
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 WaMbaṇi: 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Shengyu Tao Haizhou Liu Chongbo Sun Haocheng Ji Guanjun Ji Zhiyuan Han Runhua Gao Jun Ma Ruifei Ma Yuou Chen Shiyi Fu Yu Wang Yaojie Sun Yu Rong Xuan Zhang Guangmin Zhou Hongbin Sun Abstrakiti Batirisi zvikuru zvakasiyiwa zvisina kurongeka nezvinhu zvakasiyana zve cathode zviri kudzivisa kutorwa kwekutora kwakananga nekuda kwemhando yezvinhu zvacho zve cathode. Kuwedzera kwezvikuru zvakasiyiwa zviri kudiwa kurongeka kwakaringana kwekutora kwakananga, asi matambudziko anomuka kubva mumamiriro ekushanda akasiyana, vagadziri vakasiyana-siyana, uye zvinetso zvekuvanzika kwemiddata yevadyidzani vezvinyorwa (varidzi vemari). Pano tinoratidza, kubva kune yakasarudzika dataset ye130 lithium-ion mabhatiri anofukidza zvinhu 5 zve cathode uye vagadziri 7, nzira yekudzidza muchina yakabatana inogona kupa mabhatiri aya akasiyiwa pasina kuvimba nedata rekushanda rekare, kuchengetedza kuvanzika kwemari kwevadyidzani vezvinyorwa. Nekushandisa zvinhu zvakatorwa kubva kune yekupedzisira-ye-upenyu yekuchaja-kuburitsa kutenderera, mufananidzo wedu unoratidza 1% uye 3% yekushandisa kwe cathode zvikanganiso pasi pehomogeneous uye heterogeneous mabhatiri ekutora zvinhu zvisina kurongeka, zvichikonzerwa nehunyanzvi hwedu hweWasserstein-distance voting strategy. Zvemari, nzira yakatarwa inosimbisa kukosha kwekurongeka kwakaringana kwemabhatiri ekutorwa kweindasitiri yekutorwa kwakabudirira uye yakagadzikana. Chidzidzo ichi chinoratidza maitiro matsva ekushandisa mari yakavanzika kubva kune akasiyana-siyana masosi, achibvumira kuenderana uye kuremekedza-kuvanzika kuita sarudzo dzezvigadzirwa zvakaparadzirwa. Nhanganyaya Lithium-ion batteries (LIBs), kushanda semidziyo yekuchengetedza simba, yakagamuchira kushandiswa kwakanyanya munzvimbo dzakasiyana-siyana, kubva pakugadzirwa kwemaindasitiri kusvika kuhupenyu hwezuva nezuva senzira yakagamuchirwa tekinoroji. Zvinotarisirwa kuti huwandu hwekugadzirwa kwepasi rose kweLIBs hunopfuura 1.3 TWh na2030 apo kudiwa kunowedzera kwezvikuru kuchapfuura kuwanikwa kwezvinhu zvakakosha zvekusimudzirwa senge lithium ne cobalt , . Zvisinei, hupenyu hwazvino hwepakati hwezvigadzirwa zveLIB hunosvika makore 5-8, zvichitungamira kukuwedzera kwekushandiswa kwezvikuru kune akawanda nyika. Kana zvisina kugadziriswa nenzira kwayo, zvikuru zvakasiyiwa zvichaguma nekurasikirwa kwesimba risina kugadzikana uye kukanganiswa kwezvakatipoteredza. Nekuda kweizvi, budiriro yeunhangamire yekutorwa kwezvikuru inova yakakosha sezvo tiri kusangana nekudziya kweLIB kudzokera kudzokera . 1 2 3 4 Zvitsva zvichangobva kuitika mukutsvagurudza kwezvinyorwa zvebhatiri zvakatarisa pamitauro ye pyrometallurgical, hydrometallurgical, uye direct recycling approaches . Zvinopesana nemitauro ye pyrometallurgical ne hydrometallurgical, direct recycling inosiyaniswa senzira yakasiyana. Chirongwa ichi hachikonzere kukuvara kwechipiri pahuwandu hwezvinhu, zvichibvumira kugadziriswa kwakanyanya kwehuwandu uye kudzorerwa kwekuita. Pamusoro pezvo, direct recycling inoratidza pundutso yakakwirira, sezvo ichisanganisira kushandiswa kwesimba kushoma, kuderera kwezvipimo zve greenhouse gas, uye kuderera kwezvakatipoteredza zvekukanganiswa , . Pakugadzirwa chaiko, zvisinei, vatori vezvinyorwa vanowanzosangana neLIBs dzine zvinhu zvisingazivikanwe kana ma modules ebhatiri anoumbwa nechikamu chemarudzi akasiyana e cathode. Kufunga kuti kutora kwakananga kunogona kunyanya kuve kwakakosha kune cathode, kusawirirana kwakadai kunoita kuti kushandiswa kwekutora kwakananga kusingakwanise kuita kwekugadzirwa kwehutano hwekukosha kwezvinyorwa zvakasiyiwa . Zvakakosha kusimbisa kuti kunyange kana zvinhu zvakakosha kubva kune akasiyana-siyana e cathode zvinogona kutotorwa uchishandisa nzira dzekutora dzakajairika, kubatana pakati pemhando dzakasiyana dze cathode panguva yekutora kunogona kukanganisa kunaka kwezvigadzirwa . Naizvozvo, kunzwisisa marudzi emakemikari e cathode pamusoro pekutorwa kunokosha zvikuru kunokanganisa kusarudzwa kwe direct recycling route uye pakupedzisira kunosimudzira kunaka kwezvigadzirwa, pundutso, uye kugadzikana. 5 6 7 8 9 Human-assisted direct recycling yakakurudzirwa kuti ione marudzi emakemikari e cathode akasiyiwa mune pre-treatment link, iyo isati yave yakakwana pakubhadhara mari apo indasitiri yekutora iri kuwedzera . Kuti uwane zvirongwa zve cathode yemhando yechikuru zvakakodzera, nharaunda yesainzi neindasitiri yakatanga system yekutsvaga hupenyu hwebhatiri uye pfungwa dzinomuka se battery passport ne battery data genome . Kunyangwe mabhatiri akawanda akashandiswa vasati vava nepfungwa idzi, kune kubvumirana kunowedzera kuti ruzivo rwebhatiri runofanira kuwanikwa munharaunda yose yehupenyu kuti ribvumire kusarudza kwechipiri-hupenyu . Izvi zviri pachena kune chikamu chekutora mabhatiri, chiteshi chekupedzisira chechipiri-hupenyu chebhatiri, sezvo nzira yekutora inogona kunyanya kuve yakakosha kune cathode. Zvisinei, masystem ekutsvaga hupenyu hwebhatiri kana mapepa ebhatiri anofambiswa nemagetsi akadai macode uye near-field communications, izvo zvingangounza mari yakawanda yekudyara uye zvingasashandisane nemafurani akasiyana ebhatiri. Pamusoro pezvo, magajeti emagetsi anoramba achinetsa kugadzirisa zvakanaka muhupenyu hwavo hwose, zvichitungamira kumidziyo yakasakara uye kusakwanisa kuwana panguva yekutora sezvo nzira yemhando yekugadzira yeLIB isati yagadzirwa-kugadzirwa-kutorwa . Saka, kuputsika kwakawanda kunodiwa kuti uwane kurongeka kwemhando ye cathode yebhatiri chete uchishandisa ruzivo rwepasvondo ruri nyore kuwana , , zvakapesana nedata rekurongedzera kare kana nzira yakaitwa nevanhu, ichibatsira kutora kwe direct recycling kuti kusimudzire kunaka uye pundutso yezvigadzirwa zvakatorwa. 1 10 11 12 13 14 15 16 Munguva yemakore mashoma apfuura, machine learning yakamuka senzira inoshanda yekugadzirisa mibvunzo yakavhurika muzvikamu zvese zvebhatiri. Mune zvimwe zvinhu zvine chekuita nebhatiri, machine learning yakatibvumira kuwana otomatiki nzira dzakaoma dzehupenyu , , , kufanana neupenyu hwasara hwekuita , , , , , kuongorora mamiriro ehutano , , , kugadzirisa purogiramu yekuchaja , , kufanana nekugoverwa kwekutadza , kana kunyange kudzidzira dhizaini yebhatiri , , uye kufanana nekuita kwenguva-hupenyu ipapo mushure mekugadzirwa . Pakutora