Processing math: 100%
  • ๋Œ€ํ•œ์ „๊ธฐํ•™ํšŒ
Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers
  • COPE
  • kcse
  • ํ•œ๊ตญ๊ณผํ•™๊ธฐ์ˆ ๋‹จ์ฒด์ด์—ฐํ•ฉํšŒ
  • ํ•œ๊ตญํ•™์ˆ ์ง€์ธ์šฉ์ƒ‰์ธ
  • Scopus
  • crossref
  • orcid

  1. (Department of Electrical Engineering, Chungnam Natโ€™l University)



Electric vehicle(EV), Urban dynamometer driving schedule(UDDS), State-of-health(SOH), Machine learning, Artificial neural network(ANN), Long short term memory(LSTM)

1. ์„œ ๋ก 

ํ™”์„์—ฐ๋ฃŒ ๊ฐ€๊ฒฉ์˜ ์ธ์ƒ๊ณผ ํ™˜๊ฒฝ๋ฌธ์ œ ์ธ์‹์˜ ๋ณ€ํ™”๋กœ ์ „๊ธฐ์ž๋™์ฐจ(electric vehicle; EV)์˜ ์‹œ์žฅ๊ทœ๋ชจ๋Š” ์ฆ๊ฐ€ํ•˜๊ณ  ์žˆ๋Š” ์ถ”์„ธ์ด๋ฉฐ, ์„ธ๊ณ„์ ์ธ ์ „๊ธฐ์ž๋™์ฐจ ํ™•์‚ฐ์ •์ฑ…์œผ๋กœ ์‹œ์žฅ๊ทœ๋ชจ๋Š” ๋”์šฑ ์ปค์งˆ ๊ฒƒ์œผ๋กœ ์˜ˆ์ƒ๋œ๋‹ค(1). ์ด์—, ์ „๊ธฐ์ž๋™์ฐจ์˜ ์ฃผ์š” ๋ถ€ํ’ˆ๋“ค์˜ ์—ฐ๊ตฌ ๋ฐ ๊ฐœ๋ฐœ๋„ ํ™œ๋ฐœํžˆ ์ง„ํ–‰๋˜๊ณ  ์žˆ๋Š”๋ฐ, ์ด ์ค‘ ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ๋Š” ๊ณ ์—๋„ˆ์ง€/๊ณ ์ „๋ ฅ๋ฐ€๋„ ๋ฐ ์žฅ์ˆ˜๋ช…์˜ ์žฅ์ ์ด ์žˆ์–ด ์ „๊ธฐ์ž๋™์ฐจ์˜ ์ฃผ๋™๋ ฅ์›์œผ๋กœ์จ ์‚ฌ์šฉ๋˜๊ณ  ์žˆ๋‹ค(2). ๊ทธ๋ ‡์ง€๋งŒ, ๊ธ‰์†์ถฉ์ „ ๋ฐ ๋‹ค์–‘ํ•œ ์ „๋ฅ˜ํฌ๊ธฐ(C-rate)๋กœ ์ถฉ์ „๊ณผ ๋ฐฉ์ „์ด ๋นˆ๋ฒˆํ•˜๊ฒŒ ๋ฐ˜๋ณต๋˜๋Š” ์ „๋ฅ˜ํ”„๋กœํŒŒ์ผ์˜ ํŠน์„ฑ์ƒ ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์—ดํ™”๋Š” ๋ถˆ๊ฐ€ํ”ผํ•˜๋ฉฐ(3), ์ด๋กœ ์ธํ•œ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์ „๊ธฐํ™”ํ•™์  ํŠน์„ฑ์˜ ๋ณ€ํ™”, ์ฆ‰ ๋ฐฉ์ „์šฉ๋Ÿ‰์˜ ๊ฐ์†Œ ๋ฐ ๋‚ด๋ถ€์ €ํ•ญ์˜ ์ฆ๊ฐ€๊ฐ€ ์•ผ๊ธฐ๋œ๋‹ค(4). ์—ดํ™”๋กœ ์ธํ•œ ์ „๊ธฐํ™”ํ•™์  ํŠน์„ฑ๋ณ€ํ™”๋Š” ์ „๋ฐ˜์ ์ธ ์ „๊ธฐ์ž๋™์ฐจ์˜ ์„ฑ๋Šฅ์„ ์ €ํ•˜(์ถœ๋ ฅ์ €ํ•˜ ๋ฐ ์—ดํญ์ฃผ๋กœ ์ธํ•œ ์•ˆ์ „์„ฑ์˜ ๋ฌธ์ œ ๋“ฑ)์‹œํ‚ค๋Š” ๋งŒํผ ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์—ดํ™” ์ƒํƒœ๋ฅผ ํšจ์œจ์ ์œผ๋กœ ๋ชจ๋‹ˆํ„ฐ๋งํ•˜๊ธฐ ์œ„ํ•œ ๋ฐฐํ„ฐ๋ฆฌ๊ด€๋ฆฌ์‹œ์Šคํ…œ(battery management system; BMS)์ด ํ•„์š”ํ•˜๋ฉฐ, ํŠนํžˆ ์—ดํ™”์ƒํƒœ์˜ ์ง€ํ‘œ๋กœ ์‚ฌ์šฉ๋˜๋Š” ์ˆ˜๋ช…์ƒํƒœ(state-of-health; SOH)์˜ ์ •ํ™•ํ•œ ์ถ”์ • ๋ฐ ์˜ˆ์ธก์ด ์š”๊ตฌ๋œ๋‹ค. ํ•˜์ง€๋งŒ, ์ „๊ธฐ์ž๋™์ฐจ์— ์‚ฌ์šฉ๋˜๋Š” ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์—ดํ™”๋Š” ์™ธ๋ถ€ํ™˜๊ฒฝ ๋ฐ ์‚ฌ์šฉ์กฐ๊ฑด ๋“ฑ๊ณผ ๊ฐ™์€ ๋‹ค์–‘ํ•œ ์˜ํ–ฅ์˜ ์ƒํ˜ธ์ž‘์šฉ์œผ๋กœ ๋ฐœ์ƒํ•˜๋ฉฐ, ์—ดํ™”์— ๋”ฐ๋ฅธ ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ ๋‚ด๋ถ€ ์ „๊ธฐํ™”ํ•™์  ๋ณ€ํ™”์˜ ์ง์ ‘ ์ธก์ •์˜ ์–ด๋ ค์›€๊ณผ ๋น„์„ ํ˜•์„ฑ์œผ๋กœ ์˜ˆ์ธก์ด ์‰ฝ์ง€ ์•Š์€ ๋ฌธ์ œ์ ์ด ์žˆ๋‹ค(5). ์ด๋Ÿฌํ•œ ๋ฌธ์ œ์ ์„ ํ•ด๊ฒฐํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ๋ฐฐํ„ฐ๋ฆฌ ๋‚ด๋ถ€์˜ ๋ณต์žกํ•œ ๋ณ€ํ™”๋ฅผ ๊ณ ๋ คํ•˜์ง€ ์•Š์•„๋„ ๋˜๋ฉฐ, ๋น„์„ ํ˜• ๋ชจ๋ธ ์ถ”์ •์— ๊ฐ•์ ์„ ๊ฐ–๋Š” ์ธ๊ณต์‹ ๊ฒฝ๋ง(artificial neurnal network; ANN) ๊ธฐ๋ฒ•์„ ์ ‘๋ชฉ์‹œํ‚ค๋Š” ์—ฐ๊ตฌ๊ฐ€ ์ง„ํ–‰๋˜๊ณ  ์žˆ๋‹ค(6).

๋ณธ ์—ฐ๊ตฌ๋Š” ์ธ๊ณต์‹ ๊ฒฝ๋ง์˜ ์—ฌ๋Ÿฌ ๋ฐฉ๋ฒ• ์ค‘์— ์Œ์„ฑ์ธ์‹ ๋ฐ ์‹œ๊ณ„์—ด ๋ถ„์„์— ๊ฐ•์ ์„ ๊ฐ€์ง€๋Š” long short term memory(LSTM) ๋ชจ๋ธ์„ ์ด์šฉํ•˜์—ฌ ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ SOH ์˜ˆ์ธก์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๊ตฌํ˜„ํ•˜์˜€๋‹ค. ์ด๋ฅผ ์œ„ํ•ด, ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์—ดํ™” ์‹คํ—˜ ๋ฐ ์ „๊ธฐ์  ํŠน์„ฑ๋ถ„์„์„ ์ง์ ‘ ์‹ค์‹œํ•˜์—ฌ ๋‚ด๋ถ€ ์—ดํ™”ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ์ถ”์ถœํ•˜์˜€๊ณ  ๋ฐฉ์ „์šฉ๋Ÿ‰๊ณผ ๋‚ด๋ถ€์ €ํ•ญ์˜ ๋ณ€ํ™” ๊ธฐ๋ฐ˜ ์˜คํ”„๋ผ์ธ ์ƒ๊ด€๋ถ„์„(์ƒ๊ด€๊ณ„์ˆ˜ ์‚ฐ์ถœ)์„ ํ†ตํ•ด LSTM ๋ชจ๋ธ ๊ตฌ์„ฑ์„ ์œ„ํ•œ ๋ฐ์ดํ„ฐ์…‹์„ ํ™•๋ฆฝํ•˜์˜€๋‹ค. LSTM ๋ชจ๋ธ ๊ตฌํ˜„์„ ์œ„ํ•ด ๊ตฌ๊ธ€(Google)์—์„œ ์ œ๊ณตํ•˜๋Š” ์˜คํ”ˆ์†Œ์Šค์ธ ํ…์„œํ”Œ๋กœ์šฐ(Tensorflow) ๋ฐ ์ผ€๋ผ์Šค(Keras) ํŒจํ‚ค์ง€๋ฅผ ์‚ฌ์šฉํ•˜์˜€๊ณ , ์ด์˜ ๊ฒ€์ฆ์„ ์œ„ํ•ด recurrent neurnal network(RNN)๊ธฐ๋ฐ˜ SOH ์˜ˆ์ธก์•Œ๊ณ ๋ฆฌ์ฆ˜ ๊ฒฐ๊ณผ์™€ ๋น„๊ตํ•˜์˜€๋‹ค.

2. ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™” ์‹คํ—˜ ๋ฐ ๋ถ„์„

(์ž๋™์ฐจ ์ฃผํ–‰ ์‚ฌ์ดํด ์ด์šฉ)

2.1 SOH(State-of-health)

SOH๋Š” ๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™”๋กœ ์ธํ•œ ์„ฑ๋Šฅ ๊ฐ์†Œ์˜ ์ง€ํ‘œ๋กœ์จ ์ด๋ฅผ ํ† ๋Œ€๋กœ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์ˆ˜๋ช…์„ ํŒ๋‹จํ•œ๋‹ค. ์ผ๋ฐ˜์ ์œผ๋กœ SOH๋Š” ์‹(1)๊ณผ ๊ฐ™์€ ์ดˆ๊ธฐ ๋Œ€๋น„ ์—ดํ™”์— ์˜ํ•ด ๊ฐ์†Œ๋œ ๋ฐฉ์ „์šฉ๋Ÿ‰ ๋ฐ ์‹(2)์™€ ๊ฐ™์€ ์ดˆ๊ธฐ ๋Œ€๋น„ ์—ดํ™”์— ์˜ํ•ด ๊ฐ์†Œ๋œ ์ถœ๋ ฅ, ์ฆ‰ ๋‚ด๋ถ€์ €ํ•ญ ์ฆ๊ฐ€๋กœ ์ •์˜ํ•  ์ˆ˜ ์žˆ๋Š”๋ฐ, ์ด๊ฒƒ์€ ์‹ค์ œ ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•œ ์šฉ๋Ÿ‰์˜ ๊ฐ์†Œ์™€ ์ถœ๋ ฅ ๊ฐ์†Œ๋กœ ์ธํ•œ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์„ฑ๋Šฅ ๊ฐ์†Œ๋ฅผ ์˜๋ฏธํ•œ๋‹ค.

(1)
SOHcapacity=Cnactโˆ’CnEOLCnfreshโˆ’CnEOL

(2)
SOHpulsepower=Pactโˆ’PEOLPfreshโˆ’PEOL

์ด ๋•Œ, Cnact์™€ Pact๋Š” ํ˜„์žฌ ์‚ฌ์šฉ๊ฐ€๋Šฅ ๋ฐฉ์ „์šฉ๋Ÿ‰ ๋ฐ ์ถœ๋ ฅ, Cnfresh์™€ Pfresh๋Š” ์ •๊ฒฉ ์šฉ๋Ÿ‰ ๋ฐ ์ดˆ๊ธฐ ์ถœ๋ ฅ, CnEOL์™€ PEOL์€ ์ˆ˜๋ช… ์ž„๊ณ„์ (end of life: EOL)์—์„œ์˜ ๋ฐฉ์ „์šฉ๋Ÿ‰ ๋ฐ ์ถœ๋ ฅ์„ ์˜๋ฏธํ•œ๋‹ค. ์ˆ˜๋ช… ์ž„๊ณ„์ (EOL)์€ SOH๊ฐ€ 0์ธ ์ƒํƒœ๋ฅผ ๋‚˜ํƒ€๋‚ด๋ฉฐ, ๋ฐฉ์ „์šฉ๋Ÿ‰์˜ ๊ฒฝ์šฐ์—๋Š” ์ดˆ๊ธฐ ์šฉ๋Ÿ‰ ๋Œ€๋น„ 80%์˜ ์ž”์กด์šฉ๋Ÿ‰์ด๋ผ๋Š” ๊ธฐ์ค€์„ ์‚ฌ์šฉํ•˜์ง€๋งŒ ์ €ํ•ญ์˜ ๊ฒฝ์šฐ๋Š” ๊ธฐ์ค€์ด ๋ช…ํ™•ํ•˜์ง€ ์•Š์œผ๋ฏ€๋กœ ๋ฐฉ์ „์šฉ๋Ÿ‰์œผ๋กœ SOH๋ฅผ ํŒ๋‹จํ•˜๋Š” ๊ฒƒ์ด ์ผ๋ฐ˜์ ์ด๋‹ค(7). ํ•˜์ง€๋งŒ ๋ฐฉ์ „์šฉ๋Ÿ‰ ์ธก์ •์€ ์™„์ „ ์ถฉ์ „ ์ดํ›„ ์™„์ „ ๋ฐฉ์ „ ๋™์•ˆ ์ „๋ฅ˜์ ์‚ฐ๋ฒ•(ah-counting)์œผ๋กœ ๊ณ„์‚ฐํ•˜๊ธฐ ๋•Œ๋ฌธ์— ์„ผ์„œ์˜ ์ •ํ™•๋„์— ์˜ํ–ฅ์„ ๋งŽ์ด ๋ฐ›์œผ๋ฉฐ, ๋งŽ์€ ์‹œ๊ฐ„์„ ์†Œ๋น„ํ•ด์•ผํ•œ๋‹ค๋Š” ๋ฌธ์ œ์ ์„ ๊ฐ€์ง€๊ณ  ์žˆ๋‹ค.(8) ๋”ฐ๋ผ์„œ ์ง์ ‘์ ์ธ ๋ฐฉ์ „์šฉ๋Ÿ‰ ์ธก์ •์„ ๋Œ€์‹ ํ•˜์—ฌ ํŠน์ • ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ํ†ตํ•œ ๊ฐ„์ ‘์ ์ธ ์ธก์ • ๋˜๋Š” ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ์ถ”์ •ํ•˜๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜ ์—ฐ๊ตฌ๊ฐ€ ํ•„์š”ํ•˜๋‹ค.

2.2 ์ž๋™์ฐจ ์ฃผํ–‰ ์‚ฌ์ดํด ์—ดํ™” ์‹คํ—˜

๋ณธ ์‹คํ—˜์€ ์ „๊ธฐ์ž๋™์ฐจ(EV)์™€ ๊ฐ™์€ ๊ณ ์ถœ๋ ฅ ์–ดํ”Œ๋ฆฌ์ผ€์ด์…˜์˜ ์—ดํ™” ๋ถ„์„์„ ์œ„ํ•œ ์‹คํ—˜์œผ๋กœ ํ‘œ 1์—์„œ ๋‚˜ํƒ€๋‚œ NCA๊ณ„์—ด์˜ ๊ณ ์ถœ๋ ฅ 18650 ์›ํ†ตํ˜• ์…€์„ ์‚ฌ์šฉํ•˜์˜€๋‹ค. ๋˜ํ•œ ์‹ค์ œ ๋„์‹ฌ์—์„œ ์ฃผํ–‰๋˜๋Š” ์ „๊ธฐ์ž๋™์ฐจ์˜ ๋‹ค์–‘ํ•œ ์ถœ๋ ฅ์ด ๋ฐ˜๋ณต๋˜๋Š” ์ƒํ™ฉ์„ ๋ชจ์‚ฌํ•˜๊ธฐ ์œ„ํ•ด UDDS(urban dynamometer driving schedule) ํ”„๋กœํŒŒ์ผ์„ ์‚ฌ์šฉํ•˜์—ฌ ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. UDDS ํ”„๋กœํŒŒ์ผ ๊ธฐ๋ฐ˜ ์‹คํ—˜์€ ์ž๋™์ฐจ ์ฃผํ–‰ ์‚ฌ์ดํด์— ๋”ฐ๋ฅธ ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™” ๋ถ„์„ ๋ฐ ๋ฐฐํ„ฐ๋ฆฌ ๋‚ด๋ถ€ ํŒŒ๋ผ๋ฏธํ„ฐ ๋ฐ์ดํ„ฐ๋ฅผ ํ™•๋ณดํ•˜์—ฌ SOH ์ถ”์ • ๋ชจ๋ธ ๊ฐœ๋ฐœ์— ํŽธ๋ฆฌํ•œ ์žฅ์ ์„ ๊ฐ–๊ณ  ์žˆ๋‹ค. ๊ทธ๋Ÿฌ๋‚˜ ์‹ค์ œ ์ž๋™์ฐจ ์ฃผํ–‰ ๊ณผ์ •์—์„œ์˜ ์ง„๋™, ์ถฉ๊ฒฉ ๋“ฑ์˜ ๋ฌผ๋ฆฌ์  ์ธก์ • ๋ฐ ์‚ฌํ›„ ๋ถ„์„์ด ๋ถˆ๊ฐ€๋Šฅํ•˜๋‹ค๋Š” ํ•œ๊ณ„์„ฑ์ด ์กด์žฌํ•œ๋‹ค(9).

