• λŒ€ν•œμ „κΈ°ν•™νšŒ
Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers
  • COPE
  • kcse
  • ν•œκ΅­κ³Όν•™κΈ°μˆ λ‹¨μ²΄μ΄μ—°ν•©νšŒ
  • ν•œκ΅­ν•™μˆ μ§€μΈμš©μƒ‰μΈ
  • Scopus
  • crossref
  • orcid

  1. (Electrification System Research Division, Korea Railroad Research Institute, Korea)



High-resistance ground fault, Deep learning, CNN, Distribution system, Protection system

1. μ„œ λ‘ 

배전계톡 λ‚΄ κ³ μ €ν•­ 사고(High Resistance Fault, HRF)λŠ” 배전선이 μ–΄λ– ν•œ 원인 λ•Œλ¬Έμ— 자갈, λͺ¨λž˜, 수λͺ© λ“± 저항이 큰 λ¬Όμ§ˆμ— μ ‘μ΄‰ν•˜μ—¬ λ°œμƒν•˜λŠ” 사고이며, 단선지락사고가 배전계톡 μ‚¬κ³ μ˜ 70% 이상을 μ°¨μ§€ν•œλ‹€. κ³„ν†΅μ˜ 정상적 μš΄μ˜μ„ μœ„ν•˜μ—¬ κ³ μ €ν•­ 사고도 λ‹€λ₯Έ μ‚¬κ³ μ²˜λŸΌ μ‹ λ’°μ„± μžˆλŠ” κ²€μΆœ 및 μ œκ±°κ°€ ν•„μš”ν•˜λ‚˜, κ³ μ €ν•­ μ‚¬κ³ λŠ” 사고전λ₯˜ μ¦κ°€λŸ‰μ΄ λΆ€ν•˜ μ „λ₯˜μ— λΉ„ν•΄ 크지 μ•ŠμœΌλ©°, μ•„ν¬λ‘œ λΆ€ν•˜μ™€ μœ μ‚¬ν•œ νŠΉμ„±μ„ λ³΄μ—¬μ„œ 사고 νŒμ •μ΄ μ–΄λ ΅λ‹€(1). 이와 같은 이유둜 일반적인 κ³Όμ „λ₯˜ 계전기 λ“±μœΌλ‘œ 사고 κ²€μΆœν•˜λŠ” 것이 μ–΄λ €μš°λ©°, κ²€μΆœμ„ μœ„ν•΄ κ³ μ €ν•­ μ‚¬κ³ μ˜ λΉ„μ„ ν˜•, μ‹œλ³€μ μΈ νŠΉμ„±μ„ ν™œμš©ν•œ 기법듀이 μ—°κ΅¬λ˜μ–΄μ™”λ‹€. 특히 배전망에 λΆ„μ‚°μ „μ›μ˜ 적용이 λŠ˜μ–΄λ‚˜κ³  μžˆλŠ”λ°, 뢄산전원이 μžˆλŠ” λ°°μ „κ³„ν†΅μ—μ„œ κ³ μž₯ λ°œμƒμ΄ λ°œμƒν•˜λ©΄ κ³ μž₯ μœ„μΉ˜μ™€ 뢄산전원 μœ„μΉ˜μ˜ 관계에 따라 μ „λ₯˜ λ°©ν–₯ λ“± κ³ μž₯μ „λ₯˜μ˜ 양상이 달라진닀(2).

배전계톡 ν”Όλ”μ˜ κ³ μž₯은 μ •ν™•ν•˜κ³  μ‹ μ†ν•˜κ²Œ κ°μ§€λ˜κ³  μ œκ±°λ˜μ–΄μ•Ό κ³ μž₯의 확산을 방지할 수 μžˆλ‹€(3,4). 단선지락사고에 λŒ€ν•œ μ—¬λŸ¬ 가지 λ³΄ν˜ΈκΈ°λ²•λ“€μ΄ μ œμ•ˆλ˜μ–΄ μ™”μœΌλ©°(4), 크게 μ„Έ κ°€μ§€λ‘œ λΆ„λ₯˜ν•  수 μžˆλŠ”λ° 1) μ •μƒμƒνƒœ μ‹ ν˜Έ 기반 뢄석, 2) κ³Όλ„μƒνƒœ μ‹ ν˜Έ 기반 뢄석, 3) ITμœ΅ν•© 기술 기반 뢄석이 μžˆλ‹€. 이 쀑 μ•žμ˜ 두 가지 방법은 계톡 κ΅¬μ‘°λ‚˜ 사고가 λ³΅μž‘ν•œ 상황에 λŒ€ν•΄ ν•œκ³„κ°€ μžˆλ‹€. IT기술이 λ°œμ „ν•¨μ— 따라 λ‰΄λŸ΄ λ„€νŠΈμ›Œν¬(5), μ„œν¬νŠΈ 벑터 λ¨Έμ‹ (6), μœ μ „μ•Œκ³ λ¦¬μ¦˜, μ „λ¬Έκ°€ μ‹œμŠ€ν…œ, 퍼지 이둠(7) 등이 μ μš©λ˜μ–΄ μ™”λ‹€. Farshad와 Sadeh(8)λŠ” k-μ΅œκ·Όμ ‘ μ•Œκ³ λ¦¬μ¦˜μ„ μ΄μš©ν•œ νšŒκ·€λΆ„μ„μ„ μ μš©ν•˜μ—¬ 단선 μ§€λ½μ‚¬κ³ μ˜ μœ„μΉ˜λ₯Ό νŒμ •ν•˜μ˜€λ‹€. λ…Όλ¬Έ (9)μ—μ„œλŠ” 웨이블릿 λ³€ν™˜κ³Ό λ² μ΄μ‹œμ•ˆ 기법을 μ΄μš©ν•˜μ—¬ κ³ μž₯ 피더λ₯Ό νŒλ³„ν•˜μ˜€κ³ , (10)μ—μ„œλŠ” 인곡신경망(Artificial Neural Network, ANN)을 μ΄μš©ν•˜μ—¬ 사고 피더와 건전 피더λ₯Ό νŒλ³„ν•˜μ˜€λ‹€. ν•˜μ§€λ§Œ μ΄λŸ¬ν•œ 방법은 κ³ μž₯으둜 λ°œμƒν•˜λŠ” μ‹ ν˜Έμ— λŒ€ν•œ μ μ ˆν•œ μ‹ ν˜Έμ²˜λ¦¬ 방법이 ν•„μš”ν•˜λ‹€. μ΅œκ·Όμ—λŠ” 웨이블릿 λ³€ν™˜κ³Ό λ¨Έμ‹ λŸ¬λ‹μ„ ν†΅ν•œ νŒ¨ν„΄ λΆ„λ₯˜λ₯Ό κ²°ν•©ν•œ 방법이 μ œμ•ˆλ˜κ³  μžˆλ‹€. (11)μ—μ„œλŠ” 이산 웨이블릿 λ³€ν™˜κ³Ό λΉ„μŒμˆ˜ ν–‰λ ¬ λΆ„ν•΄λ₯Ό μŒμ„± 뢄석에 μ μš©ν•˜μ˜€κ³ , (12)μ—μ„œλŠ” 지ν–₯μ„± 웨이블릿 λ³€ν™˜μ„ μ΄μš©ν•œ 심측 CNN(Convolutional Neural Network, ν•©μ„±κ³± 신경망)에 기반 μ €μ„ λŸ‰ Xμ„  컴퓨터 λ‹¨μΈ΅μ΄¬μ˜(CT)을 μ œμ•ˆν•˜μ˜€λ‹€. μ΄λŸ¬ν•œ 웨이블릿 λ³€ν™˜κ³Ό CNN을 μ μš©ν•˜μ—¬ λ°°μ „λ§μ—μ„œ κ³ μž₯ 피더 νŒλ³„μ„ 톡해 κ³ μž₯ μœ„μΉ˜λ₯Ό νŒλ³„ν•˜κ±°λ‚˜(13), 힐베λ₯΄νŠΈ-ν™© λ³€ν™˜(Hilbert-Huang Transform)κ³Ό CNN을 μ μš©ν•˜μ—¬ 배전망 사고 μ’…λ₯˜λ₯Ό νŒλ³„ν•˜λŠ” 방법이 μ œμ•ˆλ˜μ—ˆλ‹€(14).

