
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Ababhaleli: 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 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 Isishwankathelo Ukunyanzeliswa okukhawulezayo nokuchanekileyo kwesimo sempilo (SOH) samabhetri asetyenzisiweyo yinto ebalulekileyo yokulungiselela ukusetyenziswa kwakhona kunye nokurisayikliswa. Nangona kunjalo, iindlela ezixhaswa yidatha zifuna ukucocwa kwedatha okupheleleyo phantsi kweemeko zokulahlwa kwe-SOH kunye ne-state of charge (SOC) ngokungahambanga. Apha, sibonisa ukuba ukuqikelelwa kwe-SOH okuncediswe kukufunda okunesizukulwana kunokubonelela ngokuphucula imingeni yobutyebi bedatha kunye nemixholo, eqinisekiswe ngedatha yokutofeka kwe-pulse yedatha yamasampula angama-2700 ebhetri elahlisiweyo, ebandakanya iintlobo ezi-3 zemathiriyeli yekathode, iifomathi ezisi-3 ezibonakalayo, uyilo olunobungakanani obu-4, kunye nemisebenzi yembali emi-4 enamanqanaba esi-10 e-SOC. Kusetyenziswa idatha eyenziweyo, i-regressor iphumeza ukuqikelelwa kwe-SOH okuchanekileyo, kunye neempazamo ezingaphantsi kwe-6% phantsi kwe-SOC engaziwayo. Siyacinga ukuba ngokuthi sifune ukusetyenziswa ngokufanayo kobuchwepheshe obubonelelweyo, oku kuya kugcina i-USD ye-4.9 yeebhiliyoni kwiindleko zombane kunye ne-35.8 yeebhiliyoni kg yeCO2  ukukhutshwa ngokunciphisa iindleko zokucocwa kwedatha kwisimo sehlabathi sokulahlwa kwebhetri ka-2030. Eli phepha libalisa ukusetyenziswa kwedatha elinganiselweyo ukuze kugcinwe indawo enkulu yedatha kusetyenziswa iindlela zokuzala, kuba idatha ingathatha ixesha elide, ibe neendleko kwaye ingcolise ukuyibuyisela kwimisebenzi emininzi yokuqikelela nokubala kwangaphambili. Intshayelelo Ukufakelwa kombane kwezokuhamba, isebe elibalulekileyo kuqeqesho lwamandla aphantsi kwekhabhoni , sele liguqule i-paradigm yokuhamba esekwe kwifosili ngendlela engazange ibonwe ngaphambili. Amabhetri e-lithium-ion abonisa ukubaluleka kolu qeqesho ngokubonelela ngezixokelelwano zamandla ezixabiso liphezulu, ezihlala ixesha elide, kunye nezikhuselekileyo kwiimoto zombane (EVs) , . Umthamo opheleleyo wamabhetri asebenzayo nawasebenzayo angaphaya kwee-terawatts ezingama-32–62 ngo-2050, ubukhulu kunoko kudlula uqikelelo kwiimeko ze-International Renewable Energy Agency kunye ne-Storage Lab . Nangona kunjalo, amabhetri e-EV alahlisiweyo alawulwa kakhulu ngomthamo osalayo omkhulu ongasebenzisiwanga kakuhle, ngokubanzi ngaphezu kwe-80% yomthamo wokulinganisa , . Ukuba awenzelwanga kakuhle, oko kukhokelela kwiindleko zemali  kumenzi kunye nabasebenzisi, kunye nemiba elandelayo yendalo kunye neyentlalo , kubandakanya ukuchitha izixhobo, imingeni yekhonkco lokubonelela, kunye nokukhutshwa kwekhabhoni , , . 1 2 3 4 5 6 7 8 9 10 11 Iindlela ezibonelelayo zokujongana neenkxalayi malunga nokunyuka okungazinzisiyo kokulahlwa kwebhetri kukusebenzisa kwakhona , ,  kunye nokurisayikliswa . Ekusebenziseni kwakhona, amabhetri alahlisiweyo asetyenziselwa izicelo ezinje ngeenkqubo zokugcina amandla ezixhunyiwe kwi gridi , izixhobo zombane zokuhlala , kunye neenqwelo ezihamba kancinci. Nangona kunjalo, ukusetyenziswa kwakhona kufuna ukulungiswa kwangaphambili, kubandakanya ukuhlolwa kokuhambelana , ukuhlela umthamo , kunye nokudibanisa kwakhona ,  ukudibana neemfuno zesicelo . Ngaphandle kokuqala kwezicwangciso zovavanyo , , , ukungabikho kweemilinganiselo engqongqo yezicelo zokusetyenziswa