ํ‘œ 1. 18650 ์›ํ†ตํ˜• 25R ๋ฐฐํ„ฐ๋ฆฌ ์‚ฌ์–‘

Table 1. 18650 cylindrical 25R battery specification

Characteristic

Specification

Nominal discharge capacity

2500mAh

Nominal voltage

3.6V

Standard charge

CCCV, 1.25A, 4.2ยฑ0.05V,

125mA cut-off

Discharge cut-off voltage

2.5V

๊ทธ๋ฆผ. 1. UDDS ํ”„๋กœํŒŒ์ผ (a) ๋‹จ์ž ์ „์•• (b) ์ „๋ฅ˜

Fig. 1. UDDS profile (a) terminal voltage (b) input current

../../Resources/kiee/KIEE.2019.68.10.1214/fig1.png

๊ทธ๋ฆผ. 2. ์ž๋™์ฐจ ์ฃผํ–‰ ์‚ฌ์ดํด ์—ดํ™” ์‹คํ—˜ ํ”„๋กœํŒŒ์ผ

Fig. 2. Aging experiment profile with UDDS

../../Resources/kiee/KIEE.2019.68.10.1214/fig2.png

์™ธ๋ถ€ ํ™˜๊ฒฝ์š”์ธ์œผ๋กœ ์ธํ•œ ์—ดํ™”๋ฅผ ๋ฐฉ์ง€ํ•˜๊ธฐ ์œ„ํ•ด ํ•ญ์˜จ ํ•ญ์Šต ์ฑ”๋ฒ„๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜์˜€์œผ๋ฉฐ, ์ฑ”๋ฒ„ ๋‚ด๋ถ€ ์˜จ๋„๋ฅผ ์ƒ์˜จ(25โ„ƒ)์œผ๋กœ ์„ค์ •ํ•œ ์ดํ›„ ์‹ค์ œ ๋ฐฐํ„ฐ๋ฆฌ์˜ ์˜จ๋„๋ฅผ ๋™์ผํ•˜๊ฒŒ ๋งž์ถ”๊ธฐ ์œ„ํ•ด ์ฑ”๋ฒ„ ์•ˆ์— 1์‹œ๊ฐ„๋™์•ˆ ์œ ์ง€์‹œ์ผฐ๋‹ค. ์—ดํ™” ์‹คํ—˜์€ ๊ทธ๋ฆผ 1์—์„œ์˜ UDDS ํ”„๋กœํŒŒ์ผ์„ 1์‚ฌ์ดํด ์ •์˜ํ•˜๊ณ  ๊ทธ๋ฆผ 2์™€ ๊ฐ™์ด ์‚ฌ์ดํด์„ ๋ฐ˜๋ณตํ•˜๋ฉฐ ์ง„ํ–‰ํ•˜์˜€๋‹ค. CC-CV(constant current- constant voltage)๋ฅผ ํ†ตํ•œ ๋งŒ์ถฉ(fully charge) ์ดํ›„์— SOC(state of charge) 80%๊นŒ์ง€ ๋ฐฉ์ „ ์ดํ›„ 10 ์‚ฌ์ดํด์„ ๋ฐ˜๋ณตํ•˜์˜€์œผ๋ฉฐ, ๊ฐ ์‚ฌ์ดํด ์‚ฌ์ด์— ๋ฐฐํ„ฐ๋ฆฌ ์•ˆ์ •ํ™”๋ฅผ ์œ„ํ•ด 1์‹œ๊ฐ„์˜ ํœด์ง€๊ฐ„๊ฒฉ(Rest)์„ ๋ถ€์—ฌํ•˜์˜€๋‹ค. ์—ดํ™” ์‹คํ—˜์€ 10์‚ฌ์ดํด ์ฃผ๊ธฐ๋กœ ์ด 400์‚ฌ์ดํด์˜ ์‹คํ—˜์„ ์ˆ˜ํ–‰ํ•˜์˜€๋‹ค.

๊ทธ๋ฆผ. 3. ์—ดํ™”์— ๋”ฐ๋ฅธ ์šฉ๋Ÿ‰ ํ”„๋กœํŒŒ์ผ ์ „์•• ๊ณก์„ ์˜ ๋ณ€ํ™”

Fig. 3. Variation of capacity profile voltage curve with aging

../../Resources/kiee/KIEE.2019.68.10.1214/fig3.png

2.3 ์ „๊ธฐ์  ํŠน์„ฑ ๊ธฐ๋ฐ˜ ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ ๋ถ„์„

2.3.1 ์—ดํ™”์— ๋”ฐ๋ฅธ ์šฉ๋Ÿ‰ ์‹คํ—˜ ๋ฐ ๋ฐฉ์ „์šฉ๋Ÿ‰ ์ธก์ •

์ž๋™์ฐจ ์ฃผํ–‰ ์‚ฌ์ดํด์— ๋”ฐ๋ฅธ ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™”๋กœ ๋ณ€ํ™”ํ•˜๋Š” ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ์ธก์ •ํ•˜๊ธฐ ์œ„ํ•ด ๊ทธ๋ฆผ 3๊ณผ ๊ฐ™์€ ๋งŒ์ถฉ ์ดํ›„ ๋งŒ๋ฐฉ์œผ๋กœ ์ง„ํ–‰๋˜๋Š” ์šฉ๋Ÿ‰ ํ”„๋กœํŒŒ์ผ์„ ์‚ฌ์šฉํ•˜์˜€๋‹ค. ์šฉ๋Ÿ‰ ํ”„๋กœํŒŒ์ผ์€ 1C-rate ์ถฉ์ „์ „๋ฅ˜๋กœ SOC 80%์—์„œ๋ถ€ํ„ฐ CC-CV๋กœ ๋งŒ์ถฉ ์ดํ›„์— 1์‹œ๊ฐ„์˜ ํœด์ง€์‹œ๊ฐ„์„ ์ ์šฉํ•˜๊ณ  1C-rate์˜ ๋ฐฉ์ „์ „๋ฅ˜๋กœ ํ•˜ํ•œ ์ „์••๊นŒ์ง€ ๋งŒ๋ฐฉ์˜ ์ˆœ์„œ๋กœ ์ง„ํ–‰ํ•˜๋ฉฐ ์™„์ „ ์ถฉ์ „ ์ดํ›„ ์™„์ „ ๋ฐฉ์ „์„ ํ†ตํ•ด ๋งŒ๋ฐฉ๊นŒ์ง€ ์ „๋ฅ˜๋ฅผ ์ ์‚ฐํ•˜์—ฌ ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ์ธก์ •ํ•˜์˜€๋‹ค. ์‹คํ—˜์กฐ๊ฑด์€ ์ƒ์˜จ(25โ„ƒ)์กฐ๊ฑด์˜ ์ฑ”๋ฒ„ ๋‚ด๋ถ€์—์„œ ๋ฐฐํ„ฐ๋ฆฌ ์ŠคํŽ™์— ๋”ฐ๋ผ ์ถฉ์ „ ์ƒํ•œ ์ „์••(4.2V)๊ณผ ๋ฐฉ์ „ ํ•˜ํ•œ ์ „์••(2.5V)์„ ์ ์šฉํ•˜์—ฌ UDDS ํ”„๋กœํŒŒ์ผ 10์‚ฌ์ดํด ์ดํ›„๋งˆ๋‹ค ์ง„ํ–‰ํ•˜์˜€๋‹ค.

2.3.2 ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ ์ถ”์ถœ ๋ฐ ์ƒ๊ด€๋ถ„์„

SOH ์˜ˆ์ธก์„ ์œ„ํ•œ ๋ฐฉ์ „์šฉ๋Ÿ‰ ์ธก์ •์€ ๋งŒ์ถฉ๊ณผ ๋งŒ๋ฐฉ์œผ๋กœ ์ง„ํ–‰๋˜๋Š” ํ”„๋กœํŒŒ์ผ ํŠน์ง•์ƒ ์‹œ๊ฐ„ ์†Œ๋น„๊ฐ€ ํฌ๋ฉฐ, ๋ฐฐํ„ฐ๋ฆฌ ๋ถ€๋‹ด์ด ํฌ๊ธฐ ๋•Œ๋ฌธ์— ์‹ค์ œ ์–ดํ”Œ๋ฆฌ์ผ€์ด์…˜์—์„œ๋Š” ์‹ค์งˆ์ ์œผ๋กœ ๋ถˆ๊ฐ€๋Šฅํ•˜๋‹ค. ํ•˜์ง€๋งŒ ์ด๋Ÿฌํ•œ ๋ฌธ์ œ์ ์€ ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™”์— ๋”ฐ๋ผ ๋ณ€ํ™”ํ•˜๋Š” ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ํ†ตํ•œ ์ถ”์ •์œผ๋กœ ํ•ด๊ฒฐํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์‹ค์‹œ๊ฐ„ SOH ์˜ˆ์ธก์„ ์œ„ํ•œ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๋Š” ์ ์€ ๊ณ„์‚ฐ๋Ÿ‰๊ณผ SOH ๋ฐ ๋ฐฉ์ „์šฉ๋Ÿ‰๊ณผ์˜ ๋†’์€ ์ƒ๊ด€๊ด€๊ณ„๊ฐ€ ์š”๊ตฌ๋œ๋‹ค.