λ³Έ λ…Όλ¬Έμ—μ„œλŠ” 뢄산전원이 μ„€μΉ˜λœ λ°°μ „κ³„ν†΅μ—μ„œ 고저항단선지락사고가 λ°œμƒν•˜μ˜€μ„ λ•Œ 그것을 νŒλ³„ν•˜λŠ” 방법을 μ œμ•ˆν•˜κ³ μž ν•œλ‹€. μ‹ μž¬μƒλ°œμ „ λ“± 뢄산전원 μ¦κ°€λ‚˜ μ „λ ₯λ³€ν™˜μž₯치 λ„μž… λ“±μœΌλ‘œ 인해 배전계톡이 더 λ³΅μž‘ν•΄μ Έ 사고 λ°œμƒμ‹œ κ³ μž₯μ „λ₯˜μ˜ 양상이 λ³΅μž‘ν•΄μ§€κ³  있으며, 이에 따라 κ³ μž₯μ‹ ν˜Έ μ²˜λ¦¬μ— μ μ ˆν•œ 방법을 μ°ΎκΈ°κ°€ μ–΄λ €μ›Œμ§€κ³  μžˆλ‹€. λ°°μ „κ³„ν†΅μ—μ„œ 자주 λ°œμƒν•˜λŠ” κ³ μ €ν•­ λ‹¨μ„ μ§€λ½μ‚¬κ³ λŠ” κ·Έ νŠΉμ„± λ•Œλ¬Έμ— κ·Έ νŒλ³„μ΄ λ”μš± 쉽지 μ•Šλ‹€. 이에 따라 λ‹€λ₯Έ λΆ„μ•Όμ—μ„œλ„ 많이 적용되고 μžˆλŠ” 웨이블릿 λ³€ν™˜κ³Ό CNN을 ν™œμš©ν•˜μ—¬ κ³ μ €ν•­ 지락사고λ₯Ό νŒλ³„ν•˜λŠ” 방법을 μ œμ•ˆν•˜κ³ μž ν•œλ‹€. 이λ₯Ό ν†΅ν•˜μ—¬ λ‹€μ–‘ν•œ λ°°μ „κ³„ν†΅μ˜ κ΅¬μ‘°λ‚˜ 사고 상황에 따라 λ‹¬λΌμ§€λŠ” κ³ μ €ν•­ 지락사고λ₯Ό νŒλ³„ν•  수 μžˆλ‹€.

2. κ³ μž₯상황 데이터 생성

κ³ μ €ν•­ 사고 κ²€μΆœ μ•Œκ³ λ¦¬μ¦˜μ„ κ°œλ°œν•˜κΈ° μœ„ν•΄μ„œ κ³„ν†΅μ—μ„œ μ‹€μ œλ‘œ λ°œμƒν•œ κ³ μž₯μ΄λ‚˜ μ‹€ν—˜μ„ ν†΅ν•˜μ—¬ κ΅¬ν•œ κ³ μ €ν•­ μ‚¬κ³ μ˜ μ „μ••κ³Ό μ „λ₯˜λ₯Ό 뢄석해왔닀. μ΄λŸ¬ν•œ 데이터λ₯Ό 뢄석할 λ•Œ λ™μΌν•œ μ ‘μ΄‰λ¬Όμ§ˆμ— μ˜ν•œ κ³ μ €ν•­ 사고라 ν•˜λ”λΌλ„ 사고 μ‹œμ μ˜ λΆ€ν•˜μš©λŸ‰μ΄λ‚˜ 기타 계톡상황에 μ˜ν•΄ κ³„μ‚°μ μ—μ„œμ˜ μ „μ••, μ „λ₯˜κ°€ 달라진닀(1). λ³Έ λ…Όλ¬Έμ—μ„œλŠ” μ „λ ₯계톡 λͺ¨μ˜λ₯Ό ν†΅ν•˜μ—¬ 사고 μ‹œ 데이터와 일반 운영 μ‹œμ˜ 데이터λ₯Ό μƒμ„±ν–ˆμœΌλ―€λ‘œ κ³ μ €ν•­ μ‚¬κ³ μ˜ μ „μ••, μ „λ₯˜ νŠΉμ„±μ„ 잘 λ‚˜νƒ€λ‚Ό 수 μžˆλŠ” μ „λ ₯계톡 λͺ¨λΈλ§μ΄ ν•„μš”ν•˜λ‹€. 배전계톡은 싀계톡에 κΈ°λ°˜ν•œ λͺ¨λΈμ— 뢄산전원이 μ„€μΉ˜λ˜μ—ˆλ‹€κ³  κ°€μ •ν•˜μ—¬ 계톡을 λͺ¨λΈλ§ν•˜μ˜€λ‹€. κ³ μ €ν•­ μ‚¬κ³ μ˜ νŠΉμ„±μœΌλ‘œλŠ” μ¦κ°€ν˜„μƒ, λ©ˆμΆ€ν˜„μƒ, λΉ„μ„ ν˜•μ„±, λΉ„λŒ€μΉ­μ„± 등이 μžˆλ‹€. (15), (16) λ³΄ν˜Έκ³„μ „κΈ°μ—μ„œλŠ” ν•œλ‘ 싸이클 내에 사고λ₯Ό κ²€μΆœν•΄μ•Ό ν•˜λ―€λ‘œ 짧은 μ‹œκ°„μ— λ‚˜νƒ€λ‚˜λŠ” νŠΉμ„±μΈ λΉ„μ„ ν˜•μ„±, λΉ„λŒ€μΉ­μ„±μ„ λ°˜μ˜ν•œ λͺ¨λΈμ„ κ΅¬μ„±ν•˜μ˜€λ‹€.

