38 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO. 1, FEBRUARY 2011
Optimal Placement of Fault Indicators
Using the Immune Algorithm
Chin-Ying Ho, Tsung-En Lee, and Chia-Hung Lin, Member, IEEE
Abstract—This paper examines the application of the immune algorithm for the problem of optimal placement of fault indicators to minimize the total cost of customer service outage and invest-ment cost of fault indicators. The reliability index of each service zone is derived to solve the expected energy not served due to fault contingency, and the customer interruption cost is then determined according to the customer type and power consumption within the service zone. To demonstrate the effectiveness of the proposed IA methodology and solve the optimal placement of fault indicators, a practical distribution feeder of Taiwan Power Company is selected for computer simulation to explore the cost benefit of fault indi-cator placement.
Index Terms—Distribution automation system, fault indicator, immune algorithm.
I. INTRODUCTION
W
ITH economic development and increasing computer applications, power quality has become a more and more critical concern for utility customers. Power compa-nies have to improve overall customer satisfaction through enhancing service quality to maximize customer retention. Therefore, distribution automation systems have been imple-mented at Taiwan Power Company (Taipower) as an intelligent technology to strengthen the reliability and operation efficiency of distribution systems.Among all functions to be achieved by the distribution automation system (DAS), the fault detection, isolation, and restoration (FDIR) is considered to be the most important with the objective of reducing service restoration time from an average of 58 min to less than 20 s for the permanent fault contingency of distribution feeders. However, the experience of Taipower distribution network operations has indicated a large amount of distribution network outages occurred in the laterals of distribution feeders not monitored in the current design of Taipower’s DAS. Table I illustrates the average of the feeder and lateral outages for the 24 of Taipower’s business districts in 2008. Most of the outages are from laterals for
Manuscript received February 28, 2010; revised March 04, 2010. First pub-lished May 20, 2010; current version pubpub-lished January 21, 2011. This work was supported in part by the National Science Council of Republic of China under the Contract NSC98-3114-E-214-001. Paper no. TPWRS-00157-2010.
C.-Y. Ho is with the Department of Electrical Engineering, National Kaoh-siung University of Applied Sciences, KaohKaoh-siung 807, Taiwan, and also with the Department of Electrical Engineering, Kao Yuan University, Kaohsiung, Taiwan (e-mail: zyho@cc.kyu.edu.tw).
T.-E. Lee and C.-H. Lin are with the Department of Electrical Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung 807, Taiwan (e-mail: telee@mail.ee.kuas.edu.tw, chlin@mail.ee.kuas.edu.tw).
Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRS.2010.2048725
TABLE I
AVERAGE OFFEEDER ANDLATERALOUTAGES OFTAIPOWER IN2008
overhead and underground conductor types. Due to the larger number of branches, to include feeder laterals into the DAS will require a large capital investment in the automatic switches and communication devices. Installing high voltage and customer side fault indicators (FIs) with communication capability is a solution to minimize the fault detection and identification time at the feeder and lateral levels. The fault indicator trips when it senses an inrush of fault current, then communicates back to the distribution dispatching control center (DDCC) of Taipower through the mesh network. The DDCC personnel monitoring the master station receive the fault information and dispatch a troubleshooting crew to the fault location. This significant decrease in response time allows repair crews to react swiftly to restore power to Taipower customers.
It is neither economical nor necessary to install an FI at each line segment of a wide-area distribution system. With so many line segments of a feeder in the Taipower distribution system, the placement of FIs becomes a very difficult and tedious problem to be solved by conventional optimization techniques because of the voluminous combinations to be examined. With the installation of FIs in the distribution system, the reliability indices of customer service zones can therefore be evaluated according to the installation locations of FIs. As a result, the problem of optimal FI placement (OFP) is concerned with where and how many FIs should be implemented in the dis-tribution systems to heighten reliability at a minimum number of FIs.
A genetic algorithm (GA) based procedure for solving the OFP problem was presented in [1]. The method makes it pos-sible to select the optimal placement by knowing the charac-teristics of the distribution systems. The impact of FIs on the reliability of distribution systems was examined in [2]. The pro-posed model and evaluation technique was applied to a real Ira-nian distribution network.
