9 Fuzzy Logic and AGC Systems

Search in book...
Toggle Font Controls
Create new playlist

Name your new playlist

Playlist description (optional)
Sign In

Email address

Password

Forgot Password?

or

Continue with Facebook

Continue with Google
Sign Up

Full Name

Email address

Confirm Email Address

Password

or

Continue with Facebook

Continue with Google

An overview on fuzzy-logic-based AGC systems with different configurations is given in Chapter 3. This chapter presents two multifunctional fuzzy-logic-based AGC schemes considering the MWh constraint for the power transmission through the interarea tie-line and the regulation margin for the AGC units. The first design scheme includes two control loops. The main control loop consists of a three-dimensional polar-information-based fuzzy logic control block, and the additional control loop consists of a two-dimensional polar-information-based fuzzy logic control block.^1,2 There exist additional parameters, which give the contribution factor from each subarea and another contribution factor from each unit. These contribution factors are determined according to the remaining available capacities of the AGC units in each subarea, and that of each individual AGC unit, respectively. Additional functions, such as the real power flow control scheme on the trunk lines considering the power flow constraints, and the regulation margin control scheme to maintain the remaining AGC capacity to the specified level,^3,4 have also been proposed. These additional control loops are activated only when the power flow constraints are violated, and also only when the level of the regulation margin becomes lower than the specified level.

The second design provides a particle-swarm-optimization (PSO)-based fuzzy logic AGC system. In order to improve the control performance, the PSO technique is used for tuning of the fuzzy system’s membership function parameters in the supplementary frequency control loop.⁵ The efficiency of the proposed control schemes has been demonstrated through nonlinear simulations by using appropriate interconnected power system examples.

9.1 Study Systems

9.1.1 Two Control Areas with Subareas

The study system for the first AGC synthesis approach is shown in Figure 9.1. The study system consists of two areas. Area X is the study area, and area Y is its external area. Study area X is divided into four subareas: A, B, C, and D. Each subarea has a certain number of AGC units for the load-frequency regulation and also non-AGC units. There are several trunk lines between these subareas; therefore, the power accommodation between the subareas is also possible through the trunk lines, in addition to the power accommodation between area X and area Y.

Images

FIGURE 9.1
Interconnected two-area power system.

A detailed nonlinear AGC simulator has been developed in the MATLAB/ Simulink environment.⁶ The simulator includes twenty-three thermal units to simulate the AGC performance of the entire Kyushu Electric Power System in Japan. However, the power generation from the nuclear units, from the hydro units, and also from the other utility units in the Kyushu Electric Power System is not included in the simulation model because these units are out of the AGC.⁷ In the simulations, the generation rate constraints are considered for each unit, and various types of load changes, such as step, ramp, and random changes, are utilized.

All the nonlinear simulations have been performed by using the detailed simulation program developed in the MATLAB/Simulink environment. Figure 9.2 illustrates the main Simulink block for the study system. There exist five subblocks that represent the external area EX, and the subareas A to D. Each subblock also consists of several blocks, such as a power generation block and block for the load-frequency dynamics. The main block also includes a power flow calculation block to determine the real power flow on the tie-line and on the trunk lines. This test system is used to examine the polar-information-based fuzzy logic AGC scheme.

9.1.2 Thirty-Nine-Bus Power System

To investigate the performance of the PSO-based fuzzy logic AGC design, a simulation study is provided in the SimPowerSystems environment of MATLAB software. For this purpose, a network with the same topology as the well-known IEEE ten-generators, thirty-nine-bus test system, described in Figure 6.7, is considered. The thirty-nine-test system is organized into three areas. Here, to increase the power fluctuation, all three areas are equipped with wind farms. A single line diagram of the updated test system is redrawn in Figure 9.3. This system has ten generators, nineteen loads, thirty-four transmission lines, and twelve transformers. The simulation parameters for the generators, loads, lines, and transformers of the test system are given in Bevrani et al.⁸ The installed wind farms use a double-fed induction generator (DFIG) wind turbine type.

Images

FIGURE 9.2
Main Simulink block.

The total installed wind power is about 85 MW of wind power generation. Similar to the simulation studies of Chapters 7 and 8, it is assumed that all power plants in the power system are equipped with a speed governor and power system stabilizer (PSS). However, only one generator in each area is responsible for the AGC issue: G1 in area 1, G9 in area 2, and G4 in area 3.

