isiZulu
Amakhamera e-AI okuqapha Ikhwalithi Yomoya Emadolobheni: Ukuchaza kabusha Ukuphathwa Kokungcola Ngobuhlakani Obubonakalayo
Kwadalwa ngo 01.29
Ukungcola komoya emadolobheni kuseyiphutha elikhulu kwezempilo yomphakathi emhlabeni wonke, lapho izinto ezincane ezitholakala emoyeni (PM2.5, PM10) kanye nezinto eziyigesi eziyingozi (VOCs) zibangela izigidi zokufa ngaphambi kwesikhathi minyaka yonke. Izindlela zendabuko zokuqapha ikhwalithi yomoya—ezisebenzisa iziteshi zezinzwa ezimile—sekuyisikhathi eside zilwa nezinkinga: ukufakwa okunganele, ukudluliswa kwedatha okubambezelekile, nokuhluleka ukuthola imithombo yokungcola ehambayo. Kufike amakhamera asebenzisa ubuhlakani bokwenziwa (AI): ubuchwepheshe obuguqula izinto obuhlanganisa ukubona ngamakhompyutha, ukufunda ngomshini, kanye nokuhlaziywa kwedatha yesikhathi sangempela ukuze kuguqulwe indlela amadolobha aqapha ngayo, ahlaziye, futhi aphendule ekungcoleni komoya. Ngokungafani nezinzwa ezijwayelekile ezilinganisa kuphela amazinga okungcola, amakhamera e-AI ahlinzeka ngombono obonakalayo, avumela amadolobha ukuthi ahambe kusukela ekuqoqweni kwedatha okungasebenzi ukuya ekungeneleleni okusebenzayo ekungcoleni. Lesi sihloko sihlole ukuthi kanjani Amakhamera e-AI ayiguqula indlela yokuphatha ikhwalithi yomoya emadolobheni, ubuchwepheshe bayo obuyinhloko, izicelo zayo zangempela, kanye nekusasa lobuhlakani obubonwayo ekubuseni kwemvelo.
Ukweswela Kokuqapha Okujwayelekile Kwekhwalithi Yomoya
Iminyaka eminingi, amadolobha ethembele eziteshini ezimisiwe zokuqapha ikhwalithi yomoya njengesilinganiso segolide sokulandelela izinto ezingcolisa umoya. Lezi ziteshi zinikeza ukunemba okuphezulu kodwa zihlupheka ngezinkinga ezakhelwe ngaphakathi ezithiya ukulawulwa kahle kokungcola kwase-urban. Okokuqala, ukumbozwa kwazo kunomkhawulo: idolobha elivamile lifaka iziteshi ezimbalwa nje, lishiya izindawo ezinkulu—ikakhulukazi izindawo zezimboni, izindawo zokwakha, nemigwaqo yasemakhaya—ezingaqashwanga ngokwanele. Lokhu kudala "izindawo ezingabonakali" lapho ukwanda kokungcola okwenzeka khona endaweni ethile kungabonakali. Okwesibili, idatha evela eziteshini ezimisiwe ivame ukubambezeleka ngamahora, okwenza kube nzima ukuthi abaphathi baphendule ngokushesha ezenzakalweni ezingalindelekile zokungcola, njengokuqubuka kotshani bokwakha noma ukuvuza kokukhishwa kwezimboni. Okwesithathu, izinzwa zakudala azikwazi ukubona imithombo yokungcola ngokubonwayo, zidinge uphenyo olwenziwa ngesandla ukuze kulandelwe imvelaphi yamaleveli aphezulu ezinto ezingcolisa umoya—inqubo edla isikhathi esiningi evame ukuvumela ukungcola ukuthi kuqhubeke.
Imakethe yomhlaba wonke yokuqapha ikhwalithi yomoya kulindeleke ukuthi ikhule isuka ku-R5.45 billion ngo-2023 yafinyelela cishe ku-R9 billion ngo-2030, iqhutshwa isidingo sezixazululo zokuqapha eziphelele nezisheshayo. Lokhu kukhula kuye kwavulela izikhamera ze-AI indlela, ezixazulula ububi bezindlela zakudala ngokuhlanganisa idatha enobungako nemininingwane ebonakalayo. Ngokungafani nezinzwa ezizwa nje ukungcola, izikhamera ze-AI ziyakubona—zikwazi ukukhomba umthombo ngqo, ukulandelela ngesikhathi sangempela, nokuphendula okuzenzakalelayo.
