Remote monitoring of transmission lines of a power system is significant for improved reliability and stability during fault conditions and protection system breakdowns. This paper proposes a smart backup monitoring system for detecting and classifying the type of transmission line fault occurred in a power grid. In contradiction to conventional methods, transmission line fault occurred at any locality within power grid can be identified and classified using measurements from phasor measurement unit (PMU) at one of the generator buses. This minimal requirement makes the proposed methodology ideal for providing backup protection. Spectral analysis of equivalent power factor angle (EPFA) variation has been adopted for detecting the occurrence of fault that occurred anywhere in the grid. Classification of the type of fault occurred is achieved from the spectral coefficients with the aid of artificial intelligence. The proposed system can considerably assist system protection center (SPC) in fault localization and to restore the line at the earliest. Effectiveness of proposed system has been validated using case studies conducted on standard power system networks. Keywords: Phasor measurement unit (PMU), Backup protection, Fault classification, Support vector machine (SVM), Equivalent power factor angle (EPFA) 1 Introduction Numerous researches have been presented in lit- Real time backup monitoring system for transmission erature for transmission line fault detection and clas- lines is quintessential for stable and reliable operation of sification using PMUs. Major research works are any power grid. Such backup system plays a crucial role enumerated here. Fault classification scheme based during power grid fault conditions and protection sys- on SVMs for preventing incorrect operation of con- tem breakdowns [1–5]. Hence, such methodologies are ventional distance relays was presented in reference gaining much research attraction in recent years [6–11]. . However, the paper does not discuss about dis- Emerging phasor measurement units (PMUs) facilitate crimination of faulty phases. Reference proposed realization of remote monitoring system for transmission protection method based on PMU measurements for lines using global positioning system (GPS) and wide transposed and un-transposed transmission lines. Al- area communication systems. Backup monitoring sys- though the proposed scheme detects the fault, no tems based on PMUs are being developed and tested method was presented to classify the type of fault. across the globe. Major functionalities of such remote Fault detection, classification and location using backup monitoring system are detecting and classify- PMU measurements was presented in reference . ing the transmission line faults occurred in a power But influence of fault resistance (FR) and fault grid [12–15]. inception angle (FIA) on fault classification was not studied. Fault location technique for multi-terminal transmission lines was presented in reference . * Correspondence: firstname.lastname@example.org Department of Electrical and Electronics Engineering, National Institute of The paper did not discuss about fault classification. Technology, Tiruchirappalli, Tamilnadu 620015, India Technical issues associated with fault diagnosis based Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 2 of 10 on the data acquired from intelligent electronic de- vices (IEDs) were discussed in reference . It can be perceived from the literature that majority of the fault detection and classification methods presented are focused on specific transmission line configura- tions. Fault monitoring of whole power grid require execution of these methods for all transmission lines. This may not be feasible in practice or incur in- creased overall cost when employed in as backup protection system. This paper proposes a remote backup fault monitor- ing system for detecting and classifying all types of transmission lines faults occurred anywhere in a power grid. The system can achieve the functionalities using PMU measurements at any one of the generator buses. Frequency domain analysis of equivalent power factor angle (EPFA) is employed for detecting the fault. Classification of the fault is achieved with the aid of artificial intelligence. Proposed methodologies studied and validated on WSCC-9 (Western system coordinating council) bus system and IEEE-39 bus system. Precise detection and classification of fault oc- curred at any locality of