You are on page 1of 24




The objectives of the Load Frequency Control (LFC) are to distribute the load between
generators and to control the tie-line power to pre-specified values and to maintain sensibly
uniform frequency. In order to supply reliable electric power with good quality, LFC in power
system is very important. Constant frequency is identified as the mark of a normally operating
system. A power plant got to monitor the load conditions and serve consumers entire day. It is
therefore irrelevant to consider that uniform power is generated throughout. So, depending on
load demand, power generation varies. The objective of control strategy is to deliver and
generate power in an interconnected system as reliably and economically as possible while
maintaining the frequency and voltage within the limits. The system frequency is mainly
affected due to change in load, while reactive power depends on changes in voltage magnitude
and is less sensitive to frequency. To keep the frequency constant PID controller is used which
controls the turbines used for tuning the generators and also the steady state error of system
frequency is reduced by tuning the controller gains. There are different algorithms to optimize
the controller gains for load frequency control of an interconnected power system like PSO.
Since its invention, many PSO variants have been proposed by modifying its solution update
rule to improve its performance. The subtraction-based solution update mechanism has caused
premature convergence and stagnation in particle population during the iterations. To overcome
these issues, this study presents a distribution-based update rule for PSO algorithm. The
experimental results and comparisons show that PSOd is a better version of PSO.


To maintain power with an acceptable quality is the main objective of power system operation
and control. The problem of the load frequency control (LFC) is one of the most important areas
in the interconnected power systems. Power system is a complex system, nonlinear and is
subjected to different kinds of events. Frequency of a power system needs to be kept constant
for reliable power supply despite of fluctuations in load. Recently, many different control
algorithms have been proposed for LFC. PSO a population based technique first described by
James Kennedy and Russell C. Eberhart (1995) [1]. This applies the concept of social
interaction to problem solving. Generally, PSO is opted as it is easy to implement,
computationally efficient and simple in concept. Hence PSO is successfully applied to tune the
parameters of controller which helps in achieving the objective of keeping the frequency
constant by minimising the objective function. Since its invention, many PSO variants have
been proposed by modifying its solution update rule to improve its performance. The
subtraction-based solution update mechanism has caused premature convergence and stagnation
in particle population during the iterations. To overcome these issues, a distribution-based
update rule for PSO algorithm was developed which not only minimises the objective function
but also prevent premature convergence. [2] Here PID is used for the application of Load
Frequency Control (LFC) of an interconnected power system. The Integral of Time multiplied
by Absolute Error (ITAE) is a performance index used to design control system. The index was
proposed by Graham and Lathrop (1953), who derived a set of normalized transfer function
coefficients to minimize the ITAE criterion for a step input.


As we know 50 Hertz is normal operating frequency in India and if there is a variation of ±2.5
hertz then it is going to seriously affect the entire system. For example, turbine blades are prone
to get damaged in such condition. Also, there is a relation between frequency and motor speed
which is also going to be affected by frequency variation.

The objective of this work is

 To design a controller based on the optimized parameters obtained from PSOd

algorithm for restricting the value of change in frequency to a constant against any
variation in load demand.
 The power flow through the tie line of each area must be maintained to its pre-
specified value.
 Minimise the error of the system.

 Also compare it with the PSO algorithm


Chapter 1 Reviews the literature on load frequency control (LFC) of power system and

necessity of frequency control. Motivation and objective along with brief description of the work

is presented.

Chapter 2 Describes the load frequency control (LFC) of two area interconnected power system,

need for maintenance of constant frequency. The major components of power system are

described with mathematical modeling. Objective function and PID controller are described


Chapter 3 Discusses recent algorithms on optimization PSOd. Algorithm of PSOd are included

in this.

Chapter 4 Comparison of PSO and PSOd is done. Table 4.3 Shows the simulation results of

PSO and PSOd control algorithm of two area interconnected power system. Simulations were

performed using MATLAB Simulink. Step load disturbance is applied in areas for PSOd based

controller and tie- line power flows and frequency oscillations are observed.

Chapter 5 Gives conclusion of the Project

Chapter 6 Gives references of the Project




If the system is connected to a number of different loads in a power system, then the system
frequency and speed change with the governor characteristics as the load changes. If it is not
required to keep the frequency constant in a system, then the operator is not required to change
the setting of the generator. But if constant frequency is required the operator can adjust the speed
of the turbine by changing the governor characteristic as and when required. If a change in load is
taken care by two generating stations running at parallel, then the complexity of the system
increases. The possibility of sharing the load by two machines is as follow:
 Suppose there are two generating stations that are connected to each other by tie
line. If the change in load is either at A or at B and the generation of A is alone asked
to regulate so as to have constant frequency, then this kind of regulation is called
Flat Frequency Regulation.

 The other possibility of sharing the load the load is that both A and B would regulate
their generations to maintain the constant frequency. This is called parallel
frequency regulation.
 The third possibility is that the change in the frequency of a particular area is taken
care of by the generator of that area thereby the tie-line loading remains the same.
This method is known as flat tie-line loading control.
 In Selective Frequency control each system in a group is takes care of the load
changes on its own system and does not aid the other systems un the group for
changes outside its own limits.

 In Tie-line Load-bias control all the power systems in the interconnection aid in
regulating frequency regardless of where the frequency change originates. The
equipment consists of a master load frequency controller and a tie line recorder
measuring the power input on the tie as for the selective frequency control.
The error signal i.e. Δf and ΔPtie are amplified, mixed and transformed to real power
command signal ΔPV which is sent to the prime mover to call for an increase in the
torque. The prime mover shall bring about a change in the generator output by an

amount ΔPG which will change the values of Δf and ΔPtie within the specified
tolerance. The first step to the analysis of the control system is the mathematical
modelling of the system’s various components and control system techniques [3].



Applying the swing equation of a synchronous machine to small perturbation, we have:

2𝐻 𝑑 2∆𝛿
= ∆𝑃𝑚 − ∆𝑃𝑒
𝜔 𝑑𝑡 2
𝑑∆𝜔 1
Or in terms of small deviation in speed = (∆𝑃𝑚 − ∆𝑃𝑒 )
𝑑𝑡 2𝐻

Taking Laplace Transform, we obtain

∆Ω(𝑠) = [Δ𝑃𝑚 (𝑠) − Δ𝑃𝑒 (𝑠)]

Fig.1 Mathematical modeling block diagram for generator

b) Load Model
The load on the power system consists of a variety of electrical drives. The
equipment used for lighting purposes are basically resistive in nature and the rotating
devices are basically a composite of the resistive and inductive components. The
speed-load characteristic of the composite load is given by
Δ𝑃𝑒 = Δ𝑃𝐿 + 𝐷Δ𝜔


ΔPL is the non-frequency- sensitive load change, 𝐷Δ𝜔is the frequency sensitive load
D is expressed as percent change in load by percent change in frequency.

Fig.2: Mathematical modeling Block Diagram for Load

c) Prime mover model

The source of power generation is commonly known as the prime mover. It may be
hydraulic turbines at waterfalls, steam turbines whose energy comes from burning of the
coal, gas and other fuels. The model for the turbine relates the changes in mechanical
power output ΔPm to the changes in the steam valve position ΔPV

Δ𝑃𝑚 (𝑠) 1
𝐺𝑇 = =
Δ𝑃𝑉 (𝑠) 1 + 𝜏 𝑇 𝑠

Where T, the turbine constant is, in the range of 0.2 to 2.0 seconds.

d) Governor model

When the electrical load is suddenly increased then the electrical power exceeds
the mechanical power input. As a result of this the deficiency of power in the load
side is extracted from the rotating energy of the turbine. Due to this reason the
kinetic energy of the turbine i.e. the energy stored in the machine is reduced and the
governor sends a signal to supply more volumes of water or steam or gas to
increase the speed of the prime-mover so as to compensate speed deficiency.

Fig.3: Graphical Representation of speed regulation by governor

The slope of the curve represents speed regulation R. Governors typically have a speed regulation
of 5-6 % from no load to full load.
Δ𝑃𝑔 = Δ𝑃𝑟𝑒𝑓 − Δ𝑓

Or in s- domain
Δ𝑃𝑔 (𝑠) = Δ𝑃𝑟𝑒𝑓 − ΔΩ(𝑠)

The command ΔPg is transformed through hydraulic amplifier to the steam valve position
command ΔPV. We assume a linear relationship and consider simple time constant we have the
following s-domain relation:
Δ𝑃𝑉 (𝑠) = Δ𝑃
1 + 𝜏𝑔 𝑔(𝑠)