Images

FIGURE 9.3
Single line diagram of updated thirty-nine-bus case study.

For the sake of simulation, random variations of wind velocity have been considered. The dynamics of wind turbine generators, including the pitch angle control of the blades, are also considered.

9.2 Polar-Information-Based Fuzzy Logic AGC

9.2.1 Polar-Information-Based Fuzzy Logic Control^1,2

According to the frequency change, the tie-line power change, and the integrated tie-line power deviation, the total demand of the additional generation is determined through a simple polar-information-based fuzzy logic control scheme in the main control loop shown later. Three-dimensional information is required to determine the additional generation as follows:

Z a (k) = (A C E (k) - A C E (k - 1)) / Δ T (9.1)

$Z a (k) = (A C E (k) - A C E (k - 1)) / Δ T (9.1)$

Z s (k) = A C E (k) (9.2)

$Z s (k) = A C E (k) (9.2)$

Z p (k) = Σ Δ P_{A E X} (k) (9.3)

$Z p (k) = Σ Δ P_{A E X} (k) (9.3)$

where

ACE (k) = Δ P + β Δ f (9.4)

$ACE (k) = Δ P + β Δ f (9.4)$

Here, the sampled variables Zs(k), Za(k), and Zp(k) represent the area control error (ACE), the ACE gradient, and the integrated tie-line power deviation, respectively. The area control error ACE(k) determines the frequency deviation and the total power flow change from each subarea through the tie-line and the trunk lines.

Figure 9.4 illustrates the switching surface utilized to determine the additional generation for the AGC. The state space is divided into two subspaces. In the space over the switching surface, the total generation should be reduced to keep the frequency and the tie-line power. On the contrary, the total generation should be increased in the space below the switching surface. In order to simplify the control algorithm using the three-dimensional information, the system state is replaced onto a two-dimensional phase plane, shown in Figure 9.5. Then, the system state is given by a point p(k) in the phase plane.

p (k) = [Z_{s} (k) + Z S S, A_{s} Z_{a} (k)] (9.5)

$p (k) = [Z_{s} (k) + Z S S, A_{s} Z_{a} (k)] (9.5)$

where ZSS(k) = Sg∙Zp(k).

Images

FIGURE 9.4
Switching surface for fuzzy logic control.

Images

FIGURE 9.5
Phase plane to identify system state.

The term ZSS(k) gives the shift size of the origin O to the origin O* in the transient period, and Sg is the shift gain. At the final steady state, the origin O* coincides with the origin O, because Zp(k) becomes equal to zero at the final steady state. The polar notation of the present state is given by the following two equations:

D (k) = \sqrt{(Z s (k) + Z S S)^{2} + (A s Z a (k))^{2}} (9.6)

$D (k) = \sqrt{(Z s (k) + Z S S)^{2} + (A s Z a (k))^{2}} (9.6)$

θ (k) = \tan^{- 1} (A s \cdot Z a (k) / (Z s (k) + Z S S)) (9.7)

$θ (k) = \tan^{- 1} (A s \cdot Z a (k) / (Z s (k) + Z S S)) (9.7)$

Finally, the defuzzification process to provide the control signal is described as follows:

u (k) = \frac{N (θ (k)) - P (θ (k))}{N (θ (k)) + P (θ (k))} \cdot G (D (k)) \cdot U_{\max} (9.8)

$u (k) = \frac{N (θ (k)) - P (θ (k))}{N (θ (k)) + P (θ (k))} \cdot G (D (k)) \cdot U_{\max} (9.8)$

G (D (k)) = D (k) / D r for D (k) \leq D r (9.9)

$G (D (k)) = D (k) / D r for D (k) \leq D r (9.9)$

G (D (k)) = 1.0 for D (k) \geq D r (9.10)

$G (D (k)) = 1.0 for D (k) \geq D r (9.10)$

where U_max is the maximum size of the control signal from the fuzzy logic control block.

Here, the integrated tie-line power deviation Zp(k) is utilized to determine the additional generation demand to maintain the integrated tie-line power deviation at zero. In other words, the contract MWh constraint is satisfied through the proposed fuzzy logic control scheme. The final total power demand is given by Tpower, which is the integration of the control signal u(k) in the main regulation block, shown in Figure 9.6.