a power grid with minimal measurements form the major contribution of this paper. This is a distinctive advantage compared to many of theexistingmethods which havelimited transmission line configurations as mentioned before. The minimal requirements on measurements make Fig. 1 Phasor diagram illustrating the power factor angle the proposed method ideal for providing backup and EPFA monitoring. Moreover, the proposed methodology is e e insensitive to FIA and has minor sensitivity to FR. formulation of d and q axis components of voltage s s The proposed methodology can significantly contrib- and current are as presented in references [20, 21]. ute SPC for accelerated fault localization and restor- The angular separation between equivalent voltage ! ! ation of faulty line. phasor ( V ) and equivalent current phasor ( I ), called s s EPFA (Φ ), can be calculated using eq. (1). eq e e 2 Introduction to EPFA and its dependence on V I qs qs −1 −1 Φ ¼ tan − tan ð1Þ transmission line faults e e V I ds ds Concept of EPFA and its variations during transmis- sion line faults are discussed in this section. Angular EPFA holds a constant value during normal operat- difference between equivalent three-phase voltage ing conditions, as the system frequency and rotating phasor (Vs) and equivalent three-phase current phasor frame frequencies are equal. During fault conditions (Is) is termed as EPFA (Φeq). The concept of EPFA in a transmission line, this relation gets violated and and its measurement are illustrated below with the EPFA undergoes variation, especially when measured help of Fig. 1. at generator terminals. This can be justified as fol- From Fig. 1,the axes d and q represent the sta- lows; during transmission line fault condition, the s s tionary reference frame in which PMU measurements topology of the three-phase system gets altered and ofthreephase voltagephasors (V ,V ,V )and the total reactance seen by the generator changes. R Y B current phasors (I ,I ,I ) are shown. The autono- Transient and sub-transient reactances play their re- R Y B e e mous reference frame d and q rotate at a constant spective roles within the synchronous generator s s speed corresponding to nominal system frequency followed by the contingency. Line current waveform (ω ). During normal operating conditions, both system distortion resulting from fault generates the harmonic frequency (ω)and ω are equal. Mathematical frequencies in the current. As the fault current has to s Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 3 of 10 flow from generator to fault location through the transmission line, harmonic voltages get developed due to line impedances. These voltage and current harmonics along with change in overall system reac- tances trigger variations in the power factor angle at generator terminals (ϕ) and hence in EPFA (Φ ). It eq can be observed from the case studies (as illustrated in the following section of this paper) that frequencies related to EPFA variation have hidden fault informa- tion. These variations are analysed in frequency do- main and the coefficients acquired are utilized in this paper for real time fault monitoring. Fig. 2 Proposed fault classification methodology 3 Proposed fault monitoring methodology based on EPFA The proposed fault monitoring methodology, with which transmission line fault can be detected and classified, is elaborated in this section. The spectral coefficients of Fig. 3 a EPFA variation during no-fault condition in WSCC-9bus system. b. EPFA variation during LG-Fault condition in WSCC-9bus system with FR = 0 Ω and FIA = 0°. c. EPFA variation during LL-Fault condition in WSCC-9bus system with FR = 0 Ω and FIA = 0°. 3 d. EPFA variation during LLG-Fault condition in WSCC-9bus system with FR = 0 Ω and FIA = 0°. e. EPFA variation during LLL-Fault condition in WSCC-9bus system with FR = 0 Ω and FIA = 0°. f. EPFA variation during wide area load change in WSCC-9bus system Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 4 of 10 Fig. 6 FFT spectra of EPFA variation with fault conditions of Fig. 4 FFT spectra of EPFA variation shown in Fig. 3(a) FR = 100 Ω, FIA = 0° in WSCC-9bus system to (f) The transmission line connecting buses 4 and 5 is EPFA (χ ) are extracted through FFT analysis of EPFA taken for case study in this section (this branch is variation as given in eq. (2). equivalent to a doubly fed transmission line). Faults were created one at a time at 20 km from Bus 4. The EPFA of N−1 generator Gen-1 is opted