Electric Power Systems Research 79 (2009) 1538–1545 Contents lists available at ScienceDirect Electric Power Systems Research journal homepage: www.elsevier.com/locate/epsr A new and accurate fault location algorithm for combined transmission lines using Adaptive Network-Based Fuzzy Inference System Javad Sadeh ∗ , Hamid Afradi Electrical Engineering Department, Faculty of Engineering, Ferdowsi University of Mashhad, P.O. Box: 91775-1111, Mashhad, Iran a r t i c l e i n f o Article history: Received 5 September 2008 Received in revised form 13 April 2009 Accepted 30 May 2009 Available online 26 June 2009 Keywords: Transmission line protection Fault location algorithm Combined overhead and underground transmission line Adaptive Network-Based Fuzzy Inference System (ANFIS) a b s t r a c t This paper presents a new and accurate algorithm for locating faults in a combined overhead transmission line with underground power cable using Adaptive Network-Based Fuzzy Inference System (ANFIS). The proposed method uses 10 ANFIS networks and consists of 3 stages, including fault type classification, faulty section detection and exact fault location. In the first part, an ANFIS is used to determine the fault type, applying four inputs, i.e., fundamental component of three phase currents and zero sequence current. Another ANFIS network is used to detect the faulty section, whether the fault is on the overhead line or on the underground cable. Other eight ANFIS networks are utilized to pinpoint the faults (two for each fault type). Four inputs, i.e., the dc component of the current, fundamental frequency of the voltage and current and the angle between them, are used to train the neuro-fuzzy inference systems in order to accurately locate the faults on each part of the combined line. The proposed method is evaluated under different fault conditions such as different fault locations, different fault inception angles and different fault resistances. Simulation results confirm that the proposed method can be used as an efficient means for accurate fault location on the combined transmission lines. © 2009 Elsevier B.V. All rights reserved. 1. Introduction In power systems, transmission and distribution lines are vital links that achieve the continuity of service from the generating plants to the end users. Protection systems for transmission lines are one of the most important parts in power systems. Fault location is a desirable feature in any protection scheme. The increasing complexities of the modern power transmission systems have greatly raised the importance of the fault location research studies in recent years. The restoration can be accelerated if the location of a fault is known or can be estimated with reasonable accuracy. Following the occurrence of a fault, the utility tries to restore power as quickly as possible, as rapid restoration of service reduces customer complaints, outage time, operating cost, loss of revenue and maintains system stability. To aid rapid and efficient service restorations, an accurate fault location technique is needed. Transmission line fault location techniques can be classified into two main categories: techniques based on the fundamental power frequency component [1–5], and techniques utilizing the higher frequency components of the fault signals [6–9]. The latter are also referred to as traveling wave methods, due to their use of travel- ∗ Corresponding author at: Faculty of Engineering, Electrical Department, P.O. Box: 917775-1111, Mashhad, Iran. Tel.: +98 511 8763302; fax: +98 511 8763302. E-mail addresses: [email protected], [email protected] (J. Sadeh). 0378-7796/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2009.05.007 ing wave theory. Each of these two groups can be further divided into two subgroups: to use single-end information or to use information from both ends of the faulted line. Algorithms that use measurements of current and voltage at both terminals of the line are generally more accurate than the ones using data only from the local terminal. However, in many transmission lines, a communication channel between the local and remote line terminals is not available, thus make it necessary to use data from the local terminal only. Fault location algorithms based on only local terminal current and voltage data need some simplifying assumptions to make the fault distance calculation possible, affecting the accuracy of the results. However, fault location techniques using single-ended data could be more attractive for researchers. Recently, the electricity