๊ทธ๋ฆผ 4๋Š” ์šฉ๋Ÿ‰ ์‹คํ—˜์—์„œ ์—ดํ™”์— ๋”ฐ๋ผ ๋ณ€ํ™”ํ•˜๋Š” ์ „์••๊ณก์„ ์„ ๋‚˜ํƒ€๋‚ธ ๊ทธ๋ž˜ํ”„์ด๋‹ค. ์‚ฌ์ดํด์˜ ์ฆ๊ฐ€์— ๋”ฐ๋ผ ์ „์•• ๋ณ€ํ™”๋Ÿ‰์ด ์ฆ๊ฐ€ํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด๋Š” ์—ดํ™”์— ๋”ฐ๋ฅธ ๋‚ด๋ถ€์ €ํ•ญ์˜ ์ฆ๊ฐ€๋ฅผ ๋ณด์—ฌ์ค€๋‹ค. ๋‚ด๋ถ€์ €ํ•ญ์˜ ์ฆ๊ฐ€๋Š” ๋ฐฐํ„ฐ๋ฆฌ ์ถœ๋ ฅ ๊ฐ์†Œ๋ฅผ ์˜๋ฏธํ•˜๋ฉฐ, ์ฆ‰ ์—ดํ™”๋กœ ์ธํ•œ ๋ฐฐํ„ฐ๋ฆฌ ์„ฑ๋Šฅ ๊ฐ์†Œ๋ฅผ ๋‚˜ํƒ€๋‚ด๋Š” ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ์ด๋‹ค. ๋˜ํ•œ ์ „์••์˜ ๋ณ€ํ™”์œจ์„ ์ „๋ฅ˜๋กœ ๋‚˜๋ˆ„์–ด ๊ณ„์‚ฐ ๊ฐ€๋Šฅํ•œ ์ ์€ ๊ณ„์‚ฐ๋Ÿ‰์ด ํŠน์ง•์ด๋ฉฐ ์ด๋Ÿฌํ•œ ํŠน์ง•์œผ๋กœ ์‹ค์‹œ๊ฐ„ ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๋กœ ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•˜๋‹ค.

๊ทธ๋ฆผ. 4. ์—ดํ™”์— ๋”ฐ๋ฅธ ๋ฐฉ์ „ ์ „์•• ๊ณก์„ ์˜ ๋ณ€ํ™” ๋ฐ ์ €ํ•ญ์ถ”์ถœ

Fig. 4. Variation of discharge voltage curve with aging and resistance extraction

../../Resources/kiee/KIEE.2019.68.10.1214/fig4.png

๋‚ด๋ถ€์ €ํ•ญ์€ ํŠน์ง•์— ๋”ฐ๋ผ ์ „๋ฅ˜์— ์˜ํ•œ ์ˆœ๊ฐ„์ ์ธ ์ „์••๊ฐ•ํ•˜๋กœ ๋‚˜ํƒ€๋‚˜๋Š” ์ง๋ ฌ์ €ํ•ญ(series resistance)๊ณผ ๋ถ„๊ทนํšจ๊ณผ๋กœ ์ธํ•ด ๋น„์„ ํ˜•์ ์ธ ์ „์••๊ฐ•ํ•˜๋กœ ์ด์–ด์ง€๋Š” ๋ถ„๊ทน์ €ํ•ญ(polarization resistance)์œผ๋กœ ๊ตฌ๋ถ„ํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๋”ฐ๋ผ์„œ ์‹(3)์—์„œ ๋‚˜ํƒ€๋‚ฌ๋“ฏ์ด ์ง๋ ฌ์ €ํ•ญ๊ณผ ๋ถ„๊ทน์ €ํ•ญ์œผ๋กœ ๊ตฌ๋ถ„ํ•˜์˜€๋‹ค. ์ง๋ ฌ์ €ํ•ญ์€ ๋ฐฉ์ „ ์ „์••๊ณก์„ ์—์„œ 100ms์˜ ์ƒ˜ํ”Œ๋ง์‹œ๊ฐ„๋™์•ˆ ์ˆœ๊ฐ„์ ์ธ ์ „์••๋ณ€ํ™”๋Ÿ‰์„ ๋ฐฉ์ „์ „๋ฅ˜๋กœ ๋‚˜๋ˆ„์–ด ๊ณ„์‚ฐํ–ˆ์œผ๋ฉฐ, ๋ถ„๊ทน์ €ํ•ญ์€ ์ˆœ๊ฐ„์ ์ธ ์ „์•• ๋ณ€ํ™” ์ดํ›„ ์งง์€ ๋ฐฉ์ „ ๊ตฌ๊ฐ„๋™์•ˆ 1s์™€ 60s์˜ ๋‘ ๊ตฌ๊ฐ„์œผ๋กœ ๋‚˜๋ˆ„์–ด ๊ฐ ๊ตฌ๊ฐ„์—์„œ์˜ ์ „์••๋ณ€ํ™”๋Ÿ‰์„ ๋ฐฉ์ „์ „๋ฅ˜๋ฅผ ๋‚˜๋ˆ„์–ด ๊ณ„์‚ฐํ•˜์˜€๋‹ค(10).

(3)
R(series)=โ–ณV100msI R(p1s)=โ–ณV1sโˆ’โ–ณV100msI R(p60s)=โ–ณV60sโˆ’โ–ณV1sI

๋ฆฌํŠฌ์ด์˜จ ๋ฐฐํ„ฐ๋ฆฌ๋ฅผ ์ถฉ์ „ํ•˜๋Š” ๊ณผ์ •์—์„œ CC๋ฅผ ํ†ตํ•œ ์ถฉ์ „์€ SOC 100[%]๊นŒ์ง€ ๋งŒ์ถฉ์ด ๋ถˆ๊ฐ€๋Šฅํ•˜๋ฉฐ, ๋งŒ์ถฉ์— ๊ฐ€๊นŒ์šด SOC์ผ์ˆ˜๋ก ์ €ํ•ญ์ด ์˜ฌ๋ผ๊ฐ€๋Š” ํŠน์ง•์œผ๋กœ ์ธํ•ด ๋†’์€ ์ „๋ฅ˜๋Š” ๋†’์€ ์ €ํ•ญ์„ฑ ๋ฐœ์—ด์„ ์•ผ๊ธฐ์‹œ์ผœ ๋‚ด๋ถ€์— ์†์ƒ์„ ์ผ์œผํ‚จ๋‹ค(11). ๋”ฐ๋ผ์„œ CC์ถฉ์ „์œผ๋กœ ๋ฐฐํ„ฐ๋ฆฌ ์ƒํ•œ์ „์••์— ๋„๋‹ฌํ•œ ๊ฒฝ์šฐ ์ผ์ •ํ•œ ์ „์••์œผ๋กœ ์ „๋ฅ˜๋ฅผ ์ตœ์†Œ๊นŒ์ง€ ๋–จ์–ด๋œจ๋ฆฌ๋ฉฐ ์ถฉ์ „ํ•˜๋Š” CC-CV ์ถฉ์ „๋ฐฉ์‹์„ ์ผ๋ฐ˜์ ์œผ๋กœ ์‚ฌ์šฉํ•œ๋‹ค.

๊ทธ๋ฆผ 5๋Š” ์šฉ๋Ÿ‰ ์‹คํ—˜์—์„œ ์—ดํ™”์— ๋”ฐ๋ผ ๋ณ€ํ™”ํ•˜๋Š” CC-CV ์ถฉ์ „ ์ „์••๊ณก์„ ์ด๋ฉฐ, ์‚ฌ์ดํด ์ฆ๊ฐ€์— ๋”ฐ๋ผ ๋ฐฐํ„ฐ๋ฆฌ๊ฐ€ ์—ดํ™” ๋ ์ˆ˜๋ก CC์ถฉ์ „์€ ์งง์•„์ง€๊ณ  CV์ถฉ์ „์€ ๊ธธ์–ด์ง€๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค. ์ด๋Š” ๋ฐฐํ„ฐ๋ฆฌ ์—ดํ™”์— ๋”ฐ๋ผ ๋ฐฉ์ „์šฉ๋Ÿ‰์€ ๊ฐ์†Œ๋˜๋Š” ๊ฒƒ์— ๋Œ€๋น„๋˜์–ด ์ถฉ์ „์šฉ๋Ÿ‰์˜ ๊ฒฝ์šฐ์—๋Š” ์—ดํ™”์— ์˜ํ•ด CC์ถฉ์ „์šฉ๋Ÿ‰์€ ์ฆ๊ฐ€ํ•˜๊ณ  CV์ถฉ์ „์šฉ๋Ÿ‰์€ ๊ฐ์†Œํ•˜๋Š” ๊ฒƒ์„ ์˜๋ฏธํ•œ๋‹ค. CV์ถฉ์ „์šฉ๋Ÿ‰์˜ ์ฆ๊ฐ€๋Š” ์ž”์กด์šฉ๋Ÿ‰ ๋ฐ ๋ฐฐํ„ฐ๋ฆฌ ์ถœ๋ ฅ์˜ ์†์‹ค๊ณผ ๋Œ€์‘๋˜๋ฉฐ, ๋‚ด๋ถ€ ๋ถˆ๊ท ํ˜•์œผ๋กœ ์ธํ•œ ๋†’์€ ๋ถ„๊ทนํ˜„์ƒ์„ ์•ผ๊ธฐํ•˜์—ฌ ์šฉ๋Ÿ‰ ๊ฐ์†Œ๋ฅผ ๊ฐ€์†์‹œํ‚ค๋Š” ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ์ด๋‹ค. ๋˜ํ•œ CV์ถฉ์ „์šฉ๋Ÿ‰์˜ ์ฆ๊ฐ€๋Š” CV ์ถฉ์ „์‹œ๊ฐ„์œผ๋กœ ๋Œ€๋ณ€ํ•˜์—ฌ ๊ณ„์‚ฐ๋Ÿ‰์„ ์ค„์ผ ์ˆ˜ ์žˆ์œผ๋ฏ€๋กœ ์—ดํ™” ํŒ๋‹จ ํŒŒ๋ผ๋ฏธํ„ฐ๋กœ ์‚ฌ์šฉ ๊ฐ€๋Šฅํ•˜๋‹ค(12).

๊ทธ๋ฆผ. 5. ์—ดํ™”์— ๋”ฐ๋ฅธ CC-CV ์ถฉ์ „์‹œ๊ฐ„์˜ ๋ณ€ํ™”

Fig. 5. Variation of CC-CV charging time with aging

../../Resources/kiee/KIEE.2019.68.10.1214/fig5.png

์ด์™€ ๊ฐ™์€ ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ ๋ฐ ์ž”์กด ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ์ด 200์‚ฌ์ดํด์˜ ์ž๋™์ฐจ ์ฃผํ–‰ ์—ดํ™” ์‹คํ—˜๋™์•ˆ 10์‚ฌ์ดํด๋งˆ๋‹ค ์šฉ๋Ÿ‰์‹คํ—˜์„ ํ†ตํ•ด ๊ณ„์‚ฐํ•˜์—ฌ ๊ทธ๋ฆผ 6์— ๋‚˜ํƒ€๋‚ด์—ˆ๋‹ค. (a)-(d)๋Š” ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ(์ง๋ ฌ์ €ํ•ญ, ๋ถ„๊ทน์ €ํ•ญ, CV ์ถฉ์ „์‹œ๊ฐ„), (e)๋Š” ์ž”์กด ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ๊ทธ๋ž˜ํ”„๋กœ ๋„์‹œํ•œ ๊ฒฐ๊ณผ ๊ทผ์‚ฌํ•œ ๊ฒฝํ–ฅ์„ฑ์„ ๊ฐ–๋Š”๋‹ค.

(4)
ฮฅ=SXYSXSY=โˆ‘ni=1(Xiโˆ’ห‰X)(Yiโˆ’ห‰Y)โˆšโˆ‘ni=1(Xiโˆ’ห‰X)2(Yiโˆ’ห‰Y)2

์—ดํ™” ํŒ๋‹จ ํŒŒ๋ผ๋ฏธํ„ฐ์™€ SOH์˜ ๋†’์€ ์ƒ๊ด€๊ด€๊ณ„๋Š” SOH์™€์˜ ์ง์ ‘์ ์ธ ๊ด€๊ณ„๋ฅผ ๋‚˜ํƒ€๋‚ด๋ฉฐ, ์ธ๊ณต์ง€๋Šฅ ๊ธฐ๋ฐ˜ ๋ชจ๋ธ์— ์–‘์งˆ์˜ ํ•™์Šต๋ฐ์ดํ„ฐ๋กœ์จ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋Š” ์ค‘์š”ํ•œ ์š”์†Œ์ด๋‹ค. ์ƒ๊ด€๋ถ„์„์€ ํ†ต๊ณ„์ ์œผ๋กœ ๋ณ€์ˆ˜๋“ค ์‚ฌ์ด์˜ ์„ ํ˜•๊ด€๊ณ„๋ฅผ ๋ถ„์„ํ•œ ๊ฒƒ์ด๋‹ค. ๋ณ€์ˆ˜๋“ค์€ ๋…๋ฆฝ์ ์ธ ๊ด€๊ณ„๋กœ ์„œ๋กœ ๊ด€๋ จ ์žˆ์„ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๋ณ€์ˆ˜๋“ค ์‚ฌ์ด์˜ ์ƒ๊ด€๊ด€๊ณ„์˜ ํฌ๊ธฐ๋ฅผ ์ƒ๊ด€๊ณ„์ˆ˜๋ผ๊ณ  ๋ถ€๋ฅธ๋‹ค. ์‹(4)๋Š” ํ”ผ์–ด์Šจ ์ƒ๊ด€๊ณ„์ˆ˜์— ๋”ฐ๋ผ ๊ณ„์‚ฐํ•˜๋ฉด ๊ณ„์ˆ˜์˜ ํฌ๊ธฐ๋Š” -1๋ถ€ํ„ฐ 1๊นŒ์ง€ ๋ฐ์ดํ„ฐ ์ฆ๊ฐ€, ๊ฐ์†Œ๊ด€๊ณ„์— ๋”ฐ๋ผ ์–‘์ˆ˜์™€ ์Œ์ˆ˜๋กœ ๋‚˜ํƒ€๋‚˜๋ฉฐ, ์ƒ๊ด€๊ณ„์ˆ˜์˜ ์ ˆ๋Œ€๊ฐ’์ด ํด์ˆ˜๋ก ๋‘ ๋ณ€์ˆ˜๋“ค ์‚ฌ์ด์˜ ์—ฐ๊ด€์„ฑ์ด ๋†’๋‹ค๋Š” ๊ฒƒ์„ ์˜๋ฏธํ•œ๋‹ค(13). ํ‘œ 2๋Š” SOH ์˜ˆ์ธก์„ ์œ„ํ•œ ์ž”์กด์šฉ๋Ÿ‰๊ณผ ์—ดํ™” ํŒŒ๋ผ๋ฏธํ„ฐ 4๊ฐœ ์‚ฌ์ด์˜ ์ƒ๊ด€๊ณ„์ˆ˜์ด๋‹ค, ์ „์ฒด์ ์œผ๋กœ 0.9 ์ด์ƒ์˜ ๋†’์€ ์ƒ๊ด€๊ณ„์ˆ˜๋ฅผ ๊ฐ–์œผ๋ฉฐ, CV ์ถฉ์ „์‹œ๊ฐ„์ด ๊ฐ€์žฅ ๋†’์€ ์ƒ๊ด€ ๊ณ„์ˆ˜๋ฅผ ๊ฐ–๋Š” ๊ฒƒ์œผ๋กœ ๋‚˜ํƒ€๋‚ฌ๋‹ค.

๊ทธ๋ฆผ. 6. ์—ดํ™” ํŒŒ๋ผ๋ฏธํ„ฐ (a) ์ง๋ ฌ์ €ํ•ญ (b) 1s ๋ถ„๊ทน์ €ํ•ญ (c) 60s ๋ถ„๊ทน์ €ํ•ญ (d) CV ์ถฉ์ „์‹œ๊ฐ„ (e) ๋ฐฉ์ „์šฉ๋Ÿ‰

Fig. 6. Aging parameter (a) Series resistance (b) 1s polarization resistance (c) 60s polarization resistance (d) CV charging time (e) discharge capacity

../../Resources/kiee/KIEE.2019.68.10.1214/fig6a.png../../Resources/kiee/KIEE.2019.68.10.1214/fig6b.png

ํ‘œ 2. ์ž”์กด์šฉ๋Ÿ‰๊ณผ ์—ดํ™” ํŒŒ๋ผ๋ฏธํ„ฐ ์‚ฌ์ด์˜ ์ƒ๊ด€๊ณ„์ˆ˜

Table 2. Correlation coefficient between residual capacity and aging parameter

Series resistance

1s polarization resistance

60s polarization resistance

CV charging time

Disgharge Capacity

-0.917

-0.944

-0.994

-0.996

3. ์ธ๊ณต์‹ ๊ฒฝ๋ง ๋ชจ๋ธ

๋ณธ ๋…ผ๋ฌธ์—์„œ๋Š” ๋น„์„ ํ˜•์ ์ธ SOH ์˜ˆ์ธก์— ๊ฐ•์ ์„ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ์ธ๊ณต์‹ ๊ฒฝ๋ง ๋ชจ๋ธ์„ ์„ ์ •ํ•˜์˜€์œผ๋ฉฐ, ์‹œ๊ณ„์—ด ๋ฐ์ดํ„ฐ ์ถ”์ • ๋ฐ ์˜ˆ์ธก์— ๊ฐ€์žฅ ์ตœ์ ํ™”๋˜์–ด ์ฃผ๋กœ ์ž์—ฐ์–ด ์ฒ˜๋ฆฌ์— ์‚ฌ์šฉ๋˜๋Š” RNN ๋ชจ๋ธ์„ ์ ‘๋ชฉํ•˜์—ฌ SOH๋ฅผ ์˜ˆ์ธกํ•˜์˜€๋‹ค. RNN์€ ๊ธฐ๋ณธ์ ์ธ ๋‹ค์ธต ํผ์…‰ํŠธ๋ก (multi-layer perceptron)์€ ๋ชจ๋“  ์…€ ๊ฐ„์˜ ์ž…์ถœ๋ ฅ์ด ๋…๋ฆฝ์ ์ธ ๋ฐ˜๋ฉด ๊ทธ๋ฆผ 7์—์„œ์™€ ๊ฐ™์ด ์ด์ „ ์ถœ๋ ฅ ๊ฒฐ๊ณผ์— ์˜ํ–ฅ์„ ๋ฐ›์Œ์œผ๋กœ์จ

๊ทธ๋ฆผ. 7. RNN ๊ตฌ์กฐ

Fig. 7. Architecture of the RNN

../../Resources/kiee/KIEE.2019.68.10.1214/fig7.png

๊ทธ๋ฆผ. 8. LSTM ๊ตฌ์กฐ

Fig. 8. Architecture of the LSTM

../../Resources/kiee/KIEE.2019.68.10.1214/fig8.png

๊ณผ๊ฑฐ์˜ ๋ฐ์ดํ„ฐ๊ฐ€ ๋ฏธ๋ž˜์— ์˜ํ–ฅ์„ ์ค„ ์ˆ˜ ์žˆ๋Š” ๊ตฌ์กฐ์ด๋‹ค. ํ•˜์ง€๋งŒ ์—ฐ์†ํ•ด์„œ ์ด์–ด์ง€๋Š” ์ฒด์ธํ˜•์‹ ๋ชจ๋ธ ํŠน์ง•์ƒ ๋ฐ์ดํ„ฐ๊ฐ€ ๊ธธ์–ด์งˆ์ˆ˜๋ก ๊ณผ๊ฑฐ์˜ ํ•™์Šต๊ฒฐ๊ณผ๊ฐ€ ์‚ฌ๋ผ์ง€๋Š” vanishing gradient ๋ฌธ์ œ๊ฐ€ ๋ฐœ์ƒํ•  ๊ฐ€๋Šฅ์„ฑ์ด ์žˆ๋‹ค.

