per day is modelled using MATLAB / SIMULINK. For heating / cooling
an optimum energy consumption of 200 kWh/m2 per year is considered, while for
electric lighting the level is 100 kWh/m2 per year (Fig. 2.2.b, 2.2.c.).
The daily energy consumption computed by the model is divided by the optimum
energy consumption per day to provide the parameters f4 and f5. The simulation results
for the cost function for three winter days are presented in section 3.
Figure 2.2.a. Optimum values of the cost function
Figure 2.2.b,c. Energy consumption in kWh/m2 for heating/cooling and electric lighting per day over
one year period
SIMULATION RESULTS
The described fuzzy controller is tested using MATLAB / SIMULINK. The response
of the controlled parameters are shown in Fig. 3.1. for three typical winter days. The
set point in Fig. 3.1. is PMV = 0, [CO2] = 800 ppm and illuminance = 500 lux.
The users preferences are modelled using random generators providing various set
points for the PMV, [CO2] and illuminance levels within the acceptable limits. The
PMV set point is changing once every 3 hours while the [CO2] once every 7 hours,
considering that if the pollutants concentration is kept within the acceptable limits,
people are satisfied.
The simulation results for an one day period are illustrated in Fig. 3.2. The PMV is
approaching the set point smoothly and without any overshootings or oscillations.
Fig. 3.1. Fuzzy controller response for fixed set points
The [CO2] concentration follows the desired set point accurately. As it can be
observed after the time step 600, the set point is higher than the [CO2] concentration
in the space, but due to the absence of occupants the CO2 level cannot be increased.
As far as the lighting levels are concerned, the controlled parameter follows the
desired values, quite satisfactory. The overshooting observed at time step 200 is due to
the immediate increase of lighting levels because of the daylight contribution. During
the night, the desired lighting levels cannot be reached due to low level of electric
lighting used by the model.
The simulation results for the cost function for three winter days is depicted in Fig.3.3.
The cost function is kept within the acceptable limits. This may not be always true
when the system is in actual operation. For that reason, after the installation of the
system and during the on line operation, the cost function is estimated every 1/4 hour.
When the cost function is out of the acceptable limits, the responsible parameter is
punished by reducing the weight of the rules responsible for its control.
Fig. 3.3.Cost function for three days simulation of the fuzzy controller
CONCLUSIONS
The described BEMS integrates the users preferences and ensures energy saving by
the use of a suitable cost function. The energy consumption per m2 for both
heating/cooling and lighting is below the levels recommended by the bibliography.
The users preferences are satisfied without oscillations and overshootings by the use
of the fuzzy controller. The performance of the system in real time operation will
contribute to the improvement of the overall system performance.
ACKNOWLEDGEMENTS
REFERENCES