for fault monitoring. During 1 −2πkn χ ¼ Φ ðÞ n e 0 < k < N−1 ð2Þ eq normal operating conditions, the EPFA of Gen-1 is n¼0 shown in Fig. 3(a), and has a constant value. While LG In eq. 6, N stands for number of samples. Online fault fault happened in branch 4–5 (20 km away from bus 4) detection is carried out from these spectral coeffi- at 0.04 s with FR of zero ohms and FIA of zero degrees, cients and classification of the fault is achieved with the EPFA variation is as shown in Fig. 3(b). It can be ob- the aid of nominal voltage coefficients (NVCs) using served that till fault time, EPFA was constant and after multi-class SVM as explained in following section. occurrence of fault it started varying. EPFA variation Spectral analysis of EPFA during various operating during LL, LLG and LLL faults are shown in Fig. 3(c)to conditions are studied through case studies con- (e) respectively. Variations in EPFA during load change ducted on standard WSCC-9 bus system. Functional is also studied and plotted in Fig. 3(f) (EPFA of all buses block diagram of system under study is shown in are shown for illustrating the impact of fault on all Fig. 2. WSCC-9 bus system with three load buses buses. Proposed methodology necessitate EPFA variation modeled as constant power has been taken for case at any one of the generator buses). study . The transmission lines parameters are The spectral analysis of these EPFA variations are con- taken as presented in reference . All three syn- ducted using FFT and the coefficients estimated are chronous generators (G1 to G3) are modeled with shown in Fig. 4. sixth order having steam turbine governor and IEEE To verify the impact of FR and FIA on EPFA variation, type-1 exciter. All generators are assumed to be hav- the same test conditions (Fault created between buses 4 and ing same rating. It is assumed that all computational 5at20kmfrombus4) is takenwithFRof 10 Ω, 100 Ω and and communication facilities are available for the FIA of 0°, 45°, 120° are studied. FFT spectra for these test proposed scheme. conditions are shown in Figs. 5, 6, 7 and 8 respectively. Fig. 5 FFT spectra of EPFA variation with fault conditions of Fig. 7 FFT spectra of EPFA variation with fault conditions of FR = 10 Ω, FIA = 0° in WSCC-9bus system FR = 0 Ω, FIA = 45° in WSCC-9bus system Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 5 of 10 Fig. 8 FFT spectra of EPFA variation with fault conditions of FR = 0 Ω, FIA = 120° in WSCC-9bus system Fig. 9 Schematic diagram of fault classifier module in the proposed backup fault monitoring system Figures 4, 5, 6, 7 and 8 validate that, transmission line faults of any kind can result in variation of EPFA After detecting the fault, the FFT coefficients σ, α, β, and hence reflect it as non-zero coefficients for all γ and δ are transferred to SVM classifier module. frequencies above zero Hertz. For example, in LG The module can classify type of fault from these co- fault condition with FR of 0 Ω and FIA of 0° (Fig. 4), efficients, however, the phase discrimination cannot the frequency coefficients corresponding to 0 Hz, be achieved. This is due to the loss of phase infor- 50 Hz, 100 Hz, 150 Hz and 200 Hz are found to be mation related to fault when converting the three- 0.98, 0.05, 0.14, 0.01 and 0.001 respectively. When phase phasors to synchronous reference frame. the same fault condition occurred with FR of 10 Ω Hence SVM classifier necessitates some auxiliary the FFT spectra of EPFA variation is shown in Fig. 5. parameters for phase discrimination that have the in- It is clear that the coefficients of frequencies 0 Hz, formation regarding the faulty phase, like voltage co- 50 Hz and 100 Hz have undergone slight reduction efficients, current coefficients or power coefficients. to 0.92, 0.04 and 0.13 respectively. All other coeffi- This paper utilizes NVCs (ΓR, ΓY, ΓB) for phase dis- cients have same value. When FR is further increased crimination as given in eq. (3). to 100 Ω, FFT spectra of EPFA variation is shown in Fig. 6. It is evident from the figure that the frequency A=B=C coefficients 0 Hz and 100 Hz are found to have Γ ¼ ð3Þ A=B=C 50 50 50 max η ; η ; η undergone minor reduction to 0.87 and 0.11 respect- A B C ively. When the faults are studied with FIAs of 60° and 120°, no changes in FFT spectra of 0° are ob- Where, η denotes the FFT coefficient of 50 Hz fre- served. Since frequencies of 0 Hz, 50 Hz, 100 Hz, quency component in i-phase voltage. The choice of η 150 Hz and 200 Hz (indicated as σ, α, β, γ and δ re- for estimating NVCs can be validated based on the case spectively)are found tobe dominantinall faultcon- studies conducted on the test system shown in Fig. 2. ditions, they are opted for detection and classification 0 50 100 150 NVCs estimated using η ,η ,η and η for no-fault i i i i of fault occurred. The functional block diagram of the condition and four fault conditions in WSCC system proposed backup fault monitoring methodology is il- shown in Fig. 2 are tabulated in Table 1. A balanced lustrated in Fig. 9. harmonic-free system is considered for the study. During From Fig. 9,PMU measurements from generator no-fault condition NVCs estimated using η will have bus are transferred to fault classifier through proper unity values indicating equal values of fundamental volt- communication channels. Inside fault classifier mod- 0 100 ages in all phases. NVCs estimated using η , η and ule, the acquired phasors are transformed to i i η have zero values, since the system is assumed to be synchronous reference frame using Park’stransform- free of harmonics. All faults are applied with FR of 0 Ω ation and EPFA is estimated, as illustrated in preced- and FIA of 0°, one at a time at 80 Km away from Gen-1 ing section. EPFA over time is recorded in memory in branch 4–5 of WSCC system shown in Fig. 2. During and FFT analysis is carried out. The estimated FFT coefficients σ, α, β, γ and δ are fed to fault detection fault condition, NVCs estimated using η differ in their algorithm. Fault detection algorithm detects the values based on the type of fault and NVCs estimated presence of fault based on the values of α and β. using all others remain zero. Values of these coefficients above zero is the clear Hence SVM classifier uses two inputs viz. frequency indication of fault occurred on a transmission line. coefficients of EPFA variation (σ, α, β, γ and δ) and Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 6 of 10 Table 1 NVCs for phase discrimination in the test system shown in Fig. 1 Type of FFT coefficient No Fault LG Fault (AG) LL Fault (AB) LLG Fault (ABG) LLL Fault (ABC) (Γ ,Γ ,Γ ) (Γ ,Γ ,Γ ) (Γ ,Γ ,Γ ) (Γ ,Γ ,Γ ) (Γ ,Γ ,Γ ) R Y B R Y B R Y B R Y B R Y B 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 η 1, 1, 1 0.6, 1, 1 0.6, 0.6, 1 0.5, 0.5, 1 1, 1, 1 η 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 0, 0, 0 Where, ψ is the kernel parameter that defines nonlin- NVCs estimated using η . Details of the SVM classifier th ear mapping and x stands for i input vector. is described in the following subsection. Frequency coefficients of EPFA variations (σ, α, β, γ and Since the FFT coefficients and NVCs are inter- δ)and NVCs (ΓR, ΓY, ΓB) form the input data set. Fault spersed, a simple logical function may not able to type data corresponding to each set of values in input data accomplish the classification of fault. Hence artificial set constitutes the output data set for SVM classifier. intelligence technique with fine learning capability SVM classifier is trained with 1800 data sets to afford bet- to be adopted. SVM classifiers are becoming a pre- ter boundary for fault classification. The input data sets dominant machine intelligence technique for wide are generated for different fault locations, fault resistances range of applications in power systems [23–25]. and inception angles. Since fault types are of many classes, SVM classifiers are well defined in the literature [23, one-step multi-class SVMs are employed for fault classifi- 24, 26]. A brief outline of SVM classifiers from the cation [23, 24, 26]. Schematic diagram of SVM classifier perspective of fault classification in SPG is illus- for fault classification is shown in Fig. 10. SVM algorithm trated below. used is as presented in references [23, 24, 26]. SVM classifier can be either one-class SVM or The diversity in training data has been maintained by multi-class SVM. The former can classify a single keeping the data for distinct fault locations having differ- type of data from a bunch of data, while the latter ent FRs and FIAs. can classify multiple types of data. Both approaches make a linear decision surface that secede data classes with maximum distance from border line (support 4 Results and discussion vectors). When such linear decision surface cannot be Simulation studies and the results observed in proposed achieved, then SVM draws the data to a higher di- backup fault monitoring scheme are discussed in this mensional feature space using kernel function so that section. IEEE-39 bus system (400 kV) is opted for case a separating linear decision surface exists can be de- study, as shown in Fig. 11 . All ten conventional gen- termined [23, 24, 26]. erators (G1 to G10) generators and transmission lines SVM finds the optimal separating hyperplane from are modelled in the similar way as in case-I. EPFA of all the training data set near to the border line, helps to generators during normal operating conditions are generalize the data for unseen cases [23, 24, 26]. shown in Fig. 12(a). The constant value of EPFA during LIBSVM toolbox in MATLAB/SIMULINK environ- ment with radial basis function (RBF) kernel is de- ployed in this research work. The kernel function is defined as follows; Kx ; x ¼ exp −ψ x −x i j i j Fig. 10 Functional block diagram of SVM classifier for fault classification Fig. 11 Single line diagram of IEEE-39 bus system Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 7 of 10 Fig. 12 a. EPFA variation during no-fault condition in IEEE-39 bus system. b. EPFA variation during LG-fault (AG) condition at branch 3–18, 50 km away from bus-3 with FR = 0 Ω and FIA = 0° in IEEE-39 bus system. c. EPFA variation during LL-fault (BC) condition at branch 4–5 20 km away from bus-4, with FR = 10 Ω and FIA = 45° in IEEE-39 bus system. d. EPFA variation during LLG-fault (CAG) condition at branch 23–24, 70 km away from bus-23 with FR = 100 Ω and FIA = 120° in IEEE-39 bus system. 12 e. EPFA variation during LLL-fault (ABC) condition at branch 5–6, 50 km away from bus-5 with FR = 0 Ω and FIA = 0° in IEEE-39 bus system Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 8 of 10 Table 2 FFT Frequency Coefficients under Various Operating conditions in IEEE-39 Bus System Calculated from PMU at Bus-1 Fault Location (Branch) Fault Type Fault Condition (FR, FIA) σα β γ δΓ , Γ , Γ ANN Classifier ANFIS Classifier SVM Classifier R Y B 3–18, 50 km from bus-3 AG 0 Ω, 0° 0.599 0.042 0.140 0.018 0.009 0.55, 1, 0.99 CA AG AG 4–15, 20 km from bus-4 BG 10 Ω, 45° 0.597 0.039 0.136 0.016 0.008 0.99, 0.76, 1 AB BG BG 22–23, 50 km from bus-22 CG 100 Ω, 120° 0.588 0.032 0.132 0.012 0.007 1, 0.99, 0.84 BC CG CG 9–8, 70 km from bus-9 AB 0 Ω, 0° 1.253 0.294 0.275 0.161 0.062 0.66, 0.67, 1 AB AB AB 4–5, 20 km from bus-4 BC 10 Ω, 45° 1.250 0.291 0.271 0.157 0.058 1, 0.74, 0.74 BC BC BC 19–33, 10 km from bus-19 CA 100 Ω, 120° 1.244 0.285 0.267 0.148 0.052 0.86, 1, 0.87 CA CA CA 2–3, 60 km from bus-2 ABG 0 Ω, 0° 1.385 0.270 0.230 0.132 0.047 0.59, 0.60, 1 AB AG ABG 17–27, 40 km from bus-17 BCG 10 Ω, 45° 1.382 0.266 0.227 0.129 0.046 1, 0.77, 0.77 BC BG BCG 23–24, 70 km from bus-23 CAG 100 Ω, 120° 1.378 0.260 0.220 0.121 0.038 0.87, 1, 0.87 AC CG CAG 2–30, 20 km from bus-2 ABC 0 Ω, 0° 1.498 0.219 0.055 0.027 0.019 1, 1, 1 ABC ABC ABC 5–6, 50 km from bus-5 ABC 10 Ω, 45° 1.495 0.215 0.054 0.025 0.018 1, 1, 1 ABC ABC ABC 21–22, 60 km from bus-21 ABC 100 Ω, 120° 1.489 0.210 0.051 0.021 0.016 1, 1, 1 ABC ABC ABC No Fault 0.533 00001,1,1 No Fault No Fault No Fault Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 9 of 10 this condition is reflected as value of 0.533 for σ and a From Table 2, ANN can precisely classify LL faults value of zero for all other frequency coefficients as given and LLL faults at random locations of the system. in Table 2. EPFA variations corresponding to various However, it has misclassified all LG and LLG faults. fault conditions applied at random locations of the While, ANFIS classifier classified LG, LL and LLL system are shown in Fig. 12(b)-(e) (EPFA of all gener- faults but classification result went wrong on LLG ators are shown in figure for validating the impact of faults. SVM has classified all fault conditions and fault on all generators. Proposed methodology require hence it can be presumed that SVM classifier is su- EPFA of any one of the generators alone). Frequency perior to other two for proposed fault classification coefficients of EPFA variations in fault conditions are methodology. listed in Table 2. For example, EPFA variations of all generators while LG-fault (AG) on transmission line 5 Conclusion connecting buses 3 and 18 at 50 km away from bus-3 Backup protection of transmission lines is of elevated with FR of 0 Ω and FIA of 0° is shown in Fig. 12(b). significance in power grid operation and control. Al- Time of occurrence of fault is 0.04 s, and hence, up though advanced numerical relays are available for to 0.04 s EPFAs of all generators have constant value transmission line protection, backup protection is im- and starts varying afterwards. The frequency coeffi- portant to provide adequate system performance during cients σ, α, β, γ and δ have the values of 0.599, 0. protection system breakdowns and circuit breaker fail- 042, 0.140, 0.018, 0.009 as given in Table 2. Similarly, ures. This paper proposes a fault monitoring method- EPFA variation during LL, LLG and LLL are shown ology can be adopted as backup monitoring in power in Fig. 12(c)-(e) and corresponding frequency spectral grids. The proposed system can detect and classify trans- coefficients are listed in Table 2.Itisevidentfrom mission line fault occurred anywhere in the power grid the Figs. 12(a)-(e) and Table 2 that EPFA suffers vari- with the aid of PMUs. Since any one of the generator ation during faults and the corresponding frequency bus PMU measurements are sufficient, the methodology components depend on type of fault occurred. It can is ideal for providing backup protection. This is unique be observed that, for fault occurred far away point advantage compared to conventional methods which are from the generator causes frequency coefficients in devised for specific transmission line configurations like EPFA variation. However, the magnitudes of the EPFA transmission line fed at single end or double ends. frequency components vary with fault location and Authors’ contributions this demands the necessity for artificial intelligence to GP has developed and implemented the proposed algorithm. MB has made classify the fault type. The frequency components and substantial contributions to simulate WSCC-9 bus system. MJBR has been the corresponding NVCs (as given in Table 2)are given technical adviser for the total work and DKM has supported us in interpreting the simulation results for fault analysis using PMU measurements. All authors to SVM module and the results of the fault classifica- read and approved the final manuscript. tion are also given in Table 2. As a comparative study with others classifiers, result obtained with ANFIS Competing interests and ANN classifiers are described in the succeeding The authors declare that they have no competing interests. subsection. Author details Department of Electrical Engineering, National Institute of Technology, Calicut, Kerala, India. Department of Electrical and Electronics Engineering, National Institute of Technology, Tiruchirappalli, Tamilnadu 620015, India. 4.1 Comparative study with other machine intelligence Department of Electrical and Electronics Engineering, BITS Mesra, Ranchi, techniques India. Study on performance accuracy of SVM classifier com- Received: 4 September 2017 Accepted: 8 May 2018 pared to artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) are illustrated in this section. The same test case of IEEE-39 bus system is References considered for comparative study. ANN has been devel- 1. Mallikarjuna, B., Vardhan Varma, P. V., Samir, S. D., Jaya Bharata Reddy, M., & Mohanta, D. K. (2017). An adaptive supervised wide-area backup protection oped with feed-forward back propagation method con- scheme for transmission lines protection. Protection and Control of Modern sisting of two layers of input and output with two Power Systems, 22(2), 1–16. hidden intermediate layers . While ANFIS based 2. Wang, J., Zhong, H., Xia, Q., et al. (2015). Transmission network expansion planning with embedded constraints of short circuit currents and N-1 fault classifier was developed using hybrid optimization security. Journal of Modern Power Systems and Clean Energy, 3(3), 312–320. method with 600 membership functions . The input- 3. Jamil, M., Sharma, S. K., & Singh, R. (2015). Fault detection and classification output data set is made common for ANN, ANFIS and in electrical power transmission system using artificial neural network. SpringerPlus, 334(4), 1–13. SVM classifiers. The results after case studies are tabu- 4. Majd, A. A., Samet, H., & Ghanbari, T. (2017). k-NN based fault detection and lated in Table 2, which validate the superiority of SVM classification methods for power transmission systems. Protection and classifiers over the others. Control of Modern Power Systems, 2, 32,1–11. Gopakumar et al. Protection and Control of Modern Power Systems (2018) 3:16 Page 10 of 10 5. Roostaee, S., Mini, S. T., & Mehfuz, S. (2017). Experimental studies on impedance based fault location for long transmission lines. Protection and Control of Modern Power Systems, 2:16,1–9. 6. Slootweg, J. G., & Kling, W. L. (2002). Impacts of distributed generation on power system transient stability. Proc IEEE Power Eng Soc Summer Meet, 2, 862–867. 7. Cecati, C., Citro, C., & Siano, P. (2011). Combined operations of renewable energy systems and responsive demand in a smart grid. IEEE Transactions on Sustainable Energy, 2(4), 468–476. 8. Gopakumar, P., Jaya bharata Reddy, M., & Mohanta, D. K. (2014). Letter to the editor: Stability concerns in smart grid with emerging renewable energy technologies. Electric Power Components and Systems, 42(3), 418–425. 9. Ma, J., Zhang, P., Fu, H., Bo, B., Dong, Z. (2010). Application of phasor measurement unit on locating disturbance source for low-frequency oscillation. IEEE Transactions on Smart Grid, 3(1), 340–346. 10. Seethalekshmi, K., Singh, S. N., & Srivastava, S. C. (2011). A Synchrophasor assisted frequency and voltage stability based load shedding scheme for self-healing of power system. IEEE Transactions on Smart Grid, 2(2), 221–230. 11. Hashiesh, F., Mostafa, H. E., Khatib, A., Helal, I., & Mansour, M. M. (2012). An intelligent wide area Synchrophasor based system for predicting and mitigating transient instabilities. IEEE Transactions on Smart Grid, 3(2), 645–652. 12. Phadke, A. G., Thorp, J. S., & Adamiak, M. G. (1983). A new measurement technique for tracking voltage phasors, local system frequency, and rate of change of frequency. IEEE Transactions on Power Apparatus and Systems, 102(5), 1025–1038. 13. Phadke, G., & Thorp, J. S. (2008). Synchronized Phasor Measurements and Their Applications. Power electronics and power systems series. New York: Springer Publication. 14. NERC Technical Reference Document (2010) Transmission system backup protection systems, Dec 2010. 15. Seethalekshmi, K., Singh, S. N., & Srivastava, S. C. (2012). A classification approach using support vector machines to prevent distance relay Maloperation under power swing and voltage instability. IEEE Transactions on Power Delivery, 27(3), 1124–1133. 16. Shan, C., Chih-Wen, L., & Joe-Air, J. (2002). A new adaptive PMU based protection scheme for transposed/Untransposed parallel transmission lines. IEEE Transactions Power Delivery, 17(2), 902–907. 17. Joe-Air, J., Ching-Shan, C., & Chih-Wen, L. (2003). A new protection scheme for fault detection, direction discrimination, classification, and location in transmission lines. IEEE Transactions Power Delivery, 18(1), 34–42. 18. Chih-Wen, L., Kai-Ping, L., Ching-Shan, C., & Joe-Air, J. (2008). A universal fault location technique for N-terminal (N≥3) transmission lines. IEEE Transactions Power Delivery, 23(3), 1366–1373. 19. Kezunovic, M. (2011). Smart fault location for smart grids. IEEE Transactions on Smart Grid, 2(1), 11–22. 20. Grainger, J. J., & Stevenson, W. D. (2003). Power system analysis. New Delhi: Tata McGraw-Hill Edition. 21. Kundur, P. (2012). Power system stability and control. New Delhi: Tata McGraw Hill Education Pvt Ltd. 22. Gopakumar, P., Jaya Bharata Reddy, M., & Mohanta, D. K. (2015). Transmission line fault detection and localisation methodology using phasor measurement unit measurements. IET Generation Transmission and Distribution, 11(9), 1033–1042. 23. Ravikumar, B., Thukaram, D., & Khincha, H. P. (2010). Comparison of multiclass SVM classification methods to use in a supportive system for distance relay coordination. IEEE Transactions on Power Delivery, 25(3), 1296–1305. 24. Kalyani, S., & Shanti Swarup, K. (2011). Classification and assessment of power system security using multiclass SVM. IEEE transactions on Systems, Man, and Cybernetics—Part C: Applications and Reviews, 41(5), 753–758. 25. Chang, C. C., & Lin, C. J. (2011). LIBSVM: a library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2,27:1–27:27. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm. 26. Vapnik, V. (1998). Statistical learning theory. New York: Wiley. 27. Seyedtabaii, S. (2012). Improvement in the performance of neural network- based power transmission line fault classifiers. IET Generation Transmission and Distribution, 6(3), 731–737. 28. Jaya Bharata Reddy, M., & Mohanta, D. K. (2008). Adaptive-neuro-fuzzy inference system approach for transmission line fault classification and location incorporating effects of power swings. IET Generation Transmission and Distribution, 2(2), 235–244.
Protection and Control of Modern Power Systems – Springer Journals
Published: Jun 5, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud