Kwenzekeni
I-Google DeepMind ikhiphe i-WeatherNext 3, imodeli yokubikezela isimo sezulu ye-AI eklanyelwe ngokukhethekile amapulazi omoya, abakhiqizi belanga, kanye nabaqhubi begridi kagesi. Imodeli ibuyekeza njalo ngehora futhi ikhiqize imifudlana emibili eyisisekelo yesimo sezulu: isivinini somoya nendawo ekuphakameni okungafika kumamitha ayi-100, kanye nekhava yamafu ngemisebe yelanga engenayo. Lokhu kulungiswa kwehora kuhloswe ukusiza opharetha ukuthi balindele ukushintshashintsha kokuphuma kwamandla avuselelekayo futhi balawule ukuzinza kwegridi ngempumelelo kakhudlwana.
I-Google DeepMind imemezele ukwethulwa kwe-WeatherNext 3 ngoSepthemba 14, iyichaza njengethuluzi labaqhubi bengqalasizinda yamandla. Imodeli yakhelwe ukubikezela izimo ezithinta ngokuqondile ukukhiqizwa kwamandla kagesi ovuselelekayo, ikakhulukazi imboni yomoya nelanga.
Uhlelo lusebenza nge-cadence yokuvuselela njalo ngehora, i-DeepMind ethi ibalulekile ngoba izimo zezulu ezithinta ukuphuma kwamandla zingashintsha phakathi nesikhathi sokuhweba esisodwa ezimakethe zamandla. Imodeli ikhiqiza imifudlana yedatha emibili ehlukene: eyodwa eyesivinini somoya nesiqondiso endaweni ephakeme ehlobene ne-turbine (kufika kumamitha angu-100) futhi enye isembozo samafu nemisebe yelanga ukuze kulinganiswe isizukulwane se-photovoltaic.
I-DeepMind ibeke lokhu kukhululwa njengengqalasizinda yoshintsho lwamandla, iphikisa ngokuthi isaziso sangaphambi kwesikhathi sokushintsha kwesimo sezulu sivumela amaqembu ukuthi alinganisele amagridi kagesi lapho ukukhiqiza okuvuselelekayo kushintshashintsha. Lokhu kwakhela emsebenzini wangaphambilini wenkampani ne-GraphCast, imodeli yesimo sezulu somhlaba wonke eyayisebenza ngokudlula izilinganiso zendabuko.
Imininingwane yomthombo: Google DeepMind ↗
Kungani kubalulekile
I-WeatherNext 3 ikhuluma ngenselele ebalulekile ekushintsheni kwamandla: ukuhlukahluka kwamandla omoya namandla elanga. Ngokuhlinzeka ngokulungiswa okuphezulu, izibikezelo zehora, imodeli ivumela abaqhubi begridi ukuthi bafanise kangcono ukunikezwa kwamandla ahlanzekile nesidingo sabathengi, ukunciphisa isidingo sokukhiqiza ikhophi ebiza kakhulu kanye nokwenza ngcono ukusebenza kahle kwezimakethe zamandla. Lokhu kumelela ukusetshenziswa okusebenzayo kwe-AI kungqalasizinda ebalulekile, idlulela ngale kokubikezela isimo sezulu esijwayelekile kuya ekuphathweni kwezinsiza ezikhethekile.
Amagridi kagesi ngokomlando ayethembele ekwenziweni okungathunyelwa, okufana namalahle, igesi, nezimboni zenuzi, ezingancishiswa uma kudingeka. Umoya nelanga azithutheki, okusho ukuthi ukuphuma kwazo kuncike esimweni sezulu esingalawuleki. Njengoba isabelo sezinto ezivuselelekayo sikhula, amagridi kufanele alindele ukushintshashintsha kwesizukulwane kunokuba asabele kukho.
Amaphutha esibikezelo kumandla avuselelekayo angaba nemiphumela ehlanganisiwe ekuzinzeni kwegridi. Umphumela womoya oweqile ungaholela ekushodeni kokuhlinzeka ngesikhathi sokufunwa okuphezulu, kuyilapho ukubukelwa phansi kumosha umthamo futhi kuphoqelele ukusetshenziswa komkhiqizo wokulondoloza imali obiza kakhulu. Ukubikezela kwe-AI yokulungiswa kwehora kwamahora kuhlose ukunciphisa lezi zingozi ngokuhlinzeka ngedatha enembile yokuthunyelwa kwesitoreji kanye nempendulo yesidingo.
Ukuthunyelwa kwemodeli enjalo ekhethekile ye-AI ngelebhu ehamba phambili efana ne-DeepMind kuphakamisa izinga lokuncintisana ngokunemba kwesibikezelo kulo mkhakha. Iphinde igqamise ukuhlangana okukhulayo kwe-AI nengqalasizinda ebalulekile, lapho idatha enembayo ibalulekile ekulawuleni inkimbinkimbi yegridi yamandla e-decarbonizing.
I-Interactive Mechanism: Indlela Esebenza Ngayo Ngempela
Hlola ubuchwepheshe obuyisisekelo ngemuva kwalokhu kuthuthukiswa ngokuhlanganyela.
A route planner searches possible journeys using explicit rules. What does this illustrate about AI?
Ongakubuka ngokulandelayo
Izinyathelo ezilandelayo zibandakanya ukubheka ukuthi yimaphi ama-opharetha wegridi ethile noma izinkampani zamandla ezithatha i-WeatherNext 3, njengoba i-DeepMind ingadalulanga abasebenzisi bamanje. Ukwengeza, ukunemba kwemodeli phezu kwemikhathi enwetshiwe yesibikezelo, ikakhulukazi ngaphezu kwamahora ayisithupha, kuzobaluleka ekusebenziseni kwayo ekubhideni kwemakethe yamandla yansuku zonke, lapho amaphutha esibikezelo ethwala izinhlawulo ezinkulu zezimali.
I-DeepMind ayidalulanga ukuthi yimaphi ama-opharetha wegridi, izinsiza, noma abakhiqizi bamandla abazimele abasebenzisa i-WeatherNext 3 njengamanje. Ukwemukelwa kokuqapha kwezinkampani ezinkulu zamandla kuzobonisa umthelela ongokoqobo wemodeli nokusebenza kwezohwebo.
Ukusebenza kwemodeli ekubhideni kwemakethe yamandla osuku olungaphambili kuyimethrikhi ebalulekile okufanele uyibuke. Njengoba umthamo omkhulu kakhulu wezinkontileka zikagesi uzinzile kuleli windi, ukunemba kwezibikezelo ezingaphezu kwamahora ayisithupha kuzonquma ukusetshenziswa kwayo kwezezimali kubadayisi bamandla kanye nabaqhubi.
Izimbangi, okuhlanganisa neziqalo ezifana ne-Tomorrow.io nama-ejensi kazwelonke wesimo sezulu, nazo ziqondise le makethe. Ukungena kwe-DeepMind kungase kusheshise ukuqamba okusha futhi kucindezele izinsizakalo ezikhona ukuze zithuthukise ukunemba kwazo kwesibikezelo kanye nokulungiswa.