2.1 배전계톡 λͺ¨λΈλ§

κ·Έλ¦Ό 1 뢄산전원이 μžˆλŠ” 배전계톡 λͺ¨λΈλ§

Fig. 1 Distribution system model with a distribution generation

../../Resources/kiee/KIEE.2022.71.11.1715/fig1.png

λ°°μ „κ³„ν†΅μ—μ„œ λ°œμƒν•˜λŠ” κ³ μ €ν•­ μ‚¬κ³ μ˜ 데이터λ₯Ό λ§Œλ“€κΈ° μœ„ν•΄μ„œ λͺ¨λΈ μ „λ ₯계톡을 κ΅¬μ„±ν•˜κ³  사고상황을 κ΅¬μ„±ν•˜μ˜€λ‹€. λŒ€μƒ μ „λ ₯계톡은 뢄산전원이 μžˆλŠ” λ°°μ „κ³„ν†΅μ—μ„œ κ³ μ €ν•­ 사고가 λ°œμƒν•˜λŠ” 상황을 μƒμ •ν•˜μ˜€λ‹€. 그리고 비ꡐλ₯Ό μœ„ν•˜μ—¬ λΆ€ν•˜λŸ‰μ΄ μ¦κ°€ν•˜μ—¬ μ „λ₯˜κ°€ μ»€μ§€λŠ” 상황을 μƒμ •ν•˜μ˜€λ‹€.

κ·Έλ¦Ό 1은 데이터 생성을 μœ„ν•΄ μ‚¬μš©ν•œ λͺ¨λΈ λ°°μ „κ³„ν†΅μ˜ ꡬ성도이닀. λ³€μ „μ†Œμ—μ„œ λ‚˜κ°€λŠ” 피더 ν•œ 개λ₯Ό λͺ¨λΈλ§ν•˜μ˜€μœΌλ©°, 피더 λ‚΄ 총 λΆ€ν•˜λŸ‰μ€ 10MW 이며 μ—­λ₯ μ€ 95%둜 ν•˜μ˜€λ‹€. 그리고 뢄산전원이 μžˆλŠ” 배전계톡이며, μ‹ μž¬μƒ λΆ„μ‚°μ „μ›μ˜ μ΅œλŒ€ λ°œμ „μš©λŸ‰μ€ 8MW둜 κ°€μ •ν•˜μ˜€λ‹€. 데이터 생성 μ‹œ μ—¬λŸ¬ 가지 경우λ₯Ό μƒμ •ν•˜κΈ° μœ„ν•΄, λΆ„μ‚°μ „μ›μ˜ λ°œμ „λŸ‰μ΄ 8, 4.5, 1MW인 각각의 κ²½μš°μ— λŒ€ν•΄ 사고가 λ°œμƒν•œ 상황을 λͺ¨μ˜ν•˜μ˜€λ‹€.

μ„ λ‘œ λͺ¨λΈμ€ 싀계톡 λ°μ΄ν„°μ˜ μ„ λ‘œκΈΈμ΄, 선쒅을 κ³ λ €ν•˜μ—¬ λͺ¨λΈλ§ν•˜μ˜€μœΌλ©°, PI λ“±κ°€λͺ¨λΈλ‘œ κ΅¬ν˜„ν•˜μ˜€μœΌλ©°, λΆ€ν•˜λŠ” μ •μ „λ ₯κ³Ό μ •μž„ν”Όλ˜μŠ€ λΆ€ν•˜λ‘œ κ΅¬μ„±ν•˜μ˜€λ‹€. 전원 ν”Όλ”μΈ‘μ˜ μ ‘μ§€λŠ” Y-Yg둜 ν•˜μ˜€μœΌλ©°, 뢄산전원 츑은 Yg-β–³λ‘œ ν•˜μ˜€λ‹€.

κ·Έλ¦Ό 2 전원 피더츑 접지 λͺ¨λΈλ§

Fig. 2 Grounding modeling of generation side

../../Resources/kiee/KIEE.2022.71.11.1715/fig2.png

κ·Έλ¦Ό 3 뢄산전원츑 접지 λͺ¨λΈλ§

Fig. 3 Grouding modeling of distributed generation side

../../Resources/kiee/KIEE.2022.71.11.1715/fig3.png

2.2 κ³ μ €ν•­ 사고 λͺ¨λΈλ§

뢄산전원이 μžˆλŠ” λ°°μ „κ³„ν†΅μ—μ„œ λ°œμƒν•˜λŠ” κ³ μ €ν•­ 지락사고λ₯Ό λΆ„μ„ν•˜κΈ° μœ„ν•˜μ—¬ 사고 데이터 및 그것과 λΉ„κ΅ν•˜κΈ° μœ„ν•œ 정상 데이터λ₯Ό μƒμ„±ν•˜μ˜€λ‹€. κ³ μ €ν•­ 지락사고λ₯Ό 뢄석을 μœ„ν•œ 데이터λ₯Ό κ΅¬μΆ•ν•˜λŠ” 데 μ€‘μš”ν•œ 것은 κ·Έ νŠΉμ„±μ„ κ°€λŠ₯ν•œ μ •ν™•ν•˜κ²Œ λ°˜μ˜ν•˜λŠ” 것이닀. κ·Έλ¦Ό 4λŠ” κ°• μžκ°ˆμ—μ„œ κ³ μ €ν•­ 사고 μ‹€ν—˜μ„ 톡해 얻은 μ „λ₯˜ νŒŒν˜•μ„ 보여주고 있으며, 증가 ν˜„μƒκ³Ό 멈좀 ν˜„μƒμ„ λ³Ό 수 있으며, λ‹€λ₯Έ 원인에 μ˜ν•œ κ³ μ €ν•­ 지락사고도 λΉ„μŠ·ν•œ νŠΉμ„±μ„ νƒ€λ‚˜λ‚Έλ‹€. κ·Έλ¦Ό 5λŠ” 사고 λ°œμƒ ν›„ 20주기와 40μ£ΌκΈ°μ—μ„œ μ „λ₯˜ νŒŒν˜•μ„ λΉ„κ΅ν•΄μ„œ 보여주고 있으며, λΉ„μ„ ν˜•μ„±κ³Ό λΉ„λŒ€μΉ­μ„±μ΄ 사고 λ°œμƒ ν›„ λͺ¨λ“  μ‹œκ°„ μ˜μ—­μ—μ„œ λ‚˜νƒ€λ‚˜κ³  μžˆλ‹€(1).

고쑰파 κ²€μΆœμ„ ν†΅ν•˜μ—¬ κ³ μ €ν•­ 지락사고λ₯Ό νŒλ³„ν•˜κΈ° μœ„ν•΄ μ€‘μš”ν•œ νŠΉμ„±μ€ μ΄λŸ¬ν•œ λΉ„μ„ ν˜•μ„±μ΄λ―€λ‘œ μ΄λŸ¬ν•œ νŠΉμ„±μ„ λͺ¨λΈλ§μ„ ν†΅ν•˜μ—¬ κ΅¬ν˜„ν•˜μ˜€μœΌλ©°, μ΄λ ‡κ²Œ μƒμ„±λœ λ°μ΄ν„°μ˜ νŠΉμ„±μ„ λΆ„μ„ν•˜μ—¬ κ³ μ €ν•­ 지락사고λ₯Ό νŒλ³„ν•  수 μžˆμ„ 것이닀. κ³ μ €ν•­ μ§€λ½μ‚¬κ³ μ˜ νŠΉμ„±μ„ λ°˜μ˜ν•˜κΈ° μœ„ν•΄ μ΄λŸ¬ν•œ λΉ„μ„ ν˜• νŠΉμ„±μ„ κ°€μ§€λŠ” λͺ¨λΈμ„ μƒμ„±ν•˜μ—¬ λͺ¨μ˜ν•˜μ˜€λ‹€.

κ·Έλ¦Ό 4 κ°• μžκ°ˆμ—μ„œμ˜ κ³ μ €ν•­ 사고 μ „λ₯˜ (1)

Fig. 4 HRF current on robust pebbles (1)

../../Resources/kiee/KIEE.2022.71.11.1715/fig4.png

κ·Έλ¦Ό 5 사고 λ°œμƒ ν›„ μ‹œκ°„μ— λ”°λ₯Έ μ „λ₯˜ νŒŒν˜• (1)

Fig. 5 Current with the time after a HRF (1)

../../Resources/kiee/KIEE.2022.71.11.1715/fig5.png

κ³ μ €ν•­ 사고λ₯Ό λͺ¨μ˜ν•˜κΈ° μœ„ν•˜μ—¬ μ‹œλ³€μ €ν•­μ˜ λͺ¨λΈμ„ μ‚¬μš©ν•˜μ˜€μœΌλ©°, κ·Έλ¦Ό 6κ³Ό 같이 가변저항을 μ΄μš©ν•˜μ—¬ κ³ μž₯저항을 λͺ¨λΈλ§ν•˜μ˜€λ‹€. κ·Έλ¦Ό 7(a)μ—μ„œλŠ” 이 λͺ¨λΈμ„ ν†΅ν•˜μ—¬ 얻은 κ³ μž₯μ „λ₯˜ νŒŒν˜•μ΄ κ·Έλ¦Ό 5와 μœ μ‚¬ν•œ ν˜•νƒœμž„μ„ λ³Ό 수 있고, μ΄λŸ¬ν•œ κ³ μž₯μ „λ₯˜ μž…λ ₯이 μžˆμ„ λ•Œ κ³„ν†΅μ—μ„œ 흐λ₯΄λŠ” κ³ μž₯μ „λ₯˜λ₯Ό κ·Έλ¦Ό 7(b)μ—μ„œ λ³Ό 수 μžˆλ‹€. μ΄λŸ¬ν•œ κ³ μž₯μ „λ₯˜ νŒŒν˜•μ—μ„œ λͺ¨λΈμ— μ˜ν•œ κ³ μž₯μ „λ₯˜μ˜ λΉ„μ„ ν˜•μ„±κ³Ό λΉ„λŒ€μΉ­μ„±μ„ 확인할 수 μžˆλ‹€.

κ·Έλ¦Ό 6 가변저항을 μ΄μš©ν•œ κ³ μž₯μ €ν•­ λͺ¨λΈ

Fig. 6 Fault resistance model using variable resistors

../../Resources/kiee/KIEE.2022.71.11.1715/fig6.png

κ·Έλ¦Ό 7 배전계톡 λͺ¨λΈμ—μ„œ μƒμ„±λœ κ³ μž₯μ „λ₯˜ νŒŒν˜•

Fig. 7 Fault current by simulation with the distribution system model

../../Resources/kiee/KIEE.2022.71.11.1715/fig7.png

2.3 λΆ€ν•˜κΈ‰μ¦ λͺ¨λΈλ§

κ³ μž₯이 μ•„λ‹Œ λΆ€ν•˜μ˜ 정상적인 증가 μ‹œμ—λ„ 계전기가 μΈ‘μ •ν•˜λŠ” μ „λ₯˜λŸ‰μ΄ 컀질 수 μžˆλ‹€. κ³ μ €ν•­ κ³ μž₯이 λ°œμƒν•˜λŠ” κ³ μž₯μ „λ₯˜μ˜ 양이 크지 μ•ŠκΈ° λ•Œλ¬Έμ— λ‹¨μˆœν•œ μ „λ₯˜ λ³€ν™”λŸ‰ 츑정을 ν†΅ν•΄μ„œλŠ” 정상적인 λΆ€ν•˜ 증가에 따라 λŠ˜μ–΄λ‚˜λŠ” μ „λ₯˜μ™€ κ΅¬λ³„ν•˜κΈ° 쉽지 μ•ŠμœΌλ©°, 이 λ…Όλ¬Έμ—μ„œλŠ” 웨이블릿 λ³€ν™˜κ³Ό CNN 기법을 μ μš©ν•˜μ—¬ κ³ μ €ν•­ κ³ μž₯을 νŒλ³„ν•˜κ³ μž ν•œλ‹€. 이λ₯Ό μœ„ν•œ 데이터 생성을 μœ„ν•˜μ—¬ κ³ μž₯ 상황뿐 μ•„λ‹ˆλΌ λΆ€ν•˜ 증가에 λŒ€ν•œ 데이터가 ν•„μš”ν•˜λ©°, 이λ₯Ό μœ„ν•˜μ—¬ λ‹€μŒκ³Ό 같이 λͺ¨λΈμ„ κ΅¬μ„±ν•˜μ—¬ λͺ¨μ˜ν•˜μ˜€λ‹€. μˆœκ°„μ μœΌλ‘œ μ¦κ°€ν•˜λŠ” λΆ€ν•˜λ₯Ό λͺ¨μ˜ν•˜κΈ° μœ„ν•΄ 각 μƒλ³„λ‘œ λΆ€ν•˜λŸ‰μ΄ κΈ‰μ¦ν•˜λŠ” 상황을 λͺ¨μ˜ν•˜μ˜€μœΌλ©°, 피더에 κ°€κΉŒμš΄ 지점, 쀑간 지점, 뢄산전원에 κ°€κΉŒμš΄ μ§€μ μ—μ„œ λΆ€ν•˜κ°€ κΈ‰μ¦ν•˜λ„λ‘ κ·Έλ¦Ό 8κ³Ό 같이 λͺ¨λΈλ§ν•˜μ˜€λ‹€. κ·Έλ¦Ό 9λŠ” 이 λͺ¨λΈμ— 따라 λΆ€ν•˜κ°€ 증가할 λ•Œ λͺ¨μ„ λ³„ μ „μ••μ˜ λ³€ν™”λ₯Ό λ‚˜νƒ€λ‚΄κ³  μžˆλ‹€.