In Figure 9.6, the term pf_A gives the participation factor of subarea A to the load-frequency regulation, and therefore the power generation in subarea A is changed to the value of Tpower⋅pf_A. The power generation in the other sub-areas is also changed according to the values of Tpower·pf_B, Tpower·pf_C, and Tpower·pf_D, as shown in Figure 9.7. Each participation factor is determined based on the available AGC capacity in the associated area. Figure 9.8 also includes an additional control loop to regulate the power flow on the trunk line from the associated subarea. Here, a fuzzy-logic-based control scheme is utilized, where only two-dimensional information, i.e., the ACE and its gradient for the associated area are used. The control signal is determined by using the same equations shown above after setting the shift gain Sg to zero.

Images

FIGURE 9.6
Main regulation block.

Images

FIGURE 9.7
Subregulation block for subareas B, C, and D.

Images

FIGURE 9.8
Power generation block, including AGC and non-AGC power stations.

The additional control loop is activated only when the real power flow constraint on the trunk line is violated. The configuration of the control loop is the same as that of the conventional supplementary control loop; therefore, the load change in the associated area is regulated by the generation change in the same area. Consequently, the trunk line power is maintained up to its power flow limit.

Figure 9.8 illustrates the power generation block, which includes several AGC and non-AGC power stations in each subarea. Each subarea has a specific number of AGC units and non-AGC units. Whenever the power generation from the AGC units exceeds the prespecified limit, a portion of the power generation is shifted to the non-AGC units in order to keep the regulation margin to a certain level. Namely, the regulation capacity is always kept to a certain extent by the participation of the non-AGC unit to the load-frequency regulation. Usually, the generation from the non-AGC unit can be changed quite slowly; therefore, it takes some time to finish the replacement of the required generation from the AGC unit to the non-AGC unit. However, this does not mean that the regulation requires the same length of time to reach the steady state. This issue is shown later.

Images

FIGURE 9.9
Configuration of an AGC power station.

Figure 9.9 shows the detailed configuration of one of the AGC power stations, where there exist two AGC units. Each AGC unit uses a specific participation factor, but the total sum of participation factors among an area is equal to 1.0.

9.2.2 Simulation Results

Simulations have been performed subject to various types of load changes, such as step, ramp, and random load. All the control parameters have been set to their typical values; therefore, the tuning of these parameters is not considered in this study. Throughout the simulations, the frequency deviation, tie-line power, and trunk line power are sampled every 2.5 s, and the load-frequency regulation command is renewed at every 10 s, considering the practical AGC operation.

9.2.2.1 Trunk Line Power Control

Table 9.1 shows the initial settings of the total power generation from the AGC units, non-AGC units, and all units. In subarea D, there is no non-AGC unit. Table 9.2 shows the real power flow on the tie-line and also on the trunk lines. Table 9.3 indicates the load change considered for the simulations.

When the trunk line power control is not considered, the total load change in the utility X is covered by the generation changes from all the units, including the non-AGC units, to keep the regulation capacity to a certain extent. The contribution from each unit is determined by the subarea participation factors together with the unit participation factors in each subarea. In Figure 9.10, from the top to the bottom, the real power flows on the tie-line and the trunk lines, from subareas B, C, and D to subarea A, are illustrated. In the bottom graph, the activation of the trunk line power control is indicated. The value 0 shows the nonactivation, while the value 1 shows the activation. In this simulation, the real power flow limit is not specified; therefore, the trunk line power control is not activated.

TABLE 9.1
Initial Setting of Total Generation (1 pu = 10,000 MW)

Generation	Area A	Area B	Area C	Area D
From AGC units (pu)	0.32	0.08	0.40	0.20
From non-AGC units (pu)	0.06	0.02	0.10	—
Total generation (pu)	0.40	0.10	0.50	0.20

TABLE 9.2
Tie-Line Power and Trunk Line Power (1 pu = 10,000 MW)

Tie-Line	P_AEX	P_B	P_C	P_D
Trunk Lines	A-EX	B-A	C-A	D-A
Power flow (pu)	0.10	0.00	0.20	0.10

TABLE 9.3
Load Change

	Area A	Area B	Area C	Area D
Initial load (pu)	0.60	0.10	0.30	0.10
Type and size (pu)	Random	—	Ramp: -0.1	—
Start time (s)	Always	—	50	—

Images

FIGURE 9.10
Real power flow on tie-line and trunk lines without trunk line power control.

