Isisitshisele sesikhundla sokugcina isikhundla esingaphansi esinamandla okusuka kumqondiso wokuphakanyiswa ngokuhorizontali wokuphakanyiswa kwe-AI backtracking

Ukhulu Lwesikhwama

Ama-algorithm ajwayelekile okubuyela emuva asebenzisa imodeli yejometri yesikhundla selanga nezilinganiso ze-array ukuze abale ama-engeli omugqa avimbela ukufiphaza phakathi komugqa. Lokhu kusebenza kahle endaweni eyisicaba ngokuphelele enemizila efanayo kodwa izindawo zangempela zephrojekthi zinokuhlukahluka kwe-microtopography, isikhala semigqa esingafani kusuka ekubekezelweni kokufakwa kwesisekelo, nezimo zokukhanya ezihlakazayo imodeli yejiyometri engakwazi ukuzichaza.

Isisistemu sokufunda emuva ngokusebenzisa i-AI sokufunda emuva ngokusebenzisa i-axi esinye sikhombisa ukuthi isithembiso sokufunda emuva sishintsha isibonelo sokubala ngokusebenzisa umshinzi wokufunda emuva owenziwe ngokusebenzisa i-machine, okwenzeka ngokuphathelana nokubala kwezinkampani zezinsimbi zokugcina izinsimbi zokugcina ngokushesha. Ngakho-ke, akabali kuphi ukuthi kuzoba kubhekwa kanjani, kodwa i-AI ibona ukuthi kubhekwa kanjani futhi ishintsha ngokufanele.

Ukuba uyafuna:

Ukubhalwa kwesikhathi se-DC string yezinsimbi zonke ziyabhalwa emaphakathini okuhlanganisa (akukho amasensha ayengeziwe kufuneka — ulwazi luyasebenza kwi-SCADA system yomphakathi)

Ngeminyaka yokufunda emuva (kusasa kakhulu, emuva kakhulu), umshinzi wokufunda emuva uyashintsha ngokucacile izinkampani zezinsimbi zonke

Xa isikhathi sezinsimbi siphula ngokulingana nesizinhloko sakhe — okubonisa ukuthi kubhekwa — umshinzi wokufunda emuva uyashintsha lezinsimbi zokufunda emuva kuphela, akushinthi zonke izinsimbi ngokufanayo

Ngemuva kwezinsuku, umodeli wokufunda emuva uyafunda isimo sokubhekwa esihlukene nesikhathi, isikhathi/usuku/imithwalo yezinsimbi, futhi uyapredicta izinkampani ezifanele kusikhathi/usuku/imithwalo yezinsimbi, okunciphisa ukufuna ukushintsha ngokucacile

Inzuzo elinganisiwe: Ukulahleka kokuthwebula kwehliswe ngama-90% uma kuqhathaniswa nokubuyela emuva kwejometri, okuhumusha cishe ku-24% umkhiqizo wamandla wonyaka owengeziwe kunomlandeli ojwayelekile ngokubuyela emuva kwejometri owenziwe ngezindleko zensimbi ezingeziwe ngaphandle kokuthuthukiswa kwe

Amalungiselelo Angaphakathi

Izicelo

Amaphoyisa aphakathi emagugu engafanele noma amaphepha okuhambisana kwemithombo

Amaphoyisa okukhula ngokuphelele (GCR 0.45)

Amaphoyisa aphakathi okunye kwe-1% yokuthuthukiswa kokuphatha kuyabonakala njengokuthuthukiswa okukhulu kwezimali zokubhokha izimpahla zokusebenza

Kwenzeka ukufaka kuze kuziphatha kwezindawo zokusebenza zezimpahla zokusebenza (ukuguqula ukuvunywa kwezimpahla kuphela)

Imibuzo Evame Ukubuzwa

Umbuzo. Ingabe ikhonkhroller yase-AI iyadinga ukuxhumana ne-intanethi?

Cha. Imodeli yase-ML iyasebenza kwekhonkhroller yase-edge ephakathi kwesikhumbuzo sokusebenza. Ukufunda kwale modeli kusuka kwezindawo zokusebenza kuyasebenza ngamazwi amabili–ne-nezine ezinsukwini zokusebenza kule ndawo ukuze ifunde amaphepha okuhambisana okwesikhumbuzo. Emva kokufunda, imodeli iyasebenza phakathi kwesikhumbuzo. Ukuxhumana ne-cloud kusukela kwezindawo zokusebenza kuyakwenzeka ngokuzimisayo ukuguqula ukuvunywa kwezimpahla kanye nokubhekana kwezimali zokusebenza.

Umbuzo. Kungakwenzeka ukufaka kuze kuziphatha kwezindawo zokusebenza zezimpahla zokusebenza?

Yebo, uma izimpahla zezimpahla zokusebenza (slew drive, torque tube, foundation) ziyasebenza ngendlela enjalo. Ukuguqula kuyizimpahla zokuvunywa kanye nokugcina ukuthi izimpahla zokusebenza ziyasebenza kwezimpahla zokusebenza zokusebenza. Akukho guqulo lomthwalo wokusebenza lokho.

Ukufaka kwalemyezo

Amagama okuhlela ukubuka kwe-SEO

Okokuqala: i-AI backtracking solar tracker, isistimu yekugcina ngokusha yamaphoyisa, i-machine learning PV tracker, i-intelligent backtracking

Okwesibili: ukunciphisa ukungcoliswa kwesibonelo sokugcina ngokusha, isithembiso sokubuyela emuva esivumelana, ukuphucula ukugcina ngokusha ngumhla