κ·Έλ¦Ό 8 μˆœκ°„ λΆ€ν•˜κΈ‰μ¦ λͺ¨λΈλ§

Fig. 8 Modeling for rapid increase in load

../../Resources/kiee/KIEE.2022.71.11.1715/fig8.png

κ·Έλ¦Ό 9 μˆœκ°„ λΆ€ν•˜κΈ‰μ¦μ‹œ 각 λͺ¨μ„ λ³„ μ „μ••κ°•ν•˜

Fig. 9 Voltage drop by rapid increase in load

../../Resources/kiee/KIEE.2022.71.11.1715/fig9.png

μ΄μƒκΉŒμ§€ λ”₯λŸ¬λ‹μ„ μœ„ν•œ λͺ¨μ˜λ°μ΄ν„° 생성 쑰건에 λŒ€ν•΄ μ •λ¦¬ν•˜λ©΄ ν‘œ 1κ³Ό κ°™λ‹€. μ΄λŸ¬ν•œ 쑰건에 λŒ€ν•΄μ„œ λͺ¨μ˜λ₯Ό μˆ˜ν–‰ν•˜κ³ , κ·Έ κ²°κ³Όλ₯Ό ν…μŠ€νŠΈ ν˜•μ‹μ˜ 데이터 파일둜 μ €μž₯ν•˜μ˜€λ‹€. 각 쑰건에 λŒ€ν•΄ λͺ¨μ˜λ₯Ό μˆ˜ν–‰ν•˜μ˜€μœΌλ©°, μƒ˜ν”Œλ§ νƒ€μž„μ€ μƒμš© 디지털 계전기와 κ°™κ²Œ 50ΞΌsec둜 ν•˜μ˜€λ‹€.

ν‘œ 1 λͺ¨μ˜λ°μ΄ν„° 생성 쑰건

Table 1 Conditions for generating simulation data

뢄산전원 μš©λŸ‰ [MW]

8.0, 4.5, 1.0, 0.0

κ³ μž₯ / λΆ€ν•˜ 급증 μœ„μΉ˜

2-3 λͺ¨μ„  사이 / 7-8 λͺ¨μ„  사이 / 13-14 λͺ¨μ„  사이

κ³ μž₯ μ’…λ₯˜

κ³ μ €ν•­ 단상 지락 κ³ μž₯: κ³ μž₯μ €ν•­ 100 / 500 / 1,000 Ξ©

3. λ”₯λŸ¬λ‹μ„ ν†΅ν•œ κ³ μž₯ νŒλ³„

3.1 κ³ μž₯ λ°μ΄ν„°μ˜ μ‹œκ°ν™”

λͺ¨μ˜λ₯Ό ν†΅ν•˜μ—¬ 얻은 κ³ μž₯데이터 및 μ •μƒμƒνƒœμ˜ λ°μ΄ν„°λŠ” μ „μ••κ³Ό μ „λ₯˜μ˜ μ‹œκ³„μ—΄ 데이터이닀. 이 데이터듀을 CNN을 ν†΅ν•˜μ—¬ κ΅¬λΆ„ν•˜κΈ° μœ„ν•˜μ—¬ μ‹œκ³„μ—΄ 데이터λ₯Ό μ‹œκ°ν™”ν•˜μ˜€λ‹€. κ³ μž₯을 νŒλ³„ν•˜λŠ” 일반적인 λ³΄ν˜Έκ³„μ „κΈ°μ™€ 같이 μ „λ₯˜ 뢄석을 ν†΅ν•˜μ—¬ κ³ μž₯ νŒλ³„μ„ ν•  수 μžˆλ„λ‘ μ „λ₯˜μ˜ μ‹œκ³„μ—΄ 데이터λ₯Ό λŒ€μƒμœΌλ‘œ λΆ„μ„ν•˜μ˜€λ‹€. λ³Έ λ…Όλ¬Έμ—μ„œλŠ” κ³ μ €ν•­ 사고가 μ•„λ‹Œ 일반적인 μ‚¬κ³ λŠ” κΈ°μ‘΄ 계전기 μ•Œκ³ λ¦¬μ¦˜μœΌλ‘œ νŒλ³„ν•  수 μžˆλ‹€κ³  κ°€μ •ν•˜κ³ , κ³„μ „κΈ°μ˜ κΈ°λ³Έ κΈ°λŠ₯으둜 νŒλ³„μ΄ μ–΄λ €μš΄ κ³ μ €ν•­ 사고와 λΆ€ν•˜λŸ‰ μ¦κ°€μ˜ ꡬ별 방법을 μ œμ•ˆν•˜κ³ μž ν•œλ‹€. λ”°λΌμ„œ λŒ€μƒ 데이터도 κ·Έ 두 경우의 데이터λ₯Ό λ‹€λ£¨μ—ˆλ‹€.

μ „λ ₯κ³„ν†΅μ—μ„œ κ³ μž₯이 λ°œμƒν•˜μ˜€μ„ λ•Œ 계톡을 λ³΄ν˜Έν•˜κΈ° μœ„ν•΄μ„œλŠ” λΉ λ₯΄κ²Œ κ³ μž₯ μ—¬λΆ€λ₯Ό νŒλ‹¨ν•  수 μžˆμ–΄μ•Ό ν•œλ‹€. 일반 κ³„μ „κΈ°λŠ” 두 μ£ΌκΈ° 내에 νŒλ‹¨ν•  수 μžˆμ–΄μ•Ό ν•˜λ©°, λ³Έ λ…Όλ¬Έμ—μ„œλŠ” ν•œ μ£ΌκΈ° 내에 νŒλ³„ν•  수 μžˆλ„λ‘ ν•œ 주기의 μ „λ₯˜ 데이터λ₯Ό μ‚¬μš©ν•˜μ—¬ νŒλ³„ν•  수 μžˆλ„λ‘ ν•˜μ˜€λ‹€.

μ‹€μˆ˜ λͺ¨λ₯Όλ › μ›¨μ΄λΈ”λ¦Ώμ˜ λͺ¨ν•¨μˆ˜λŠ” λ‹€μŒκ³Ό κ°™λ‹€.