demand density has increased rapidly in metropolitan areas. All over the world, large scale underground power cable installations start to replace overhead transmission lines due to environmental concerns in densely populated areas. To operate combined transmission lines (lines consisting of overhead and underground cable portions) with high efficiency, numerous related technologies are required. One of them is the fault location and protection. Underground cable faults may be series faults in which the cable is cut without the electrical insulation being broken, or shunt faults in which a break in the electrical insulation occurs without the conductor itself being cut. On the other hand, overhead line faults may be caused by lightning strokes, falling trees, fog, and salt spray on contaminated insulators. In J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 transmission systems consisting of an overhead line in combination with an underground power cable, the main problem in the fault location techniques is the unequal impedances of the underground cables and the overhead lines. Various fault location algorithms for overhead transmission lines and underground cable transmission lines have been developed so far [1–9] but the field of fault location in combined lines is not matured yet. In Ref. [10] a digital distance relaying algorithm is presented for combined overhead transmission lines with underground power cables. This reference describes an accurate calculation algorithm of the line impedance with the compensation factor at the relay point, as well as the method to discriminate the fault section between both sections when the single line-toground fault has occurred. In Ref. [11] a fault location scheme for transmission systems consisting of an overhead line combined with an underground power cable is proposed that requires synchronized phasor measurements data at one end of the transmission line and at the far end of the power cable. Fault location is determined using distributed line model, modal transformation theory and Discrete Fourier Transform. Wavelet-based traveling wave fault location algorithm is presented in Ref. [12] for combined transmission lines. With the great developments of ANN training and design methods, a series of contributions have been introduced to estimate the fault distance of the faulted line. This is mainly accomplished by those ANN types that use supervised training to map the hidden relations between the available data and the unknown fault distance through the collected training data [13–15]. So far, Fuzzy 1539 Logic (FL) systems have no solid contributions for fault location purposes. The main reason behind this is the difficulty in optimizing the constructed FL network parameters. However, the recent training methods for optimizing the parameters for FL networks are expected to change the situation drastically and turn the FL into efficient and successful tools. As is clear from the above explanations, the fault location problem, especially for the combined overhead line and underground cable, is the very type of the problems which ANN-FL tool is designed to handle in an efficient manner. In general, a neuro-fuzzy system is an intelligent model which has both the learning capabilities of the neural networks plus the knowledge illustration of fuzzy logic systems using linguistic expressions. Application of the artificial neural network to fault location in the combined transmission line is proposed in Ref. [16]. The algorithm consists of two parts. In the first part the faulty section is determined, and in the second part the exact location of fault is calculated. Wavelet and neuro-fuzzy-based fault location algorithm is proposed in Ref. [17]. In this reference, neuro-fuzzy system consists of two parts as in Ref. [16]. The algorithm has the ability to consider the high impedance fault in overhead section and the simulation results confirm the accuracy of the algorithm. In this paper an accurate fault location algorithm for combined overhead transmission line with underground power cable is proposed which is based on Adaptive Network-Based Fuzzy Inference System (ANFIS). Four inputs are used to train the ANFISs in order to classify the fault type, detect the faulty section and accurately locate the faults on each part of the combined line. The proposed method is tested under different fault conditions such as different fault locations, different fault inception angles and different fault resistances. The simulation results confirm the validity and high accuracy of the new algorithm. 