
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Abagoliga: 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 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 Nokuti Nokuti na imfumu ya kuwona mpaka umoyo (SOH) wa mabatire yakugwiliskiridwa ntchito kale ndi chinhu chachikulu cha ntchito yakusungirako pakati pa kugwiliskiridwa ntchito kwamburu na kuwundura. Kweni, njira yakukhazikika pa deta imfuna kusankha kwa deta yosaya mu SOH na state of charge (SOC) mu nthawi yakugwiliskiridwa ntchito. Pa malo ghano, tikuwonetsa kuti kusankha kwa SOH kukuthandizidwa na generative learning kuli kwakuyambira mu kulimbana na vuto la kusowa na kusiyana kwa deta, kwakuyesedwa na pulse injection dataset ya 2700 ma batire ya lithiamu-ion yakugwiliskiridwa ntchito kale, yakupakika mitundu 3 ya ma cathode, mitundu 3 ya minofu, ma desaini 4 ya mphamvu, na ma ntchito yakale 4 na 10 ma SOC levels. Kugwiliskira ntchito deta yakupangika, regressor yikupereka kusankha kwa SOH kwakukwanira, na MAPE yakuyana na 6% mu SOC yosayende. Tikulalikira kuti kugwiliskira ntchito njira iyi kungapulumuse 4.9 biliyoni USD mu ndalama za umagetsi na 35.8 biliyoni kg CO2 emissions mwakudenga vuto la kusankha deta kwa 2030 global battery retirement scenario. Sumu iyi yikupereka chidwi cha kugwiliskira ntchito deta yakusowa pakufufuza deta yakukwana kwa njira za generative, cifukwa deta ingakhale yambura nthawi, yakudura, na yakufukula kwa ntchito zinyake zakusankha na kupanga. Kuyamba Kugwiliskira ntchito kwamagetsi kwamagetsi, chigawo chofunika kwambiri mu kusintha kwa mphamvu zopanda mpweya  , kwasintha kwambiri njira yakumanga mafuta. Mabatire ya lithiamu-ion yili yakufunika kwambiri pa kusintha uku popereka mphamvu zakudula, zakukhazikika, na zakutetezeka kumagalimoto yamagetsi (EVs)  ,  . Mphamvu yakufika ya ma batire ya EV yakugwiliskiridwa ntchito na yakumala ntchito idzapitilira 32–62 tera-watts mpaka 2050, yakukwera kwambiri kusiyana na kusankha mu International Renewable Energy Agency na Storage Lab scenarios  . Kweni, ma batire ya EV yakugwiliskiridwa ntchito kale yili yovomerezedwa mwamphamvu na mtengo wakumanga wakumanga wakukwana, kaŵelenji pamwamba pa 80% ya mphamvu yakukwanira  ,  . Ngati sichikusamaliridwe makonda, chikubala mikangano yachuma   kwa opanga na ogwiriskira, komanso mavuto ena pa chilengedwe na chikhalidwe  , kusankha kwa nthaka, ngozi za kuphatikizika kwa zinthu, na mpweya wa carbon  ,  ,  . 1 2 3 4 5 6 7 8 9 10 11 Njira zakuyambira zakufuna kusamalira mavuto a kusankha kwa batire yakukwana ndi kugwiliskira ntchito kwambiri  ,  ,   na kuwundura  . Mu kugwiliskira ntchito, ma batire yakugwiliskiridwa ntchito kale yikuphatikizidwa kumagwiriskira ntchito ngati ma grid-connected energy storage systems  , ma mphamvu zamudzi  , na magalimoto yakudula kutsinuka. Kweni, kugwiliskira ntchito kukufuna kusankha kwakuyambira, kusankha kusankha makonda  , kusankha mphamvu  , na kuphatikizika  ,   kuti zigwirizane na zofunika kumagwiriskira ntchito  . Ngakhale kuti pulogalamu yikuyambira yikuyamba  ,  ,  , kusowa kwa malamulo wakukwana kwa malo yakugwiliskira ntchito na njira zakumala ntchito kukulepheretsa kugwiliskira ntchito makonda a mphamvu yakusungika  . Kuwundura m'malo mwake kumagwiliskira ntchito nthaka yakusungika ya ma batire yakugwiliskiridwa ntchito kale mwakubuka zinthu kapena kukonza mawonekedwe  ,  , yakuyenera kumabatire yakugwiliskiridwa ntchito kale yakudwala mwamphamvu. Pakuyerekeza na pyrometallurgy na hydrometallurgy  ,  , kuwundura kwachindunji kukuyambirira pakugwiritsidwa ntchito kwakukulu, mphamvu yakudula, na mpweya wa carbon wakudula  ,  ,  . Kugwiliskira ntchito kusintha kwa lithiamu na