(1)
$\omega(t)=\dfrac{1}{K\sigma}e^{-(\sigma t)^{2}}\cos\left(2\pi f_{0}t\right)$

λ³€ν™˜ 결과의 κ³„μˆ˜λ₯Ό κ΅¬ν•˜κΈ° μœ„ν•΄μ„œ pywt λͺ¨λ“ˆμ˜ pywt.cwt() ν•¨μˆ˜λ₯Ό μ΄μš©ν•˜μ˜€κ³ , matplotlib.pyplot을 μ‚¬μš©ν•˜μ—¬ κ³„μˆ˜λ“€μ„ κ·Έλž˜ν”½μœΌλ‘œ λ‚˜νƒ€λ‚΄μ—ˆλ‹€. 상전λ₯˜μ˜ ν•œ μ£ΌκΈ° 데이터에 λͺ¨λ₯Όλ › 웨이블릿 λ³€ν™˜(Morlet wavelet transform)을 μ μš©ν•˜κ³ , κ·Έ κ³„μˆ˜μ— λŒ€ν•˜μ—¬ κ·Έλž˜ν”½ 데이터λ₯Ό μƒμ„±ν•˜μ˜€λ‹€. κ·Έλ¦Ό 10은 μƒμ„±λœ μ „λ₯˜ 데이터λ₯Ό κ·Έλž˜ν”½ λ°μ΄ν„°λ‘œ λ³€ν™˜ν•œ μ˜ˆμ΄λ‹€. λ³€ν™˜ν•˜λŠ” μ „λ₯˜ μ‹œκ³„μ—΄ λ°μ΄ν„°μ˜ κ°œμˆ˜λŠ” ν•œ 주기에 ν•΄λ‹Ήν•˜λŠ” 32개의 μ‹œκ³„μ—΄ 값이며, λͺ¨λ₯Όλ › 웨이블릿 λ³€ν™˜μ˜ κ³„μˆ˜λ₯Ό μƒ‰μœΌλ‘œ ν‘œν˜„ν•˜μ—¬ 32Γ—48 ν”½μ…€μ˜ μ‹œκ°„-주파수 κ·Έλž˜ν”½ λ°μ΄ν„°λ‘œ κ΅¬μ„±ν•˜μ˜€λ‹€.

κ·Έλ¦Ό 10 μ „λ₯˜ λ°μ΄ν„°μ˜ μ‹œκ°„-주파수 κ·Έλž˜ν”½ 데이터 λ³€ν™˜

Fig. 10 Time-frequency energy picture of current data

../../Resources/kiee/KIEE.2022.71.11.1715/fig10-1.png

../../Resources/kiee/KIEE.2022.71.11.1715/fig10-2.png

λͺ¨μ˜λ₯Ό 톡해 얻은 고저항지락사고 상황과 λΆ€ν•˜λŸ‰ 증가에 μ˜ν•œ μ „λ₯˜ 변화상황에 λŒ€ν•œ 데이터에 λŒ€ν•˜μ—¬ 웨이블릿 λ³€ν™˜μ„ ν†΅ν•˜μ—¬ κ·Έλž˜ν”½ λ°μ΄ν„°λ‘œ λ³€ν™˜ν•˜μ˜€κ³ , λ³€ν™˜λœ κ·Έλž˜ν”½λ°μ΄ν„°μ— λŒ€ν•˜μ—¬ κ³ μ €ν•­ 사고와 λΆ€ν•˜μ¦κ°€λ‘œ λΌλ²¨λ§ν•˜μ˜€λ‹€. ν‘œ 2에 뢄석 λŒ€μƒ 데이터에 λŒ€ν•œ λ‚΄μš©μ„ μ •λ¦¬ν•˜μ˜€λ‹€.

ν‘œ 2 CNN ν•™μŠ΅μ— μ‚¬μš©ν•œ 데이터

Table 2 Amount of data for CNN

데이터 μ’…λ₯˜

총 개수

ν›ˆλ ¨ (Train)

검증 (Validation)

ν…ŒμŠ€νŠΈ (Test)

고저항사고

7,200

3,600

2,160

1,440

λΆ€ν•˜μ¦κ°€

2,400

1,200

720

480

합계

9,600

4,800

2,880

1,920

3.2 CNNλͺ¨λΈ ꡬ성 및 ν•™μŠ΅ κ²°κ³Ό

κ·Έλ¦Ό 11 CNN ν•™μŠ΅ μˆ˜ν–‰ κ³Όμ •

Fig. 11 Structure of CNN algorithm

../../Resources/kiee/KIEE.2022.71.11.1715/fig11-1.png

../../Resources/kiee/KIEE.2022.71.11.1715/fig11-2.png

κ·Έλ¦Ό 12 CNN ν•™μŠ΅ 곑선

Fig. 12 Learning curve of CNN

../../Resources/kiee/KIEE.2022.71.11.1715/fig12-1.png

../../Resources/kiee/KIEE.2022.71.11.1715/fig12-2.png

μƒμ„±λœ κ·Έλž˜ν”½ 데이터에 λŒ€ν•΄ CNN을 μ μš©ν•˜μ˜€μœΌλ©°, λ¬Έμ œλŠ” 이진 λΆ„λ₯˜λ¬Έμ œλ‘œ κ΅¬μ„±ν•˜μ˜€λ‹€. (κ³ μ €ν•­ 지락고μž₯ / 정상 μš΄μ „) κ·Έλ¦Ό 11은 λ”₯λŸ¬λ‹ μ΄μš©ν•œ κ³ μ €ν•­ 지락 κ³ μž₯을 λΆ„μ„ν•˜λŠ” μˆœμ„œλ„μ™€ μ μš©ν•œ CNN의 계측 ꡬ성이닀. 파이썬(Python) ν™˜κ²½μ—μ„œ tensorflow, keras λͺ¨λ“ˆμ„ ν™œμš©ν•΄ CNN ν•™μŠ΅μ„ μˆ˜ν–‰ν•˜μ˜€μœΌλ‹€. κ·Έ ν•™μŠ΅κ³Όμ •μ€ κ·Έλ¦Ό 12와 κ°™μœΌλ©°, 각각 ν•™μŠ΅μ— λ”°λ₯Έ 정확도(accuracy)와 손싀(loss)의 λ³€ν™”λ₯Ό λ‚˜νƒ€λ‚΄κ³  μžˆλ‹€. ν•™μŠ΅μ„ 톡해 μƒμ„±λœ λͺ¨λΈμ„ ν…ŒμŠ€νŠΈ 데이터에 μ μš©ν•œ κ²°κ³Ό, 정확도 98.29%둜 고저항지락고μž₯을 νŒλ³„ν•˜μ˜€λ‹€. μ΄λŸ¬ν•œ μ •ν™•λ„λŠ” 계전기에 적용 κ°€λŠ₯ν•œ μˆ˜μ€€μœΌλ‘œ νŒλ‹¨λœλ‹€.

μ΄λ ‡κ²Œ μƒμ„±λœ κ³ μž₯νŒλ‹¨ λͺ¨λΈμ„ 계전기에 νƒ‘μž¬ν•¨μœΌλ‘œμ¨ κ³ μ €ν•­ 사고λ₯Ό νŒλ³„ν•˜κ²Œ λœλ‹€. μ΄λ ‡κ²Œ ν˜•μ„±λœ κ³ μ €ν•­ νŒλ³„ μ•Œκ³ λ¦¬μ¦˜μ„ ν¬ν•¨ν•œ λ°°μ „κ³„ν†΅μ—μ„œμ˜ κ³ μž₯νŒλ³„ 과정을 μ •λ¦¬ν•˜λ©΄ κ·Έλ¦Ό 13κ³Ό κ°™λ‹€. λ³΄ν˜Έκ³„μ „κΈ°μ— μ΄λŸ¬ν•œ μ•Œκ³ λ¦¬μ¦˜μ„ μ μš©ν•¨μœΌλ‘œμ¨ κ³ μ €ν•­ 사고가 μ•„λ‹Œ κ³ μž₯μ—λŠ” 기쑴의 일반적인 κ³ μž₯계전기 μ•Œκ³ λ¦¬μ¦˜μ„ ν†΅ν•˜μ—¬ λŒ€μ‘ν•˜κ³ , κ³ μ €ν•­ 사고에 λŒ€ν•΄μ„œλŠ” 이 λ…Όλ¬Έμ—μ„œ μ œμ•ˆν•œ μ•Œκ³ λ¦¬μ¦˜μœΌλ‘œ λŒ€μ‘ν•  수 μžˆλ‹€.

κ·Έλ¦Ό 13 κ³ μ €ν•­ κ³ μž₯νŒλ³„μ„ ν¬ν•¨ν•œ 배전계톡 λ³΄ν˜Έκ³„μ „κΈ° λ™μž‘

Fig. 13 Protction relay operation including with HRF detection

../../Resources/kiee/KIEE.2022.71.11.1715/fig13.png

4. κ²° λ‘ 

λ³Έ λ…Όλ¬Έμ—μ„œλŠ” 뢄산전원 μ„€μΉ˜ λ“±μœΌλ‘œ ꡬ성이 λ³΅μž‘ν•΄μ§„ λ°°μ „κ³„ν†΅μ—μ„œ 고저항지락사고가 λ°œμƒν•˜μ˜€μ„ λ•Œ 이λ₯Ό νŒλ³„ν•˜κΈ° μœ„ν•œ 방법을 μ œμ‹œν•˜μ˜€λ‹€. κ³ μ €ν•­μ§€λ½μ‚¬κ³ λŠ” 사고전λ₯˜μ˜ 크기가 크지 μ•Šμ•„ 일반적인 λ³΄ν˜Έκ³„μ „κΈ°μ˜ μ•Œκ³ λ¦¬μ¦˜μœΌλ‘œ νŒλ³„ν•˜κΈ° 쉽지 μ•Šμ•„μ„œ 고저항지락사고 μ‹œ λ°œμƒν•˜λŠ” κ³ μž₯μ „λ₯˜μ˜ νŠΉμ„±μ„ ν™œμš©ν•œ μ£ΌνŒŒμˆ˜λΆ„μ„ λ“±μ˜ 기법이 μ μš©λ˜μ–΄ μ™”λ‹€. 뢄산전원 λ„μž… λ“±μœΌλ‘œ κ³ μž₯μ „λ₯˜μ˜ 양상이 λ³΅μž‘ν•΄μ§€λ©΄ κ³ μ €ν•­μ‚¬κ³ μ˜ νŒλ³„μ΄ λ”μš± μ–΄λ €μ›Œμ§€κ²Œ 되며, μ΄λ•Œμ—λ„ λ³€ν•˜μ§€ μ•ŠλŠ” κ³ μž₯νŠΉμ„±μ„ ν™œμš©ν•  수 μžˆλŠ” κΈ°λ²•μœΌλ‘œμ„œ κ³ μž₯μ‹œ λ°œμƒν•˜λŠ” μ „λ₯˜λ₯Ό μ‹œκ°λ°μ΄ν„°λ‘œ λ³€ν™˜ν•˜κ³ , 이에 λŒ€ν•΄ CNN 기법을 μ μš©ν•˜μ—¬ 고저항지락사고λ₯Ό νŒλ³„ν•˜λŠ” 방법을 μ œμ‹œν•˜μ˜€λ‹€.

CNN 기법을 ν•™μŠ΅ν•˜κΈ° μœ„ν•œ λ°μ΄ν„°λŠ” λͺ¨λΈκ³„ν†΅μ˜ λͺ¨μ˜λ₯Ό ν†΅ν•˜μ—¬ μƒμ„±ν•˜μ˜€λ‹€. 데이터 생성을 μœ„ν•΄ λͺ¨λΈκ³„ν†΅μ—μ„œ 고저항지락사고가 λ°œμƒν•œ κ²½μš°μ™€ λΆ€ν•˜κ°€ μ¦κ°€ν•œ κ²½μš°μ— λŒ€ν•΄μ„œ κ³ μž₯μ €ν•­μ˜ 크기, λ°œμƒ μœ„μΉ˜, λΆ„μ‚°μ „μ›μ˜ λ°œμ „λŸ‰ 등을 λ³€ν™”ν•˜λ©΄μ„œ λͺ¨μ˜λ₯Ό μˆ˜ν–‰ν•˜μ˜€λ‹€. μ΄λ ‡κ²Œ μƒμ„±λœ 데이터에 λͺ¨λ₯Όλ › 웨이블릿 λ³€ν™˜μ„ μ μš©ν•˜μ—¬ κ·Έλž˜ν”½ λ°μ΄ν„°λ‘œ λ³€ν™˜ν•œ λ‹€μŒ CNN을 μ μš©ν•˜μ—¬ ν•™μŠ΅μ„ μˆ˜ν–‰ν•˜μ˜€λ‹€. ν•™μŠ΅ κ²°κ³Ό 98.29%의 μ •ν™•λ„λ‘œ 고저항지락사고λ₯Ό νŒλ³„ν•˜μ˜€μœΌλ©°, 이λ₯Ό κΈ°μ‘΄ λ³΄ν˜Έκ³„μ „κΈ° μ•Œκ³ λ¦¬μ¦˜κ³Ό μ‘°ν•©ν•¨μœΌλ‘œμ¨ λ°°μ „κ³„ν†΅μ—μ„œ λ°œμƒν•œ 고저항지락사고에 λŒ€μ‘ν•  수 μžˆλŠ” λ³΄ν˜Έκ³„μ „ μ•Œκ³ λ¦¬μ¦˜μ„ μ œμ•ˆν•˜μ˜€λ‹€. μΆ”ν›„ μ‹€μ œ κ³„ν†΅μ—μ„œ λ°œμƒν•œ 데이터λ₯Ό ν†΅ν•˜μ—¬ μ œμ•ˆ λ°©λ²•μ˜ 검증과 κ°œμ„ μ— λŒ€ν•œ 연ꡬλ₯Ό μ œμ•ˆν•œλ‹€.

Acknowledgements

This work was supported by KETEP (Korea Institute of Energy Technology Evaluation and Planning) grant funded by the Korea government (MOTIE) (No. 20191210301890)

References

1 
Soon-Ryul Nam, Yong-Chul Kang, Jong-Keun Park, Nov 1999, Modeling of a high impedance fault in a distribution system using time-varying resistance, Proc. KIEE Fall Conf. Korea, pp. 113-115Google Search
2 
Won-Wook Jung, Hak-Ju Lee, Seong-Chul Kwon, Woo-Kyu Chae, Sep 2010, Application of directional over current protection schemes considering the fault characteristics in the distribution system with dispersed generation, Journ. KIIEE, Vol. vol 24, No. 9, pp. 97-107DOI
3 
Z. Zhixia, L. Xiao, P. Zailin, Feb 2014, Fault line detection in neutral point ineffectively grounding power system based on phase-locked loop, IET Generat. Transmiss., Distrib., Vol. 8, No. 2, pp. 273-280DOI
4 
Y. Wang, J. Zhou, Z. Li, Z. Dong, Y. Xu, Aug 2015, Discriminant-analysisbased single-phase earth fault protection using improved PCA in distribution systems, IEEE Trans. Power Del., Vol. 30, No. 4, pp. 1974-1982DOI
5 
T. Ji, Q. Pang, X. Liu, Aug 2006, Study on fault line detection based on genetic artificial neural network in compensated distribution system, Proc. IEEE Int. Conf. Inf. Acquisition, Vol. weihai, No. china, pp. 1427-1431DOI
6 
X. Deng, R. Yuan, Z. Xiao, K. L. L. Wang, Feb 2015, Fault location in loop distribution network using SVM technology, Int. J. Electr. Power Energy Syst., Vol. 65, pp. 254-261DOI
7 
Y. Y. Wang, X. J. Zeng, Z. Y. Dong, Y. Xu, J. Yuan, Y. Huang, Dec 2013, Stator single-phase-to-ground fault protection for bus-connected powerformers based on hierarchical clustering algorithm, IEEE Trans. Energy Convers., Vol. 28, No. 4, pp. 991-998DOI
8 
M. Farshad, J. Sadeh, Oct 2012, Accurate single-phase fault-location method for transmission feeders based on K-nearest neighbor algorithm using one-end voltage, IEEE Trans. Power Del., Vol. 27, No. 4, pp. 2360-2367DOI
9 
N. I. Elkalashy, A. M. Elhaffar, T. A. Kawady, N. G. Tarhuni, M. Lehtonen, Oct 2010, Bayesian selectivity technique for earth fault protection in medium-voltage networks, IEEE Trans. Power Del., Vol. 25, No. 4, pp. 2234-2245DOI
10 
S. Zheng, L. C. Wang, H. Zhang, Oct 2016, A fault line selection method for small current grounding system based on big data, in Proc. IEEE PES Asia.Pacific Power Energy Eng. Conf. (APPEEC), Vol. xi’an, No. china, pp. 2470-2474DOI
11 
S.-S. Wang, A. Chern, Y. Tsao, J. Hung, X. Lu, Y.-H. Lai, B. Su, Aug 2016, Wavelet speech enhancement based on nonnegative matrix factorization, IEEE Signal Process. Lett., Vol. 23, No. 8, pp. 1101-1105DOI
12 
E. Kang, J. Min, J. C. Ye, Oct 2016, A deep convolutional neural network using directional wavelets for low-dose X-ray CT reconstruction, Comput. Vis. Pattern Recognit., Vol. 44, No. 10, pp. e360-e375DOI
13 
M. F. Guo, X. D. Zeng, D. Y. Chen, N. C. Yang, Feb 2018, Deep-learning-based earth fault detection using continuous wavelet transform and convolutional neural network in resonant grounding distribution systems, IEEE Sensors Journ., Vol. 18, No. 3, pp. 1291-1300DOI
14 
M. F. Guo, N. C. Yang, W. F. Chen, April 2019, Deep-learning- based fault classification using Hilbert-Huang transform and convolutional neural network in power distribution systems, IEEE Sensors Journ., Vol. 19, No. 16, pp. 6905-6913DOI
15 
D. I. Jeerings, J. R. Linders, April 1989, Ground resistance- revisited, IEEE Trans. Power Deliv., Vol. 4, No. issue 2, pp. 949-956DOI
16 
A. Aljohani, I. Habiballah, High-Impedance Fault Diagnosis: A Review, Energies 2020, 13, 6447DOI

μ €μžμ†Œκ°œ

λ°•μ’…μ˜ (Jong-young Park)
../../Resources/kiee/KIEE.2022.71.11.1715/au1.png

Jong-young Park received the B.S., M.S., and Ph.D. degrees from Seoul National University, Seoul, Korea, in 1999, 2001, and 2007, respectively.

He was a Senior Researcher at LS Electric Co., Ltd., Korea from 2009 to 2013.

Currently, he is a Senior Researcher at Korea Railroad Research Institute (KRRI) since 2013.

His recent research interests include the optimal operation of power systems in railway with the smart grid technology.

μ΄ν•œλ―Ό (Hanmin Lee)
../../Resources/kiee/KIEE.2022.71.11.1715/au2.png

Hanmin Lee received the M.S. and Ph.D. degrees from Korea University, Seoul, Korea, in 2006.

Currently, he is a chief Researcher at Korea Railroad Research Institute (KRRI) since 2000.

His research interests include power quality and energy storage systems.

μ‘°κ·œμ • (Gyu-Jung Cho)
../../Resources/kiee/KIEE.2022.71.11.1715/au3.png

Gyu-Jung Cho (S’14) was born in South Korea, in 1986.

He received the B.S., M.S. and Ph.D. degrees, in 2012, 2014 and 2019, respectively, from the College of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, South Korea.

He is currently a Senior Researcher with the Smart Electrical & Signaling Division, Korea Railroad Research Institute, Uiwang, South Korea.

His research interests include power system dynamics, electric railway system operation and protection, integration of renewable energy resources, and distribution system planning.

μ •ν˜Έμ„± (Hosung Jung)
../../Resources/kiee/KIEE.2022.71.11.1715/au4.png

He received a B.S and M.S. degree in Electrical engineering from Sungkyunkwan University, Republic of Korea, in 1995 and 1998, respectively.

He received a Ph.D. degree from the Electrical Electronic and Computer Engineering from Sungkyunkwan University in 2002.

He is currently a chief Researcher with the Smart Electrical & Signaling Division, Korea Railroad Research Institute, Uiwang, South Korea.

His research interests are railway electrification, energy system and power protection system.

ν•œλ¬Έμ„­ (Moonseob Han)
../../Resources/kiee/KIEE.2022.71.11.1715/au5.png

He received a B.S and M.S. degree in Electrical engineering from Inha University, Republic of Korea, in 1987 and 1989, respectively.

He is currently a Researcher with the Smart Electrical & Signaling Division, Korea Railroad Research Institute, Uiwang, South Korea.