Indlela Izikhamera Ze-AI Ezisebenza Ngayo: Ubuchwepheshe Ngemuva Kokuqapha Ukungcola Okubonakalayo
Izikhamera ze-AI zokuqapha ikhwalithi yomoya zihlanganisa ubuchwepheshe obuyisisekelo obuthathu: ukuthwebula izithombe eziningi zemibala (multi-spectral imaging), izindlela zokufunda ngomshini (machine learning algorithms), kanye nokudluliswa kwedatha okusekelwa yi-IoT. Ngokuhlangene, lezi zingxenye zakha isiyingi esithi "bona-hlaziya-yenza" esiguqula idatha ebonakalayo ibe imininingwane yemvelo engenziwa kuyo isenzo.
1. Ukuthwebula Izithombe Eziningi Zemibala (Multi-Spectral Imaging): Ukubona Ngaphezu Kwakho Okubonakalayo
Ngokungafani namakhamera ajwayelekile okuphepha, amakhamera ekhwalithi yomoya asebenzisa izinzwa eziningi ezibamba kokubili ukukhanya okubonakalayo nokukhanya okungaphansi kwesikhumba. Lokhu kubavumela ukuba bangene enkungeni, inkungu, nezimo zokukhanya okuphansi—bengahlulwa "yindawo eyimpumputhe yasebusuku" ehlasele izinto zokubeka iso zakudala. Ngokwesibonelo, ikhamera yothuli i-Zhujiang Eagle Eye isebenzisa imodi ehlakaniphile yokukhanya okuphindwe kabili eshintsha ngokuzenzakalelayo phakathi kokukhanya okungaphansi kwesikhumba nokukhanya okumhlophe okugcwalisa ukukhanya ngemizuzwana engu-0.1, iqinisekisa ukubeka iso amahora angu-24 ngosuku ngokunemba kokuqashelwa kothuli okungu-99.7%. Izibalo zokuchayeka ezivumelana nezimo zithuthukisa ukucaca kwesithombe ngokulungisa isivinini sesivalo kanye ne-ISO ngokusekelwe ekugxileni komswakama; lapho amazinga othuli enyuka ngo-10mg/m³, isivinini sesivalo sifushane ngo-1/3 ukuze kuvinjelwe ukuchayeka ngokweqile futhi kugcinwe ukubonakala okucacile kwezikhukhula zokungcola.
2. Ukufunda Komshini: Ukuhumusha Izithombe Zibe Yidatha Yokungcola
Amandla wangempela amakhamera e-AI atholakala ekuguquleni kwedatha ebonakalayo ibe izilinganiso zamanani ezithinta ukungcola. Amamodeli athuthukisiwe wokufunda komshini—aqeqeshwe ezithombeni eziyizigidi zezinhlobo ezahlukene zokungcola (utshwala, intuthu, umusi)—ahlaziya izindinganiso zokuhlanganiswa kwezithombe, izindlela zokuhamba kwezinhlayiyana, nokuhlakazeka kokukhanya ukuze kubalwe amazinga okungcola. Ngokwesibonelo, ama-algorithm okuguqula amazinga ahlanganisa idatha yesithombe nezinga lezimo zezulu (isivinini somoya, izinga lokushisa, umswakama) ukuze kulinganiselwe amazinga e-PM2.5 kanye ne-PM10 ngephutha elingaphansi kuka-±15%. Amamodeli wokufunda okujulile, njenge-ResNet-50, angakwazi ukuhlukanisa izinhlobo eziyi-10 zezimo ezikhiqiza utshwala (ukuqhuma, ukulayisha, ukubeka) ngezinga lokuthola elingaphezu kuka-98%, okwenza ukukhomba imithombo yokungcola ngqo.
