
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Author: 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Shengyu Tao Ruifei Ma Zixi Zhao Guangyuan Ma Lin Su Heng Chang Yuou Chen Haizhou Liu Zheng Liang Tingwei Cao Haocheng Ji Zhiyuan Han Minyan Lu Huixiong Yang Zongguo Wen Jianhua Yao Rong Yu Guodan Wei Yang Li Xuan Zhang Tingyang Xu Guangmin Zhou Panya ka bo kadi kava Ho lekana le ho leka-leka hantle maemo a bophelo (SOH) a libeteri tse sebelisitsoeng ke ntho ea bohlokoa bakeng sa ts'ebetso pele li sebelisoa hape kapa li sebelisoa hape. Leha ho le joalo, mekhoa e sebelisang data e hloka ho hlophisoa ha data ka botlalo tlas'a maemo a sa reroang a SOH le a state of charge (SOC). Mona, re bontša hore ho lekana ha SOH ho sebelisa generative learning ho ntse ho khona ho nolofatsa mathata a khaello le ho fapana ha data, e netefatsoa ka data ea lithapo tsa litaelo tse tsoang ho sampole tse 2700 tsa libeteri tsa lithium-ion tse sebelisitsoeng, tse kenyelletsang mefuta e 3 ea lintho tsa cathode, lifomate tse 3 tsa 'mele, meralo ea matla a 4, le ts'ebeliso ea nalane ea 4 e nang le maemo a 10 a SOC. Ho sebelisa data e hlahisoeng, regressor e fihlela ho lekana hantle ha SOH, ka karolelano e fosahetseng ea karolelano e ka tlase ho 6% tlas'a SOC e sa bonoeng. Re bolela esale pele hore haeba mokhoa o hlahisitsoeng o sebelisoa ka ho lekana, o tla boloka litšenyehelo tsa motlakase tse libilione tse 4.9 tsa liranta le mesi ea CO2 ea liranta tse libilione tse 35.8 ka ho fokotsa litšenyehelo tsa ho hlophisoa ha data bakeng sa boemo ba 2030 ba libeteri tsa lefatše. Sengoloa sena se totobatsa ho sebelisa data e lekanyelitsoeng bakeng sa ho hlahloba sebaka se eketsehileng sa data se sebelisa mekhoa ea generative, ha ho fuoa hore data e ka ba nako e telele, ea theko e boima ebile e silafatsa ho fumana bakeng sa mesebetsi e mengata ea ho lekana le ea ho bolela esale pele. Selelekela Ho fetoha ha lipalangoang, karolo ea bohlokoa phetohong ea matla a theko e tlase ea carbon , esale e fetola mokhoa oa lipalangoang o thehiloeng ho khase ea oli ka tsela e sa tloaelehang. Libeteri tsa lithium-ion li bohlokoa ho phetoho ena ka ho fana ka matla a theko e tlaase, a tšoarellang, le a sireletsehileng ho likoloi tsa motlakase (EVs) , . Matla a akaretsang a libeteri tsa EV tse teng le tse sebelisitsoeng li tla feta 32–62 tera-watts ka selemo sa 2050, ho phahama haholo ho feta teka-tekano ho International Renewable Energy Agency le Likamoreng tsa Polokelo . Leha ho le joalo, libeteri tsa EV tse sebelisitsoeng li laoloa haholo ka boleng bo boholo ba matla a setseng bo sa sebelisoeng hantle, hangata bo feta 80% ea matla a thathamisitsoeng , . Haeba li sa sebetse hantle, li lebisa litšenyehelong tsa moruo  bakeng sa baetsi le basebelisi, hammoho le litaba tse latelang tsa tikoloho le sechaba , ho kenyelletsa tahlehelo ea lisebelisoa, likotsi tsa phepelo ea thepa, le mesi ea carbon , , . 1 2 3 4 5 6 7 8 9 10 11 Mekhoa e tšepisang ea ho rarolla litaba mabapi le ho phahama hoa libeteri tse sebelisitsoeng ke ho li sebelisa hape , ,  le ho li sebelisa hape . Boemong ba ho li sebelisa hape, libeteri tse sebelisitsoeng li fetisetsoa litšebeletsong tse kang polokelo ea matla e hokahane le marang-rang , lisebelisoa tsa motlakase tsa bolulo , le likoloi tse lebelo le tlase. Leha ho le joalo, ho li sebelisa hape ho hloka litokiso tsa pele, ho kenyelletsa ho hlahloba ho ts'oana , ho hlophisa matla a , le ho li kopanya hape ,  ho fihlela litlhoko tsa tšebeliso . Ho sa tsotellehe ho qaleha ha mananeo a teko , , , ho hloka maemo a thata bakeng sa maemo a tšebeliso le litsela tsa ho sebelisa hape ho sitisa tšebeliso e utloahalang ea matla a setseng . Ho sebelisa hape ho sebelisa litefiso tse setseng tsa libeteri tse sebelisitsoeng ka ho ntša lintho kapa ho lokisa mehaho , , e loketseng libeteri tse sebelisitsoeng tse senyehileng ka ho sa feleng. Ha e bapisoa le pyrometallurgy le hydrometallurgy , , ho sebelisa hape ka ho toba ho hlahisa phaello e phahameng, ts'ebeliso e tlase ea matla, le mesi e fokolang ea carbon , , . Ho sebelisa ho tsosolosoa ha lithium le ho hlophisoa ha lintho tsa cathode, ho sebelisa hape ka ho toba ho fihlela ho lokisoa ha meaho ea lintho ka katleho, le ho tsosolosoa ha ts'ebetso . Leha ho le joalo, boemo ba bophelo (SOH) bo khetholla li-reagent tsa lik'hemik'hale tse hlokahalang le litefiso tse lebelletsoeng tsa ho eketsa lithium maemong a leano la ho sebelisa hape ka ho toba , , . Ho hloka litefiso ho ka lebisa ho lokisoeng ho sa feleng, ha litefiso tse feteletseng li ka lebisa ho hlahisoeng ha alkali e setseng holim'a cathode, ho senya ts'ebetso ea ho tsosolosoa , . Ka bomalimabe, ho fumana SOH ho hloka ho hlahlojoa ha lintho ka tsela e senyang ,  le liteko tse telele tsa ho lekanya matla, ho fumana SOH ka tsela e sa senyeng, ka lebelo, le ka ho ts'oarellang ho ntse ho le mathata.  6 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 26 27 30 31 32 33 34 27 35 36 37 Ho netefatsa tlhaiso-leseling e ts'oanang ea SOH, libeteri tsa EV li hlahlojoa, 'me data ea ts'ebetso e ngoloa ke li-platform tsa leru  le li-passport tsa beteri . European Parliament e amohetse Batteries Directive ka la 14 Phuptjane 2023, ho netefatsa hore libeteri tse sebelisitsoeng li ka sebelisoa hape, tsa lokisoa, kapa tsa sebelisoa hape qetellong ea bophelo ba tšebeliso . Leha ho le joalo, mokhoa oa ho hlahlojoa le ho ngoloa o hlokoang o fumaneha feela nakong ea tšebeliso ea EV, libeteri tse sebelisitsoeng ha li sa hokahane le yuniti ea ho hlahlojoa ka hare ho koloi. Ka hona, ho thata ho fumana data ea SOH e fumanehang sebakeng. Botšepehi ba data ea bophelo bohle bo ntse bo le mathata a maholo, bo kopa ho lekana ha SOH ho sebelisa feela data ea sebaka, ho fapana le data ea nalane kapa maemong a ka laolehang , , . Tharollo e 'ngoe ke ho etsa liteko tsa ho lekanya matla sebakeng sa pokello ea libeteri tse sebelisitsoeng, ke hore, ho lokisa boemo ba motlakase (SOC), e leng ntho e bonolo, leha ho le joalo, e hloka nako e sa theko e tlase ea liteko le litšenyehelo tse eketsehileng tsa motlakase. Liteko tsa ts'ebetso ea matla a hybrid ke mokhoa o mong oa ho lekana ha SOH ka ho kenya li-pulse tsa dynamic, empa tatellano e felletseng ea liteko e nka lihora tse fetang 12. Tao le ba bang ba sebelisitse li-pulse tse khuts'oane bakeng sa ho lekana ha SOH ho libeteri tse sebelisitsoeng, ba nka hore ho etsoa phetoho ea SOC ho ea ho 5% SOC, eo ho seng bonolo ho e hokahanya le maemo a sa reroang a ho sebelisa hape SOC . Litholoana tsa morao tjena tse fumanehang ka ho lekola ka theknoloji li kenyelletsa X-ray imaging , impedance ea motlakase , , sensing ea optical fiber , , acoustic sensing , , ho tjhaja karoloana , , le ho kenya li-pulse , . Mekhoa e thehiloeng ho sensing e boemong ba lab ka lebaka la ho senya, ha ho kenyeletsoa ha li-pulse ho fapana le ho senya beteri ntle le tšenyo ea 'mele, ka nako e le 'ngoe ho feta ho tjhaja karoloana le mekhoa e thehiloeng ho impedance ea motlakase , , , , , . Leha ho le joalo, ho kenyeletsoa ha li-pulse ke mokhoa o tsepamisitsoeng