2012 International Conference on Artificial Intelligence and Soft Computing Lecture Notes in Information Technology, Vol.

12

A Hybrid Evolutionary Computation Algorithm for Optimal Reactive Power Dispatch Considering Voltage Magnitude Deviation
Chao-Ming Huang1,a, Yann-Chang Huang 2,b,* and Kun-Yuan Huang3,c
1 2 3

Department of Electrical Engineering, Kun Shan University, Tainan 710, Taiwan

Department of Electrical Engineering, Cheng Shiu University, Kaohsiung 833, Taiwan Department of Electrical Engineering, Cheng Shiu University, Kaohsiung 833, Taiwan
a

h7440@ms21.hinet.net, b huangyc@csu.edu.tw, c kyhuang@csu.edu.tw

Keywords: Optimal reactive power dispatch, Evolutionary computation, Differential evolution, Ant system.

Abstract. This paper proposes a hybrid evolutionary computation (HEC) algorithm for solving the optimal reactive power dispatch (ORPD) problem. Traditionally, ORPD is defined as the minimization of active power transmission losses by controlling a number of control variables. In this paper, the deviation of bus voltage magnitude which influences the security operation of power transmission systems is also considered as an objective function. ORPD is typically a nonlinear constrained optimization problem. The proposed HEC algorithm combines basic differential evolution (DE) algorithm and probabilistic state transition rule used in the ant system to deal with the ORPD problem. To verify the performance of the proposed method, the similar evolution approaches such as evolutionary programming (EP) and particle swarm optimization (PSO) are also implemented using the same database. The proposed method has been verified on the IEEE 30-bus 6-generator system. Testing results indicate that the proposed HEC can obtain better results than the other methods in terms of active power transmission losses, voltage magnitude deviation, and convergence performance. 1. Introduction The ORPD is an important issue on secure and economic operation of power transmission systems. Traditionally, ORPD is defined as the minimization of active power transmission losses by controlling the generator terminal voltages, transformer tap settings, and shunt capacitors/ reactors. The purpose of ORPD is to improve the voltage profile in the power transmission systems by using a number of control variables. Since the shunt capacitors/reactors and the tap settings of transformer have the discrete nature, the ORPD is then formulated as discrete and nonlinear optimization problem. Much research has been devoted to cope with the ORPD problem. These techniques include the nonlinear programming method [1], mixed-integer programming method [2], interior point method [3], genetic algorithm [4], particle swarm optimization [5], and adaptive immune algorithm [6], which have served as effective tools for solving the ORPD problem. In this paper, four evolutionary computation methods including EP, PSO, DE, and the proposed HEC algorithms are separately used to solve the ORPD problem. EP [7] is a global search technique that simulates the natural evolution process and constitutes a stochastic optimization algorithm. By simulating the behavior of a swarm as a simplified social system, PSO [8] has the fast convergence and easy implementation characteristics. DE was first introduced by Storn and Price in 1995 [9]. It
978-1-61275-009-5/10/$25.00 2012 IERI
178

ICAISC2012

has the parallel search and rapid convergence naturals. Recently, DE has been successfully applied to solve the reactive power dispatch [10] and the multi-objective power flow [11] problem. This paper proposes a HEC algorithm which combines the basic DE and probabilistic state transition rule used in the ant system [12] to solve the ORPD problem. Based on rapid convergence and global search ability, HEC can offer higher probability of converging toward global solution than the other methods. 2. Problem formulation 2.1 Objective functions Typically, ORPD is to minimize active power transmission loss subject to a number of constraints. In this paper, the voltage magnitude deviation at each load bus is also considered an objective function. The objective of active power transmission loss can be expressed as follows.

fQ =

k( i , j )

l ,k

= g ij Vi + V j 2 Vi V j cos( i j )
2 2 i =1 j =1

[ (

)]

(1)

where fQ is the active power transmission loss, Pl,k is the active power transmission loss of branch k, L is the number of transmission lines, |Vi| is the voltage magnitude at bus i, gij is the conductance between bus i and j, and i is the voltage phase angle of bus i. The bus voltage magnitude is one of the important indices in security operation of power transmission systems. The objective of voltage magnitude deviation is given as follows [13].
Vdev = Vi Vi ref
i =1 N pQ

(2)

where Vi ref is the ith specified reference value of the voltage magnitude for load buses which is generally set at 1.0 p.u. and NPQ is the number of load buses. The objective functions in (1) and (2) can be converted into a single objective optimization problem by giving the relative weighting values as follows.