3. Ukuhlanganiswa kwe-IoT: Ukwabelana ngedatha ngesikhathi sangempela kanye nempendulo ezenzakalelayo
Amakhamera e-AI adlulisa idatha yesikhathi sangempela kumapulatifomu asekelwe emafini ngezingosi ze-IoT (5G, Mesh, noma i-fiber optics), avumela izikhulu zedolobha ukuthi ziqaphe ikhwalithi yomoya kusuka ku-dashboard eyodwa. Okubaluleke kakhulu, angahlanganiswa nemishini yokulawula ukungcola (imishayo yenkungu, izifafazayo, izikrini zothuli) ukuze kwakhiwe uhlelo lokuphendula oluzenzakalelayo. Lapho amazinga okungcola edlula imingcele esethiwe (isb., PM10 > 150μg/m³), ikhamera ithumela isignali ukuze kuvulwe imishayo yenkungu, ngamandla okufutha alungiswa ngokulinganayo namazinga okungcola—kunciphisa ukumosha kwamanzi ngo-40% uma kuqhathaniswa nezinhlelo zokufutha ezimile. Lolu hlelo oluvalekile luyasusa isidingo sokungenelela komuntu, luqinisekisa ukusabela okusheshayo ekukhuphukeni kokungcola.
Umthelela wangempela: Ama-AI Cameras esebenza emadolobheni omhlaba
Amadolobha emhlabeni wonke asevele asebenzisa ama-AI cameras ukuguqula ukuphathwa kwekhwalithi yomoya, athola ukuthuthukiswa okubonakalayo ekulawulweni kokungcola nasezempilo yomphakathi. Ngezansi kunezifundo ezimbili ezivelele ezibonisa inani elisebenzayo lobuchwepheshe.
Uphenyo Lwecala 1: Uhlelo Lokubheka I-Micro-Grid eHangzhou – Ukuhlonza Imithombo Yokungcoliswa Okusheshayo Ngama-300%
I-Hangzhou, China, isithuthukise uhlelo lokuhlola ikhwalithi yomoya olusekelwe ku-AI oluhlukanisa idolobha libe izindawo ezincane ezingama-48,000 (300m × 300m), ludala inethiwekhi yokuhlola eseduze kakhulu. Lolu hlelo luhlanganisa amakhamera e-AI anezinzwa zeselula ezifakwe kumabhasi kagesi namatekisi, lucubungula amaphuzu edatha angaphezu kuka-500,000 ngehora ukuze kutholwe izindawo ezinokungcola. NgoFebhuwari 2024, uhlelo luthole ukwanda okungazelelwe kwe-PM10—kaphinda kathathu kunesilinganiso sedolobha, namanani aphezulu adlula u-700μg/m³—endaweni enkulu yomgwaqo. Ukuhlaziywa kwe-AI kwezithombe zamakhamera kuthole ngokushesha indawo yokulahla imfucumfucu yokwakha eseduze njengomthombo. Iziphathimandla zendawo zazisiwe futhi zaphendula kungakapheli imizuzu engama-30, zisebenzisa izinyathelo zokulawula uthuli ezanciphisa amazinga e-PM10 kakhulu kungakapheli ihora.
Kusukela lapho kufakwa uhlelo, uHangzhou ufinyelele "izixwayiso zezinga lesibili, ukuphendula kwezinga leminithi, kanye nezixazululo zezinga lehora" zokungcoliswa kukadust. Ngonyaka ka-2024, isilinganiso se-PM10 sedolobha sehle ngo-6.7% unyaka nonyaka sathola u-47.1μg/m³, futhi izikhalazo ezihlobene nokungcoliswa kukadust zehla ngo-80%. Ukuqinisekiswa kokuxwayisa kwangaphambili kwe-95% kohlelo kanye nezinga lokuphumelela lokulandela imithombo le-80% kukhombisa ukuthi kanjani amakhamera e-AI angaguqula idatha ehlakazekile ibe yizinhlelo zokuphatha ezihlosiwe, ezisebenzisekayo.
Isifundo Sokuphumelela 2: Ukuqapha Indawo Yezimboni YaseWuhan – 60% Ukusebenza Okungcono Kokuphatha Utshwala
E-Wuhan, idolobha elikhulu lezimboni e-China, selisebenzise amakhamera e-Zhujiang Eagle Eye AI ezindaweni ezinokungcola okuningi njengezimboni zensimbi, iziteshi zemikhumbi, kanye nezimayela. E-Wuhan Qingshan Iron and Steel Plant, amakhamera e-AI avimbela ukuqhuma afakwe ezindaweni zokuphuma ezichoboza izinto, asebenza ngokuthembekile ezingeni lokushisa elidlulele (-40℃ kuya ku-85℃) futhi emelana namandla okushayisana angu-1000g. Amakhamera axhunywe ezithunjini zenkungu nasezindaweni zokuvikela uthuli, avumela amazinga othuli ukuthi abuyele emingceleni ephephile kungakapheli umzuzu owodwa ngemuva kokuqhuma—ugwema ukuvalwa kokukhiqiza okubizayo ngenxa yokungathobeli.
Echwebeni iWuhan Yangluo, amakhamera e-AI axazulule inkinga yokungcola kwezintuli okwesikhashana ngesikhathi kulayishwa impahla. Amakhamera aguqukela kumodi yamandla aphansi ngaphandle kwamahora okusebenza futhi avuke ngokuzenzakalelayo ihora elilodwa ngaphambi kokuba umkhumbi ubophe. Ngokuxhuma ku-cannons ezine zenkungu nge-Mesh networks, uhlelo lunciphise ukugcwala kwezintuli ngo-75% nokusetshenziswa kwamanzi ngo-60% uma kuqhathaniswa nokusebenza ngesandla. Abaphathi bezwele babika ukonga ngonyaka okungaphezu kwezigidi ezi-5 zama-yuan ezindlekweni zokuhambisana nemvelo.
Ngaphezu kokugadwa: Inani lezinhloso ze-AI Cameras emadolobheni
Amakhamera e-AI ahlinzeka ngokungaphezu kokulandelela ukungcola ngesikhathi sangempela—anikeza izinhlaka zedolobha imininingwane yamasu ukuze kuthuthukiswe izinqubomgomo zemvelo zesikhathi eside. Okunye okubalulekile ukumodela ukungcola okubikezelayo. Ngokuhlaziya idatha yamakhamera esikhathini esidlule, izitayela zezulu, namaphethini omgwaqo, i-AI ingabikezela ukuqubuka kokungcola kuze kube amahora angu-72 kusengaphambili. Ngokwesibonelo, uhlelo lwe-AI lwaseWuhan lusebenzisa idatha yomlando ukulingisa umthelela wokuvinjelwa kwethrafikhi noma amaparamitha we-fog cannon ekusebenzeni komoya, kusiza izikhulu ukuthi zakhe izinqubomgomo eziqondisiwe. Izifaniso zibonise ukuthi ukuvinjelwa kwezimoto ezisebenzisa uphethiloli ezihlangabezana ne-pre-Euro III endaweni enkulu yomgwaqo kunganciphisa amazinga e-PM2.5 ngo-14%—isiphakamiso esamukelwa futhi sagcinwa.
Enye inzuzo yamasu ukuhlanganiswa kwe-digital twin. Ama-AI cameras angakwazi ukudlulisa idatha kumamodeli we-digital twin emadolobheni, adale "amamephu okushisa okungcola" ashukumisayo abonisa ukusakazeka komoya ongcolile ngesikhathi sangempela. Emalokishini ezimboni aseWuhan, lawa makhadi okushisa abonise ukuthi ukugcwala kotshani kwakukhuphuka emahoreni okuqala ekuseni, okuphoqe iziphathimandla ukuthi ziqale ukuhlola okuhloswe ezindaweni zokwakha ezisebenza ebusuku. Ngenxa yalokho, amazinga e-PM2.5 esifunda ehla ngo-22%.
Ezikhalweni zamadolobha ezihlose ukufeza imigomo ye-carbon neutrality, amakhamera e-AI nawo anendima ebalulekile ekulandeleleni ukukhishwa kwe-greenhouse gas. Ithimba labacwaningi laseShayina lathuthukisa inkundla ye-AI ehlanganisa izithombe zamakhamera nokuqapha okunganyakazisi ukuze kumephwe ukukhishwa kwe-CO₂ emigwaqweni yedolobha ngesinqumo sika-30-meter. Le nkundla, eyasetshenziswa eShenzhen, ibona ngokunembayo imithombo yokukhishwa futhi ilinganise umthelela wokuhamba kwezimoto, isimo sezulu, kanye nesakhiwo sedolobha ekukhishweni—ihlinzeka ngedatha yokuhlola ukusebenza kwezinyathelo zokunciphisa i-carbon.
Izinselelo kanye neziqondiso zesikhathi esizayo
Nakuba amakhamera e-AI enikeza izinzuzo ezinkulu, ukwamukelwa kwawo kabanzi kubhekene nezinselele. Izindleko ziyisithiyo esibalulekile: amakhamera e-AI asezingeni eliphezulu namapulatifomu efu adinga utshalomali olukhulu lokuqala, olungase lube nzima emadolobheni amancane. Ukuvikeleka kwedatha kanye nobumfihlo nakho kuyizinto ezikhathazayo, njengoba izithombe zamakhamera zingase zithathe ulwazi olubucayi. Ukubhekana nalezi zinkinga, abakhiqizi bayathuthukisa amamodeli angabizi kakhulu, anomandla aphansi, futhi amadolobha ayasebenzisa ubuchwepheshe be-blockchain ukuvikela idatha futhi aqinisekise ukuba khona obala.
Ikusasa lokuqapha ikhwalithi yomoya nge-AI lisekhona ezintweni ezintathu ezibalulekile: ukusetshenziswa kwezindlela eziningi, ukubambisana komphakathi, kanye nokubambisana kwamanye amazwe. Okokuqala, amadolobha azohlanganisa amakhamera e-AI amile nama-drones kanye nezinzwa ezihambayo ukuze kwakhiwe inethiwekhi yokuqapha "eyehlanganisa isibhakabhaka nomhlaba", okuzosusa izindawo ezingaqaphwe kahle. Okwesibili, ukubamba iqhaza komphakathi kuzongezwa ngezinhlelo zesayensi yezakhamizi—njengezinhlelo zokusebenza ezivumela izakhamizi ukuthi zibike ukungcola kusetshenziswa omakhalekhukhwini bazo, ngekhamera ye-AI eqinisekisa imibiko futhi iqale izimpendulo. Ekugcineni, ukubambisana kwamanye amazwe kuzovumela amadolobha ukuthi ahlanganyele izinhlelo ze-AI kanye nedatha yokungcola, ebhekana nezinselelo zokungcola komoya ezingeqi imingcele.
Ngo-2030, imakethe yomhlaba wonke yokuqapha ikhwalithi yomoya cishe izobuso ngobuchwepheshe be-AI obusekelwe ezithombeni, njengoba amadolobha eqaphela ukuthi "ukubona kuyalawula" uma kukhulunywa ngokungcola. Ukuhlanganiswa kwamakhamera e-AI namapulatifomu amadolobha ahlakaniphile kuzodala uhlelo lokubusa kwemvelo olusabela ngokushesha, olusebenza kahle, futhi olugxile ezakhamini—oluguqula idatha yekhwalithi yomoya ibe ngcono obonakalayo empilweni yomphakathi kanye nekhwalithi yokuphila.
Isiphetho: Ukwamukela Ubuhlakani Bezithombe ukuze Kutholwe Imizi Ehlanzekile
Amakhamera e-AI awusona nje isisetshenziswa esisha sokuqapha ikhwalithi yomoya—ngokuyishintsha indlela amadolobha aphatha ngayo imvelo. Ngokuhlanganisa imininingwane ebonakalayo nokuhlaziywa kwedatha yesikhathi sangempela, avumela amadolobha ukuthi aqale ekulawuleni ukungcola ngokuphendula abe ngokuqondisa, anciphise ubungozi kwezempilo kanye nezindleko zokuhambisana nemithetho. Izindaba zempumelelo ezivela eHangzhou, Wuhan, naseShenzhen zibonisa ukuthi amakhamera e-AI aletha imiphumela ecacile: ukutholwa okusheshayo komthombo, ukusebenza kahle kokuqondisa okuphezulu, kanye nezinqubomgomo eziqondiswe kakhulu.
Njengoba ukuhlela kwemizi kusheshisa futhi ushintsho lwezulu lukhula, isidingo sokuhlola ikhwalithi yomoya ngezixazululo ezihlakaniphile sizokhula kuphela. Izixeko ezamukela amakhamera e-AI namuhla zizoba sezingeni elingcono lokuhlangabezana nemithetho yemvelo, ukufezekisa imigomo yokungabi ne-carbon, nokwakha imiphakathi enempilo, enezinzuzo ezihlala isikhathi eside kusasa. Ikusasa lekhwalithi yomoya emadolobheni libonakalayo—futhi lisebenza nge-AI.
Uxhumane
Sicela uxhumane nathi uhambele
WhatsApp
WeChat