The effectiveness of the immune algorithm (IA) to solve com-plicated optimization problems has been illustrated in many pre-vious case studies [3]–[8]. In this paper, an economically based fitness function is used for the IA to determine the optimal lo-cations of FIs with communilo-cations capability for the existing
HO et al.: OPTIMAL PLACEMENT OF FAULT INDICATORS USING THE IMMUNE ALGORITHM 39
distribution system to illustrate the simulation process. To alle-viate the degeneration phenomenon in the original GA and in-crease convergence speed, the proposed IA uses the prior knowl-edge of the problem in the search process. This notion is ap-plied to the OFP problem since some line segments that are or are not required to install FI can be initially determined. Three effective vaccines are abstracted using the rules associated with Taipower’s FI installation rules and the levels of the priority cus-tomers. By comparing to the genetic algorithm, the IA provides the following advantages to solve the optimization problems.
1) The memory cell is maintained without applying operators such as recombination, selection, etc. to the population. 2) It operates on the memory cell, guaranteeing rapid
conver-gence.
3) The diversity of the immune system is embedded using an affinity calculation.
4) The injection of vaccines into the individuals of genera-tions reveals a remarkably increased convergence process. In this paper, the objective function for the OFP is expressed as the antigen inputs. The feasible solutions are represented as the antibody for the IA to solve the optimization problem. The genetic operators including crossover and mutation are then pro-cessed for producing antibodies in a feasible space. Through op-erating IA on the memory cell, a very rapid convergence will be obtained during the searching process by applying the infor-mation entropy as a measure of diversity for the population to avoid falling into a local optimal solution. The effectiveness of the proposed IA to solve FI placement is then verified by com-paring to the GA.
This paper is organized as follows. Section II describes the FI placement problem description and formulation. Section III provides a brief overview of the optimization algorithm and its applications in the paper. Simulation results for the practical Taipower distribution system are provided in Section IV. Dis-cussion and conclusion are presented in Section V to explain the effectiveness of the obtained results by the proposed method.
II. TECHNICAL WORK PREPARATION PROBLEM
DESCRIPTION ANDFORMULATION
To evaluate the service reliability of the Taipower distribution system, the number of customers affected and outage duration time for each fault contingency is generated in the data logging of the outage management system (OMS). By performing the statistical analysis of service outage, the customer interruption cost of a distribution system is expressed as follows:
(1)
where
total number of line segments;
interruption cost per year due to outages in line
Segment ;
outage rate (failure per year/Km) of line Segment ;
length of line Segment ;
interruption cost of load at Segment due to an outage at Segment ;
total load of line Segment .
The in (1) represents the integrated interruption costs of different types of customers derived for the residential, commer-cial, and industrial customers, respectively [9]. Besides, three different categories of key customers with high service priority levels are considered in this paper. The higher hierarchy level of customers indicates the power service is more critical to them.
Level 1) the customers with power outage could be affected by inconvenience or public concern (schools, supermar-kets, sport and entertainment facilities, etc.)
Level 2) the customers with power outage could result in serious financial damage (banks, oil refinery plants, high technology plants, etc.)
Level 3) the customers with power outage could jeopardize public security (hospitals, police stations, fire stations, im-portant telecommunications, etc.)
(2) where
, , , load percentage of residential, commercial, industrial, and key customers;
, , , interruption cost function of residential, commercial, industrial, and key customers;
duration of service interruption of Segment due to a outage at
Segment ;
hierarchy level of key customers. To solve the load percentages of residential, commercial, industrial, and key customers within each service zone, the customer-to-transformer mapping is retrieved from the facility database of OMS. Besides, the daily load patterns of different customer classes are derived by load survey study [10], and the energy consumption of each customer is retrieved from the customer information system (CIS) database. The hourly loading of each service zone is then obtained by integrating the power profiles of all customers served. Fig. 1 shows the overall structure of reliability assessment for the distribution systems.
As described previously, the objective of service reliability improvement is to reduce customer service outage cost by proper placement of FIs. To solve the problem, FIs have to be installed at the feeder and lateral levels to improve distribution system reliability. In this paper, the total cost of reliability ( ) to be minimized is defined as (3):
(3) where is customer interruption cost and is the in-vestment cost of FIs.
To reduce the search space and provide the installation knowledge for the abstraction of vaccines of IA, the following
40 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 26, NO. 1, FEBRUARY 2011
Fig. 1. Reliability assessment for distribution systems.
Taipower’s FI installation rules are considered in the proposed IA.
1) The feeders or laterals serving high technology plants, industrial areas, business metropolises, and outage fre-quently service zones have higher priority to install FIs. 2) The FIs must be installed on the non-automatic switches,
three-phase, four-way sectionalizing cabinets, or three-phase, feed-through, pad-mounted transformers. 3) The FIs are not required to be installed on the normally
open-tie nodes.