kunye neendlela zokulahlwa zithintela ukusetyenziswa okuchanekileyo komthamo osalayo . Ukurisayikliswa endaweni yoko kusebenzisa amaxabiso aselweyo amabhetri alahlisiweyo ngokukhutshwa kwezinto okanye ukulungiswa komkhandi , , efanelekileyo kumabhetri alahlisiweyo axinezelekileyo ngokungenakupheliswa. Kuthelekiswa ne-pyrometallurgy kunye ne-hydrometallurgy , , ukurisayikliswa ngqo kugqama ngeenzuzo eziphantsi, ukusetyenziswa kwamandla okuphantsi, kunye nokukhutshwa kwekhabhoni okuncitshisiweyo , , . Ukusebenzisa ukuthethelelaniswa kwe-lithium kunye nokuphunyezwa kwemathiriyeli yekathode emva kokuphela kokuphunyezwa, ukurisayikliswa ngqo kufezekisa ukulungiswa komkhandi okusebenzayo, kunye nokubuyiselwa kokusebenza . Nangona kunjalo, imeko yempilo (SOH) ichaza izixhobo zamachiza ezifunekayo kunye needosi zokubonelela nge-lithium ezilindelekileyo kwizicwangciso zokurisayikliswa ngqo , , . Idosi engonelanga ingakhokelela kulungiso olungaphelelanga, ngelixa idosi eninzi kakhulu ingakhokelela ekubeni kubekho i-alkali esalayo kumacandelo e-cathode, ilungisa ukubuyiselwa kwendalo , . Ngelishwa, ukufunyanwa kwe-SOH kufuna ukuhlolwa kwezinto ezijongene ngqo ,  kunye neevavanyo ezide zokulinganisa umthamo, ukufunyanwa kwe-SOH okungajongene ngqo, okukhawulezayo, nokuzinzileyo kuhlala kuyimfuneko engabalulekanga. 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 Ukukhusela ulwazi oluqinisekileyo lwe-SOH, amabhetri e-EV ayalandelelwa, kwaye idatha yokusebenza irekhodiswe ngezixokelelwano zamafu  kunye nezipasipoti zebhetri . IPalamente yaseYurophu yamkele iBhili yeBhhetri ngoJuni 14, 2023, ukuqinisekisa ukuba amabhetri alahlisiweyo angasetyenziswa, alungiswe, okanye alingiswe xa sele engasebenzi . Nangona kunjalo, inkqubo yokulandelwa nokurekhodwa efunekayo ifumaneka kuphela kwisigaba senkonzo ye-EV, amabhetri alahlisiweyo awasekho axhunyiwe kwi-unit yokulandelwa kwenqwelo. Ke ngoko, kunzima ukubuyisela idatha ye-SOH ekhoyo kwindawo. Ubungqwalela bedatha yobomi buhlala buyimfuneko enkulu, ibiza ukuqikelelwa kwe-SOH kusetyenziswa kuphela idatha yendawo, ngokuchaseneyo nedatha yembali okanye phantsi kweemeko ezilawulwayo , , . Isisombululo esinye kukwenza uvavanyo lokulinganisa umthamo kwindawo yokuqokelela ibhetri elahlisiweyo, oko kukuthi, ulungiso lwesimo sokutshaja (SOC), olulula, nangona kunjalo, kufuna ixesha elingafikelelekiyo lovavanyo kunye neendleko zombane ezongezelelweyo. Uvavanyo lwesici samandla e-hybrid pulse yindlela yokukhetha yokuqikelelwa kwe-SOH ngokutofeka kwe-pulse, kodwa inkqubo epheleleyo yovavanyo ithatha ngaphezu kweeyure ezili-12. U-Tao kunye nabanye basebenzise ii-pulses ezimfutshane zokuqikelelwa kwe-SOH kwamabhetri alahlisiweyo, befuna ukuba ulungiso lwe-SOC ukuya kwi-5% ye-SOC lwenziwa, nto leyo ebengangqinelani neziphumo zokulahlwa kwe-SOC ngokungahambanga . Ukuqhubela phambili kweteknoloji yokulinganisa esekwe kwi-sensory kubandakanya imifanekiso ye-X-ray , i-impedance ye-electrochemical , , ukuchaswa kwe-fiber optical , , ukuphawulwa kwe-acoustics , , ukutshaja okuyinxenye , , kunye nokutofeka kwe-pulse , . Iindlela zokuchasa zikwindawo yovavanyo ngenxa yokujongana, ngelixa i-pulse injection ibonisa ukuhambelana kokwehla kwebhetri ngaphandle komonakalo womzimba, ngaxeshanye ngokukhawuleza kunokutshaja kunye neendlela ezisekelwe kwi-impedance ye-electrochemical , , , , , . Nangona kunjalo, i-pulse