IEEE Systems, Man and Cybernetics Magazine - April 2021 - 31

We adopt PSO, ABC, and ICS as the strategies to conduct
10 independent experiments on different types of classrooms and occupancies. Then, the average power consumption of each strategy in the 10 experiments is obtained, as
depicted in Figures 12-14.
1)	 Lecture halls: According to our questionnaire statistics,
the average occupancy of lecture halls does not exceed 20
persons per room. The current power consumption of lecture halls on campus without the scheduling of electricity
is calculated as follows: the lecture halls with 20 persons
per room are powered on for 5 h to accommodate 1,000,
2,000, and 3,000 occupants, and their power consumption
is correspondingly about 1,087.5, 1,174.5, and 1,174.5  KWh,
respectively. However, if one of the three evolutionary
algorithms is employed to schedule the aforementioned
same classrooms, their power consumption can be saved
by more than 82.2, 66.7, and 49.2%, respectively.
2)	 Large classrooms: The average occupancy of large classrooms does not exceed 10 persons per room. C
- ompared
with their current power consumption under no scheduling, if one of the three evolutionary algorithms is employed,
the power consumption can be saved by more than 82.0,
80.3, and 67.3%, when the c- orresponding numbers of
occupants are 500, 1,000, and 1,500, respectively.

◆◆ PSO: The particle population is 50. The weight is 0.9 and,

the two constants for the particle velocity update are 1.5
and 2, respectively.
◆◆ ABC: The number of the colony size is 50. The number
of both the employed and onlooker bees is 25. The number of scout bees is 1. The number of food sources is 25.
We set 100 as the total number of occupants in small
classrooms. Three simulations via different scheduling strategies were conducted for the small classrooms as follows.
1)	 When the PSO scheduling strategy is applied, we have
i 1 = 4 and H 1 = 7. When rs 1 # i 1 and S 1 # H 1 are satisfied, one additional small classroom is scheduled to be
powered on.
2)	 When the ABC scheduling strategy is applied, we have
i 1 = 3 and H 1 = 7. When rs 1 # i 1 and S 1 # H 1 are satisfied, one additional small classroom is scheduled to be
powered on.
3)	 When the ICS scheduling strategy is applied, we have i 1 = 3
and H 1 = 8. When rs 1 # i 1 and S 1 # H 1 are satisfied, one
additional small classroom is scheduled to be powered on.
In the case of other numbers of occupant traffic and the corresponding type of classrooms, the scheduling also follows
the strategy provided in Table 1, and if the conditions are
met, a classroom of the same type will be powered on.

Table 1. The simulation and scheduling strategies under different occupant traffic.
Types

Small Classroom

Occupants

100

Large Classroom

200

300

500

Lecture Hall

1,000

1,500

1,000

2,000

3,000

Threshold

i3

H3

i3

H3

i3

H3

i2

H2

i2

H2

i2

H2

i1

H1

i1

H1

i1

H1

PSO

4

7

3

10

5

12

5

19

3

30

8

45

3

30

8

55

8

81

ABC

3

7

3

8

4

13

6

17

8

29

7

44

8

29

9

60

9

79

ICS

3

8

4

7

4

12

5

18

7

32

6

40

7

32

9

50

10

72

Lecture Halls
1,174.5

592.62

596.17
391.07
193.87

389.98
191.47

3,000

2,000

592.04
385.19

588.19
383.38

191.11

3,000

189.81

2,000

N

S
IC

S

N

+

o

LS

Sc

TM

-R

N

IC

C
AB

O
PS

lin
g

1,000

Number of Occupants

1,000

1,087.5

1,200
1,000
800
600
400
200
0

he
du

Power Consumption (kWh)

1,174.5

Figure 12. The power consumption under different occupancy levels in lecture halls (5 h).

	

Ap ri l 2021

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE	

31
IEEE Systems, Man and Cybernetics Magazine - April 2021

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - April 2021