2. Adaptive Network-Based Fuzzy Inference System [18] 2.1. ANFIS architecture An adaptive network is a multilayer feedforward network in which each node performs a particular function on the incoming signals, as well as on the set of parameters pertaining to this node. The formula for the node functions may vary from node to node, and the choice of each node function depends on the overall input–output function which the adaptive network is required to carry out. The parameter set of an adaptive network is the union of the parameter sets of each adaptive node. In order to achieve a desired input–output mapping, these parameters are updated according to given training data and a gradient-based learning procedure. In the rest of this section, the architecture of ANFIS as an adaptive network is described. For the purpose of illustration, the simplifying assumption is considered that the fuzzy inference system under consideration has two inputs x and y and one output f. Suppose that the rule base contains two fuzzy if-then rules of Takagi–Sugeno’s type [19]: • Rule 1: if x is A1 and y is B1 then f1 = p1 x + q1 y + r1 . • Rule 2: if x is A2 and y is B2 then f2 = p2 x + q2 y + r2 . Then the fuzzy reasoning is illustrated in Fig. 1(a) and the corresponding equivalent ANFIS architecture is shown in Fig. 1(b). The node functions in the same layer are of the same function family, as described below: • Layer 1: Every node i in this layer is an adaptive node with a node function: Fig. 1. (a) Fuzzy reasoning and (b) equivalent ANFIS. Oi1 = Ai (x) (1) 1540 J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 where x is the input node i, and Ai is the linguistic label associated with this node function. In other words, Oi1 is the membership function of Ai and satisfies the quantifier Ai . Usually Ai (x) is chosen to be bell-shaped or Gaussian with maximum equal to 1 and minimum equal to 0, such as Ai (x) = 1 (2) 2 bi 1 + [((x − ci )/ai ) ] or x − ci 2 A (x) = exp − (3) ai where {ai , bi , ci } is the parameter set. As the values of these parameters change, the bell-shaped or Gaussian functions vary accordingly, thus exhibiting various forms of membership functions on linguistic label Ai . In fact, any continuous and piecewise differentiable functions, such as commonly used trapezoidal or triangular-shaped membership functions, are also qualified candidates for node functions in this layer. Parameters in this layer are referred to as premise parameters. • Layer 2: Every node in this layer is a fixed node labeled ˘, which multiplies the incoming signals and sends the product out. For instance, wi = Ai (x) × Bi (y) i = 1, 2 (4) Each node output represents the firing strength of a rule. • Layer 3: Every node in this layer is a fixed node labeled N. The ith node calculates the ratio of the ith rule’s firing strength to the sum of all rules’ firing strengths: w̄i = wi w1 + w2 i = 1, 2 (5) For convenience, outputs of this layer will be called normalized firing strengths. • Layer 4: Every node i in this layer is an adaptive node with a node function: Oi4 = w̄i fi = w̄i (pi x + qi y + ri ) (6) where w̄i is the output of layer 3, and {pi , qi , ri } is the parameter set. Parameters in this layer will be referred to as consequent parameters. • Layer 5: The single node in this layer is a fixed node labeled ˙ that computes the overall output as the summation of all incoming signals, i.e.: Oi5 = overall outpur = wi fi w̄i fi = i i i wi (7) 2.2. ANFIS learning algorithm ANFIS employs two modes of learning. First, a forward pass is made using current premise parameters to optimize rule consequent parameters using least square estimation based on output error. This is possible since outputs are a linear function of consequent parameters. Second, a backward pass is made to alter premise parameters using gradient-based learning. This process of learning is named Hybrid Learning. The backward pass employs learning in a similar way as to the back-propagation in neural networks. For each pass, each rule antecedent parameter ˛ is changed according to ˛ = − ∂E ∂˛ (8) Fig. 2. The proposed ANFIS structure for fault type classification and faulty section detection. and = k (∂E/∂˛) ˛ (9) 2 where E is the output error, is the learning rate parameter and k is a parameter which is automatically varied during learning process to adapt to the learning rate. k is increased if four consecutive learning epochs reduce output error and is decreased if two consecutive learning epochs result in non-monotonic changes in error. ∂E/∂␣ is calculated using the chain rule [18]. 