injection yindlela ejoliswe kwidatha kuphela efanelekileyo phantsi komlinganiselo ofanelekileyo wokuba idatha efunyenweyo ngokomzimba ibandakanya iimeko zokulahlwa kwesigaba sokusetyenziswa kwemodeli, eyaziwa njengomngeni omkhulu wokujongana kwidomain kuluntu lwesayensi yedatha. Ngaphandle kokuqhubela phambili kobuchwepheshe bokufunda kunye neendlela zokulungelelanisa idomains , , , imingeni iyanda kuba idomains ekujoliswe kuyo ekufuneka ihambelaniswe isafuna ulwazi olungaphambili xa isetyenziswa, kwaye iindlela zokufunda ezikhoyo akunakwenzeka ukuba zihlaziywe kwiimeko zokulahlwa ezingahambaniyo . Indlela enye kukwandisa isikali sovavanyo lwedatha phantsi kweemeko zokulahlwa ezingekhonjwanga ukunciphisa ubutyebi bedatha kunye nemixholo, nangona oko kungakhokelela ekwandeni kweendleko. U-Wang kunye nabanye bavule indlela yokucima ubushushu ukudibanisa ukhetho lwe-kinetic yokubala kubushushu kwiimeko ezibandakanya izithuthi ngexesha lenkqubo ye-thermochemical, evumela ukwakhiwa kweziseko zedatha zobulungisa kumlinganiselo omncinci we-thermochemical experimental kunye neendleko njengokongezwa kwedatha yoqeqesho lwemodeli ye-machine learning . Ngaphandle koko, ukufunda okunesizukulwana kubonise amathuba okuqikelelwa kwe-SOH kunye nedatha eyongeziweyo evela kwii-profiles ezixhonywe ngokuyinxenye, kugcina idatha efunekayo efunyenweyo , . Nangona kunjalo, ubutyebi bedatha kunye nemixholo kwisimo sokusetyenziswa kwakhona kunye nokurisayikliswa kwebhetri kubukhulu kakhulu ngenxa yodibaniso lweentlobo zemathiriyeli yekathode, iifomathi ezibonakalayo, uyilo olunobungakanani, kunye nemisebenzi yembali, kuthintela ukudibanisa okuyimpumelelo kolwazi olukhethekileyo malunga nokwehla kwiimodeli ze-machine learning. Ke ngoko, kufanelekile ukucinga ngobudlelwane phakathi kweemeko zokulahlwa kunye neempendulo zombane we-pulse zokuvelisa idatha, kungabikho kubaluleka kwexesha elide lokulinganisa umthamo kunye novavanyo olongezelelweyo lomzimba ngaxeshanye, ukuphucula ukusebenza kokuqikelelwa kwe-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 Kwesi sifundo, senza ukulungiswa kwangaphambili kwebhetri elahlisiweyo, oko kukuthi, ukuqikelelwa kwe-SOH, kusetyenziswa idatha yempendulo yombane we-pulse eyenziwe yi-autoencoder ethatha inxaxheba kwixesha elisindiswe lokulinganisa lomzimba kunye neendleko. Inkqubo ephezulu ye-general pretreatment steps, yoqeqesho lwemodeli yesizukulwana, kunye nezigaba zokusetyenziswa zemodeli ziboniswe kwiFig.   . Ingcamango yophando kukusebenzisa ukufunda okunesizukulwana ukusebenzisa idatha esele ilinganisiweyo yokugcina iimpendulo zombane we-pulse kwiimeko zokulahlwa eziqhubekayo, oko kukuthi, ukwahlulwa kwe-SOC kwisigaba sokuqokelelwa kwebhetri elahlisiweyo. Ukuqikelelwa kwe-SOH okudityaniswe kunye kuphunyezwa ngaphandle kwedatha yempendulo yombane we-pulse efotografiywe ngokomzimba, kugcina iindleko zombane zokulungisa kwangaphambili kunye nokukhutshwa kwekhabhoni okufuneka kungenjalo kuvela ekulungelelaniseni i-SOC ukuya kwinqanaba elilindelekileyo. Kuthelekiswa nendlela yokubala ye-ampere-hour kunye ne-pulse injection ye-machine learning, ukufunda okunesizukulwana kugqamisa isantya esikhawulezileyo sokubonelela kunye nokuqikelelwa okuchanekileyo. Ngedatha eyenziweyo, i-regressor elula iphumeza i-SOH ngempazamo ephantsi phantsi kweemeko zokulahlwa ezingaziwayo okanye ezingafikelelekiyo kwisiseko sedatha, umba onzima wokungaphaya kwedatha (OOD). Umba we-OOD usonjululwe ngokufunda imodeli yendalo phakathi kwesimo se-SOC kunye nempendulo yombane ekhutshwe yi-pulse ekhawulezileyo. Sibonisa ukubanzi kwendlela yethu ngokuqinisekisa ngamasampula angama-2700 afotografiywe ngokomzimba, aqhubela iintlobo ezi-3 zemathiriyeli yekathode, iifomathi ezibonakalayo ezi-3, uyilo olunobungakanani obu-4, kunye nemisebenzi yembali emi-4 (1 yovavanyo oluqheliswe kwi-lab, 3 zokuhamba kwe-EV, kubandakanya 1 imowudi yokuhamba kwe-powertrain yombane kuphela kunye ne-2 iimowudi zokuhamba ezixubileyo). Ukufunda okunesizukulwana kufanelekile kwi-interpolation kunye ne-extrapolation ngokudibanisa ulwazi olungaphambili lweemeko zokulahlwa kwisikali sedatha, kugcina ixesha lokulinganisa, iindleko zombane, kunye nokukhutshwa kwekhabhoni ngelixa kungalungisi ukuchaneka kokuqikelelwa kwe-SOH. Isifundo esigqibeleleyo sehlabathi esijolise kwisimo sokulahlwa kwebhetri ka-2030 sibonisa ukubaluleka kokuqikelelwa kwe-SOH okukhawulezayo nokuzinzileyo kwiimeko zokurisayikliswa kwe-pyrometallurgical, i-hydrometallurgical, kunye neyona nto ifanelekileyo. Ngokuhluzwa kwezobuchwepheshe-nemali, ukuqikelelwa kwe-SOH okuncediswe kukufunda okunesizukulwana kungagcina i-4.9 yeebhiliyoni zeedola ngeendleko zombane kunye ne-35.8 yeebhiliyoni yeekhilogram zeCO2  ukukhutshwa ngo-2030 kwihlabathi jikelele. Sixubusha ngokuqondakala kwemodeli, iziphumo zokulungiswa kokurisayikliswa, kunye nemiba ebanzi yezikhokelo zokulungiswa kwesixhobo sokurisayikliswa esizayo ezidibaniswe ne-machine learning. 1    Izigaba zokulungiswa kwangaphambili kwebhetri elahlisiweyo zibandakanya ukulungelelaniswa kwe-SOC, ukulinganisa komzimba, i-machine learning yokuqonda i-SOH, kunye nenkqubo yesibini yokugqiba isigqibo ngaphambi kokusetyenziswa kwakhona okanye ukurisayikliswa. Ukulungelelaniswa kwe-SOC kubhekisa ekuhlengahlengiseni amanqanaba e-SOC amabhetri alahlisiweyo ukuya kumanqanaba afunwayo kusetyenziswa indlela yesintu ye-constant current constant voltage (CCCV).   Umfanekiso we-attentional variational autoencoder wokuvelisa idatha yempendulo yombane we-pulse.   Ukusetyenziswa okusebenzayo kwemodeli yokwenza idatha yokuziqikelela i-SOH kufuna ukuba kungabikho kulungiso olongezelelweyo lwe-SOC. Imodeli ye-machine learning eqeqeshwe ixhaswa yidatha eyenziweyo kwaye ifumana iimpendulo zombane ze-pulse ezikhawulezileyo xa isetyenziswa kwisigaba sokuthunyelwa.   >Ezona zinzuzo zemali nezendalo kunye nesantya esikhawulezileyo sokwenziwa kwendlela yethu yokufunda eyenesizukulwana ngokugcina iindleko zokulungelelaniswa kwe-SOC kwisigaba sokulahlwa kwebhetri ka-2030 kwihlabathi jikelele. a b c d Izimvo Ukufunyanwa kwedatha Ubutyebi bedatha kunye nemixholo yimiba emikhulu ekufundeni okuxhaswa yidatha yokuchonga kunye nokuchaza ibhetri, ngakumbi kumabhetri alahlisiweyo. Siqokelele iisampulu ezingama-2700 kumabhetri angama-270 afotografiywe ngokomzimba kwizigaba ezibanzi ze-SOC, oko kukuthi, ukusuka kwi-5% ukuya kuma-50% ngomlinganiselo we

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

Le audio iveliswe ngolwimi lokuqala lwebali!

Ukuveliswa kwe-AI kuQhuba ukuSetyenziswa okung sustainability kweeBhetri ze-Lithium-Ion ezikhutshiweyo

About Author

IZIMVO

ZIJONGE IIMPAWU

ELI NQAKU LINIKEZELWE KU

Related Stories

Light-Mode

Classic

Newspaper

Minty

Dark-Mode

Neon Noir

Minty

HN StartUps