mabhatiri, mashoma mashandisi akaongorora machine learning maererano nezvinhu zve cathode , , izvo zvinopesana nehuwandu hwebhatiri data, kunyanya kune avo vakachajwa kusvika kumagumo-e-hupenyu. Huwandu hwezvidzidzo zvakabudiswa zvinoratidza huwandu hwedhata hushoma kwazvo  uye zvakanyanya kushomeka mukusiyana kwebhatiri kwe cathode . Kushomeka kunokonzerwa nekubhadhara kwakanyanya, nguva yakareba yekuyedza , uye, chakanyanya kukosha, kuvanzika kwedata nekuda kwezvikonzero zvekutengesa kana zvekufarira. Nekudaro, dambudziko rekuvanzika rinoruramisa dambudziko umo data rebhatiri riripo, kunyangwe rakakura muhuwandu uye kusiyana kubva kune akawanda mapato senge vagadziri vebhatiri, vanoshandisa, masangano edzidzo, ne mapuratifomu echitatu, haagone kugovaniswa. Dambudziko rakadaro rinoda kudzidza marudzi emakemikari e cathode kuti agadzirise nzira yekutora bhatiri mune nzira yakabatana asi inoremekedza kuvanzika. 17 18 19 20 21 22 23 24 19 25 26 27 28 29 30 31 32 33 34 35 36 37 Federated machine learning, senzira yakaparadzirwa uye yekuchengetedza-kuvanzika, ine kugona kugadzirisa zvese kuenderana kwemapato akawanda (zvakaenzana, huwandu hwebhatiri data uye kusiyana) uye zvinetso zvekuvanzika kuburikidza nekuenderana kwekuidza muchina , , . Muchikamu chega chekudzidzira, varidzi vedata vakaparadzirwa vanoita kudzidzira kwemuno nemidziyo yavo yemuno, vanovharira maparamita/mhedzisiro emufananidzo wakadzidzirwa, uye vanoitumira kune mutongi wepakati kuti vaunganidze. Zvinhu kuti nzvimbo dzakachena hadzimbofi dzakasiya varidzi vadzo vedata uye kuti maparamita/mhedzisiro yakatumirwa yakavharidzirwa zvakanaka kuchengetedza kuvanzika kwedata. Federated machine learning yakadzidzirwa zvakanyanya munzvimbo dzakawanda dzinoshandiswa, kusanganisira hutano hweruzhinji , , kuongorora kwezvekurapa , , , e-commerce , Internet of Things , mobile computing , uye smart grid , , . Nzira iyi inogona kushandura nzira yekutsvagurudza inofambiswa nedata munzvimbo dzakawanda dzesimba nekubvumira kubatana kunochengetedza kuvanzika, kunyanya kune avo vane dataishoma. Maererano nechikamu chekutora mabhatiri, zvidzidzo zve federated machine learning zvinoratidza mikana yakajeka yekushandisa huwandu hukuru hwebhatiri data iripo asi haigone kugovaniswa nekuda kwezvinetso zvekuvanzika. Nechechi yakabatana asi inochengetedza kuvanzika, kurongeka kwemabhatiri akasiyiwa kunogona kuitwa nekunaka kukuru, kugona, kuwedzera, uye kuenderana, kugadzirisa kunaka uye pundutso yezvigadzirwa zvakatorwa. Kuti tizive, zvidzidzo zve federated machine learning zvakanangana nekutora mabhatiri hazvisati zvamboonekwa. 38 39 40 41 42 43 44 45 46 47 48 49 50 51 Muchidzidzo chino, tinoita kurongeka kwemakemikari e cathode akasiyiwa, tichishandisa data rebhatiri riripo kubva kune vanobatana vakawanda, senge vagadziri vebhatiri, vanoshandisa, masangano ekutsvagurudza, ne mapuratifomu echitatu, mune nzira yakabatana asi yakavanzika-yekuenderana nemuchina sezvakaratidzwa paFig. 1. Mufananidzo wedu we federated machine learning wakadzidziswa uchishandisa chete kutenderera kumwe chete kwe data rekuyedza repasvondo kuburikidza ne yakajairika feature extraction process, pasina ruzivo rwekare nezve mamiriro ekushanda ekare. Tinoshandura simba rekufanira remufananidzo wedu we federated machine learning neiyo yemhando dzemunharaunda dzakadzidzirwa dzakavakirwa pane dzawo data pasi pehomogeneous uye heterogeneous mabhatiri ekutora zvinhu zvakasungirirwa. Dambudziko reheterogeneity rinogadziriswa nehunyanzvi hwedu hweWasserstein-distance voting strategy. Kuongorora kwemari kwekutora mabhatiri akasiyiwa uchishandisa nzira yedu yakatarwa ye federated machine learning inoonekwa kukosha uye kudiwa kwekurongeka kwakaringana kwezvikuru zvakasiyiwa. Tinoshandisa zvakanyanya kutorwa kwemufananidzo, zvinoreva mabhatiri ekutora, uye zvimwe zvakajeka zvezvakadzika zvebhatiri rekutora.    Masosi mazhinji emari, senge vagadziri vebhatiri (Mufananidzo nekuda kweAddionics), vanoshandisa (bhatiri pack mupasi pe Tesla. Mufananidzo nekuda kweTesla), masangano ekutsvagurudza, ne mapuratifomu echitatu, anogona kuve vanopa data. Data rebhatiri harisati rapiwanwa pakati pevanopa kana kutumirwa kumutori webhatiri. Panzvimbo pezvo, vanopa data vanodzidzira mafirimu emuno uye vanopa maparamita efirimu kune mutori webhatiri kuti vagadzire mufananidzo wepasi rose. Chirongwa cheWasserstein-distance voting chakagadzirwa chinobatanidza mafirimu emuno mufirimu repasi rose, iro rinogona kushandiswa kune kusagadzikana kwedata uye ruzha. Vatori vebhatiri vanogona kushandisa mufananidzo wakabatana wakavakirwa kuti vatore mabhatiri, vakabatana nedata rekuyedza repasvondo riri nyore kuwana.   Chirongwa chedu che federated machine learning chinokurudzira vadyidzani kugovana data uchichengetedza kuvanzika kwedata sekutarisana nechirongwa chekare che data islanding. a b Mhedzisiro Kuunganidzwa kwedata uye standardization Kusiyana kwakasiyana kwemhando dzakasiyana dzebhatiri kunowanzo kuve kwakakura uye kunonetsa kutsanangura nekuda kwemamiriro akasiyana ekushanda, kusiyana kwevagadziri, uye kushandiswa kwenguva yakapfuura . Kuwana mhinduro kune dambudziko iri, takunganidza uye takajairisa 130 mabhatiri akasiyiwa ane 5 mhando dze cathode kubva kune vagadziri 7 kuti tigadzire dataset isina kubva pane chero chipi zvacho, yakafanana neheterogeneous. Tichitarisa kushandiswa kwenguva yakapfuura kwakasiyana, hunyanzvi hwebhatiri hwakanga husina kusvika 90% yehunyanzvi hwekutanga. Zvinhu zve cathode zviri lithium cobalt oxide (LCO), nickel manganese cobalt (NMC), lithium ferrophosphate (LFP), nickel-cobalt-aluminum oxide (NCA), uye NMC-LCO yakavhenganiswa, izvo zvakakamurwa kuita mapoka 9 zvichienderana nevagadziri (Supplementary Table  ). Tinodzidzisa nemaune mabhatiri ane kushandiswa kwenguva yakapfuura kwakasiyana-siyana, kubva pakuyedzwa kwelabhoritari kusvika kune mamiriro ekutyaira emotokari, kuti tidzidzise mufananidzo wakajairika kune mutori webhatiri asina ruzivo rwekushandisa kwenguva yakapfuura uye marudzi ebhatiri. 52 1 Kuti tive neyakajairika, zvese zvinodiwa kubva kumutori zvinongova zvekutenderera kwazvino (field-testing) nekuchaja kumwe chete uye kuyedzwa kwekufambisa, izvo zviri nyore kuita munzvimbo chaidzo. Deta rakabudiswa rinotanga raiswa pasi nekuzadza zvisipo, kutsiva zvisina kurongeka, uye kuita median filtering. Zvakaipa zvakakonzerwa nevanhu uye zvinhu zvinokonzerwa ne cathode-heterogeneity zvinochengetedzwa nemaune, zvisinei, kuti uite kuti mufananidzo uenderane nezvinhu zvisina kunaka. Deta rinozoiswa linearly kuti rive curve filling (Supplementary Fig.  ) uye feature engineered kuti kuderedzwe kwemhando, neyakajairika seti yemafara ekugadzirisa (Supplementary Note  ). Zvinhu zvakatorwa kubva kune standardization pipeline zviri nyore kunzwisisa, dambudziko rekunakidza kukuru kwekutengesa. Kuti tizive, iyi ndiyo yekutanga nguva iyo data rebhatiri rakasiyana-siyana kubva kune akawanda masosi uye kushandisa kwenguva yakapfuura zvakashandiswa kubatsira dhizaini yemaitiro ekutora mabhatiri. 1 1 Chifananidzo  a,   inoratidza maitiro echiito chechiito. Tinotarisa pane voltage-capacity uye dQ/dV curve yekuchaja nekuburitsa kwemabhatiri akasiyiwa mukutenderera kwekupedzisira, ndiko kuti, kutenderera kumwe chete kwekutengesa nekuburitsa (Supplementary Figs.  – ). Mukutenderera kwekutengesa, zvinhu 15 zvinotorwa kubva kuvhoriji-hunyanzvi uye dQ/dV curve, uko V naQ vanoreva vhoriji nehunyanzvi values, zvichiteerana. Iwo akafanana zvinhu zvinotorwa kweyekuburitsa kutenderera. Nekuda kweizvozvo, zvinhu 30 zvinotorwa zvachose, sezvinoratidzwa kubva kuF1 kusvika kuF30.ONA Supplementary Table   uye Supplementary Note   kune tsananguro yakadzama yezvinhu. Chifananidzo  c inoratidza absolute uye relative feature values ezvikamu zvakasarudzwa kubva muchikamu chimwe nechimwe. Zvikuru zvinhu zvakafananidzwa mumakirasi akasiyana-siyana zvinobatana munzvimbo ye −1 kusvika 0 (nechekiroyo ye green), uye hazvizivikanwe, zvichiratidza kunetseka mukududzirwa kwemhando yebhatiri uchishandisa kutenderera kumwe chete kwebhatiri data. Kunetseka kunotarisirwa nekuti mamiriro ekushanda akasiyana-siyana anogona kukanganisa kuchaja-kuburitsa kinetics yemabhatiri kuitira kuti zvinhu zvakatorwa zvigone kuve zvakabatanidzwa kunyangwe marudzi akasiyana ebhatiri (Supplementary Fig.  ). Pane kungodudzira zvinhu zvakatorwa uchishandisa ruzivo rwemudzi, tinoita nzira yakasarudzika inobva kune data kuti ichengetedze otomatiki maitiro akaomarara pakati pemhando dzakasiyana dzebhatiri. 2 b 2 5 2 2 2 6    Kune maitiro ekuchaja, zvinhu 15 zvinotorwa kubva ku voltage-capacity (kuruboshwe) ne dQ/dV curve (kurudyi).   Iwo akafanana zvinhu ndezve maitiro ekuburitsa se F16 kusvika kuF30.   Zvibvumirano zvakaratidzwa nemakirasi, zvichitevera fomati CxBn, zvichiratidza bhatiri re n kubva kuchikamu x. Kukura kwe denderedz 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 ena e sireletsehang boprivacy e ka fetola litšila tsa betri hore li be phaello e kholo

About Author

FAALLO

HANG TAGS

MAQAALKAN AYAA LAGU SOO BANDHIGAY

Related Stories

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

Startups Winners, Turn Your Win into a Story Worth Sharing

Bitcoin NFTs are Getting Easier Every Day: Introducing Ordinals

69 Stories To Learn About Productivity Tools

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

Startups Winners, Turn Your Win into a Story Worth Sharing

Bitcoin NFTs are Getting Easier Every Day: Introducing Ordinals

69 Stories To Learn About Productivity Tools

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps