Professional Documents
Culture Documents
AbstractThe optimal coordination of Directional The coordination of DOCRs is achieved using graphical
Overcurrent Relays (DOCRs) is a nonlinear and non-convex approach such as curve fitting technique mentioned in [5] and
optimization problem integrating large number of constraints. In [6], Fixed Point Coordination Curve Adjustment (FPCCA) [7]
this paper, the hybrid Cuckoo Search Algorithm (CSA) - Firefly and computer programming software [8]. In the next stage of
Algorithm (FFA) approach is implemented to solve coordination invention, deterministic methods have been pertained for
problem of DOCRs. The Artificial Intelligence (AI) based method optimal coordination of DOCRs. The coordination problem is
such as FFA searches large solution space with large deviation in formulated as Mixed Integer Non-Linear Programming
results with different number of simulations. Also it is not (MINLP) and solved using General Algebraic Modeling
assured that result given by FFA is global best. To solve this
System software [3]. Due to intricacy in MINLP, the
problem, the preliminary optimal value of Time Multiplier
Setting (TMS) and pickup current (Ip) are determined using
coordination problem of DOCRSs is constructed with Linear
CSA. The values of these variables are used in FFA as upper Programming (LP) techniques [4], [9]. These methods involve
bounds which reduce the solution space and give a global optimal assumption in either pickup current settings or TMS, allowing
solution with very minimum deviation. The obtained results for Time of Operation (TOP) of each relay as a linear function
using proposed method are compared with hybrid Genetic of TMS or pickup current settings.
Algorithm-Nonlinear Programming (GA-NLP) as well as with The deterministic optimization methods involve a large
conventional CSA and FFA methods. The outcome demonstrates number of iterations and computational resources. Also, these
that the proposed method can obtain realizable and global best
methods ensnare in a local optimal solution due to its
solution with minimum deviation in results and improved
computational efficiency for this complex problem.
dependence on initial point. The AI techniques are proficient to
prevail over the drawbacks of deterministic techniques by
KeywordsCuckoo Search Algorithm; Directional Overcurrent generating an initial population of feasible solutions. The
Relays; Firefly Algorithm; power system protection; relay seeker algorithm [10] and analytical approach [11] are used to
coordination optimize relay settings for minimizing the operating time of
DOCRs. To overcome shortcomings of conventional Particle
I. INTRODUCTION Swarm Optimization (PSO), modifications are suggested in the
The ever increasing demand of electrical energy has original PSO to solve coordination problem of DOCRs [12],
rendered power systems with increased complexities and [13]. Hybridization of AI methods with deterministic method
design challenges. The exercise of load flow analysis, fault increases the computational efficiency. The operating time of
calculation and listing of Primary and Backup (P/B) relay pairs DOCRs is optimized using hybrid GA-LP [14] and hybrid GA-
is tedious and time consuming [1]. The DOCRs are an NLP [15]. The setup of P/B relay pairs is attended in [16] using
appropriate choice in protective systems because of its easy LINKNET structure and optimization of DOCRs is achieved
functionality and cost effectiveness. The prime objective of using Teaching Learning Based Optimization (TLBO)
DOCRs in a power system is to identify fault timely, algorithm. The continuous and discrete version of Informative
effectively and confiscate them from the network to avoid Differential Evolution (DE) Algorithm with self-adaptive re-
possible outages by operating all P/B relay pairs within a clustering technique is used in [17] to optimize TMS and Plug
prescribed time interval [2]. At first, challenges in coordination Setting Multiplier (PSM) for minimizing operating time of
were met with manual calculations which made it a protracted DOCRs and eliminating the miscoordination of P/B relay pairs.
and strenuous task. Later in 1960s, graphical and analytical To overcome the drawbacks of DE, coordination of DOCRs is
approaches were introduced relying on developing computing resolved using opposition based chaotic DE method which uses
capacities. Optimization technique was first applied for two variants supported by opposition based learning and
coordination of DOCRs by [3]. The advent of optimization chaotic scale factor for obtaining an optimum solution [18]. To
techniques, to obtain optimal settings, eliminates the need to improve convergence rate, five modified versions of DE
determine breakpoint set and thus complex topological analysis algorithm are formed to obtain optimal coordination of DOCRs
programs are not required [2], [4]. [19], [20].
All the fireflies are unisex. A firefly is attracted to other IV. A STUDY
CASE
fireflies irrespective of its sex. The proposed implementtation analyzes the effect of
Attractiveness and brightness are prooportional to each hybridization of FFA and CSA. The results for test case
other. A less bright firefly will movee towards the one demonstrate the effectiveness of hybrid CSA-FFA to find a
which is brighter. global optimum solution. The simulation is carried out using
MATLAB version 7.8.0.347 (R2009a)
( on a Windows Vista
As the distance between two fireffly increases, the platform with 2 GB RAM andd an Intel Core 2 Duo processor
attractiveness and brightness will decrease. In a clocked at 2 GHz.
situation when no firefly is brighter than the other, it
will move randomly. A. Problem Formulation
The proposed method is applied to an eight bus, nine
The brightness of firefly is determined by b view of the OF. branch network as shown in Figg. 2. The test case has 14 IDMT
The other forms of brightness can be definned in an identical type DOCRs. The detailed system data is available in [13]. The
way to the fitness function in genetic algorithhms. short-circuit current for near-ennd 3 fault and P/B relay pairs
C. Hybrid CSA-FFA is presented in Table I. Thee directional IDMT relays are
considered in the case study anda the coordination problem is
The AI methods such as CSA and FFA F search large
carried out for phase relays.. A similar procedure can be
solution space and does not lead to a gloobal solution. The
followed for ground relays.
statistical analysis of the results given in secttion IV proves that
the deviation in fitness values obtained using FFA is large. The CSA parameters are coonsidered as the number of nests
Hence a global optimum solution cannot bee assured by FFA. n 15 and discovery rate P 0.25. The parameters of
To reduce the deviation, FFA is hybridized with CSA. In this FFA are considered as numb ber of firefies 40, 0.25,
paper, the TMS and I are considered as conntrol variables. The
x to x are variables of TMS and x to x are considered as
variables of I .
Upper Bound x ,UB , x ,UB , x ,UB , x ,UB (7)
Lower Bound x ,LB , x ,LB , x ,LB , x ,LB (8)
Initially, the optimal value of control variiables are obtained
using CSA with considering conventional upper and lower
bounds as given in (3), (4), (7) and (8). Thee value of optimal
variables found using CSA is presented as,
x ,CS , x ,CS , x ,CS , x ,CS (9)
These optimal variables are used as an uppper bound in FFA
which reduces the searching space of this method
m as shown in
Fig. 1. The lower bound is considered same as
a in case of CSA.
Upper Bound x ,CS , x ,CS , x ,CS , x ,CS (10)
Lower Bound x Fig. 2. Single line diagram of an 8 buus system
,LB , x ,LB , x ,LB , x ,LB (11)
0.2 and 1. The continuous TMS value lies in the TABLE II. OPTIMIZED RELAY SETTINGS FOR CONVENTIONAL METHODS
interval 0.1, 1.1 . The minimum operating time of each relay Relay CSA FFA
and CTI are assumed to be 0.2 and 0.3 seconds respectively. No. TMS TMS
Therefore the OF to be minimized is, 1 0.1000 550.04 0.1088 527.91
. 2 0.3071 529.43 0.3645 364.18
min Z . (12) 3 0.3409 200.12 0.2480 418.51
I,
4 0.2446 300.03 0.2389 386.21
5 0.1001 598.64 0.1483 436.81
where I , is the fault current under consideration, for 6 0.1761 556.66 0.2019 465.75
nearest fault (fault in its primary protection zone). The 7 0.2452 449.96 0.2754 383.61
coordination constraints are formed using (2) and (6), and relay 8 0.2394 300.00 0.2327 343.43
operating time constraints are formed using (5) and (6). The 9 0.1566 441.19 0.1599 375.27
10 0.2232 470.46 0.2260 425.62
constraints due to bounds on TMS and I of relays are formed
11 0.3066 301.82 0.2087 555.33
using (3) and (4) respectively. 12 0.3011 598.56 0.3216 458.95
13 0.1001 552.60 0.1125 502.65
B. Implementation of proposed method 14 0.3929 200.02 0.2468 431.53
In CSA, the constraints due to bounds on TMS are taken OF 9.2525 sec 9.2222 sec
care of by defining the lower and upper limit of variables x to
x and the constraints due to bounds on I are taken care of by
defining the lower and upper limit of variables x to x . The
TABLE III. OPTIMIZED RELAY SETTINGS FOR HYBRID METHODS
OF value can be calculated as the sum of the operating time of
each relay for the fault in its primary zone of protection. The Relay GA-NLP CSA-FFA
optimal values of TMS and I are found using CSA to No. TMS TMS
minimize the operating time of DOCRs. These values are used 1 0.2585 353.81 0.1000 516.15
as upper bound of TMS and I in FFA, which results in global 2 0.3272 521.35 0.2538 599.57
3 0.2623 450.00 0.2052 449.98
best solution for the given relay coordination problem.
4 0.2210 600.00 0.1553 599.83
C. Results and discussion 5 0.2380 312.08 0.1000 559.26
6 0.3260 423.48 0.1651 599.71
The results of optimized TMS, I and OF value using 7 0.3068 382.23 0.2201 449.85
conventional CSA and FFA, hybrid GA-NLP and CSA-FFA 8 0.3099 434.37 0.1614 599.66
are presented in Table II and Table III respectively. The 9 0.2895 209.95 0.1262 450.00
obtained value of CTI of P/B relay pairs by these methods is 10 0.2265 600.00 0.1663 599.99
11 0.2236 600.00 0.1798 599.98
shown in Table IV. It is clear that the CTI value is maintained 12 0.3624 423.35 0.2596 597.82
as minimum value of 0.3 seconds for almost all P/B relay pairs 13 0.2649 310.40 0.1000 517.12
in case of CSA-FFA but it is higher in GA-NLP, CSA and 14 0.2901 437.12 0.2223 449.95
FFA. The minimized relay operating time using these methods OF 11.6616 sec 8.0283 sec
is presented in Table V. The obtained OF value using CSA-
FFA is less than the other stated methods.
TABLE I. NEAR-END FAULT CURRENT AND P/B RELAY PAIRS OF 8 TABLE IV. OPTIMIZED VALUE OF CTI
BUS SYSTEM
P/B CTI (sec)
Primary Fault Backup Fault Relay
Pair GA-NLP CSA FFA CSA-FFA
Relay Current (A) Relay Current (A)
1 3232 6 3232 1-6 0.3000 0.3001 0.3028 0.3000
2 5924 1 996 2-1 0.8106 0.3030 0.3026 0.3000
2 5924 7 1890 2-7 0.4026 0.3102 0.3000 0.3000
3 3556 2 3556 3-2 0.3000 0.3017 0.3003 0.3000
4 3783 3 2244 4-3 0.3000 0.3050 0.3002 0.3000
5 2401 4 2401 5-4 0.3000 0.3087 0.3000 0.3000
6 6109 5 1197 6-5 0.3905 0.5022 0.4840 0.4266
6 6109 14 1874 6-14 0.5426 0.6996 0.6241 0.5890
7 5223 5 1197 7-5 0.4228 0.3213 0.3000 0.3000
7 5223 13 987 7-13 0.7843 0.5184 0.4402 0.4629
8 6093 7 1890 8-7 0.5223 0.6393 0.6393 0.5819
8 6093 9 1165 8-9 0.3623 0.5780 0.4268 0.4437
9 2484 10 2484 9-10 0.3000 0.3002 0.3000 0.3000
10 3883 11 2344 10-11 0.3000 0.3010 0.3000 0.3000
11 3707 12 3707 11-12 0.3000 0.3005 0.3001 0.3000
12 5899 13 987 12-13 0.6464 0.3014 0.3001 0.3001
12 5899 14 1874 12-14 0.4370 0.3016 0.3000 0.3000
13 2991 8 2991 13-8 0.3029 0.3038 0.3027 0.3000
14 5199 1 996 14-1 0.9302 0.3550 0.5152 0.4374
14 5199 9 1165 14-9 0.3623 0.3009 0.3001 0.3000
TABLE V. OPTIMIZED RELAY OPERATING TIME TABLE VI. STATISTICAL ANALYSIS
Relay Parameter CSA FFA CSA-FFA
GA-NLP CSA FFA CSA-FFA
No.
Simulations 50 50 50
1 0.8000 0.3883 0.4128 0.3746
Best value 9.2525 9.2222 8.0283
2 0.9197 0.8689 0.8896 0.7580
Mean value 9.5622 9.7725 8.0326
3 0.8701 0.8058 0.7942 0.6805
Worst 10.5315 10.9144 8.0355
4 0.8246 0.6587 0.7164 0.5797
Range 1.2790 1.6922 0.0720
5 0.8000 0.4976 0.5987 0.4735
Standard Deviation 0.2650 0.4309 0.0026
6 0.8323 0.5022 0.5251 0.4863
7 0.8000 0.6832 0.7191 0.6130
8 0.8000 0.5398 0.5503 0.4760
9 0.8000 0.6233 0.5812 0.5083
10 0.8331 0.7248 0.7000 0.6116
11 0.8440 0.8344 0.7551 0.6786
12 0.9379 0.9002 0.8592 0.7758
13 0.8000 0.4082 0.4338 0.3919
14 0.8000 0.8170 0.6770 0.6205