III. IMMUNEALGORITHM
The immune algorithm (IA) has been widely used to solve optimization problems by applying the same operating principle as the human immune system. The capability of IA method for pattern recognition and memorization does provide a more effi-cient way to solve the optimization problem compared to the ge-netic algorithm. The objective function is represented as antigen inputs, while the solution process is simulated by the antibody production in the feasible space through the genetic operation mechanism (i.e., crossover and mutation). During the actual op-eration, IA prevents the degenerative phenomena arising from the crossover and mutation processes, thus making the fitness of population increase steadily [11]. The immune operators in-cluding vaccine injection and immune selection are performed in the IA process, as shown in Fig. 2. The vaccine is used for increasing fitness and the immune selection is for preventing deterioration. The calculation of affinity between antibodies is embedded within the immune selection to determine the pro-motion and suppression of antibody production. Through the IA computation, the antibody best fitting the antigen is considered the solution to the optimization problem.
An immune algorithm based decision making [12] is pro-posed in this study to find the optimal FI number and their place-ment for distribution systems. The population of memory cells
Fig. 2. Flowchart to find the optimal solution by IA. TABLE II
BINARYSTRUCTURE ANDCORRESPONDINGDESCRIPTION FOREACHGENE
is a collection of the antibodies (feasible solutions) accessible toward the optimality, which is the key factor to achieve rapid convergence for global optimization. In this paper, the genetic coding structure for the immune algorithm is adopted and the diversity and affinity of the antibodies are calculated during the decision making process to discover the FI placement. By ap-plying the immune algorithm to solve the optimal placement of FIs, the attributes of each gene represent installation status (with/without FI) and type (FI only/manual switch with FI) at the candidate installing locations, as illustrated in Table II. The data structure of the genes is represented as two bits with binary coding in each gene structure. For a feeder with pos-sible strategies of FI placement with FIs, it will generate antibodies having genes in the antibody pool as shown in Fig. 3.
A. Abstraction of Vaccines
As mentioned earlier, vaccines are abstracted from prior knowledge or local information of the problem. According
HO et al.: OPTIMAL PLACEMENT OF FAULT INDICATORS USING THE IMMUNE ALGORITHM 45
TABLE VI
TCROF THETESTEDTAIPOWERDISTRIBUTIONSYSTEM
$83 980/year. Although almost the same results were obtained by both algorithms, higher efficiency of the IA algorithm with vaccines to solve the optimization problem has been illustrated. It is noted that the vaccination operator has some overhead com-putation in each generation, so the execution time of a genera-tion in IA with vaccines is a little longer than that in IA without vaccines. However, the efficiency of IA with vaccines is still superior to that of IA without vaccines while the required gen-erations decrease to achieve an acceptable solution.
V. CONCLUSIONS
In this paper, the application of the IA method to solve the optimal placement of FIs for distribution systems is presented. The main idea of the proposed IA is to utilize prior knowledge to associate with the considered problem. The a priori knowledge of the OFP problem is abstracted as some vaccines based on the Taipower’s FI installation rules and the levels of the priority customers. Rapid convergence is obtained during search process by injecting these vaccines into the individuals of generations. With the proposed placement of FIs, the customer interruption time has been reduced very effectively with enhanced of service reliability.
To demonstrate the effectiveness of the proposed immune al-gorithm to solve the OFP problem, one Taipower feeder within the service area of Fengshan DAS project was selected for com-puter simulation. The number and installation locations of FIs were determined after solving the optimization problem by the proposed IA algorithm. It is found that four and 27 FIs are in-stalled at the primary feeder/lateral levels, and four FIs corre-sponding with manual switches are added for a lateral with key customers. The expected customer interruption cost due to ser-vice outage was derived to examine the effect of the proposed OFP on system service reliability. It is found the customer inter-ruption cost of the test feeder decreased by 32% or $38 846 per year with an annualized investment of $3456 for the proposed placement of FIs. It is concluded that the optimal placement of FIs by the proposed immune algorithm can therefore strengthen the FDIR function of distribution systems to reduce customer in-terruption cost for fault contingency in a very cost-effective way.
REFERENCES
[1] D. P. Cong, B. Raison, J. P. Rognon, S. Bonnoit, and B. Manjal, “Op-timization of fault indicators placement with dispersed generation in-sertion,” in Proc. 2005 IEEE Power Eng. Soc. General Meeting, pp. 355–362.
[2] H. Falaghi, M. R. Haghifam, and M. R. Osouli-Tabrizi, “Fault indica-tors effects on distribution reliability indices,” in Proc. 2005 18th Int.
Conf. Electricity Distribution..