In order to check the efficiency of the proposed trunk line power control, the real power limit is specified at 0.22 pu from subarea C to the subarea A. The simulation results are shown in Figures 9.11 and 9.12. Figure 9.11 shows the real power flow on the tie-line and also on the trunk lines. As shown in the variation of the real power flows and in the bottom graph, the trunk line power control is activated whenever the trunk line power exceeds the specified limit of 0.22 pu, and in the final steady state the trunk line power is kept at its limit of 0.22 pu. Figure 9.12 shows the frequency deviation in each subarea. The frequency deviation is within the range of tolerance, i.e., 0.05 Hz. Furthermore, the profile of frequency change in each subarea has almost the same profile as shown in Figure 9.12.

9.2.2.2 Control of Regulation Margin

Images

FIGURE 9.11
Real power flow on tie-line and trunk lines with trunk line power control.

In the following simulations, the efficiency of the regulation margin control has been demonstrated. The load change utilized for the simulation is shown in Table 9.4. In the simulation, the power generation limits from the AGC units in subarea A are specified at 0.33 pu. Whenever the generation from the AGC units exceeds this limit, a certain power generation is shifted from the AGC units to the non-AGC units to keep the generation from the AGC units to its limit of 0.33 pu in the final steady state.

Images

FIGURE 9.12
Frequency deviation.

TABLE 9.4
Load Change

	Area A	Area B	Area C	Area D
Initial load (pu)	0.60	0.10	0.30	0.10
Type and size (pu)	Random-ramp: 0.03 pu		^—	^—
Start time (s)	50 (s)	—	—	—

Figure 9.13 shows the variations of the total power generation from the AGC units in subareas A to D. The bottom graph illustrates the activation of the control of the regulation margin, where the value zero indicates the non-activation, while the value 1 indicates the activation. As shown in this figure, the power generation from the AGC units in subarea A is reduced by the proposed control scheme to have the specified regulation margin. Here, it must be noted that the total power generation from the AGC units exceeds the specified limit in the transient stage after having the load change. Figure 9.14 shows the total generation from the non-AGC units in subareas A to D. In the subareas, a certain amount of power comes from the non-AGC units in the final steady state to keep the regulation margin.

Images

FIGURE 9.13
Total generation from AGC units in subareas A to D.

Images

FIGURE 9.14
Total generation from non-AGC units in subareas A to D.

Images

FIGURE 9.15
Total load, total generation from AGC units, total generation from non-AGC units, and integrated tie-line power deviation.

Figure 9.15 illustrates the total load, the total generation from the AGC units, the total generation from the non-AGC units, and the integrated tie-line power deviation from the top to the bottom. As shown in the figure, the integrated tie-line power deviation becomes equal to zero in the final steady state. Namely, the MWh constraint is satisfied by using the proposed control scheme.

9.3 PSO-Based Fuzzy Logic AGC

9.3.1 Particle Swarm Optimization

Particle swarm optimization (PSO) is a population-based stochastic optimization technique. It belongs to the class of direct search methods that can be used to find a solution to an optimization problem in a search space. The PSO was originally presented based on the social behavior of bird flocking, fish schooling, and swarming theory.^9,10 In the PSO method, a swarm consists of a set of individuals, with each individual specified by position and velocity vectors (x_i(t), v_i(t)) at each time or iteration. Each individual is named a particle, and the position of every particle represents a potential solution to the under-study optimization problem. In an n-dimensional solution space, each particle is treated as an n-dimensional space vector and the position of the ith particle is presented by v_i = (x_i1, x_i2, …, x_in); then it flies to a new position by the velocity represented by v_i = (v_i1, v_i2, …, v_in). The best position for ith particle represented by p_best,i = (p_best,i1, p_best,i2, …, p_best,in) is determined according to the best value for the specified objective function.