F ( x) = w1 f Q + w2Vdev
where F(x) is the objective value and wi (wi=1) is the weight of the ith objective. 2.2 Constraints

(3)

In addition to the power balance constraint, the ORPD of model (3) also needs to meet the following constraints. 2.2.1 Reactive generation constraint The reactive generation constraint is usually used on the generation bus (PV bus) as follows.

Q Bi , min Q Bi Q Bi , max
where QBi,min and QBi,max are the lower and upper limits of the injected reactive power at bus i.

(4)

2.2.2 Bus voltage constraint To remain the stable operation of the power system, each bus voltage must be controlled within the constraints. The bus voltage constraint is generally utilized on the load bus (PQ bus) as follows.

Vi ,min Vi Vi ,max
where Vi,min and Vi,max are the lower and upper limits of the bus voltage at bus i.

(5)

179

2.2.3 Capacitor and transformer tap setting constraints The constraints of the capacitor and transformer tap setting are defined as follows.

QC ,min QC QC ,max

(6) (7)

Tk ,min Tk Tk ,max

where Qc is the reactive power installed by capacitor, Qc,min and Qc,max are the lower and upper limits of the capacitor, Tk is the position of transformer k, and Tk,min and Tk,max are the lower and upper positions of transformer k. In this paper, the capacitor and transformer tap setting are regarded as continuously adjustable variables. 3. The proposed HEC algorithm The evolutionary computation algorithms including EP, PSO, DE, and the proposed HEC are artificial intelligent methods based on the mechanics of natural evolution. These methods share many similarities. Based on the basic evolutionary strategies, the proposed approach achieves the fittest individual after repeated initialization, mutation, recombination, and selection operations as follows. 3.1 Initialization Let pi=[pi1, pi2, , piM] be a trial vector representing the ith individual (i =1, 2, , P) of the population to be evolved, where P is the population size and M is the dimension of each individual. The elements in vector pi represent the decision variables (genes) which are randomly generated as follows.

pij = pij ,min + ( pij ,max pij ,min ), j = 1, 2, ..., M

(8)

where pij represents the jth gene of the ith individual, pij,min and pij,max mean the lower and upper bounds of pij, respectively; and represents the uniform random number between 0 and 1. In this paper, the trial vector pi represents the desired values of voltage magnitude at generation bus, transformer tap settings, and shunt capacitors. 3.2 Mutation The mutation operation of basic DE is performed by adding a differential vector to the parent individual as follows.

pi ' = pi + f m ( pia pib )

represents the mutation factor. In this paper, f m (t + 1) = e 0.1t f m (t ) , t is the number of iteration. 3.3 Recombination

(9)

where p ia and p ib are the randomly selected individuals in the parent population and f m [ 0,1]

In essence, the mutant individual in (9) is a noisy replica of pi. When the population diversity is small, the candidate individuals will rapidly gather together so the individuals cannot be further improved. To extend the local diversity of the mutant individuals, a recombination operation is introduced as follows.
pij , pij ' = pij ' , if randij > Rr if randij Rr

(10)

where pij is the jth gene of the ith individual before mutation, pij ' represents the jth gene of the ith offspring individual following mutation, randij is a random number with normal distribution, and Rr [0,1] is a recombination factor. Equation (10) indicates each gene of the ith individual is reproduced from the current gene pij or the mutant gene pij ' .
180

3.4 Selection using ant system The basic DE used one-to-one competition to retain its offspring that gives rise to a rapid convergence rate. This rapid convergence may lead to a higher probability of obtaining a local optimum point because the diversity of the population descends more rapidly during the optimization processes. To increase the global search capability, a probabilistic state transition rule used in the ant system is utilized to replace the selection operation in the basic DE algorithm. The ant system was first applied to the traveling salesman problem [11]. Informally, ants prefer to move to cities which are connected by short distance with a high amount of pheromone. However, the cities with short distance and high pheromone are not absolutely selected by ants. Each ant generates a complete tour by choosing the cities according to a probabilistic state transition rule as follows.
Pri (t ) =