kusankha kwa zinthu za cathode, kuwundura kwachindunji kukukwaniritsa kusintha kwa mawonekedwe a zinthu, na kusintha kwa ntchito  . Kweni, umoyo wa batire (SOH) ukutsogolera ku mtundu wa mankhwala wakufunika na kuchuluka kwa lithiamu yakufunika pakupanga njira zakuwundura kwachindunji  ,  ,  . Kusakwanira kwa kuchuluka kungabweretse kusintha kwakusowa, pamene kuchuluka kwakukulu kungayambitse kusakhazikika kwa alkali pa pamwamba pa cathode, kukulitsa kusintha kwa ntchito  ,  . Mwatsoka, kupeza SOH kukufuna kuyeza zinthu zakuvulaza  ,   na kuyezetsa mphamvu kwakudza, kusakhala kwakuvulaza, mwamphamvu, na kosalekeza kwa SOH kukukhalabe vuto lakukwanira. 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 Kuti tipeze makonda a SOH, ma batire ya EV yikuyang'aniridwa, na deta yakugwira ntchito yikulembeka ndi mapulatifomu ya makompyuta   na malemba ya batire  . Nyumba ya Msonkhano ya European Parliament yikuvomereza Batteries Directive pa June 14, 2023, kuti ma batire yakugwiliskiridwa ntchito kale yikugwilitsidwe ntchito, kukonzedwa, kapena kuundulidwa kumapeto kwa ntchito  . Kweni, njira yakufunika kuyang'anira na kulemba yikupambana pakati pa nthawi yakugwiliskira ntchito, ma batire yakugwiliskiridwa ntchito kale yikuphatikizidwa kale ku chipangizo cha kuyang'anira mu galimoto. Nthenda, nchito yovuta kupambana deta ya SOH yakumunda. Kukhazikika kwa deta ya umoyo kukulibe vuto, kucema kusankha kwa SOH pogwiliskira ntchito deta yokha yakumunda, kusiyana na deta yakale kapena mu nthawi zakulamulirika  ,  ,  . Chimodzi mwa njira ndi kuyesedwa kwa mphamvu yakukwanira pa malo yakusonkhanira ma batire yakugwiliskiridwa ntchito, ndiko kuti, SOC conditioning, icho chili champhamvu, kweni chikufuna nthawi yoyezetsa yambura mphamvu na ndalama zowonjezera za umagetsi. Kuyezetsa mphamvu ya hybrid pulse power characterization ni njira yina ya SOH estimation pogwiliskira ntchito pulse injection, koma kuyezetsa kwakukwana kumanga pa ma ola 12. Tao et al. adagwiliskira ntchito mapulisi yayifupi kuti asankhe SOH ya ma batire yakugwiliskiridwa ntchito kale, kuganizira kuti SOC conditioning mpaka 5% SOC yikachitika, yomwe yikugwirizana basi na nthawi yakugwiliskira ntchito yakusowa  . Kusintha kwapang'onopang'ono mu kuyeza kwa sensory kumaphatikizapo X-ray imaging  , impedance yamagetsi  ,  , optical fiber sensing  ,  , acoustics sensing  ,  , kusankha kwapakati  ,  , na pulse injection  ,  . Njira zakuchokera ku sensa zili pachigawo cha labu chifukwa cha kuvulaza, pamene pulse injection ikupanga kusankha kufotokozera kuwonongeka kwa batire popanda kuvulaza, mwamsanga kuposa njira zakusankha kwapakati na njira zakuchokera kumagetsi  ,  ,  ,  ,  ,  . Kweni, pulse injection ni njira yakukhazikika pa deta yomwe yikugwira ntchito pokhapokha pakuganizira kuti deta yakuyezedwa imakhala ndi malo yakugwiliskiridwa ntchito, yimanyama yichidziwika kuti vuto la domain shift mu gulu la machine learning. Ngakhale kuti pali kusintha kwa transfer learning na njira zakusankha domain  ,  ,  , mavuto akukhalabe cifukwa chakuti malo yakugwiritsiridwa ntchito kuti igwirizane yikufuna chidziwitso chakale pakugwiritsidwa ntchito, ndipo njira zakusowa zakusowa zikufunika kusintha kuti zigwirizane na nthawi yakugwiliskira ntchito yakusowa  . Njira yina ndiyakufalitsa kukula kwa deta yakuyezedwa mu malo yakugwiliskira ntchito yakusowa kuti ikwaniritse kusowa kwa deta na kusiyana kwa deta pomwe ingabweretse ndalama zowonjezera. Wang et al. adatulukira njira ya kutentha kuti isankhe zochitika zamankhwala pamiyeso yapakati pa njira ya thermochemical, kulola kupanga ma database akukwanira oyezetsa ndi mtengo wochepa wa thermochemistry ngati kuwonjezera deta pakupanga mtundu wa machine learning  . Kuphatikiza apo, generative learning yikuwonetsa mwayi wosankha SOH ndi deta yakulengedwa kuchokera kumiyeso yophatikizika, kusunga deta yakuyezedwa  ,  . Kweni, kusowa na kusiyana kwa deta mu vuto la kugwiliskira ntchito na kuwundura kwa batire kuli kovuta kwambiri chifukwa cha kusakaniza kwa mitundu ya zinthu za cathode, mitundu ya minofu, ma desaini ya mphamvu, na ma ntchito yakale, kukulimira kugwilizana kwa chidziwitso cha kuwonongeka kwa nthawi yakugwilira ntchito mu mitundu ya machine learning. Nthenda, nchito yeniyeni ndiyakuganizira za kudalira pakati pa malo yakugwilira ntchito na ma yankho a pulse voltage kuti apange deta, popanda kulepheretsa nthawi yaitali ya kukonza mphamvu na kuyesedwa kwina kwakuthupi nthawi imodzi, kuti ikwaniritse kusankha kwa 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 Mu kafukufuku uwu, tikuchita pretreatment ya batire yakugwiliskiridwa ntchito, ndiko kuti, kusankha SOH, pogwiliskira ntchito deta ya pulse voltage yopangidwa ndi attentional variational autoencoder pa nthawi yakuyezedwa yakupulumutsidwa na mtengo. Chigawo chachikulu cha njira za pretreatment, kupanga mtundu wa generative, na njira zakugwilitsira ntchito zakugwiritsiridwa ntchito zikufotokozedwa mu Chithunzi.  . Lingaliro la kafukufuku ndikugwiliskira ntchito generative learning kufufuza deta yakuyezedwa kale pakufufuza kwa pulse voltage responses m'malo yakugwilira ntchito, ndiko kuti, kugwiritsidwa ntchito kwa SOC mu gawo la kusonkhanira kwa batire yakugwiliskiridwa ntchito. Kusankha kwa SOH kukuchitika popanda deta yakuyezedwa ya pulse, kupulumutsa ndalama za pretreatment za umagetsi na mpweya wa carbon zomwe zikufunika kuchokera pakusintha SOC mpaka pamlingo wakuyembekezeredwa. Poyerekeza na ampere-hour integral method na machine-learning pulse injection, generative learning imapereka liwiro lakupanga lamphamvu na kusankha kwakukwanira. Ndi deta yakupangika, regressor yosavuta yikukwaniritsa SOH na cholakwika chaching'ono mu malo yakugwilira ntchito yosayende kapena yosakwanira, vuto lalikulu lopanda mphamvu (OOD). Vuto la OOD likutsukidwa powaphunzira kudalira pakati pa SOC condition na yankho la voltage lomwe likubwera chifukwa cha pulse yachangu. Tikusonyeza kuti kusintha kwa njira yathu pochita kuyezedwa ndi ma samples 2700 yakuyezedwa kwakuthupi, kufalitsa mitundu 3 ya zinthu za cathode, mitundu 3 ya minofu, ma desaini 4 ya mphamvu, na ma ntchito yakale 4 (1 laboratory accelerated aging, 3 pure battery EV driving, and 2 hybrid EV driving). Generative learning yikuyenera interpolation na extrapolation pogwilizana chidziwitso cha kale cha malo yakugwilira ntchito mu kukula kwa latent space yamitundu yamitundu, kupulumutsa nthawi yakuyezedwa, ndalama za umagetsi, na mpweya wa carbon wakutsatiridwa popanda kuwononga kusankha kwa SOH. Chigawo chachikulu cha kafukufuku wa padziko lonse wa batire yakugwiliskiridwa ntchito mu 2030 chikufotokozera kufunika kwa SOH estimation mwamphamvu na kosalekeza mu pyrometallurgical, hydrometallurgical, na njira zakuwundura zachindunji. Kupyolera mu kafukufuku wa umisiri-chuma, generative learning-assisted SOH estimation ingapulumuse $4.9 biliyoni ndalama za umagetsi na 35.8 biliyoni kg CO2 emissions mpaka 2030 padziko lonse. Tikukambirana za kusonyezedwa kwa mtundu, zotsatira za pretreatment ya kuwundura, na mbali zambiri za njira zamtsogolo zanzeru za pretreatment zomwe zikugwirizana na machine learning. 1    Njira za pretreatment ya batire yakugwiliskiridwa ntchito zikuphatikizapo SOC conditioning, kuyezedwa kwakuthupi, machine learning ya SOH diagnosis, na gawo lachiwiri la kusankha kwa moyo pamaso pa kugwiliskira ntchito kapena kuwundura. SOC conditioning imatanthauza kusintha kwa magawo a SOC a ma batire yakugwiliskiridwa ntchito kumayiko ofunika pogwiliskira ntchito njira yakale ya constant current constant voltage (CCCV).   Chithunzi cha attentional variational autoencoder cha kupanga deta ya pulse voltage response.   Kugwiliskira ntchito kwamphamvu kwa mtundu wakupanga deta kwa kusankha kwa SOH kukufuna kusintha kwina kwa SOC. Mtundu wa machine learning wapangidwa ndi deta yakupangika ndipo umalandira ma yankho a pulse voltage nthawi yomweyo pakugwiritsidwa ntchito.   Usiro wa umisiri-chiwonetsero na liwiro lalikulu la kupanga la njira yathu ya generative learning pogwiliskira ntchito ndalama za SOC conditioning mu 2030 battery retirement scenario padziko lonse. a b c d Zotsatira Kusonkhanira kwa Dataset Kusowa kwa deta na kusiyana kwa deta ni mavuto akulu pakati pa data-driven battery diagnosis na prognosis, makamaka kwa ma batire yakugwiliskiridwa ntchito kale. Tinasonkhanitsa ma samples 2700 kuchokera ku ma batire yakugwiliskiridwa ntchito kale yakuyezedwa kwakuthupi 270 pa magawo akukulu a SOC, ndiko kuti, kuchokera ku 5% mpaka 50% ndi 5% grain. Mu Chithunzi.  , dataset ikuphatikiza mitundu 3 ya ma cathode yomwe yafala, kuphatikiza 119 ya NMC (nickel manganese cobalt oxide), 56 LFP (lithium iron phosphate), na 95 LMO (lithium manganese oxide) yakugwiliskiridwa ntchito kale, yofalikira mitundu 3 ya minofu (cylindric, pouch, prismatic) na ma njira 4 zakugwilira ntchito kale (1 laboratory accelerated test, 3 EV usages, kuphatikizapo 1 purely electrified power-train mode na 2 hybrid power-train modes). Tikugogomeza kuti kuphatikiza ma samples oyeserera kwambiri kumapangitsa deta yathu kukhala deta yayikulu kwambiri ya pulse yachangu ya SOH estimation ya ma batire yakugwiliskiridwa ntchito kale mpaka lero, kulola kusintha kwa mtundu kuposa ma dataset yakuyezedwa kwakuthupi. 2a    Chithunzi cha Sankey cha kusankha kwa batire yakugwiliskiridwa ntchito, chikupanga mitundu ya zinthu za cathode, mitundu ya minofu, ma desaini ya mphamvu, na ma ntchito yakale.   Pulse current na yankho la voltage la ma batire yakugwiliskiridwa ntchito, kumene ma chigawo amachokera ku malo otembuka, ndiko kuti, malo okhala ndi zero second-order derivative ya voltage response curve. Ma point 21 a chigawo, kuchokera ku U1 mpaka U21, amachokera. Kulemba deta yakuyambira ndi 100 Hz. Nthawi yotsala ndi masekondi 25 pakati pa pulse iliyonse mu C-rate. Dziwani kuti C imayimira charge (discharge) rate pamene maola 1 a charge (discharge) yachitika. Kutentha kwa kuzungulira kukuphatikizidwa pa 25 .   Kusankha kwa umoyo wa batire (SOH) (Gaussian fit) ya ma batire ya 2.1 Ah na 21 Ah yakugwiliskiridwa ntchito kale kuti afotokoze, motsatira. Kuti muwone SOH distribution ya mitundu ina ya batire, mutha kuwona Supplementary Fig. a b c

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

Codkan waxa lagu soo saaray afkii asalka ahaa ee sheekada!

AI Gjenerative Lishona Kufanya Kutumika Tena kwa Betri za Lithium-Ion Zilizostaafu

About Author

FAALLO

HANG TAGS

MAQAALKAN AYAA LAGU SOO BANDHIGAY

Related Stories

"We need to change our mindset about wealth" ft. Linh Dao Smooke & Host An Truong of Vietcetera

Scary-good Writing Tips: Learn How to Write a Listicle

Is Your ICP An Offensive Stereotype?

What to Do If You Don't Like Google Search

"We need to change our mindset about wealth" ft. Linh Dao Smooke & Host An Truong of Vietcetera

Scary-good Writing Tips: Learn How to Write a Listicle

Is Your ICP An Offensive Stereotype?

What to Do If You Don't Like Google Search

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps