Intelligent Energy Management Technologies: ICAEM 2019 (Algorithms for Intelligent Systems)

38 Detection and Classification of Transmission Line … 457 Fig. 3 Detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribution function results for line to ground fault a current waveform b ST-index c H-index d WDF-index e proposed fault index 3.3 Double Line Fault and Double Line to Ground Fault (LLG)

Fig. 4 Detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribution 458 T. Bhati and H. K. Khyani function results for double line fault and double line to ground fault a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index

38 Detection and Classification of Transmission Line … 459 3.4 Three Phase to Ground Fault Fig. 5 Detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribution function results for three phase to ground fault a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index

460 T. Bhati and H. K. Khyani Fig. 6 Effect of fault location on detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribution function results for line to ground fault at 25% of line length a current waveform, b ST-index c H-index, d WDF-index, e proposed fault index 3.5 Effect of Fault Location To generalize the proposed algorithm, it is tested at different locations of fault on the transmission line. All the faults have been tested at 25 and 75% of line length from the sending end of the line. However, the results of LG fault are shown as 25 and 75% of line length in the following sections. The results have been discussed for all the phases at 50% of line length in the above sections. 3.5.1 Line to Ground Fault at 25% Line Length This can be observed from Fig. 6a that current in Phase-A increases after the faulty event. It is observed from Fig. 6b that the proposed ST-index has zero values throughout the range corresponding to Phases-B and C whereas these values corre- sponding to Phase-A are high indicating that there is a fault on this phase. Figure 6c indicates that the value of proposed H-index is constant over the entire time range for the Phases-B and C whereas it has high values for the Phase-A indicating the faulty

38 Detection and Classification of Transmission Line … 461 phase. Figure 6d indicates that the value of proposed WDF-index has very high values corresponding to Phase-A near the central location of time whereas these values are low for Phases-B and C. It is depicted from Fig. 6e that proposed fault index corre- sponding to Phase-A has higher values above the threshold indicating the presence of fault. The values of fault index corresponding to Phases-B and C are below the threshold and approximately equal to zero indicating the healthy phases. Hence, LG fault on Phase-A at a line length of 25% from sending end of the transmission line has been successfully detected using the proposed algorithm based on Stockwell Transform, Hilbert Transform, and Wigner Distribution Function. 3.5.2 Line to Ground Fault at 75% Line Length This can be observed from Fig. 7a that current in Phase-A increases after the faulty event. It is observed from Fig. 7b that the proposed ST-index has zero values throughout the range corresponding to Phases-B and C whereas these values corre- sponding to Phase-A are high indicating that there is a fault on this phase. Figure 7c Fig. 7 Effect of fault location on detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform and Wigner distribution function results for line to ground fault at 75% of line length a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index

462 T. Bhati and H. K. Khyani indicates that the value of proposed H-index is constant over the entire time range for Phases-B and C whereas it has high values for Phase-A indicating the faulty phase. Figure 7d indicates that the value of proposed WDF-index has very high values corresponding to Phase-A near the central location of time whereas these values are low for Phases-B and C. It is depicted from Fig. 7e that proposed fault index corre- sponding to Phase-A has higher values above the threshold indicating the presence of fault. The values of fault index corresponding to Phases-B and C are below the threshold and approximately equal to zero indicating the healthy phases. Hence, LG fault on Phase-A at a line length of 75% from sending end of the transmission line has been successfully detected using the proposed algorithm based on Stockwell Transform, Hilbert Transform, and Wigner Distribution Function. 3.6 Effect of Reverse Power Flow This can be observed from Fig. 8a that current in Phase-A increases after the faulty event. It is observed from Fig. 8b that the proposed ST-index has zero values throughout the range corresponding to Phases-B and C whereas these values corre- sponding to Phase-A are high indicating that there is a fault on this phase. Figure 8c indicates that the value of proposed H-index is constant over the entire time range for Phases-B and C whereas it has high values for Phase-A indicating the faulty phase. Figure 8d indicates that the value of proposed WDF-index has very high values corresponding to Phase-A near the central location of time whereas these values are low for Phases-B and C. It is depicted from Fig. 8e that proposed fault index corre- sponding to Phase-A has higher values above the threshold indicating the presence of a fault. The values of the fault index corresponding to phases-B and C are below the threshold and approximately equal to zero indicating the healthy phases. Hence, LG fault on Phase-A at line length of 50% from sending end of the transmission line has been successfully detected using the proposed algorithm based on Stockwell Trans- form, Hilbert Transform, and Wigner Distribution Function during the condition of reverse power flow on the transmission line. 3.7 Effect of Switching Transients To generalize the proposed algorithm, it is tested for the switching transients. The results related to the switching of inductive load and capacitive load have been discussed in the following sections.

38 Detection and Classification of Transmission Line … 463 Fig. 8 Effect of reverse power flow on detection of transmission line faults using hybrid algorithm based on combined features of Stockwell transform, Hilbert transform and Wigner distribution function results for line to ground fault at 50% of line length a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index 3.7.1 Switching of Capacitive Load This can be observed from Fig. 9a that current contains switching transients due to switching the capacitive load. It is observed from Fig. 9b that the proposed ST- index has zero values before switching on the capacitive load and finite values at the time of switching the capacitor. Figure 9c indicates that the value of the proposed H-index is constant before switching on the capacitive load and finite values at the time of switching the capacitor. Figure 9d indicates that the value of proposed WDF- index has high values corresponding to all phases near the central location of time. It is depicted from Fig. 9e that proposed fault index corresponding to all phases has higher values but below the threshold indicating that the event is switching and not the faulty. Hence, switching of the capacitive load has been differentiated from the faulty events using the proposed algorithm based on Stockwell Transform, Hilbert Transform, and Wigner Distribution Function.

464 T. Bhati and H. K. Khyani Fig. 9 Simulation results using hybrid algorithm based on combined features of Stockwell trans- form, Hilbert transform, and Wigner distribution function for the effect of capacitive load switching a current waveform, b ST-index, c H-index, d WDF-index, e proposed fault index 3.8 Switching of Inductive Load This can be observed from Fig. 10a that current in all phases increases due to increased load after switching the inductive load. It is observed from Fig. 10b that the proposed ST-index has zero values before switching on the inductive load and finite values at the time of switching the load. Figure 10c indicates that the value of the proposed H-index is constant before switching on the inductive load and finite values at the time of switching the load. Figure 10d indicates that the value of the proposed WDF- index has high values corresponding to all phases near the central location of time. It is depicted from Fig. 10e that proposed fault index corresponding to all phases has higher values but below the threshold indicating that the event is switching and not the faulty. Hence, switching of the inductive load has been differentiated from the faulty events using the proposed algorithm based on Stockwell Transform, Hilbert Transform, and Wigner Distribution Function.

38 Detection and Classification of Transmission Line … 465 Fig. 10 Simulation results using hybrid algorithm based on combined features of Stockwell trans- form, Hilbert Transform and Wigner Distribution Function for the effect of inductive load a current waveform b ST-index c H-index d WDF-index e proposed fault index 3.9 Classification of Faults The peak values of the proposed fault index based on the current-based features are provided in Table 2. It can be observed that if values of fault index are below 5000 for all the phases then there is no fault on the system and this indicates the healthy condition. If there are only values of fault index is above 5000 corresponding to one Table 2 Peak values of current-based fault index Peak values of fault index Sr. No. Type of fault Phase-A Phase-B Phase-C 1 Healthy condition 2 Line to ground fault 00 0 3 Double line fault 100 4 Double line to ground fault 8000 100 100 5 Three-phase fault involving ground 4 × 104 3 × 104 100 2.5 × 104 3.5 × 104 4 × 104 4.5 × 104 3 × 104

466 T. Bhati and H. K. Khyani phase, then this indicates the presence of LG fault. If the fault index corresponding to two phases is above the threshold then this indicates the presence of LL or LLG fault. These can be discriminated from each other by using certain ground fault index. If the fault index corresponding to all the three phases is above the threshold then this indicates the presence of LLLG fault. 4 Conclusions In this paper, an accurate protection system against various types of power system faults like line to ground, double line, double line to ground, and three-phase faults including ground three phase to ground have been developed. An algorithm based on combined features of Stockwell transform, Hilbert transform, and Wigner distribu- tion function has been proposed for the detection and classification of transmission line faults to design effective protection scheme. The method is tested on a test transmission line for different cases which includes change in line length, reverse power flow condition and switching tran- sients using MATLAB/Simulink. The advantage of this method is its effectiveness for different cases where many traditional algorithms have failed. From the simula- tion results, it has been found that the normal value of fault index in healthy condition stays zero. In case of faulty event the value of fault index varies between 6.121e+03 as minimum value and 45.36e+03 as maximum value in case of current-based analysis when threshold is 5000. In case of voltage-based analysis, the value of fault index varies between 1.290e+03 as minimum value and 9.101e+3 as maximum value when threshold is 1000. Hence, there exist an accurate difference between a healthy and a faulty condition. References 1. Yang L, Xiao H, Zidong W, Dong-Hua X (2015) Fault Detection and Diagnosis for a Class of Nonlinear Systems with Decentralized Event-triggered Transmissions. IFAC-Pap Line (Elsevier) 48–21:1134–1139 2. Durga Prasad Ch, Srinivasu N (2015) Fault detection in transmission lines using instantaneous power with ED based fault index. Proceedia Technol 21:132–138 3. Ferreira VH, Zanghi R, Fortes MZ, Sotelo GG, Silva RBM, Souza JCS, Guimarães CHC, Gomes Jr.Fsjjglds S (2016) A survey on intelligent system application to fault diagnosis in electric power system transmission lines Elect. Power Syst Res 136:135–153 4. Hussain S, Osman AH (2016) Fault location scheme for multi-terminal transmission lines using unsynchronized measurements. Int J Elect Power Energy Syst 78:277–284 5. Gafoor Shaik A, Pulipaka RRV (2015) A new wavelet based fault detection, classification and location in transmission lines. Int J Elect Power Energy Syst 64:35–40 6. Dehghani M, Khooban MH, Niknam T (2016) Fast fault detection and classification based on a combination of wavelet singular entropy theory and fuzzy logic in distribution lines in the presence of distributed generations. Int J Elect Power Energy Syst 78:455–462

38 Detection and Classification of Transmission Line … 467 7. Morais AP, Cardoso Júnior G, Mariotto L, Marchesan G (2016) Fault location scheme for multi- terminal transmission lines using unsynchronized measurements. Int. J. Elect. Power Energy Syst 78:277–284 8. Swetapadma A, Yadav A (2016) Directional relaying using support vector machine for double circuit transmission lines including cross-country and inter-circuit faults. Int J Electr Power Energy Syst 81:254–264

Author Index A G Agarwal, Kusum Lata, 251, 335, 343, 357 Gehlot, Kishore Singh, 409 Agarwal, Nitesh, 227 Gupta, Utkarsh, 1, 139 Akter, Morium, 95 Ali, Md. Hayder, 53, 65 J Ali, Mohammad Hanif, 53, 65 Jangid, Aisha, 159 Aseri, Ramesh, 425 Jha, Aashish Kumar, 111 Jindal, Aditya, 335 B Joshi, Ashwani Kumar, 425 Bhandari, Sanjay, 191 Joshi, Yogesh, 381 Bhansali, Pratik, 303 Bhati, Tanmay, 449 K Kanwar, Khamma, 285 C Kapoor, Gaurav, 73, 119 Chaudhary, Laxmi, 159 Karia, Megha C., 25 Choudhary, Arjun, 9 Khyani, Harish, 295 Choudhary, Ranjana, 397 Khyani, Harish Kumar, 199, 219, 449 Kulshrestha, Abhijit, 171 D Kumar, Amit, 181 Dave, Sushma, 101 Kumar, Nitesh, 181 Dhanraj, 251 Kushwaha, Ajay, 267 Dhariwal, Saraswati Chand, 209 L Lehri, Divam, 9 E M Eva, Mahmuda Najnin, 95 Mangal, Rajat, 209 Mathur, Akleshwar, 219 F Mathur, Akshat, 101 Fahmida Islam, Sk., 95 Mathur, Ashish, 171 Ferdous, Jannatul, 95 Mathur, Geetika, 171 Mehta, Sandip, 303 © Springer Nature Singapore Pte Ltd. 2021 469 M. Shorif Uddin et al. (eds.), Intelligent Energy Management Technologies, Algorithms for Intelligent Systems, https://doi.org/10.1007/978-981-15-8820-4

470 Author Index Mehta, Vinit, 319, 381 Sharma, Santosh Kumar, 139 Mohanty, A. M., 181 Sharma, Shubham, 343 Sharma, Yatendra Mohan, 133 N Shorif, Sumaita Binte, 95 Namjoshi, Sadanand A., 45 Shringi, Surbhi, 139 Nehiwal, Jayesh, 295 Singhal, Amit, 85, 149 Singh, Bablu Kumar, 191, 277 P Singh, Chandershekhar, 267, 295 Panchauli, Dheeraj, 1 Singh, Samit Kumar, 45 Pandya, Bhavik J., 25 Singh, Sudeep Kumar, 181 Panwar, Kapil, 335 Swain, Suvam Sourav, 181 Patel, Shrawan Ram, 397 Patra, Prashanjeet, 181 T Purohit, Rajendra, 227 Tandon, Ankit, 85, 149 R U Ram Patel, Shrawan, 295 Uddin, Mohammad Shorif, 95 Ram, Shrawan, 409 V S Vajpai, Jayashri, 199, 285, 319 Saini, Pawan Kumar, 133 Vishnoi, Shyam Lal, 357 Sangani, Kamlesh B., 25 Vyas, Supriya, 277 Shalini, 133 Sharma, Komal, 159 Y Sharma, Nidhi, 111 Yadav, D. K., 1