maikutlo ho data o sebetsang feela tlas'a teka-tekano e loketseng ea hore data e lekantsoeng ka 'mele e akaretsa maemo a ho sebelisoa hape a nako ea ts'ebetso ea mohlala, e tsejoang e le mathata a tloaelehileng a domain shift sechabeng sa machine learning. Ho sa tsotellehe litholoana tse tsoelang pele ho ithuteng ka transfer le ho ikamahanya le domain , , , mathata a ntse a le teng kaha domain ea sepheo ea ho hokahanya e ntse e hloka tlhaiso-leseling pele ha e sebelisoa, 'me mekhoa ea hajoale ea ho ithuta e ka thatafalloa ho ntlafatsoa bakeng sa maemo a sa reroang a ho sebelisa hape . Mokhoa o mong ke ho otlolla tekanyo ea liteko tsa data tlas'a maemo a sa hlahlojoang a ho sebelisoa hape ho nolofatsa khafello le ho fapana ha data ha e ntse e ka lebisa ho eketseho ea litšenyehelo. Wang le ba bang ba ile ba hlahisa mokhoa oa mocheso oa ho epela ho kopanya likhetho tsa nako ea ketso ea k'hemik'hale maemong a fapaneng a lipakeng nakong ea ts'ebetso ea thermochemical, ho lumella ho thehoa ha database tse ngata tsa koetliso ka liteko tse fokolang tsa thermochemistry le litšenyehelo joalo ka augmentation ea data bakeng sa koetliso ea mohlala oa machine learning . Ntle ho moo, generative learning e boetse e bontša menyetla ea ho lekanya SOH ka data e eketsehileng e tsoang ho li-profile tse kentsoeng ka karoloana, ho boloka data e hlokahalang e lekiloeng ka 'mele , . Leha ho le joalo, khafello le phapano ea data maemong a ho sebelisa hape le ho sebelisa libeteri li rarahane le ho feta ka lebaka la motsoako oa mefuta ea lintho tsa cathode, lifomate tsa 'mele, meralo ea matla, le tšebeliso ea nalane, ho fokotsa phapanyetsano ea tsebo ea boemo bo ikhethang ba tšenyo ho mekhoeng ea machine learning. Ka hona, ho ka utloahala ho nahana ka kamano ea ho itšetleha lipakeng tsa maemo a ho sebelisa hape le likarabo tsa pulse voltage bakeng sa ho hlahisoa ha data, ho sa fane ka liteko tse telele tsa nako ea ho lekanya matla le liteko tse eketsehileng tsa 'mele ka nako e le ngoe, ho matlafatsa ts'ebetso ea ho lekana ha SOH. 38 24 39 30 40 41 41 42 43 44 45 46 47 48 49 50 17 41 17 51 52 53 54 55 56 57 58 59 60 61 62 Thutong ena, re etsa phetoho ea pele ea beteri e sebelisitsoeng, ke hore, ho lekana ha SOH, ho sebelisa data ea karabelo ea pulse voltage e hlahisitsoeng ke mofuta o hlokang tlhokomelo ka transformer ho boloka nako ea liteko tsa 'mele le litšenyehelo. Mokhoa o phahameng oa mehato e akaretsang ea phetoho, koetliso ea mohlala oa generative, le mehato ea ts'ebetso ea mohlala li bontšitsoe ho Setšoantšo.  . Khopolo ea lipatlisiso ke ho sebelisa generative learning ho sebelisa data e seng e lekantsoe bakeng sa ho hlahlojoa ha likarabo tsa pulse voltage maemong a sa reroang a ho sebelisoa hape, ke hore, kabo ea SOC sebakeng sa pokello ea libeteri tse sebelisitsoeng. Ho lekana ha SOH ho latelang ho etsoa ntle le data e lekiloeng ka 'mele ea karabelo ea pulse, ho boloka litšenyehelo tsa motlakase tsa phetoho le mesi ea carbon e neng e tla hlaha ho tsoa ho lokisoeng ha SOC ho ea boemong bo lebelletsoeng. Ha e bapisoa le mokhoa oa ho kopanya ampere-hour le mohlala oa machine-learning pulse injection, generative learning e totobatsa lebelo le lebelo la ho hlahisa le ho lekana hantle. Ka data e hlahisitsoeng, mohlala o bonolo oa regressor o atleha ho lekanya SOH ka phoso e tlase tlas'a maemo a sa bonoeng kapa a sa fumaneheng a ho sebelisoa hape a SOC database, mathata a tsoang kantle ho kabo (OOD) a thata. Mathata a OOD a rarolloa ka ho ithuta mokhoa oa 'mele o kopanetsweng