3. Architecture of the proposed fault location algorithm In this section the architecture of the proposed fault location algorithm is presented. The proposed method consists of three stages, including fault type classification, faulty section detection, as well as exact fault location. The presented algorithm contains 10 ANFISs, 1 for fault type classification, 1 for faulty section detection, and the other 8 networks for accurate fault location (2 for each fault type). Figs. 2 and 3 shows the schematic diagrams of the algorithm. As a first step in any pattern classification technique, feature extraction is used to reduce the dimension of the raw data and extract useful information in a concise form. It is important to note that different fault locations produce different frequency components in voltage and current signals. This also means that these signals vary with fault type, location and inception angle. The crisp inputs to ANFIS comprise a set of features based on the voltage and current. An acceptable and simple criterion used here is that a variable as a feature for the ANFIS input should provide more information for fault location than those not selected. In this respect, for fault type classification (ANFIS1), the amplitude of the fundamental frequency of three phases currents and neutral current are selected as inputs and the outputs have been termed as A, B, C and G, which represent the three phases and ground. Any one of the outputs A, B, C approaching 1 indicates a fault in that phase, and if G is taken 1 indicates a fault involve ground. When the phases which are involved in fault are determined, four features are extracted from digitized data and are fed to faulty section detection block as shown in Fig. 2. Normalized values of post-fault peaks of the fundamental component of voltage and current, the phase difference between voltage and current and the dc component of current are obtained as features of the measured signals and are used as inputs to ANFIS2 for detecting the faulty section. These inputs are estimated from sampled values of the current/voltage signal using least square error estimation (LSE) method [20] within one cycle after fault inception. This network determines whether the fault is on the overhead or on the underground section of the transmission line. J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 1541 Fig. 4. System under study. ters, an input pattern corresponding to a particular fault condition is fed to the ANFIS and its output is compared with the desired output pattern corresponding to that fault condition. The ANFIS parameters are updated after all patterns have been presented to the ANFIS. 4. Performance evaluation In this section performance of the proposed algorithm is evaluated under different fault conditions and some of the results are presented. 4.1. Power system model The single line diagram of a 100 km, 220 kV combined overhead transmission line with underground power cable is shown in Fig. 4. The line consists of a 90 km overhead transmission line ended with 10 km underground cable. Distributed parameter model is used for modeling of the overhead line and the underground cable. The proposed fault location algorithm requires only the three phase voltages and currents at the sending end of the overhead transmission line. This system has been simulated using the well known Alternative Transient Program (ATP) and the collected data is processed using MATLAB to prepare the inputs for the ANFISs. 4.2. Fault type classification and faulty section detection Fig. 3. The proposed ANFIS structure for exact fault location. In the next step, eight ANFISs are trained for each fault type and each faulty section, as shown in Fig. 3. Once the fault is classified and the faulty section is determined, the relevant ANFIS for fault location is activated. The inputs for these networks are the same as the inputs of ANFIS2 and the output is the normalized distance of the fault point from the sending end of the overhead line. In order to design an ANFIS, it is crucial to train it efficiently and correctly. The training set must be carefully chosen so that all the different fault conditions such as different fault inception angles, different fault resistances and different fault locations are considered. The performance of the ANFIS is then tested using both patterns within and outside of the training set. The approach adapted here is based on the Hybrid Learning that is described in