Combining all the block diagrams from earlier block diagrams for a single are system we get the

Fig.4: Mathematical Modelling of Block Diagram of single system consisting of Generator, Load, Prime
Mover and Governor

C. Two Area Interconnected Power System

The connection between power systems is made possible via tie-lines. Tie-line allows
the flow of electric power between areas. Area will obtain energy with the help of tie-
lines from other areas, when load change occurs in that area. Hence LFC also needs to
control the tie-line power exchange error. Tie-line power error are the integral of the
frequency difference in between two areas. Tie-line power can be written
mathematically as

|𝑉10 ||𝑉20 |
𝑃12 = sin(𝛿10 − 𝛿20 )

Where, 𝛿10 𝛿20 =power angles of equivalent machines

For small deviations in the angles the tie-line power changes to

P12=T12 1 2

|𝑉10 ||𝑉20 |
Where, 𝑇12 = cos(𝛿10 − 𝛿20 ) is the synchronizing coefficient

1 𝑑
Frequency deviation ∆f is related to reference angle by ∆𝑓 = (𝛿 0 + ∆𝛿)
2𝜋 𝑑𝑡
1 𝑑
∆𝑓 = (∆𝛿)
2𝜋 𝑑𝑡

Δ𝛿 = 2𝜋 ∫ Δ𝑓𝑑𝑡

Δ𝑃12 = 2𝜋𝑇12(∫ Δ𝑓1𝑑𝑡 − ∫ Δ𝑓2 𝑑𝑡)

Taking Laplace transformation of above formula gives

∆𝑃12 (𝑠) = (∆𝑓1(𝑠) − ∆𝑓2(𝑠))

Fig.5: Linear representation of tie-line

Similarly, T21 can be written in terms ofT12 as T21 a12T12

So, for control area 2
Δ𝑃21 (𝑠) = (Δ𝑓1(𝑠) − Δ𝑓2 (𝑠)

Fig. 6 Block diagram of two area interconnected system

D. Area Control Error
Control error of each area consists of linear combination of tie line flows and frequency.
ACE represents a mismatch between area generation and load (AGC). The objective of
LFC is to minimize the error in frequency of each area as well as to keep the tie-line error
to scheduled value which is quite difficult in presence of fluctuating load. If we control the
error in frequency back to zero, any steady state errors in the frequency of the system would
result in tie-line power errors because the error in tie-line power is the integral of the
frequency change between each pair of areas. Therefore, it is needed to consider the
information of the tie-line power deviation in control input. As a result, an error called ACE
is defined as

𝐴𝐶𝐸𝑖 = ∑𝑛𝑗=1 ∆ 𝑃𝑡𝑖𝑒,𝑖𝑗 + 𝐵𝑖 ∆𝑓𝑖


𝐴𝐶𝐸𝑖 is the 𝑖 𝑡ℎ area control error

∆𝑓𝑖 =𝑖𝑡ℎ area frequency error
∆𝑃𝑡𝑖𝑒,𝑖𝑗 =power flow error in tie line between 𝑖 𝑡ℎ and 𝑗𝑡ℎ area

Bi=𝑖𝑡ℎ area frequency bias coefficient.

The input to controller is area control error having objective of controlling the ACE and the
frequency deviation. Now which controller is to be taken into consideration depends on
performance of controller and the requirement of process.

E. Controllers
The fundamental control loop can be simplified for a SISO (single-input-single-output)
system as in Fig. 2.5 Here we are ignoring the disturbances in the system.

Fig.7 Basic control loop

The controller may have different structures. But the one of most popular controller in all is
Proportional-Integral-derivative (PID) type controller. In fact, more than 95% of the
industrial controllers are of PID type.
The transfer function of the controller is given by:

𝐶 (𝑠) = 𝐾𝑝 (1 + 𝜏𝑑 𝑠 + )
𝜏𝑖 𝑠

𝐾𝑝 =Proportional gain

𝜏𝑑 =Derivative time

𝜏𝑖 =Integral time

In this the effects of the individual components- proportional, derivative and integral on the
closed loop response of this system are explained.
1. In case of proportional controller, the time response improves (i.e. the time constant
decreases) and there is offset between the output response and desired response. By
increasing the proportional gain, this offset can be reduced; but that may also cause
increase oscillations for higher order systems.

2. When only integral action of controller is considered with integral controller, the order
of the closed loop system increases by one. This increase in order may cause instability
of the closed loop system, if the process is of higher order dynamics. The major
advantage of this integral control action is that it reduces steady state error to zero due
to step input. But simultaneously, the system response is in general oscillatory, slow as
well as even sometimes unstable.
3. PI gives the double advantages of fast response due to P-action and the zero-steady
state error because of I-action. By using P-I controller, the steady state error can be
carried down to zero, and simultaneously, the transient response can be improved.
4. P-D controller apparently is not very useful, since it cannot reduce the steady state
error to zero. But for higher order processes, it can be shown that the stability of the
closed loop system can be improved using P-D controller.

Suitable combination of proportional, integral and derivative actions can provide all the
desired performances: fast response, zero steady state error and less offset. In this order is
low, but is universally applicable as it can be used in any type of system. PID controllers
have also been found to be robust, and that is the reason, it finds wide acceptability for
industrial processes.

F. Objective function
In this adjustment of parameters is done using optimization and objective function which is
a function of error and time and the function used is integral of time-multiplied absolute
error criterion (ITAE). Another objective function which is also a function of error and time
is there known as integral of the square of the error criterion (ISE) but this performance
index is not taken into consideration because this one is computationally not comfortable
and is less sensitive in comparison to ITAE. Whereas ITAE has the benefits of producing
less oscillations and smaller overshoots, maintain robustness and in addition to that it is
most sensitive i.e. best selectivity this makes the ITAE index the desirable criterion used for
design of control system. Focus is on minimizing the ITAE criterion. ITAE is composed of
tie line power and frequency deviation of both areas. The objective function is

𝐽 = ∫ 𝑡( |∆𝑓1| + |∆𝑓2| + |∆𝑃𝑡𝑖𝑒 |)𝑑𝑡

Where, ∆f1 and ∆f2 are frequency deviations; ∆𝑃𝑡𝑖𝑒 is the change in tie line power Since
integrating up to infinity is not practicable a large value of T should be chosen such that
error is negligible. Here T=10 seconds istaken.


A. Optimization Algorithm

Keeping in view the effort required and time consumed to get optimum values of
the controller by the conventional methods motivates us to go for an advanced method
which includes optimization algorithm based on natural processes. There are many
optimization algorithms that have been proposed or developed till date.

B. Particle Swarm Optimization (PSO)

Particle swarm optimization (PSO), originated by James Kennedy and R.C. Eberhart in
1995. It is a stochastic (connection of random variable) evolutionary computation method
used to explore search space. This technique is based on swarm’s intelligence and
movement. As this is based on swarm behavior, is a population based technique. The bird
generally follows the shortest path for food searching. Based on this behavior, this
algorithm is developed. It uses a number of particles where every particle is considered as a
point in N-dimensional space. Each particle keeps on accelerating in the search space
depending on the knowledge it has about the appreciable solution comparing its own best
value and the best value of swarm obtained so far. It is well described by the concept of
social interaction because each particle search in a particular direction and by interaction
the bird with best location so far and then tries to reach that location by adjusting their
velocity this require intelligence.
Given below are the two main equations of PSO algorithm:
Velocity modification equation:

𝑣𝑖𝑘+1 = 𝑤𝑣𝑖𝑘 + 𝑐1𝑟𝑎𝑛𝑑1 × (𝑝𝑏𝑒𝑠𝑡1 − 𝑠𝑖𝑘 ) + 𝑐2 𝑟𝑎𝑛𝑑2 × (𝑔𝑏𝑒𝑠𝑡1 − 𝑠𝑖𝑘 )

Where, 𝑣𝑖𝑘 = velocity of agent i at iteration k

w =weighing function
𝑐1=weighing factor
𝑟𝑎𝑛𝑑1 =random number between 0-1
𝑝𝑏𝑒𝑠𝑡1 =personal best of agent i
𝑠𝑖𝑘 =current position of agent i at iteration k
𝑔𝑏𝑒𝑠𝑡1 =g-best of the group


Step1: The initial particles are set to some linear position in the range of 𝐾𝑝 and𝐾𝑖 .

Step2: Their velocities are set to zero.

Step3: Initial ITAE is set to some values.

Step4: Evaluate the ITAE for the particles at their corresponding positions.

Step5: Initialize 𝑝𝑏𝑒𝑠𝑡1 for each particle.

Step6: Find 𝑔𝑏𝑒𝑠𝑡1 based on minimum ITAE.

Step7: Start iteration 1.

Step8: Update the positions.

Step9: Then calculate ITAE at their corresponding position.

Step10: Accordingly update 𝑝𝑏𝑒𝑠𝑡1 and 𝑔𝑏𝑒𝑠𝑡1 basedon ITAE.

Step11: Update velocity.

Step12: Iteration=iteration+1.

Step13: If iteration<=maximum iteration, go to step 8 otherwise continue.

Step14: The obtained 𝑔𝑏𝑒𝑠𝑡1 is the optimum set of parameters of PI controller.

C. Distributed Particle Swarm Optimization (PSOd)
Proportional-plus-integral-plus-derivative controller is designed here based on Distributed
Particle Swarm Optimization (PSOd) for controlling the frequency deviation which is a
major problem of a two-area interconnected power system. In order to improvise the
performance of supplying power of a power system, error function is minimized. The
objective function taken into consideration over here is Integral Time multiplied with
Absolute Error (ITAE). To optimize the gain values of controller, the PSOd algorithm is
used. For D-dimensional search space and N particles, Let X be particle the population,
𝑝𝑏𝑒𝑠𝑡1 be the personal information or self-best solution obtained so far, 𝑔𝑏𝑒𝑠𝑡1 be the
best solution obtained by the particle population so far and V be the velocities of the
particles. X, 𝑝𝑏𝑒𝑠𝑡1 and V are N×D matrix and 𝑔𝑏𝑒𝑠𝑡1 is 1×D vector. For initialization
of the particle population, Eq. (3.1) is used.

𝑋𝑖,𝑗 (0) = 𝐿𝑗𝑚𝑖𝑛 + 𝑟𝑖 × (𝑈𝑗𝑚𝑎𝑥 − 𝐿𝑗𝑚𝑖𝑛 )

i= 1, 2, ..., N and j = 1, 2, ..., D ………………………………(Eq.3.1)

where, 𝐿𝑗𝑚𝑖𝑛 the lower bound of the search space for jth dimension, 𝑈𝑗𝑚𝑎𝑥 is the upper
bound of the search space for jth dimension and 𝑟𝑖 is a random number produced for each
particle’s each dimension, in range of [0,1].

[𝐿𝑗𝑚𝑖𝑛 + 𝑟𝑖 × (𝑈𝑗𝑚𝑎𝑥 − 𝐿𝑗𝑚𝑖𝑛 ) − 𝑋𝑖,𝑗 (0)]
𝑉𝑖,𝑗 (0) =
i=1, 2…., N and j=1, 2,…,D ..........……………………………………..…...(Eq.3.2)

The initialization of velocities of the particles depends on both the upper and lower bound
of search space and the current particle positions. In the initialized stage, the current
particle positions are assigned as self-best solution 𝑝𝑏𝑒𝑠𝑡1 of the particles by using Eq.

𝑝𝑏𝑒𝑠𝑡𝑖 (0) = 𝑋𝑖 (0)…………………………………………….…(Eq.3.3)

The best solution of the population in the initialized phase is determined using Eq. (3.4).

𝑔𝑏𝑒𝑠𝑡𝑖 = arg min {f (𝑝𝑏𝑒𝑠𝑡𝑖 )} i= 1, 2, ..., N…………………………………(Eq.3.4)

PSO algorithm is an iterative algorithm, and until a pre-determined termination is met,

Eqs.(3.5), (3.6), (3.7) and (3.4) are executed.

1 2
𝑉𝑖,𝑗 (𝑡 + 1) = 𝑤 × 𝑉𝑖,𝑗 (𝑡) + 𝑟𝑖,𝑗 × 𝑐1 × (𝑝𝑏𝑒𝑠𝑡𝑖,𝑗 (𝑡) − 𝑋𝑖,𝑗 (𝑡)) + 𝑟𝑖,𝑗 × 𝑐2 ×

(𝑔𝑏𝑒𝑠𝑡𝑗 (𝑡) − 𝑋𝑖,𝑗 (𝑡))………………………………………….…(Eq.3.5)

𝑋𝑖,𝑗 (𝑡 + 1) = 𝑋𝑖,𝑗 (𝑡) + 𝑉𝑖,𝑗 (𝑡 + 1) 𝑖 = 1,2, … . . 𝑁 𝑎𝑛𝑑

𝐽 = 1,2, … . . 𝐷………………………………………………. Eq. (3.6)

𝑋𝑖 (𝑡 + 1) 𝑖𝑓 𝑓(𝑋𝑖 (𝑡 + 1))𝑖𝑠 𝑏𝑒𝑡𝑡𝑒𝑟 𝑡ℎ𝑎𝑛 𝑝𝑏𝑒𝑠𝑡𝑖 (𝑡)

𝑝𝑏𝑒𝑠𝑡𝑖 (𝑡 + 1) = { (Eq.3.7)
𝑝𝑏𝑒𝑠𝑡𝑖 (𝑡) 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

In Eq. (3.5), w is inertia weight and it is not in the basic PSO algorithm, but it is used in all
contemporary versions of PSO algorithm.
To cope with loss of diversity in the particle population, a new position update rule based
on the normal distribution is proposed. In the new approach the velocity update rule is
removed and Eq. (3.8) is used for obtaining a new position for a particle, based on the
normal distribution given as follows:

𝑋𝑖,𝑗 (𝑡 + 1) = 𝜇 + 𝜎 × 𝑍………………………………………………. (Eq.3.8)

where, µ (mean) is calculate by using Eq. (3.9), σ (standard deviation) is calculated by

using Eq. (3.10) and Z is calculated by using Eq. (3.11).

𝑋𝑖,𝑗 (𝑡)+𝑝𝑏𝑒𝑠𝑡𝑖,𝑗 (𝑡)+𝑔𝑏𝑒𝑠𝑡𝑗 (𝑡)

𝜇= ………………………………..……(Eq.3.9)

𝜎 = √ × [(𝑋𝑖,𝑗 (𝑡) − 𝜇)2 + (𝑝𝑏𝑒𝑠𝑡𝑖,𝑗 (𝑡) − 𝜇)2 + (𝑔𝑏𝑒𝑠𝑡𝑗 (𝑡) − 𝜇)2 ]…(Eq.3.10)

𝑍 = (−2 ln 𝑘1 )2 × cos(2 π 𝑘2 )………………………………………….…. (Eq.3.11)

where, 𝑘1 and𝑘2 are uniform random numbers produced in range of [0,1].

To solve a continuous optimization problem, after Eq. (3.1)– (3.4) are calculated, Eqs.
(3.11), (3.10), (3.9), (3.8), (3.7), (3.4) are iterated until a pre-defined termination condition
is met. As seen from the steps of the algorithm, the simple structure of the basic PSO
algorithm is kept in the proposed approach. In the proposed approach, it is an issue the
standard deviation is0. To solve this issue, we used a threshold value (T) for the standard
deviation. If T is less than 1E-05, the standard deviation is accepted as half of the mean.
Therefore, the new position obtained for the particle can be different from current position
of the particle.



Then simulation work is done for two area interconnected power system according to its
block diagram as shown in Fig.6 and considering transfer function of each block simulation
is done. The parameter values used in the simulation are taken from [2] and tabulated in the
appendix. Then PSO and PSOd algorithm are applied to get the value of parameters of
controller where ITAE value is minimum. The values of different parameters used in PSO
and PSOd algorithms are mentioned in the Table 1. The table 2 gives the values of
optimized control parameters and error for each method applied.

Table 1 Parameter values for PSO and PSOd Algorithm

Parameters Values

Population Size 50

Numbers of iterations 10

Inertia Weight (w) 1.0

Cognitive Coefficient (C1) 1.5

Social Coefficient (C2) 2.0


 F1(Hz)



-0.15 PSO PID

0 5 10 15
Fig 8 Frequency deviation of Area 1



 F2(Hz)




-0.1 PSOd PID

0 5 10 15 20 25 30

Fig 9 Frequency deviation of Area 2


 Ptie(p.u)


0 5 10 15 20 25 30

Fig 10 Tie line power deviation of Area 2

The model of the system under study as shown in Fig 6 is developed in
MATLAB/SIMULINK environment and PSO and PSOd program written in (m.file).
The developed model is simulated in a separate program (by .m file) considering a 10%
step load change in area 1. The objective function is calculated in the .m file and used in the
optimization algorithm. For the implementation of PSO and PSOd, the parameters required
are taken from Table 4. The optimization was repeated 30 times and the best results were
chosen and tabulated.
To show superiority of the proposed PSOd algorithm the results obtained by taking the
appropriate parameters [9] were compared with the results of PSO [10]. From the above
simulation and result we found the values of frequency deviation of area -1, frequency
deviation of area-2 and the tie line power deviation. The results were tabulated in Table 2.
Comparing the values, we see that the ITAE value of PSOd is much smaller than PSO.
Also, the settling time and overshoot value was calculated for both PSO and PSOd
algorithms. The results were tabulated in Table 3. On comparing the values of settling time
and overshoot we found that the respective values were less for PSOd than PSO.
From the above we can observe clearly that in PSOd based controller the Settling Time, Peak and
Peak time, overshoot values are much more improved than the PSO based controller. It is clear that
the PSOd applied PID gives a much better controlling results then the normal PSO applied PID
controlling. These results help us to have an optimal control of the two-area system.

Table 2 Error values for corresponding methods

Method Kp Ki Kd ITAE
PSO -0.223 1.423 0.426 0.743
PSOd 0.166 1.8357 0.495 0.399

Table 3 Settling time and overshoot for frequency deviation and tie line power deviation


Δf1 Δf2 ΔPtie Δf1 Δf2 ΔPtie
0.0428 0.0291 0.8480 8.3955 8.1582 8.0821
PSOd 0.0368 0.0222 0.4002 5.4333 5.9110 6.0124



Controlling of power systems in order to meet the demands of consumers is a challenging

task that motivates to design optimum controllers. They should have the capability of
monitoring the power system like maintenance of frequency and voltage in no time. Many
optimization techniques are used in the design of controllers. Chapter 1 gives a brief
introduction to load frequency control (LFC) of power system and necessity of frequency
control. In Chapter 2 the major components of power system were described with
mathematical modeling. Objective function was also described. In Chapter 3 we discussed
PSOd based algorithm applied to PID. Simulations were performed using MATLAB
Simulink In Chapter 4 we compared the results of PSO and PSOd. Different plots of
frequency deviation were obtained by varying the load demand of areas. From simulation
and result we found that PSOd based PID controller is having improved system
performance in terms of settling time, maximum overshoot compared to PSO based PID
controller. It is clear that the PSOd applied PID gives a much better controlling results than
the normal PSO applied PID controlling. This signifies the superiority of PSOd over PSO.


Nominal parameters of Two-Area System [10]

Parameter Value

B1, B2 0.425 p.u. MW/Hz

R1, R2 2.4 Hz/p.u.

TG1, TG2 0.08 s.

TT1, TT2 0.03 s.

TPS1, TPS2 20 s.

T12 0.545p.u.

KPS1, KPS2 120 Hz/p.u.

a12 -1


[1] R. C. Eberhart, and J. Kennedy, A new optimizer using particle swarm theory.
Proceedings of the Sixth International Symposium on Micro-machine and Human
Science, Nagoya, Japan. pp. 39-43, 1995.

[2] Mustafa Servet Kiran, “Particle swarm optimization with a new update mechanism”,

[3] Elgerd. O. I., “Energy Systems Theory: an introduction”, New York: McGraw- Hill,

[4] K.Ogata, “Modern Control Engineering”, New Jersey, Prentice Hall,2008.

[5] C.L.Wadhwa, “Electrical Power system”, Sixth Edition, New Age International
Publisher, New Delhi.

[6] I.J. Nagrath and M.Gopal “Control System Engineering” Fifth Edition, New Age
International Publisher, New Delhi.

[7] Rabindra Kumar Sahu , Sidhartha Panda, Umesh Kumar Rout, Dillip Kumar Sahoo,
“Teaching learning based optimization algorithm for automatic generation control of
power system using 2-DOF PID controller” SCIENCEDIRECT. pp. 2-13, 2015.

[8] James Blondin, “Particle Swarm Optimization: A Tutorial”, September 4,2009.

[9] Elgerd OI, Fosha CE. Optimal megawatt-frequency control of multi-area electrical
energy systems. IEEE Trans PAS 1970.

[10] Mohanty B, Panda S, Hota PK. Hybrid BFOA–PSO algorithm for automatic
generation control of linear and nonlinear interconnected power systems. Applied Soft
Computing 2013; 13:47 18-30