[ (t )] [F (t )] [ (t )] [F (t )]

i i P

(11)

i =1

where i (t) is the pheromone concentration of the ith ant at tth iteration, Fi(t) is the objective value of the ith ant at tth iteration, is the pheromone constant and is the constants of the objective value. In addition, the pheromone concentration is updated according to the following formula:

i (t + 1) = i (t ) + i

(12)

where [0,1] is a pheromone decay parameter and i means the variation of pheromone. Equations (11) to (12) show if the objective value of the offspring individual is better than the other individuals, the pheromone concentration i is increased and it has more probability of surviving as a new individual in the next generation. Compared to the one-to-one competition method, the probabilistic state transition rule has the advantages of retaining the diversity of the population and escaping from local optimal solutions. 4. The scheme of HEC to the ORPD problem The proposed HEC algorithm for searching the ORPD is described in the following steps:

Step 1: Determine the corresponding weighting values of each objective for the ORPD problem. Step 2: Randomly generate the initial parent trial vector pi as shown in (8). Step 3: Evaluate the objective value of each parent individual using (3) subject to a number of constraints shown from (4) to (7). Step 4: Execute the mutation and recombination operations as described in (9) and (10), respectively. Step 5: Calculate the objective value of each offspring individual as described in (3). Step 6: Utilize the probabilistic state transition rule described from (11) to (12) to select P sets of the better individuals in the population. Step 7: Repeat steps 4 to 6 until the optimization process converges or the maximum iteration is reached.
5. Numerical results The proposed approach was verified on the IEEE 30-bus 6-generator system. For comparison, the EP [7], PSO [8], and DE [9] methods implemented using the commercial MATLAB package were also tested using the same database. In the system, bus 1 is slack bus, buses 2, 5, 8, 11, and 13 are generation bus, and the others are load bus. The control variables are the desired values of voltage magnitude at the PV bus, transformer tap settings, and shunt capacitors. Table 1 shows the reactive generation constraints of each PV bus. The lower and upper limits of voltage magnitude at PV bus are 0.90 and 1.10 p.u., respectively, while the
181

transformer tap settings are varied between 0.95 and 1.05 p.u. The weighting values for both active power transmission loss and voltage magnitude deviation are 0.70 and 0.30, respectively. For each method, the population size is set at 30. The maximum iteration of optimization process is set at 500. Due to the optimization processes of diverse methods are almost converged within 25 iterations through many trials with different random seeds, the numbers of iteration for the optimization process is then presented within 50 iterations. Fig. 1 shows the optimization process of different methods. After about 13 iterations, the proposed HEC converges toward the best one, while the DE, EP, and PSO methods require about 19, 18, and 15 iterations, respectively. Table 2 shows the results of reactive power dispatch for different methods, where the basic load flow solution is regarded as benchmark. The results reveal the proposed HEC can save more active power transmission losses with lower voltage deviation than the other methods. To verify the convergence performance of different methods, simulations of executing time through 30 trials using different random numbers are implemented. The average executing time for the proposed HEC is 3.24 seconds, while the DE, EP, and PSO methods need about 3.91, 3.88, and 3.55 seconds, respectively.
0.055 0.05 0.045 0.04 Objective value 0.035 0.03 0.025 0.02 0.015 0.01 DE method EP method PSO method HEC method

10

15

20

25 Iteration

30

35

40

45

50

Fig. 1 Optimization processes of different methods. Table 1. Reactive generation constraints of each PV bus. Bus 2 5 8 11 13 QBi,min (p.u.) -0.40 -0.40 -0.10 -0.06 -0.06 QBi,max (p.u.) 0.50 0.40 0.40 0.24 0.24