Ref. [18]. In order to find the optimum values of the ANFIS parame- Different fault types at various locations of each section of the system under study with different inception angles and fault resistances are used for training and testing the ANFIS1 for fault type classification and the ANFIS2 for faulty section determination. During training of the ANFIS1 any phase/ground involved in the fault is assigned ‘1’ otherwise ‘0’, whereas during testing if the output is less than 0.25 then it would be classed as ‘0’, i.e., a healthy phase indication, and if it is greater than 0.75 then it is classed as unity, i.e., a faulty phase indication. Also during training of ANFIS2 the faults on the overhead line is assigned ‘0’ and the faults on the underground cable is assigned ‘1’. Table 1 shows some of the test results for different system conditions which were not presented to the ANFIS during the training process. For each case it can be seen that the ANFISs’ outputs converge to the desired values, and are either very close to zero or to one. 4.3. Accurate fault location using ANFISs All types of faults with different inception angles, different locations and different fault resistances in both sections of the combined overhead line and underground cable are simulated to evaluate the performance of the proposed fault location algorithm. The system under study is shown in Fig. 4. In the following sections some of the results are presented. The percentage error is defined as the formula given below [11]: % Error = Actual fault location − Calculated fault location Total faulty section lenght × 100 (10) 1542 J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 Table 1 Results of the fault type classifier and the fault section detector. Fault type Fault inception angle Faulty section – a–g a–g b–g c–g a–b b–c c–a ab–g bc–g ca–g abc abc–g – 0 90 0 90 90 45 45 90 0 0 0 90 – 0 1 0 1 0 1 0 1 0 1 0 1 Output results of ANFIS1 Output results of ANFIS2 A B C G 0.000 0.996 1.023 0.132 0.013 1.063 0.033 1.039 0.899 −0.165 1.002 1.185 1.210 0.000 0.015 −0.022 1.019 −0.191 0.978 0.991 −0.001 1.028 1.112 −0.199 0.905 1.054 −0.000 0.021 0.125 0.012 0.992 0.011 0.973 1.195 0.019 0.972 1.035 1.036 0.899 0.000 1.031 1.029 0.985 1.081 −0.109 −0.053 0.134 1.103 1.055 0.961 −0.023 1.143 4.3.1. Single line to ground faults In this study, with considering the 5 and 0.5 km as fault distance steps in overhead and underground cable, respectively, 20 and 10◦ as fault inception angle steps in overhead and underground cable, respectively, and four different fault resistances (0, 10, 50 and 100 ) for overhead line and solid short circuit on the underground cable, 2132 patterns are produced for training the ANFISs, among which 1520 patterns are associated with the overhead line and the rest are for the underground cable. The percentage error during – −0.078 1.001 0.091 0.956 −0.106 1.213 0.146 0.978 −0.089 1.129 0.019 1.023 training for this type of fault on the overhead line and underground cable are shown in Figs. 5 and 6, respectively. As shown in these figures, the maximum percentage error for the overhead line and the underground cable are 0.0109 and 0.031, respectively, which means that the training process is completed successfully. Once the training procedure is done completely, the networks are tested using simulated fault patterns not presented during the training process. The percentage error for 70 patterns (35 patterns for the overhead line and the same number for the underground Table 2 Fault conditions and percentage error for single line to ground faults on the overhead line. Fault location 8 12 12 27 27 44 44 58 58 67 67 73 73 86 86 19 19 19 Fault inception angle 325 25 125 225 75 45 145 25 275 195 95 5 335 15 155 10 25 70 Fault resistance Percentage error Fault location 35 55 55 55 75 35 55 7 7 47 15 40 40 10 1 10 70 35 0.018 0.004 0.018 0.001 0.006 0.009 0.003 0.013 0.003 0.005 0.005 0.018 0.002 0.010 0.002 0.003 0.016 0.009 19 19 37 37 47 47 47 62 62 62 81 85 85 85 85 87 87 Fault inception angle 40 35 135 0 15 40 10 215 15 115 75 35 55 275 195 95 35 Fault resistance Percentage error 55 55 55 75 35 7 7 47 47 40 40 30 2 15 1 10 25 0.003 0.010 0.0277 0.014 0.006 0.003 0.003 0.002 0.005 0.012 0.009 0.004 0.005 0.004 0.013 0.011 0.013 Table 3 Fault conditions and percentage error for single line to ground faults on the underground cable. Fault location Fault inception angle Percentage error Fault location Fault inception angle Percentage error 91.2 91.2 93.7 93.7 95.6 95.6 97.4 97.4 98.2 98.2 91.3 98.8 95.1 96.6 96.6 97.7 97.7 95.6 3 333 235 7 77 125 265 25 95 215 125 45 175 345 355 85 185 230 0.027 0.005 0.008 0.034 0.005 0.026 0.022 0.015 0.017 0.036 0.002 0.004 0.038 0.034 0.010 0.029 0.010 0.022 95.6 93.3 93.3 93.3 93.7 93.7 93.7 97.2 97.2 92.9 92.9 92.4 92.4 95.5 95.5 91 91 330 18 357 45 14 18 185 118 30 105 33 113 15 45 165 58 158 0.002 0.008 0.014 0.021 0.001 0.034 0.032 0.032 0.015 0.008 0.004 0.020 0.001 0.020 0.026 0.024 0.011 J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 1543 Fig. 8. Percentage error during testing (underground cable). error for the overhead line and the underground cable are 0.0277 (about 24.9 m) and 0.038 (about 3.8 m), respectively. Fig. 5. Percentage error during training (single line to ground fault on the overhead line). cable) are indicated in Figs. 7 and 8. Also, the simulation conditions and the percentage error for these patterns are presented in Tables 2 and 3, for faults on the overhead line and on the underground cable, respectively. The presented results in these figures and tables show that the proposed algorithm has a high degree of accuracy such that during testing process the maximum percentage 4.3.2. Three phase faults In this case, the fault distance steps are assumed to be 5 and 0.25 km in overhead and underground cable, respectively, and fault inception angle steps are 20 and 10◦ in overhead and underground cable, respectively. Four different fault resistances (3, 10, 30 and 50 ) for overhead line and solid short circuit on the underground cable are considered. For this condition, 3026 patterns are produced, among which 1805 patterns are associated with the overhead line and the rest are for the underground cable. These patterns are applied to train the ANFISs. The percentage error during training for this type of fault on the overhead line and the underground cable are shown in Figs. 9 and 10, respectively. As shown in these figures, the maximum percentage errors are 0.0085 and 0.0739, respectively. These values indicate that the training process is done successfully. After executing the training procedure, the networks are tested using simulated fault patterns not presented during the training process. The simulation conditions and the percentage error for 65 patterns, 35 patterns for the overhead line and 30 patterns for the underground cable are shown in Tables 4 and 5, respectively. Based on the presented results in these tables, it can be seen that similar to the previous case, the proposed method has a high degree of accuracy such that during testing the maximum percentage error for the overhead line and the underground cable are 0.0081 (about 7.3 m) and 0.071 (about 7.1 m), respectively. The obtained results, for Fig. 6. Percentage error during training (single line to ground fault on the underground cable). Fig. 7. Percentage error during testing (overhead line). Fig. 9. Percentage error during training (three phase fault on the overhead line). 1544 J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 Table 4 Fault conditions and percentage error for three phase faults on the overhead line. Fault location 6 6 6 6 16 16 16 18 22 22 22 23 23 26 26 26 26 Fault inception angle 23 325 235 235 115 15 15 10 165 165 165 45 45 25 0 165 0 Fault resistance Percentage error Fault location 10 10 45 65 1 7 1 7 5 14 24 35 45 5 15 30 35 0.0030 0.0002 0.0004 0.0005 0.0027 0.0036 0.0013 0.0016 0.0027 0.0005 0.0007 0.0081 0.0004 0.0017 0.0005 0.0023 0.0009 47 47 47 54 67 67 67 67 72 72 77 77 82 82 87 87 Fault inception angle 70 70 240 65 260 326 100 5 45 95 23 0 240 240 345 34 Fault resistance Percentage error 5 15 15 45 5 10 15 45 35 35 3 33 5 15 13 37 0.0023 0.0011 0.0009 0.0036 0.0010 0.0013 0.0005 0.0014 0.0018 0.0030 0.0029 0.0004 0.0006 0.0010 0.0044 0.0001 Table 5 Fault conditions and percentage error for three phase faults on the underground cable. Fault location Fault inception angle Percentage error Fault location Fault inception angle Percentage error 91.0 91.25 91.75 91.9 92.0 92.2 92.2 92.9 93.3 93.7 94.1 94.1 94.7 94.7 95 63 12 50 110 41 115 120 26 335 200 25 315 10 115 185 0.071 0.006 0.065 0.001 0.062 0.006 0.006 0.0005 0.005 0.015 0.040 0.002 0.005 0.008 0.017 95.3 95.6 96.2 96.4 96.7 97.2 97.5 97.7 97.9 98.1 98.2 98.4 98.6 98.8 98.8 35 135 335 345 15 75 145 215 10 90 190 75 0 10 225 0.025 0.040 0.059 0.010 0.052 0.009 0.011 0.052 0.026 0.0005 0.025 0.011 0.019 0.025 0.049 based on the Adaptive Network-Based Fuzzy Inference System. The algorithm consists of three stages, including fault type classification, faulty section determination and fault location. To estimate the accuracy of the proposed algorithm, a wide variety of conditions such as different fault types, different fault inception angles and different fault resistances on the overhead line and on the underground cable are simulated and some of the results are presented. Based on the obtained results, it can be concluded that the proposed method is very effective to classify the fault type, to determine the faulty section and to find the exact location of fault such that the maximum percentage error is kept below 0.07%. References Fig. 10. Percentage error during training (three phase fault on the underground cable). the same fault conditions, are comparable to or better than those as published in Ref. [11]. 5. Conclusion This paper proposed a new algorithm for fault location in a combined overhead transmission line with underground power cable [1] L. Eriksson, M.M. Saha, G.D. Rockefeller, An accurate fault locator with compensation for apparent reactance in the fault resistance resulting from remote-end infeed, IEEE Trans. Power Appar. Syst. PAS 104 (1985) 424–435. [2] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondou, T. Matsushima, Development of a new type fault locator using one terminal voltage and current data, IEEE Trans. Power Appar. Syst. PAS 101 (1982) 2892–2898. [3] C.E.M. Pereira, L.C. Zanetta, An optimization approach for fault location in transmission lines using one terminal data, Int. J. Electr. Power Energy Syst. 29 (2007) 290–296. [4] V.N. Gohokar, M.K. Khedkar, Faults locations in automated distribution system, Electr. Power Syst. Res. 75 (2005) 51–55. [5] J. Mora-Florez, J. Melendez, G. Carrillo-Caicedo, Comparison of impedance based fault location methods for power distribution systems, Electr. Power Syst. Res. 78 (2008) 657–666. [6] D. Spoor, J.G. Zhu, Improved single-ended traveling-wave fault-location algorithm based on experience with conventional substation transducers, IEEE Trans. Power Deliv. 21 (2006) 1714–1720. [7] W. Zhao, Y.H. Song, W.R. Chen, Improved GPS traveling wave fault locator for power cables by using wavelet analysis, Int. J. Electr. Power Energy Syst. 23 (2001) 403–411. J. Sadeh, H. Afradi / Electric Power Systems Research 79 (2009) 1538–1545 [8] A. Borghetti, M. Bosetti, M. Di Silvestro, C.A. Nucci, M. Paolone, Continuouswavelet transform for fault location in distribution power networks: definition of mother wavelets inferred from fault originated transients, IEEE Trans. Power Deliv. 23 (2008) 380–388. [9] F.H. Magnago, A. Abur, Fault location using wavelets, IEEE Trans. Power Deliv. 13 (1998) 1475–1480. [10] J.B. Lee, C.W. Ha, C.H. Jung, Development of digital distance relaying algorithm in combined transmission lines with underground power cables, in: IEEE Power Engineering Society Summer Meeting 1, 2001, pp. 611–616. [11] El Sayed Tag El Din, M.M. Abdel Aziz, D.K. Ibrahim, M. Gilany, Fault location scheme for combined overhead line with underground power cable, Electr. Power Syst. Res. 76 (2006) 928–935. [12] M. Gilany, El Sayed Tag El Din, M.M. Abdel Aziz, D.K. Ibrahim, Traveling wave based fault location scheme for aged underground cable combined with overhead line, Int. J. Emerg. Electr. Power Syst. 2 (2005) 1–18. [13] M. Joorabian, S.M.A. Taleghani Asl, R.K. Aggarwal, Accurate fault locator for EHV transmission lines based on radial basis function neural network, Electr. Power Syst. Res. 71 (2004) 195–202. 1545 [14] Zh. Chen, J.C. Maun, Artificial neural network approach to single-ended fault locator for transmission lines, IEEE Trans. Power Syst. 15 (2000) 370–375. [15] J. Gracia, A.J. Mazon, I. Zamora, Best ANN structures for fault location in single- and double-circuit transmission lines, IEEE Trans. Power Syst. 20 (2005) 2389–2395. [16] J. Sadeh, H. Afradi, Fault location scheme for combined transmission lines using artificial neural networks, in: 22nd Int. Power Syst. Conf. (PSC’07), November, Tehran, Iran (in Persian), 2007. [17] C.K. Jung, K.H. Kim, J.B. Lee, B. Klöckl, Wavelet and neuro-fuzzy based fault location for combined transmission systems, Int. J. Electr. Power Energy Syst. 29 (2007) 445–454. [18] J.S. Jang, ANFIS: adaptive network based fuzzy inference system, IEEE Trans. Syst. Man Cyber. 23 (1993) 665–684. [19] T. Takagi, M. Sugeno, Fuzzy identification of systems and its applications to modeling and control, IEEE Trans. Syst. Man Cyber. 15 (1985) 116–132. [20] A.T. Johns, S.K. Salman, Digital Protection for Power Systems, Peter Peregrinus Ltd., IEE Power Series 15, 1995.
© Copyright 2026 Paperzz