Furthermore, the best position found by all particles in the population (global best position) can be represented as g_best = (g_best,1, g_best,2, …, g_best,n). In each step, the best particle position, global position, and corresponding objective function values should be saved. For the next iteration, the position x_ik and velocity v_ik corresponding to the kth dimension of ith particle can be updated using the following equations:

v_{i k} (t + 1) = w . v_{i k} + c_{1} . r a n d_{1, i k} (p_{b e s t, i k} (t) - x_{i k} (t)) + c_{2} . r a n d_{2, i k} (g_{b e s t, k} (t) - x_{i k} (t)) (9.11)

$v_{i k} (t + 1) = w . v_{i k} + c_{1} . r a n d_{1, i k} (p_{b e s t, i k} (t) - x_{i k} (t)) + c_{2} . r a n d_{2, i k} (g_{b e s t, k} (t) - x_{i k} (t)) (9.11)$

x_{i k} (t + 1) = x_{i k} (t) + v_{i k} (t + 1) (9.12)

$x_{i k} (t + 1) = x_{i k} (t) + v_{i k} (t + 1) (9.12)$

where i = 1, 2, …, n is the index of particles, w is the inertia weight,¹¹ rand_1,ik and rand_2,ik are random numbers in the interval [0 1], c₁ and c₂ are learning factors, and t represents the iterations.

Usually, a standard PSO algorithm contains the following steps:

Initialize all particles via a random solution. In this step, each particle position and its associated velocity are set by randomly generated vectors. The dimension of the position should be generated within a specified interval, and the dimension of the velocity vector should also be generated from a bounded domain using uniform distributions.
Compute the objective function for the particles.
Compare the value of the objective function for the present position of each particle with the value of the objective function corresponding to the prespecified best position, and replace the prespecified best position with the present position, if it provides a better result.
Compare the value of the objective function for the present best position with the value of the objective function corresponding to the global best position, and replace the present best position by the global best position, if it provides a better result.
Update the position and velocity of each particle according to Equations 9.11 and 9.12.
Stop the algorithm if the stop criterion is satisfied. Otherwise, go to step 2.

In the present section, the PSO algorithm is used to find the optimal value for membership function parameters of a fuzzy-logic-based AGC system.

9.3.2 AGC Design Methodology

Inherent nonlinearity, increasing in size and complexity of power systems as well as emerging wind turbines and their effects on the dynamic behavior of power systems, caused conventional proportional-integral-derivative (PID) and proportional-integral (PI) controllers to be incapable of providing good dynamical performance over a wide range of operating conditions.¹² In this section, to track a desirable AGC performance in the presence of high-penetration wind power in a multiarea power system, a decentralized PSO-based fuzzy logic control design is proposed. Decreasing the frequency deviations due to fast changes in output power of wind turbines, and limiting tie-line power interchanges to an acceptable range, following disturbances, are the main goals of this effort. The proposed control framework is shown in Figure 9.16.

The inputs and output are brought into an acceptable range by multiplying in proper gains. In each control area, ACE and its derivative are considered input signals, and the provided control signal is used to change the set points of AGC participant generating units. The mamdani type inference system is applied, and as shown in Figure 9.17, symmetric seven-segment triangular membership functions are used for input (a) and output (b) variables. The membership functions are defined as zero (ZO), large negative (LN), medium negative (MN), small negative (SN), small positive (SP), medium positive (MP), and large positive (LP).

Images

FIGURE 9.16
The proposed control framework: (a) area components and (b) controller structure.

Fuzzy rule is the basis of fuzzy logic operation to map the input space to the output space. Here, a rule base including forty-nine fuzzy rules is considered (Table 9.5). The rule base works on vectors composed of ACE and its gradient ΔACE. Using Table 9.5, fuzzy rules can be expressed in the form of if-then statements, such as:

Images

FIGURE 9.17
Symmetric fuzzy membership functions: (a) inputs pattern and (b) output pattern.

TABLE 9.5
Fuzzy Rule Base

Images

I f A C E is SN and Δ A C E is MP, t h e n output is SN .

$I f A C E is SN and Δ A C E is MP, t h e n output is SN .$

As explained in Chapter 3, since fuzzy rules are stated in terms of linguistic variables, crisp inputs should also be mapped to linguistic values using fuzzification. As the number of inputs is more than one (two inputs), fuzzified inputs in the if part of the rules must be combined to a single number. Here, the combination is done based on interpreting the and operator as a product of the membership values, corresponding to measured inputs. The obtained single number from the if part of each rule is used to compute the consequence of the same rule (implication), which is a fuzzy set. In this work, the product method has considered the implication method. In order to combine rules and make a decision based on all rules, the sum method is used. Finally, for converting an output fuzzy set of the fuzzy system to a crisp value, the centroid method is used for defuzzication.¹³

9.3.3 PSO Algorithm for Setting of Membership Functions

Like the performance of a fuzzy system influenced by membership functions, in order to achieve good performance by the controller, a PSO algorithm is established to find the optimal value for membership function parameters and the exact tuning of them. As can be seen in Figure 9.17, each set of input membership functions can be specified by parameters a and b, where min < a < b < max. Also, for control output, one parameter needs to be specified. Therefore, five parameters should be optimized for input membership functions using the PSO algorithm: a_{in, ACE}, b_{in, ΔACE}, a_{in, ΔACE}, b_in,ΔACE, and b_out.

For the sake of the PSO algorithm in the present AGC design, the objective function (f) is considered as given in Equation 9.13. The number of particles, particle size, and v_min, v_max, c₁, and c₂ are chosen as 10, 6, –0.5, 0.5, 2.8, and 1.3, respectively. Following use of the PSO algorithm, the optimal values for membership function parameters are obtained as listed in Table 9.6.

f = \frac{1}{3} \sum_{i = 1}^{3} (\int t (| Δ f_{i} | + | M_{t i e, i} |) d t) (9.13)

$f = \frac{1}{3} \sum_{i = 1}^{3} (\int t (| Δ f_{i} | + | M_{t i e, i} |) d t) (9.13)$

9.3.4 Application Results

TABLE 9.6
Optimal Values of Membership Function Parameters

a_{in, ACE}	b_{in, ACE}	a_{in, ΔACE}	b_{in, ΔACE}	b_out
0.267747	0.947038	0.013716	0.059880	0.986659

To demonstrate the effectiveness of the proposed fuzzy-logic-based AGC design, some simulations were carried out. In these simulations, the proposed controllers were applied to the model described in Section 9.1.2, and also used in the literature.^5,8,14,15 In the performed simulation, the performance of the closed-loop system uses the well-tuned conventional PI controllers, compared to the designed fuzzy-logic-based controllers. As a serious test scenario, the following load disturbances (step increase in demand) are applied to three areas at 5 s simultaneously: 3.8% of the total area load at bus 9 in area 1, 4.3% of the total area load at bus 18 in area 2, and 8.01% of the total area load at bus 24 in area 3. The simulation is implemented by using the MATLAB SimPowerSystems program. The simulation results are shown in Figures 9.18 to 9.21.

The total generated wind power, wind speed deviation, and overall frequency deviation of the system are shown in Figure 9.18. The area control error signals of three areas, following the load disturbances, are also shown in Figure 9.19. The produced mechanical power of AGC participant generating units and the tie-line powers are illustrated in Figures 9.20 and 9.21, respectively.

These figures show the superior performance of the proposed fuzzy logic method to the conventional PI controller in deriving area control error and frequency deviation close to zero. Interested readers can find more details on the proposed PSO-based fuzzy logic AGC design method with numerous simulation results for the same case study in Daneshmand.⁵

Images

FIGURE 9.18
System response: (a) total wind power generation, (b) wind speed deviation, and (c) frequency deviation (solid, proposed fuzzy control; dotted, conventional PI control).

Images

FIGURE 9.19
ACE signal in three control areas, following simultaneous disturbances; solid (proposed fuzzy control), dotted (conventional PI control).

Images

FIGURE 9.20
Mechanical power of AGC participating units, following simultaneous disturbances; solid (proposed fuzzy control), dotted (conventional PI control).

Images

FIGURE 9.21
Tie-line powers, following simultaneous disturbances; solid (proposed fuzzy control), dotted (conventional PI control).

9.4 Summary

Two fuzzy-logic-based AGC design methodologies have been presented for the frequency and tie-line power regulation in multiarea power systems. These methodologies are polar-information-based fuzzy logic AGC and PSO-based fuzzy logic AGC designs. The efficiency of the proposed control schemes has been demonstrated through nonlinear simulations by using the detailed models developed in the MATLAB/Simulink and SimPowerSystems environments.

By using the proposed polar-information-based fuzzy logic AGC scheme, the MWh constraint is satisfied to avoid the MWh contract violation. The regulation capacity is always kept to a certain level by the replacement of the required generation from the AGC units to the nonAGC units. In addition, the power flow on the trunk lines can be regulated whenever the power flow exceeds the specified limit. The PSO-based fuzzy logic AGC design is used for frequency and tie-line power regulation in the presence of wind turbines. The proposed method is applied to an updated version of the ten-generator, thirty-nine-bus system, and the results are compared with those from a conventional PI control design.

References

1. T. Hiyama, S. Oniki, H. Nagashima. 1996. Evaluation of advanced fuzzy logic PSS on analog network simulator and actual installation on hydro generators. IEEE Trans. Energy Conversion 11(1):125–31.

2. T. Hiyama, T. Kita, T. Miyake, H. Andou. 1999. Experimental studies of three dimensional fuzzy logic power system stabilizer on damping of low-frequency global mode of oscillation. Fuzzy Sets Systems 102(1):103–11.

3. T. Hiyama. 1982. Optimisation of discrete-type load-frequency regulators considering generation rate constraints. IEE Proc. C 129(6):285–89.

4. T. Hiyama. 1981. Design of decentralised load-frequency regulators for interconnected power systems. IEE Proc. C 129(1):17–23.

5. P. R. Daneshmand. 2010. Power system frequency control in the presence of wind turbines. MSc dissertation, Department of Electrical and Computer Engineering, University of Kurdistan, Sanandaj, Iran.

6. T. Hiyama, Y. Yoshimuta. 1999. Load-frequency control with MWh constraint and regulation margin. In Proceedings of IEEE Power Engineering Society, 1999 Winter Meeting, New York, vol. 2, pp. 803–8.

7. H. Bevrani, G. Ledwich, Z. Y. Dong, J. J. Ford. 2009. Regional frequency response analysis under normal and emergency conditions. Elect. Power Syst. Res. 79:837–45.

8. H. Bevrani, F. Daneshfar, P. R. Daneshmand. 2010. Intelligent power system frequency regulation concerning the integration of wind power units. In Wind power systems: Applications of computational intelligence, ed. L. F. Wang, C. Singh, A. Kusiak, 407–37. Springer Book Series on Green Energy and Technology. Heidelberg: Springer-Verlag.

9. J. Kennedy, R. Eberhart. 1995. Particle swarm optimization. In Proceedings of IEEE International Conference, Neural Networks, vol. 4, pp. 1942–48.

10. R. Eberhart, J. Kennedy. 1995. A new optimizer using particle swarm theory. In Proceedings of Sixth International Symposium, Micro Machine and Human Science, Nagoya, Japan, pp. 39–43.

11. Y. Shi, R. Eberhart. 1998. A modified particle swarm optimizer. In Proceedings of IEEE International Conference, Evolutionary Computation, IEEE World Congress, Computational Intelligence, Anchorage, AK, May 4–9, 1998, pp. 69–73.

12. H. Bevrani. 2009. Robust power system frequency control. New York: Springer.

13. H. Bevrani, Syafaruddin, T. Hiyama. Intelligent control in power systems. Lecture notes. http://www.cs.kumamoto-u.ac.jp/epslab/sub1.html.

14. H. Bevrani, F. Daneshfar, P. R. Daneshmand, T. Hiyama. 2010. Reinforcement learning based multi-agent LFC design concerning the integration of wind farms. In Proceedings of IEEE International Conference on Control Applications, Yokohama, Japan, CD-ROM.

15. H. Bevrani, F. Daneshfar, P. R. Daneshmand, T. Hiyama. 2010. Intelligent automatic generation control: Multi-agent Bayesian networks approach. In Proceedings of IEEE International Conference on Control Applications, Yokohama, Japan, CD-ROM.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.

Table of Contents for 9 Fuzzy Logic and AGC Systems

Create new playlist

Sign In

Sign Up

9.2.1 Polar-Information-Based Fuzzy Logic Control1,2

Table of Contents for
9 Fuzzy Logic and AGC Systems

9.2.1 Polar-Information-Based Fuzzy Logic Control^1,2