Table 2. Results of ORPD for different methods. Objective value Qloss PG QG Ploss(%) (p.u.) Ploss(p.u.) Vdev(p.u.) (p.u.) (p.u.) Load flow sol. 2.8868 1.0276 -0.2343 0.0528 0.0243 0.00 DE 2.8631 0.9945 -0.2674 0.0291 0.0160 44.92 EP 2.8666 0.9989 -0.2630 0.0326 0.0156 38.29 PSO 2.8576 0.9649 -0.2970 0.0236 0.0101 55.33 HEC 2.8512 0.9181 -0.3440 0.0172 0.0032 67.44 Note: 1. The total active and reactive power demands are 2.834 p.u. and 1.262 p.u., respectively. 2. The term Ploss(%) means the percentage that active power loss saves when compared to the load flow solution. Methods

182

6. Conclusions A novel variation of basic DE method to solve the ORPD problem has been proposed and implemented in this paper. The paper first introduces the formulation of ORPD problem. Then the proposed HEC for determining the ORPD problem is briefly reviewed. Owing to the efficient global search scheme, the proposed HEC algorithm can offer higher probability of converging toward global solution than the other methods. Testing on the IEEE 30-bus 6-generator system has shown that the proposed HEC can save more active power transmission losses, obtain better security operation of power transmission systems, and achieve faster convergence performance than the basic DE, EP, and PSO methods. 7. Acknowledgement This work was supported by the National Science Council of Taiwan, under the Grant No. NSC 99-3113-P-006 -004. References [1] S. S. Sachdeva and R. Billinton, Optimum network VAR planning by nonlinear programming, IEEE Trans. Power Appar. and Syst., vol. PAS-92, no. 4, pp. 1217-1225, Jul. 1973. [2] K. Aoki, M. Pan, and A. Nishikori, Optimum VAR planning by approximation method for recursive mixed-integer linear programming, IEEE Trans. Power Syst., vol. 3, no. 4, pp. 1741-1747, Nov. 1988. [3] S. Granville, Optimal reactive dispatch through interior point methods, IEEE Trans. Power Syst., vol. 9, no. 1, pp. 136-146, Feb. 1994. [4] K. Iba, Reactive power optimization by genetic algorithm for optimal planning based upon successive linear programming, IEEE Trans. Power Syst., vol. 9, no. 2, pp. 685-692, May 1994. [5] B. Zhao, C. X. Guo, and Y. J. Cao, A multiagent-based particle swarm optimization approach for optimal reactive power dispatch, IEEE Trans. Power Syst., vol. 20, no. 2, pp. 1070-1078, May 2005. [6] J. Lin and X. Wang, Reactive power optimization based on adaptive immune algorithm, Int. J. Emerging Elect. Power Syst., vol. 10: Iss. 4, article 1, 2009. [7] D. B. Fogel, System identification through simulated evolutionary: a machine learning approach to modeling, Ginn Press, Needham, MA, 1991. [8] J. Kennedy and R. Eberhart, Particle swarm optimization, Proc. IEEE Int. Conf. Neural Networks, vol. 4, pp. 1942-1948, 1995. [9] R. Storn and K. Price, Differential evolution: a simple and efficient adaptive scheme for global optimization over continues space, Int. Comput. Sci. Instit., Berkeley, Technique Report TR-95-012, 1995. [10] A. Alireza, I. Tabatabaii, and S. H. Hosseini, Optimal reactive power dispatch in electricity markets using a multiagent-based differential evolution algorithm, " Int. Conf. Power Eng., Energy and Elect. Drives, Setubal, Portugal, pp. 249-254, Apr. 12-14, 2007. [11] M. Varadarajan and K. S. Swarup, Solving multi-objective optimal power flow using differential evolution, IET Gener. Transm. Distrib., vol. 2, no. 5, pp. 720-730, 2008. [12] M. Dorigo, V. Maniezzo, and A. Colorni, Ant system: optimization by a colony of cooperative agents, IEEE Trans. Sys., Man and Cyber., Part B, vol. 26, no. 1, pp. 29-41, 1996.

183

[13] C. Zhang, M. Chen, and C. Luo, A multi-objective optimization method for power system reactive power dispatch, Proc. of the 8th IEEE Congress on Intel. Cont. and Auto., Jinan, China, pp. 6-10, July 2010.

184