Вы находитесь на странице: 1из 3

2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Identification of KEMA Arc Model Parameters in High Voltage Circuit Breaker by using of Genetic Algorithm
Vahid rashtchi*, Abbass Lotfi** and Ali mousavi*** Electrical engineering department of Zanjan University, Iran, Rashtchi@znu.ac.ir ** Roozbeh Institute of technology, Zanjan, Iran, abbass_lotfi_ee@yahoo.com *** Electrical engineering department of Tehran University, Iran, a.mousavi@ece.ut.ac.ir
*

Abstract- A circuit breaker is properly modeled by KEMA model. This model is created on the basis of Cassie and Mayr classical equations. The precision of this model depends strongly on the precision of its parameters. The calculation of these parameters needs the complex physical analysis and structural data of C.B. In this paper a novel method based on genetic algorithm (GA) is introduced as a method to identify the parameters of KEMA model by using a simple experiment. Keywords: KEMA arc model, Circuit breaker, Genetic algorithm

Since the precision of the model depends on precision of calculated parameters of the model, these parameters must be calculated as precisely as possible. On the other hand, because of the above mentioned reasons and non existence of structural data of CB in some cases, it is necessary to improve the analytically calculated values or to find a method other than analytical method to obtain the parameters of the KEMA model of the C.B. In the present work, it was attempted to solve this problem with the help of genetic algorithm (GA). II. PROPOSED PARAMETER IDENTIFICATION METHOD

I.

INTRODUCTION

Simulation of arc that occurred in interrupter box of circuit breakers is the main section of a CB modeling. The physical complexity of behavior in an electrical arc during the process, have always provided the motivation for researchers to develop suitable models that can describe this process [1]. In these models, the differential equations, which describe the arc conductivity, are extracted as famous Cassie and Mayer Models. The Cassie equation models the arc in high level currents and Mayer equation models it in low and nearly zeros currents. Recently, new models have been developed and presented in different articles, which are the corrected forms of these two equations. The KEMA model, corrected form of Mayer and the hybrid MayerCassie model, is of these types described in detail in [2, 3, & 4]. Among the recent corrected models, the KEMA arc model shows good results [5]. In all of the developed models, there are some parameters that influenced the arc performance; these depend on CB design and operating conditions. One of the key problems in utilizing these models is identifying the unknown parameters used in them. Considering the fact that accurate identification of these parameters needs the complex physical analysis, then exact and complicated numerical calculations will be essential. Coupled field finite element analysis and computational fluid dynamic are used for this purpose [7, 8]. In addition, these methods require having structural data of CB. However such data are confidential for manufacturers and they usually do not release these data to customers. Thus, engineers always attempt to develop a method that does not need structural data for CB modeling [8, 9].

The idea on which the method is based is to determine all the parameters of a mathematical model of the transformer simultaneously, so that the model is capable of matching the inputoutput behavior of the C.B. This can be achieved using the following approach [10]. Step 1: An experimental test is carried out consisting of an input excited by a current, producing arc voltage. Step 2: KEMA model of the C.B is implemented with the aim of simulating the experimental test of step1. Step 3: A fitness function is computed as the inverse of the weighted sum of the square differences of the output variables acquired experimentally and those computed by simulation at the same times. Step 4: The unknown parameters of the C.B are iteratively updated so as to maximize the above fitness function. To perform steps 3 and 4, a real coded genetic algorithm (RCGA) is used. III. KEMA ARC MODEL

KEMA arc model had been investigated in detail in [5]. In this section a brief review of it is given. This model is based on classical Cassie and Mayer Models. It utilized three corrected sub-models and a combination of those classical models. This model is valid in both high current and low current conditions. The differential equation that describes sub models is: dGi 1 1 (1) = G i i . U i Gi i .Ti dt Ti Wherein: G Is the electrical conductivity

1-4244-2405-4/08/$20.00 2008 IEEE

1515

2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

Is the arc voltage Is a dimensionless parameter T Is the time constant of arc Is the power loss in interrupter box i Is the index of i th sub model The electrical conductivity and the arc voltage of main arc are given by: 3 1 1 (2) = G i =1 Gi 3 U = Ui (3) i =1 The first sub model is combined of Cassie and Mayr equations. The second one is similar to Mayr equation and the third is exactly Mayr equation. Therefore the first sub model concerns with high currents and the third sub model describes situations that current is adjacent to zero. There are 9 parameters in this model that 6 parameters are dependent together by flowing equations: 1 = 1.4, 2 = 1.9, 3 = 2 (4)

V.

ARC VOLTAGE CALCULATION

For calculating of arc voltage, the differential equation of electrical conductivity should be solved. In order to the fourth order Runge-Kutta method is used. The discrete form of this equation is given by (8).
G k k k k U
I i ,n + 1

= G

i ,n

+ G

1 6
i ,n

(k I
i

+ 2k
2 i

II

+ 2k

III

+ k

IV

= = = = =

t Ti t Ti

U U

t G Ti

i ,n

II

G
i

i
i ,n

2 i

III

t Ti t Ti i brea G
ker i

G
i

i
i ,n

U U

IV

G
i

i
i ,n

2 i

t k G i ,n + I 2 Ti t k G i , n + II 2 Ti t (G i , n + k III ) Ti

(8)

T (5) T2 = 1 K2 T (6) T3 = 2 K2 3 = K3 2 (7) K1, K 2 and K3 are parameters that depend on CB design. In this model there are 6 independent parameters that describe arc behavior. These are T1, 1, 2 , K1, K 2 and K3 .

Wherein i is the index of i th sub model and n is the sampling time. Solution of these equations is done by time step of 5*10-7 sec. VI. RCGA FOR PARAMETER IDENTIFICATION Genetic algorithm is one of the methods which may be used to solve an optimization problem. This algorithm is based on natural selection using random numbers and does not require a good initial estimate. For identification of the unknown parameters of a model, parameters are regarded as the genes of a chromosome, and the fitness function that reflects the degree of goodness of the chromosome is optimized (maximized). Typically, a chromosome in GA is structured by string of values in binary form. Since in our parameter identification problem, the parameters of the C.B are real numbers, an RCGA similar to [10] is used. The parameters of the previously presented mathematical model can be represented by the vector X. (9) X = [ K 1 , K 2 , K 3 , 1 , 2 , T1 ] To identify the parameters, current of C.B in synthetic test is taken as the input and the arc voltage is considered as the output of the system. The fitness of a chromosome used in the process of simulation is defined in the following equation.

IV.

TEST OBJECT AND DATA ACQUISION

In this work for validation of proposed method, a commercial C.B. with the following specifications is used. 24KV 25KA Gas type with puffer interrupter All measurements were executed in the time domain. A synthetic test is carried out on the mention C.B. In the time domain, the arc voltage and arc current measured under 25KA short circuit current. The measurement is done in high current laboratory of Pars Switch Co. in Zanjan-Iran. Measured arc voltage in time domain is shown in figure 1.

fitness _ function =

1 (U meas (t ) U cal (t )) 2

(10)

Figure1. Measured arc voltage

Genetic algorithm tries to maximize the fitness function and in repeating process the measured voltage will gradually coincide with the curve of measured voltage. Here in this article, the GATOOL toolbox, available in MATLAB version 7.04, is being used. The algorithm had been run by doing these adjustments on toolbox: 1. the population size is 5 2. linear mutation function by the rate of 0.5

1516

2nd IEEE International Conference on Power and Energy (PECon 08), December 1-3, 2008, Johor Baharu, Malaysia

3. linear selection function After performing such adjustments, the algorithm was run and in an approximately 2-hours time period and after 2000 repetition, the convergence of algorithm has been acquired. The parameters resulted from the algorithm running is shown on table 1.
Table I. The identified parameters by GA X Identified K1 1.35 K2 0.73 K3 298.1 2 696.95 T1 0.00053 1 4.68e6

[5]

In order to examine the validity and accuracy of identified parameters, the arc voltage wave form resulted from simulation of model by means of identified parameters and measured one are shown in figure 2. This figure shows the correctness and accuracy of the identified parameters. It should be mentioned that the identified parameters can simulate behavior of the C.B with a negligible error.

Niklas Gustavsson, "Evaluation and simulation of black box arc models for high voltage circuit-breakers", M.s. thesis Lith-ISY-EX-3492 -linkopings university, march 2004 [6] Liu Xiaoming, Wang Erzhi, Cao Yundong "Dynamic arc modeling based on computation of couple electric field and flow field for high voltage SF6 interrupter", IEEE Transaction on magnetics, Vol. 40, NO. 2, March 2004 [7] J.C.Verite, T.Boucher, A.Comte, C.Delalondre, O.Simonin, "Arc modeling in circuit breakers: coupling between electromagnetics and fluid mechnics", IEEE Transaction on magnetics. Vol. 31, No. 3, May 1995 [8] Lionel R. Orama-Exclusa, Bienvenido Rodriguez-Medina, "Numerical arc model parameter extraction for SF6 circuit breaker simulation"International Conference on Power systems Transients IPST 2003, new Orleans, USA [9] Grigore A. CIVIDJIAN, Natalia G.SILVIS-CIVIDJIAN, "Kema arc model parameters evaluation", Optimization of electrical and electronic equipments-Brasov 1998 [10] Raie A, Rashtchi V (2002) Using a genetic algorithm for detection and magnitude determination of turn faults in an induction motor. Electric Eng 84:275279

Figure2. Measured and calculated arc voltage

VII.

CONCLUSION

This paper presents a systematic method for accurate parameter identification of the KEMA arc model. In this method a simple experimental set-up is used to conduct an easy test on the desired C.B, and then an RCGA identifies the parameters using the collected data. The validity of the scheme, the correctness and accuracy of the obtained parameters are shown by comparing the experimental and simulation results.

References
[1] [2] Thomas E.Brown ,Jr. "Circuit interruption, theory and techniques ,New york and basel, Marcel Dekker, INC, 1984 J.L.Guardado, S.G.Maximov, E. Melgoza, J.L Naredo, P.Moreno, " An inproved arc model before current zero based on the combined mayr and cassie arc models "IEEE Transaction on power delivery,Vol.20,No.1, January 2005 Pieter H.Schavemarker, Louu van der Sluis, "An improved mayrtype arc model based on current zero measurents",IEEE Transactions on power delivery, Vol.15, No.2, April 2000 Smeets, R.p.p, V. Kertesz, "Evaluation of high voltage circuit breaker performance with a validated arc model, IEE Proc., Gener. Transm.Distrib. Vol . 147, No.2, March 2000

[3] [4]

1517

Вам также может понравиться