[3] P. Mitra and G. K. Venayagamoorthy, “An adaptive control strategy for DSTATCOM applications in an electric ship power system,” IEEE
Trans. Power Electron., vol. 25, no. 1, pp. 95–104, Jan. 2010.
[4] L. Xu, M. -Y. Chow, J. Timmis, and L. S. Taylor, “Power distribution outage cause identification with imbalanced data using artificial im-mune recognition system (AIRS) algorithm,” IEEE Trans. Power Syst., vol. 22, no. 1, pp. 198–204, Feb. 2007.
[5] E. G. Carrano, F. G. Guimaraes, R. H. C. Takahashi, O. M. Neto, and F. Campelo, “Electric distribution network expansion under load-evo-lution uncertainty using an immune system inspired algorithm,” IEEE
Trans. Power Syst., vol. 22, no. 2, pp. 851–861, May 2007.
[6] X. Hao and C. -X. Sun, “Artificial immune network classification al-gorithm for fault diagnosis of power transformer,” IEEE Trans. Power
Del., vol. 22, no. 2, pp. 930–935, Apr. 2007.
[7] V. Cutello, G. Nicosia, M. Pavone, and J. Timmis, “An immune algo-rithm for protein structure prediction on lattice models,” IEEE Trans.
Evol. Comput., vol. 11, no. 1, pp. 101–117, Feb. 2007.
[8] S. J. Huang, “An immune-based optimization method to capacitor placement in a radial distribution system,” IEEE Trans. Power Del., vol. 15, no. 2, pp. 744–749, May 2000.
[9] G. Toefson, R. Billinton, G. Wacker, E. Chan, and J. Aweya, “A Cana-dian customer survey to assess power system reliability worth,” IEEE
Trans. Power Syst., vol. 9, no. 1, pp. 443–450, Feb. 1994.
[10] C. S. Chen, J. C. Hwang, Y. M. Tzen, C. W. Huang, and M. Y. Cho, “Determination of customer load characteristics by load survey system at Taipower,” IEEE Trans. Power Del., vol. 11, no. 3, pp. 1430–1436, Jul. 1996.
[11] L. Jiao and L. Wang, “A novel genetic algorithm based on immunity,”
IEEE Trans. Syst., Man, Cybern., vol. 30, no. 5, pt. A, pp. 552–561,
Sep. 2000.
[12] C. H. Lin, C. S. Chen, C. J. Wu, and M. S. Kang, “Application of immune algorithm to optimal switching operation for distribution-loss minimization and loading balance,” Proc. Inst. Elect. Eng., Gen.,
Transm., Distrib., vol. 150, no. 2, pp. 183–189, Mar. 2003.
[13] Y. Tsujimura and M. Gen, “Entropy-based genetic algorithm for solving TSP,” in Proc. 1998 2nd Int. Conf. Knowledge-Based
Intelli-gent Electronic System.
[14] Distribution System Planning Technical Manual (1), Dept. Business, Taiwan Power Co., Dec. 2008.
[15] D. E. Goldberg, Genetic Algorithms in Search, Optimization, and
Ma-chine Learning. Reading, MA: Addison-Wesley, 1989.
Chin-Ying Ho is currently pursuing the Ph.D. degree
in electrical engineering of National Kaohsiung Uni-versity of Applied Sciences, Kaohsiung, Taiwan.
Since August 1994, she has been with the Depart-ment of Electrical Engineering, Kao Yuan University, Kaohsiung, where she is currently a Lecturer. Her re-search interest is in the areas of load survey study and computer applications to power systems.
Tsung-En Lee received the B.Sc. degree in
elec-trical engineering from the National Institute of Technology, Taipei, Taiwan, in 1985, the M.Sc. de-gree in electrical engineering from National Taiwan University, Taipei, in 1989, and the Ph.D. degree in electrical engineering from National Sun Yat-Sen University, Kaohsiung, Taiwan, in 1995.
He is presently an Associate Professor at National Kaohsiung University of Applied Sciences. His re-search interest is the application of geographic infor-mation system to power system.
Chia-Hung Lin (S’95–M’98) received the B.S.
de-gree from National Taiwan Institute of Technology, Taipei, Taiwan, in 1991, the M.S. degree from the University of Pittsburgh, Pittsburgh, PA, in 1993, and the Ph.D. degree in electrical engineering from the University of Texas at Arlington in 1997.
He is presently a full Professor at National Kaoh-siung University of Applied Sciences, KaohKaoh-siung, Taiwan. His area of interest is distribution automa-tion and computer applicaautoma-tions to power systems.
