this a soft computing algorithm for solving global minima and maxima problem. we used this algorithm to solve the economic load dispatch problem.to minimize the cost .

1. TEACHING LEARNING BASED
OPTIMIZATION ALGORITHM
(a solution to find global optimization)
Guided by: Prof. L.N. Pathy
Biswaranjan (1321209035)
Jayaprakash(1201209161)
Prajna (120120181)
Rajnikanta(1201209207)
Sherin(1201209158)
Swostik(1201209191)

2. Contents
• Motivation
• Economic load dispatch
• Generators used in power plant
• Methods for solving economic load dispatch
• Introduction to TLBO
• Teacher phase, learner phase, self earning phase
• Advantages and disadvantages
• Progress
• Future work

3. MOTIVATION
This algorithm is purely based to our day-to-day
life , how a student behaves inside a class,
What he learns from the teacher & from his
friends & viceversa overall how it affects him to
optimize his performance (positively).So we
choose to do our project on this concept.

4. ECONOMIC LOAD DISPATCH
Economic load dispatch is a process of scheduling
the required load demand among available
generation units so that the overall cost of generation
is minimized.

5. TYPES OF GENERATORS USED IN
POWER PLANT
1.Hydro power plant – Zero operating cost.
So it is not included in ELD but can be used for
hydro thermal scheduling.
2.Nuclear power plant- Operates at constant load So it
not included in ELD.
3.Thermal power plant
So it come under Economic Load Dispatch
Cost of generation of thermal power plant:
Fi(Pgi)=ai*Pgi
2+bi*Pgi+ci
𝑹𝒔
𝒉𝒓
where Pgi=output of ith unit
ai,bi,ci=constant coefficients for ith unit.

6. Problem formulation
OBJECTIVE FUNCTION :-
Min F(Pg)=total cost
= 𝑖=1
𝑁𝑔
𝐹𝑖 𝑃 𝑔𝑖
subjected to:
1. Equality constraint
Pd = 𝑖=1
𝑁𝑔
𝑃 𝑔𝑖
2.Inequality constraint
Pgi(min) ≤ Pgi ≤ Pgi(max)

7. Where
𝐹𝑖 𝑃 𝑔𝑖 =cost of generation of ith unit
Ng=number of generators
Pd=total load or demand
Pgi(min)= minimum output of ith unit
Pgi(max)=maximum output of ith unit

8. TYPES OF METHODS TO SOLVE ECONOMIC
LOAD DISPATCH PROBLEM
-- CONVENTIONAL METHOD:-
- Lagrangian multiplier method.
- Non-linear based algorithm.
- Integer Programming problem
- Hessian Matrix
- SOFT-COMPUTING METHODS:-
- particle swarm optimization .
- TLBO(Teacher learning based optimization ).
- Genetic algorithm etc.

9. PROBLEM FOR ECONOMIC LOAD
DISPATCH
• PROBLEM:-The fuel cost functions for three
thermal plants in rupees/h are given by
C1 = 500 + 5.3 P1 + 0.004 P1^2 ; P1 in MW
C2 = 400 + 5.5 P2 + 0.006 P2^2 ; P2 in MW
C3 = 200 + 5.8 P3 + 0.009 P3^2 ; P3 in MW
The total load , Pd is 800MW.
Generation limits:
200 =< P1 =< 450 MW
150 =< P2 =< 350 MW
100 =< P3 =< 225 MW

10. TEACHING LEARNING BASED
OPTIMIZATION
• Every individual learns from other individuals
to improve themselves.
• Inspired from class room teaching process
• This algorithm simulates three fundamental
modes of learning
1. Through the teacher (Teacher phase)
2. Interacting with other learners (Learner phase)
3. Through self learning (self learning phase)

11. • TLBO A Population Based
Algorithm
• Group of students Population(any
feasible solution)
• Different subjects Different design
variable
• Result scores Fitness value of
problem
• Teacher Best solution

12. INITIAL POPULATION CREATION
Pgi=(Pgi)min+ Rand (Pgimax-Pgimin)
for i= 1,2,………,(Ng-1)
(Pgi)Ng=Pd - 𝑖=1
𝑁𝑔−1
𝑃 𝑔𝑖
We have taken 20 students in our program but here we have shown the
initial population creation of 3 students.
P1 P2 P3 Cost
Student1 300 300 200 6760
Student2 325 335 140 6749.25 Teacher
Student3 250 350 200 6855

13. Teacher phase
• During this phase teacher gives knowledge to
student .
• Students modify themselves.
Xi,new=Xi,old+r1 (Xteacher- TFXmean)
• Xmean =mean result of the class .
• XTeacher =best learner
• TF = teaching factor=round[1+rand(0,1){2,-1}]
• r1 is the random number

14. • Xmean=[275 325 200]
• Xteacher= [325 335 140]
• X3,new=X3,old+r1 (Xteacher- TFXmean)
=[250 350 200]
+1*([325 335 140]-1*[275 325 200])
=[250 350 200]+[50 10 -60]
=[300 360 140]
Here r1=1 and TF=1
(the value is improved)
Simple Calculation

15. LEARNER PHASE OF TLBO ALGORITHM
• Learners learn from other learners.
• They are chosen randomly or from the
neighbourhood positions.
• Learning from neighbours is easy and
compatible .
• While learning from non-neighbour learners
though difficult improve the search ability
thereby improving the global performance.

16. • The learners are arranged in a M*N vector
• This vector is called position matrix
• Our assumption is position=the number i.e position
of each learner is fixed (for ex. Exam hall sitting
arrangement)
POSITION MATRIX
1 2 3 4
5 6 7 8
9 10 11 12
13 14 15 16
17 18 19 20

17. • Every learner is coded with an integer.
• Thus every learner modified its position by
looking best nearby position as follows
if (Xj > Xk)
Xj,new=Xj,old +rj (Xj-Xk)
else
Xj,new=Xj,old +rj (Xk-Xj)
LEARNER NO. NEAR BY
POSITION
BEST NEAR BY
1 2 , 5 5
2 1, 3, 6 3
3 2, 4, 7 4

18. SELF LEARNING PHASE
• Not every learner includes in this process
• Searching is ambiguous as it is a self
motivated process.
• The equation is
Xi,new(K)=Xi,old(K)+r4(Xi,old(K)-Xi,old(K-1))
• K=iteration number
• r4=random number[0,1]

19. ALGORITHM

20. Advantage of TLBO in comparison to
other conventional methods
 More accurate
 Does not require any derivative.
 Follows the entire path to find its solution.

21. Disadvantages of TLBO
• It consumes lot of memory space.
• It involves lot of iterations so is a time
consuming method.

22. PROGRESS
• Programming Completed
1. population generation
2. Teacher phase
3. Learner phase

23. FUTURE WORKS AND APPLICATION OF
TLBO
• To apply the TLBO in different power system
problems

24. [1] M.Dorigo .v. Maniezzo, A.colorni,Ant system:optimization by a colony of
cooperating agents, IEEE Trans.Syst., Man, Cybern. Part B:Cybern. 26(1)
(1996)29-41
[2] C.S.Suresh, N.Anima, Data clustering Based on Teaching-Learning-Based
optimization SEMCCO 2011 part II, LNCS 7077, 2011, pp. 148-156.
[3] T.Vedat, Design of planer steel frames using teaching-learning based
optimization. Eng. Struct. 34(2012) 225-232.
[4] R.Venkata Rao, V.D. Kalyankar, Parameter optimization of mordern
Machining processes using teaching-learning-based optimization
algorithm.

25. THANK YOU