๋”ฐ๋ผ์„œ ๋ณธ ์—ฐ๊ตฌ์—์„œ๋Š” vanishing gradient ๋ฌธ์ œ์˜ ํ•ด๊ฒฐ ๋ฐฉ์•ˆ์œผ๋กœ ์ œ์‹œ๋œ LSTM(long short term memory)๋ชจ๋ธ์„ ํ†ตํ•ด ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ์ƒ์Šน์‹œ์ผฐ์œผ๋ฉฐ. RNN๊ณผ์˜ ์„ฑ๋Šฅ์„ ๋น„๊ตํ•˜์˜€๋‹ค. ๊ทธ๋ฆผ 8์˜ LSTM ๊ตฌ์กฐ๋Š” ๊ธฐ์กด RNN๋ณด๋‹ค ๋ณต์žกํ•œ ๋‚ด๋ถ€๊ตฌ์กฐ๋ฅผ ๊ฐ€์ง€๊ณ  ์žˆ๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ์œผ๋ฉฐ, ์ด 4๊ฐœ์˜ ๋ ˆ์ด์–ด๋กœ ๊ตฌ๋ถ„๋œ๋‹ค. ์ฒซ ๋ฒˆ์งธ๋กœ forget gate layer๋Š” ์ด์ „ ์ถœ๋ ฅ๊ณผ ํ˜„์žฌ์˜ ์ž…๋ ฅ์„ sigmoid ํ•จ์ˆ˜๋ฅผ ํ†ตํ•ด ์‚ญ์ œํ•˜๊ฑฐ๋‚˜ ๋ณด๊ด€ํ•  ์ •๋ณด๋ฅผ ๊ฒฐ์ •ํ•œ๋‹ค. ๋‘ ๋ฒˆ์งธ๋กœ input gate layer๋Š” ์ด์ „ ์ถœ๋ ฅ๊ณผ ํ˜„์žฌ์˜ ์ž…๋ ฅ์„ sigmoid ํ•จ์ˆ˜์™€ tanh ํ•จ์ˆ˜๋ฅผ ํ†ตํ•œ ๊ฒฐ๊ณผ ๊ฐ’์˜ ๊ณฑ์œผ๋กœ ์…€ ์Šคํ…Œ์ดํŠธ์— ์ €์žฅ๋  ๋ฐ์ดํ„ฐ๋ฅผ ๊ฒฐ์ •ํ•œ๋‹ค. ์„ธ ๋ฒˆ์งธ๋กœ forget gate layer์™€ input gate layer์—์„œ ์–ป์€ ์ •๋ณด๋ฅผ ํ† ๋Œ€๋กœ ์…€ ์Šคํ…Œ์ดํŠธ๋ฅผ ์—…๋ฐ์ดํŠธํ•œ๋‹ค. ๋„ค ๋ฒˆ์งธ๋กœ output gate layer์—์„œ๋Š” ์ด์ „ ์ถœ๋ ฅ๊ณผ ํ˜„์žฌ์˜ ์ž…๋ ฅ์˜ sigmoid ํ•จ์ˆ˜ ๊ฒฐ๊ณผ ๊ฐ’๊ณผ ์—…๋ฐ์ดํŠธ๋œ ์…€ ์Šคํ…Œ์ดํŠธ์˜ tanh ํ•จ์ˆ˜ ๊ฒฐ๊ณผ ๊ฐ’์˜ ๊ณฑ์œผ๋กœ ์ตœ์ข… ์ถœ๋ ฅ ๊ฐ’์„ ๊ฒฐ์ •ํ•œ๋‹ค(14).

4. ์‹คํ—˜ ๋ฐ ๊ฒฐ๊ณผ

์ธ๊ณต์‹ ๊ฒฝ๋ง ๊ธฐ๋ฐ˜ ๋ชจ๋ธ์€ ๋†’์€ ์ถ”์ • ๋ฐ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ๊ธฐ๋Œ€ํ•  ์ˆ˜ ์žˆ์ง€๋งŒ, ํ•™์Šต๋ฐ์ดํ„ฐ์˜ ์–‘๊ณผ ์งˆ์— ๋งŽ์€ ์˜ํ–ฅ์„ ๋ฐ›๋Š”๋‹ค. ๋”ฐ๋ผ์„œ ๋ณธ ๋…ผ๋ฌธ์˜ ๋ฐ์ดํ„ฐ ์…‹์€ ์ด 400 ์‚ฌ์ดํด์˜ ์—ดํ™”์‹คํ—˜ ๋ถ„์„์„ ํ†ตํ•ด 10 ์‚ฌ์ดํด ์ดํ›„ ์ง„ํ–‰๋œ ์šฉ๋Ÿ‰ ์‹คํ—˜์—์„œ ๋†’์€ ์ƒ๊ด€๊ด€๊ณ„์˜ ์—ดํ™” ํŒ๋‹จ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ์ถ”์ถœํ•˜์—ฌ ๊ตฌ์„ฑํ•˜์˜€๋‹ค. ์ž…๋ ฅ๊ฐ’ ๋ฐ์ดํ„ฐ๋Š” 4๊ฐ€์ง€ ์—ดํ™” ํŒŒ๋ผ๋ฏธํ„ฐ(์ง๋ ฌ์ €ํ•ญ, 1s ๋ถ„๊ทน์ €ํ•ญ, 60s ๋ถ„๊ทน์ €ํ•ญ, CV ์ถฉ์ „์‹œ๊ฐ„)๋กœ ๊ฐ๊ฐ 40๊ฐœ์˜ ๋ฐ์ดํ„ฐ์™€ ์ถœ๋ ฅ ๊ฐ’ ๋ฐ์ดํ„ฐ๋Š” 40๊ฐœ์˜ ์ž”์กด ๋ฐฉ์ „์šฉ๋Ÿ‰์œผ๋กœ ํ•™์Šต ๋ฐ์ดํ„ฐ ์…‹์„ ๊ตฌ์„ฑํ•˜์˜€๋‹ค.

์„œ๋กœ ํฌ๊ธฐ์™€ ๋‹จ์œ„๊ฐ€ ๋‹ค๋ฅด๊ฒŒ ๊ตฌ์„ฑ๋˜์–ด์žˆ๋Š” ํ•™์Šต ๋ฐ์ดํ„ฐ ์…‹์˜ ํŒŒ๋ผ๋ฏธํ„ฐ ํŠน์ง•์ƒ ๋ชจ๋ธ์ด ํ•™์Šตํ•˜๋Š” ๊ณผ์ •์—์„œ ๊ฐ€์ค‘์น˜๊ฐ€ ๋‹ค๋ฅด๊ฒŒ ์ ์šฉ๋˜๋ฉฐ, ํ•™์Šต ์„ฑ๋Šฅ์„ ์ €ํ•˜ ์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค. ๋”ฐ๋ผ์„œ ์‹(5)์™€ ๊ฐ™์ด ์ •๊ทœํ™”์˜ ํ•œ ์ข…๋ฅ˜์ธ MinMaxScaler๋ฅผ ํ†ตํ•ด ํฌ๊ธฐ๋ฅผ ์ „๋ถ€ 0๋ถ€ํ„ฐ 1์‚ฌ์ด์˜ ๊ฐ’์œผ๋กœ ๋ณ€ํ™˜ ์ดํ›„ ํ•™์Šต์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค.

(5)
MinMaxScaler=Xโˆ’min(X)max(X)โˆ’min(X)

๋ณธ ์—ฐ๊ตฌ์˜ ์ธ๊ณต์‹ ๊ฒฝ๋ง ๋ชจ๋ธ์€ ๊ตฌ๊ธ€์—์„œ ์ œ๊ณตํ•˜๋Š” ์˜คํ”ˆ์†Œ์Šค์ธ ํ…์„œํ”Œ๋กœ์šฐ, ์ผ€๋ผ์Šค ํŒจํ‚ค์ง€๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ๊ตฌ์„ฑํ•˜์˜€๋‹ค. ์ œ์•ˆํ•˜๋Š” ์ธ๊ณต์‹ ๊ฒฝ๋ง ๋ชจ๋ธ์€ RNN๊ณผ LSTM ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ๋น„๊ตํ•˜์˜€๋‹ค. ๋‘ ๋ชจ๋ธ ์ „๋ถ€ ๋‹จ์ผ ์€๋‹‰์ธต ๊ตฌ์„ฑ์œผ๋กœ ๋ชจ๋ธ์˜ ๊ณผ์ ํ•ฉ์„ ๋ฐฉ์ง€ํ•˜๊ธฐ ์œ„ํ•ด ๋ฌด์ž‘์œ„๋กœ ๋…ธ๋“œ์˜ ์—ฐ๊ฒฐ์„ ๋Š์–ด์„œ ํ•™์Šตํ•˜๋Š” Dropout์„ ์ ์šฉํ•˜์˜€๋‹ค. ๋ชจ๋ธ์˜ ํ•™์Šต๊ณผ์ •์€ ์ •๋ฐฉํ–ฅ ํ•™์Šต๊ณผ ์—ญ๋ฐฉํ–ฅ ํ•™์Šต ๊ณผ์ •์„ ํ†ตํ‹€์–ด ์ „์ฒด ๋ฐ์ดํ„ฐ ์…‹์˜ ํ•™์Šต์™„๋ฃŒ ํšŸ์ˆ˜๋ฅผ ํ‘œํ˜„ํ•˜๋Š” Epoch์™€ ๊ฐ Epoch ๋งˆ๋‹ค ์ž…๋ ฅ๋˜๋Š” ๋ฐ์ดํ„ฐ ์…‹์˜ ํฌ๊ธฐ๋ฅผ ํ‘œํ˜„ํ•˜๋Š” Batch๋กœ ๋‚˜๋ˆŒ ์ˆ˜ ์žˆ์œผ๋ฉฐ, ๋ณธ ๋ชจ๋ธ์˜ ํŒŒ๋ผ๋ฏธํ„ฐ๋“ค์˜ ์กฐํ•ฉ์€ ํ‘œ 3์— ์ œ์‹œํ•˜์˜€๋‹ค.

ํ‘œ 1. SOH ์˜ˆ์ธก ๋ชจ๋ธ์˜ ํŒŒ๋ผ๋ฏธํ„ฐ ์กฐํ•ฉ

Table 1. Parameter combinations for SOH prediction model

Learning rate

Hidden neuron

Batch size

Epochs

Dropout

0.01

20

1

20

20%

RNN ๋ชจ๋ธ๊ณผ LSTM ๋ชจ๋ธ์„ ์‚ฌ์šฉํ•˜์—ฌ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ๋น„๊ตํ•˜์˜€์œผ๋ฉฐ, ์ด 400์‚ฌ์ดํด์˜ ์—ดํ™” ์‹คํ—˜์„ ํ†ตํ•œ ๋ฐ์ดํ„ฐ ์…‹์—์„œ ํ•™์Šต ๋ฒ”์œ„์— ๋”ฐ๋ผ ๋ชจ๋ธ ์„ฑ๋Šฅ์„ ๋น„๊ตํ•˜์˜€๋‹ค. ์ „์ฒด ๋ฐ์ดํ„ฐ ์…‹์˜ 50%์ธ 200์‚ฌ์ดํด์˜ ํ•™์Šต ์ดํ›„ ๋‚˜๋จธ์ง€ 200์‚ฌ์ดํด์— ๋Œ€ํ•œ ์˜ˆ์ธก ์„ฑ๋Šฅ๊ณผ ์ „์ฒด ๋ฐ์ดํ„ฐ ์…‹์˜ 80%์ธ 320์‚ฌ์ดํด์˜ ํ•™์Šต ์ดํ›„ ๋‚˜๋จธ์ง€ 80์‚ฌ์ดํด์— ๋Œ€ํ•œ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ๋น„๊ตํ•˜์˜€๋‹ค. ๊ทธ๋ฆผ 9์—์„œ RNN ๋ชจ๋ธ๊ณผ ๊ทธ๋ฆผ 10์—์„œ LSTM ๋ชจ๋ธ์˜ ํ•™์Šต๋ฒ”์œ„์— ๋”ฐ๋ฅธ ์˜ˆ์ธก ๊ฐ’๊ณผ ์‹ค์ œ ๊ฐ’์˜ ๊ทธ๋ž˜ํ”„๋ฅผ ๋‚˜ํƒ€๋‚ด์—ˆ๋‹ค.

๊ทธ๋ฆผ. 9. (a), (b) 200 ์‚ฌ์ดํด ํ•™์Šต RNN ๋ชจ๋ธ ์˜ˆ์ธก ๊ฐ’ ๊ทธ๋ž˜ํ”„ (c), (d) 320 ์‚ฌ์ดํด ํ•™์Šต RNN ๋ชจ๋ธ ์˜ˆ์ธก ๊ฐ’ ๊ทธ๋ž˜ํ”„

Fig. 9. (a), (b) 200 cycle training RNN model prediction graph (c), (d) 320 cycle training RNN model prediction graph

../../Resources/kiee/KIEE.2019.68.10.1214/fig9a.png../../Resources/kiee/KIEE.2019.68.10.1214/fig9b.png

(6)
RMSE =โˆš1nโˆ‘ni=1e2i

๋ชจ๋ธ์˜ ์˜ˆ์ธก ์„ฑ๋Šฅ์€ ํ‰๊ท  ์ œ๊ณฑ๊ทผ ์˜ค์ฐจ์œจ(root mean square error: RMSE)์œผ๋กœ ๊ณ„์‚ฐํ•˜์—ฌ ํ‘œ 4์—์„œ ์ˆ˜์น˜๋กœ ๋‚˜ํƒ€๋‚ด์—ˆ๋‹ค. ํ‰๊ท  ์ œ๊ณฑ๊ทผ ์˜ค์ฐจ์œจ์€ ์˜ˆ์ธก๊ฐ’๊ณผ ์‹ค์ œ๊ฐ’์˜ ์ฐจ์ด๋ฅผ ์‹(6)์— ๋”ฐ๋ผ ๊ณ„์‚ฐํ•˜์—ฌ ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ์„ ํŒ๋‹จํ•˜๋Š” ์ง€ํ‘œ๋กœ ์‚ฌ์šฉ๋œ๋‹ค. ์ด์— ๋”ฐ๋ฅธ ๋ชจ๋ธ ์„ฑ๋Šฅ ๋น„๊ต ๊ฒฐ๊ณผ, RNN ๋ชจ๋ธ๋ณด๋‹ค LSTM๋ชจ๋ธ์ด ์˜ˆ์ธก ์„ฑ๋Šฅ์ด ๋›ฐ์–ด๋‚˜๋ฉฐ, ํ•™์Šต๋ฒ”์œ„๊ฐ€ ๋Š˜์–ด๋‚ ์ˆ˜๋ก ์˜ˆ์ธก ์„ฑ๋Šฅ์€ ์ฆ๊ฐ€ํ•˜๋Š” ๊ฒƒ์„ ํ™•์ธํ•  ์ˆ˜ ์žˆ๋‹ค.

๊ทธ๋ฆผ. 10. (a), (b) 200 ์‚ฌ์ดํด ํ•™์Šต LSTM ๋ชจ๋ธ ์˜ˆ์ธก ๊ฐ’ ๊ทธ๋ž˜ํ”„ (c), (d) 320 ์‚ฌ์ดํด ํ•™์Šต LSTM ๋ชจ๋ธ ์˜ˆ์ธก ๊ฐ’ ๊ทธ๋ž˜ํ”„

Fig. 10. (a), (b) 200 cycle training LSTM model prediction graph (c), (d) 320 cycle training LSTM model prediction graph

../../Resources/kiee/KIEE.2019.68.10.1214/fig10a.png../../Resources/kiee/KIEE.2019.68.10.1214/fig10b.png

ํ‘œ 4. RNN ๋ชจ๋ธ๊ณผ LSTM ๋ชจ๋ธ์˜ ํ•™์Šต๋ฒ”์œ„์— ๋”ฐ๋ฅธ ์„ฑ๋Šฅ ๋น„๊ต(RMSE)

Table 4. Comparison of performance by learning rates between RNN model and LSTM model(RMSE)

200 cycle training

(50% data)

320 cycle training

(80% data)

RNN model

0.0507

0.0179

LSTM model

0.0407

0.006

5. ๊ฒฐ ๋ก 

๋ณธ ์—ฐ๊ตฌ๋Š” ์‹ค์ œ ๊ณ ์ถœ๋ ฅ ์–ดํ”Œ๋ฆฌ์ผ€์ด์…˜์˜ ๋‹ค์–‘ํ•œ ํฌ๊ธฐ์˜ ์ถœ๋ ฅ๊ณผ ๊ทธ์— ๋”ฐ๋ฅธ ์—ดํ™” ๋ถ„์„์„ ์œ„ํ•ด UDDS ํ”„๋กœํŒŒ์ผ์„ ์‚ฌ์šฉํ•˜์—ฌ ์ด 400์‚ฌ์ดํด์˜ ์—ดํ™” ์‹คํ—˜์„ ์ง„ํ–‰ํ•˜์˜€๋‹ค. ์‹ค์ œ ์–ดํ”Œ๋ฆฌ์ผ€์ด์…˜์—์„œ SOH ์ถ”์ •์„ ์œ„ํ•œ ๋ฐฉ์ „์šฉ๋Ÿ‰ ์ธก์ •์€ ์–ด๋ ค์›€์ด ์žˆ์œผ๋ฉฐ, ๊ทธ์— ๋”ฐ๋ผ SOH์™€ ์ƒ๊ด€๊ด€๊ณ„๊ฐ€ ๋†’์€ ์—ดํ™” ํŒ๋‹จ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ํ†ตํ•ด ๋ฐฉ์ „์šฉ๋Ÿ‰์„ ์ถ”์ •ํ•˜๋Š” ๊ฒƒ์ด ์š”๊ตฌ๋œ๋‹ค. ๋”ฐ๋ผ์„œ ์ „๊ธฐ์  ํŠน์„ฑ ๊ธฐ๋ฐ˜ ์—ดํ™” ํŒŒ๋ผ๋ฏธํ„ฐ ๋ถ„์„์„ ํ†ตํ•ด ๊ณ„์‚ฐ์— ์šฉ์ดํ•˜๊ณ  ์ƒ๊ด€๊ณ„์ˆ˜๊ฐ€ ๋†’์€ ํŠน์„ฑ ํŒŒ๋ผ๋ฏธํ„ฐ๋ฅผ ์ถ”์ถœํ•˜๊ณ  ์ด๋ฅผ ํ†ตํ•ด SOH ์˜ˆ์ธก ๋ชจ๋ธ ๊ฒ€์ฆ ๋ฐ์ดํ„ฐ ์…‹์„ ๊ตฌ์„ฑํ•˜์˜€๋‹ค. ๋น„์„ ํ˜•์ ์œผ๋กœ ๋ณ€ํ™”ํ•˜๋Š” SOH๋ฅผ ์˜ˆ์ธกํ•˜๊ธฐ ์œ„ํ•˜์—ฌ ์‹œ๊ณ„์—ด ๋ฐ์ดํ„ฐ ์ถ”์ •์— ์ฃผ๋กœ ์‚ฌ์šฉ๋˜๋Š” RNN, LSTM ์ธ๊ณต์‹ ๊ฒฝ๋ง ๋ชจ๋ธ์„ ์ œ์•ˆํ•˜์˜€์œผ๋ฉฐ, ๊ตฌ๊ธ€์—์„œ ์ œ๊ณตํ•˜๋Š” ํ…์„œํ”Œ๋กœ์šฐ, ์ผ€๋ผ์Šค ํŒจํ‚ค์ง€๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ ๋ชจ๋ธ์„ ๊ตฌ์„ฑํ•˜์˜€๋‹ค. RNN๊ณผ LSTM ๋‘ ๋ชจ๋ธ์˜ ์˜ˆ์ธก ์„ฑ๋Šฅ๊ณผ ๋ฐ์ดํ„ฐ ์…‹์˜ ํ•™์Šต ๋ฒ”์œ„์— ๋”ฐ๋ผ ์˜ˆ์ธก ์„ฑ๋Šฅ์„ ๋น„๊ตํ•œ ๊ฒฐ๊ณผ ํ•™์Šต ๋ฒ”์œ„๊ฐ€ ์ฆ๊ฐ€ํ• ์ˆ˜๋ก ๋ชจ๋ธ ์„ฑ๋Šฅ์€ ์ฆ๊ฐ€ํ•˜๋ฉฐ, RNN ๋ชจ๋ธ์— ๋น„๊ตํ•˜์—ฌ LSTM ๋ชจ๋ธ์˜ ์„ฑ๋Šฅ์ด ๋” ๋›ฐ์–ด๋‚œ ๊ฒƒ์„ ๊ฒ€์ฆํ•˜์˜€๋‹ค.

Acknowledgements

์ด ์—ฐ๊ตฌ๋Š” ์ถฉ๋‚จ๋Œ€ํ•™๊ต ํ•™์ˆ ์—ฐ๊ตฌ๋น„์— ์˜ํ•ด ์ง€์›๋˜์—ˆ์Œ.

References

1 
A. Barrรฉ, B. Deguilhem, S. Grolleau, M. Gรฉrard, F. Suard, D. Riu, November 2013, A review on lithium-ion battery ageing mechanisms and estimations for automotive ap- plications, Journal of Power Sources, Vol. 241, pp. 680-689DOI
2 
T. Kodama, H. Sakaebe, September 1999, Present status and future prospect for national project on lithium batteries, Journal of Power Sources, Vol. 81, pp. 144-149DOI
3 
L. Su, J. Zhang, C. Wang, Y. Zhang, Z. Li, Y. Song, Z. Ma, February 2016, Identifying main factors of capacity fading in lithium ion cells using orthogonal design of experi- ments, Applied Energy, Vol. 163, pp. 201-210DOI
4 
X. Han, M. Ouyang, L. Lu, J. Li, Y. Zheng, Z. Li, April 2014, A comparative study of commercial lithium ion battery cycle life in electrical vehicle: Aging mechanism identi- fication, Journal of Power Sources, Vol. 251, pp. 38-54DOI
5 
M. Berecibar, I. Gandiaga, I. Villarreal, N. Omar, J. van Mierlo, P. van den Bossche, April 2016, Critical review of state of health estimation methods of Li-ion batteries for real applications, Renewable and Sustainable Energy Reviews, Vol. 56, pp. 572-587DOI
6 
H. Chaoui, C. C. Ibe-Ekeocha, June 2017, State of charge and state of health estimation for lithium batteries using re- current neural networks, IEEE Transactions on Vehicular Technology, Vol. 66, No. 10, pp. 8773-8783DOI
7 
A. Nuhic, T. Terzimehic, T. Soczka-Guth, M. Buchholz, K. Dietmayer, October 2013, Health diagnosis and remaining useful life prognostics of lithium-ion batteries using data-driven methods, Journal of Power Sources, Vol. 239, pp. 680-688DOI
8 
R. Xiong, L. Li, J. Tian, November 2018, Towards a smarter battery management system: a critical review on battery state of health monitoring methods, Journal of Power Sources, Vol. 405, pp. 18-29DOI
9 
Y. Zhang, C. Y. Wang, X. Tang, 2011, Cycling degradation of an automotive LiFePO4 lithium-ion battery, Journal of Power Sources, Vol. 196, No. 3, pp. 1513-1520DOI
10 
Y. Gao, J. Jiang, C. Zhang, W. Zhang, Y. Jiang, October 2018, Aging mechanisms under different state-of-charge ranges and the multi-indicators system of state-of-health for lithium-ion battery with Li (NiMnCo) O2 cathode, Journal of Power Sources, Vol. 400, pp. 641-651DOI
11 
A. S. Mussa, M. Klett, M. Behm, G. Lindbergh, R. W. Lindstrรถm, October 2017, Fast-charging to a partial state of charge in lithium-ion batteries: A comparative ageing study, Journal of Energy Storage, Vol. 13, pp. 325-333DOI
12 
S. S. Zhang, October 2006, The effect of the charging protocol on the cycle life of a Li-ion battery, Journal of Power Sources, Vol. 161, No. 2, pp. 1385-1391DOI
13 
K. Pearson, 1895, Notes on Regression and Inheritance in the Case of Two Parents, Proceedings of the Royal Society of London, Vol. 58, pp. 240-242Google Search
14 
Q. Lyu, J. Zhu, December 2014 publish- ed, Revisit long short-term memory:, in Advances in Neural Informa- tion Processing Systems Workshop on Deep Learning and Representation Learning, pp. 1-9DOI

์ €์ž์†Œ๊ฐœ

๊ถŒ์ƒ์šฑ (Sanguk Kwon)
../../Resources/kiee/KIEE.2019.68.10.1214/au1.png

1995๋…„ 4์›” 22์ผ์ƒ.

2014~ ํ˜„์žฌ ์ถฉ๋‚จ๋Œ€ํ•™๊ต ์ „๊ธฐ๊ณตํ•™๊ณผ ํ•™๋ถ€๊ณผ์ •

ํ•œ๋™ํ˜ธ (Dongho Han)
../../Resources/kiee/KIEE.2019.68.10.1214/au2.png

1993๋…„ 1์›” 27์ผ์ƒ.

2018๋…„ ์ถฉ๋‚จ๋Œ€ ์ „๊ธฐ๊ณตํ•™๊ณผ ์กธ์—….

2018๋…„๏ฝžํ˜„์žฌ ์ถฉ๋‚จ๋Œ€ ์ „๊ธฐ๊ณตํ•™๊ณผ ์„์‚ฌ๊ณผ์ •.

๋ฐ•์„ฑ์œค (Seongyun Park)
../../Resources/kiee/KIEE.2019.68.10.1214/au3.png

1991๋…„ 7์›” 10์ผ์ƒ.

2016๋…„ ํ•œ๊ตญ๊ธฐ์ˆ ๊ต์œก๋Œ€ ๋ฉ”์นดํŠธ๋กœ๋‹‰์Šค๊ณตํ•™๋ถ€ ์ƒ์‚ฐ์‹œ์Šคํ…œ์ „๊ณต ์กธ์—….

2018๋…„โˆผํ˜„์žฌ ์ถฉ๋‚จ๋Œ€ ์ „๊ธฐ๊ณตํ•™๊ณผ ์„์‚ฌ๊ณผ์ •.

๊น€์ข…ํ›ˆ (Jonghoon Kim)
../../Resources/kiee/KIEE.2019.68.10.1214/au4.png

1979๋…„ 4์›” 22์ผ์ƒ.

2005๋…„ ์ถฉ๋‚จ๋Œ€ ์ •๋ณดํ†ต์‹ ๊ณตํ•™๋ถ€ ์ „๊ธฐ์ „์ž์ „ํŒŒ์ „๊ณต ์กธ์—….

2012๋…„ ์„œ์šธ๋Œ€ ์ „๊ธฐ์ปดํ“จํ„ฐ๊ณตํ•™๋ถ€ ์กธ์—…(๊ณต๋ฐ•;์„๋ฐ•ํ†ตํ•ฉ๊ณผ์ •).

2012๋…„โˆผ2013๋…„ ์‚ผ์„ฑSDI ES์‚ฌ์—…๋ถ€ ์ฑ…์ž„์—ฐ๊ตฌ์›.

2013๋…„โˆผ2016๋…„ ์กฐ์„ ๋Œ€ ์ „๊ธฐ๊ณตํ•™๊ณผ ์กฐ๊ต์ˆ˜.

2016๋…„โˆผํ˜„์žฌ ์ถฉ๋‚จ๋Œ€ ์ „๊ธฐ๊ณตํ•™๊ณผ ์กฐ๊ต์ˆ˜.

2018๋…„โˆผํ˜„์žฌ ํ•œ๊ตญ๊ณผํ•™๊ธฐ์ˆ ์› ์นœํ™˜๊ฒฝ์„ผํ„ฐ์Šค๋งˆํŠธ์ž๋™์ฐจ์—ฐ๊ตฌ์„ผํ„ฐ ๊ฒธ์ง๊ต์ˆ˜.

2015๋…„โˆผํ˜„์žฌ JPE Associate Editor.

2016๋…„โˆผํ˜„์žฌ ๋‹น ํ•™ํšŒ ํŽธ์ง‘์œ„์›. 2017๋…„ ๋‹น ํ•™ํšŒ ํ•™์ˆ ์œ„์›.

2019๋…„โˆผํ˜„์žฌ IEEE Senior Member.