lipakeng tsa maemo a SOC le karabelo ea motlakase e bakiloeng ke pulse e potlakileng. Re bontša kakaretso ea mokhoa oa rona ka ho netefatsa ka sampole tse 2700 tse lekiloeng ka 'mele, tse akaretsang mefuta e 3 ea lintho tsa cathode, lifomate tse 3 tsa 'mele, meralo ea matla a 4, le tšebeliso ea nalane ea 4 (1 teko ea lab e potlakileng, 1200 kg CO2 kg ea sephethe-phethe sa EV, le 2 tsa sephethe-phethe sa EV). Generative learning e loketse ho kopanya le ho atolosa ka ho kopanya tsebo ea pele ea maemo a ho sebelisoa hape ho kabo ea data ea mohlala oa latent space, ho boloka nako ea ho lekanya, litšenyehelo tsa motlakase, le mesi ea carbon e latelang ha ho ntse ho sa senyehe ts'ebetso ea ho lekana ha SOH. Teko ea lefatše ea conservative ea ho sebelisoa ha libeteri ka 2030 e totobatsa bohlokoa ba ho lekana ha SOH ka lebelo le tšoarellang maemong a pyrometallurgical, hydrometallurgical, le direct recycling. Ka tlhahlobo ea theknoloji-moruo, ho lekana ha SOH ho thusoang ka generative learning ho ka boloka $4.9 billion litšenyehelong tsa motlakase le liraba tsa 35.8 billion kg CO2 mesele ka 2030 lefatšeng ka bophara. Re tšohla kutloisiso ea mohlala, litlamorao tsa phetoho ea ho sebelisoa hape, le maemo a pharaletseng a litharollo tsa nako e tlang tsa ho sebelisoa hape tse kopantsoeng le machine learning. 1    Mehato ea phetoho ea beteri e sebelisitsoeng e kenyelletsa ho lokisoa ha SOC, liteko tsa 'mele, machine learning bakeng sa ho fumana SOH, le ts'ebetso ea bobeli ea ho etsa qeto pele e sebelisoa hape kapa e sebelisoa hape. Ho lokisoa ha SOC ho bolela ho fetola litefiso tsa SOC tsa libeteri tse sebelisitsoeng ho ea maemong a hlokahalang ho sebelisoa mokhoa oa setso oa constant current constant voltage (CCCV).   Moaho oa mofuta o hlokang tlhokomelo ka transformer bakeng sa ho hlahisoa ha data ea karabelo ea pulse voltage.   Ts'ebetso ea mots'eare ea mohlala oa ho hlahisa data bakeng sa ho lekana ha SOH e hloka ho lokisoa ha SOC ho eketsehileng. Mohlala o koetlisitsoeng oa machine learning o tsamaisoa ke data e hlahisitsoeng 'me e amohela likarabo tsa pulse voltage hang-hang ha e le sebakeng sa ts'ebetso.   Melemo ea moruo le tikoloho le lebelo le lebelo la ho hlahisa la mokhoa oa rona oa generative learning ka ho boloka litšenyehelo tsa ho lokisoa ha SOC sebakeng sa ho sebelisoa ha libeteri tsa lefatše ka 2030. a b c d Liphetho Ho fumana data Khaello le phapano ea data ke mathata a mantlha ho machine-based battery diagnosis le prognosis, haholo-holo bakeng sa libeteri tse sebelisitsoeng. Re bokelletse sampole tse 2700 ho tsoa ho libeteri tse sebelisitsoeng ka 'mele tse 270 maemong a pharaletseng a SOC, ke hore, ho tloha ho 5% ho ea ho 50% ka seemelong sa 5%. Setšoantšong.  , data e akaretsa mefuta e 3 ea lintho tsa tsohle tse tsebahalang tsa cathode, ho kenyelletsa NMC (nickel manganese cobalt oxide) e 119 e bokelletsoeng ka 'mele, LFP (lithium iron phosphate) e 56, le LMO (lithium manganese oxide) e 95 tse sebelisitsoeng, tse akaretsang lifomate tse 3 tsa 'mele (cylindric, pouch, prismatic) le mekhoa ea tšebeliso ea nalane e 4 (1 teko ea lab e potlakileng, 3 tsa EV, ho kenyelletsa 1 ka mokhoa oa pure electrified power-train le 2 tsa hybrid power-train modes). Re hatisa hore ho kenyelletsoa ha li-sampole tse fapaneng tsa liteko ho etsa hore data ea rona e be data e kholo ka ho fetisisa ea pulse e potlakileng bakeng sa ho lekana ha SOH ho libeteri tse sebelisitsoeng ho fihlela joale, ho thusa kakaretso ea mohlala ho feta data e lekiloeng ka 'mele. 2a  

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

ສຽງນີ້ຖືກຜະລິດເປັນພາສາຕົ້ນສະບັບຂອງເລື່ອງ!

Mishimu AI i Mphamvu za Kubwezeretsanso Zogwiritsidwa Ntchito za Mabatire a Lithium-Ion Omwe Amapuma Pantchito

About Author

ຄຳເຫັນ

ວາງປ້າຍ

ບົດຄວາມນີ້ໄດ້ຖືກນໍາສະເຫນີໃນ

Related Stories

Microsoft's AI Bet Fails to Dent Google's Search Dominance

Design Shifts at HackerNoon: Streamlined Top Nav, Quick Search and Settings Dashboard 🚀

Starting 2025 with New Features: Settings Dashboard, HackerNoon Decoded, Updated Search UI, and More

10 Lessons I Learned as a First-time Tech Product Founder

Microsoft's AI Bet Fails to Dent Google's Search Dominance

Design Shifts at HackerNoon: Streamlined Top Nav, Quick Search and Settings Dashboard 🚀

Starting 2025 with New Features: Settings Dashboard, HackerNoon Decoded, Updated Search UI, and More

10 Lessons I Learned as a First-time Tech Product Founder

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps

Mishimu AI i Mphamvu za Kubwezeretsanso Zogwiritsidwa Ntchito za Mabatire a Lithium-Ion Omwe Amapuma Pantchito

About Author

ຄຳເຫັນ

ວາງປ້າຍ

ບົດ​ຄວາມ​ນີ້​ໄດ້​ຖືກ​ນໍາ​ສະ​ເຫນີ​ໃນ

Related Stories

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps

ບົດຄວາມນີ້ໄດ້ຖືກນໍາສະເຫນີໃນ