Ngokusho kombiko wemakethe we-Industrial AI kanye ne-AI 2021-2026 okhishwe muva nje, izinga lokwamukelwa kwe-AI ezindaweni zezimboni likhuphuke lisuka kumaphesenti ayi-19 laya kumaphesenti angama-31 eminyakeni engaphezu kwemibili. Ngaphezu kwamaphesenti angama-31 abaphendulile asebeqalise i-AI ngokugcwele noma ngokwengxenye emisebenzini yabo, amanye amaphesenti angama-39 okwamanje ahlola noma ahlola ubuchwepheshe.
I-AI ivele njengobuchwepheshe obubalulekile kubakhiqizi nezinkampani zamandla emhlabeni jikelele, futhi ukuhlaziywa kwe-IoT kubikezela ukuthi imakethe yezixazululo ze-AI zezimboni izobonisa izinga lokukhula eliqinile lonyaka ngemuva kobhubhane (CAGR) elingama-35% ukuze lifinyelele ku-$102.17 billion ngo-2026.
Isikhathi sedijithali sizale i-Internet of Things. Kungabonakala ukuthi ukuvela kobuhlakani bokwenziwa kusheshise ijubane lokuthuthukiswa kwe-Internet of Things.
Ake sibheke ezinye zezici ezibangela ukwanda kwe-AI yezimboni kanye ne-AIoT.
Isici 1: Amathuluzi esofthiwe engeziwe e-AIoT yezimboni
Ngo-2019, lapho ukuhlaziywa kwe-Iot kuqala ukumboza i-AI yezimboni, kwakunemikhiqizo embalwa yesofthiwe ye-AI ezinikele evela kubathengisi bezobuchwepheshe bokusebenza (i-OT). Kusukela ngaleso sikhathi, abathengisi abaningi be-OT bangene emakethe ye-AI ngokuthuthukisa nokuhlinzeka ngezixazululo zesofthiwe ye-AI ngesimo samapulatifomu e-AI esitezi sefektri.
Ngokusho kwedatha, cishe abathengisi abangu-400 banikeza isofthiwe ye-AIoT. Inani labathengisi besofthiwe abajoyina imakethe ye-AI yezimboni likhuphuke kakhulu eminyakeni emibili edlule. Ngesikhathi socwaningo, i-IoT Analytics ithole abahlinzeki abangu-634 bobuchwepheshe be-AI kubakhiqizi/amakhasimende ezimboni. Kulezi zinkampani, abangu-389 (61.4%) banikeza isofthiwe ye-AI.
Ipulatifomu entsha yesofthiwe ye-AI igxile ezindaweni zezimboni. Ngale kwe-Uptake, i-Braincube, noma i-C3 AI, inani elikhulayo labathengisi bezobuchwepheshe bokusebenza (i-OT) banikeza amapulatifomu esofthiwe ye-AI azinikele. Izibonelo zifaka phakathi i-ABB's Genix Industrial analytics kanye ne-AI suite, i-Rockwell Automation's FactoryTalk Innovation suite, ipulatifomu yokubonisana ngokukhiqiza ye-Schneider Electric, kanye muva nje, izengezo ezithile. Amanye ala mapulatifomu ahlose izinhlobo eziningi zokusetshenziswa. Isibonelo, ipulatifomu ye-ABB's Genix inikeza ukuhlaziya okuthuthukisiwe, kufaka phakathi izinhlelo zokusebenza ezakhiwe ngaphambilini kanye nezinsizakalo zokuphathwa kokusebenza kokusebenza, ubuqotho bempahla, ukusimama kanye nokusebenza kahle kochungechunge lokuhlinzeka.
Izinkampani ezinkulu zibeka amathuluzi azo esofthiwe ye-AI endaweni yokuthengisa.
Ukutholakala kwamathuluzi esofthiwe ye-ai kuqhutshwa futhi ngamathuluzi esofthiwe amasha aqondene nokusetshenziswa athuthukiswe yi-AWS, izinkampani ezinkulu ezifana ne-Microsoft ne-Google. Isibonelo, ngoDisemba 2020, i-AWS ikhiphe i-Amazon SageMaker JumpStart, isici se-Amazon SageMaker esinikeza isethi yezixazululo ezakhiwe ngaphambilini nezingenziwa ngezifiso zamacala asetshenziswa kakhulu ezimbonini, njenge-PdM, umbono wekhompyutha, kanye nokushayela okuzenzakalelayo, i-Deploy ngokuchofoza okumbalwa nje.
Izixazululo zesofthiwe eziqondene nokusetshenziswa zithuthukisa ukusebenziseka kalula.
Ama-software asetshenziswayo aqondene nokusetshenziswa, njengalawo agxile ekulungiseni izibikezelo, aya ngokuya evame kakhulu. I-IoT Analytics ibone ukuthi inani labahlinzeki abasebenzisa izixazululo zesofthiwe yokuphathwa kwedatha yomkhiqizo (i-PdM) esekwe ku-AI lenyuke laya ku-73 ekuqaleni kuka-2021 ngenxa yokwanda kwemithombo yedatha ehlukahlukene kanye nokusetshenziswa kwamamodeli okuqeqeshwa kwangaphambi kokuqeqeshwa, kanye nokwamukelwa kabanzi kobuchwepheshe bokuthuthukisa idatha.
Isici 2: Ukuthuthukiswa nokugcinwa kwezixazululo ze-AI kwenziwa lula
Ukufunda komshini okuzenzakalelayo (i-AutoML) sekuyinto ejwayelekile.
Ngenxa yobunzima bemisebenzi ehlobene nokufunda komshini (i-ML), ukukhula okusheshayo kwezinhlelo zokusebenza zokufunda komshini kudale isidingo sezindlela zokufunda komshini ezingasetshenziswa ngaphandle kobuchwepheshe. Insimu yocwaningo ephumelelayo, ukuzenzekela okuqhubekayo kokufunda komshini, ibizwa ngokuthi i-AutoML. Izinkampani ezahlukahlukene zisebenzisa lobu buchwepheshe njengengxenye yeminikelo yazo ye-AI ukusiza amakhasimende ukuthuthukisa amamodeli e-ML nokusebenzisa amacala okusetshenziswa kwezimboni ngokushesha. NgoNovemba 2020, isibonelo, i-SKF yamemezela umkhiqizo osuselwa ku-automL ohlanganisa idatha yenqubo yomshini nedatha yokudlidliza kanye nokushisa ukuze kuncishiswe izindleko futhi kunikwe amandla amamodeli ebhizinisi amasha kumakhasimende.
Imisebenzi yokufunda komshini (i-ML Ops) yenza kube lula ukuphathwa nokugcinwa kwemodeli.
Umkhakha omusha wokusebenza kokufunda komshini uhlose ukwenza lula ukugcinwa kwamamodeli e-AI ezindaweni zokukhiqiza. Ukusebenza kwemodeli ye-AI kuvame ukwehla ngokuhamba kwesikhathi njengoba kuthintwa yizici eziningana ngaphakathi kwesitshalo (isibonelo, izinguquko ekusakazweni kwedatha kanye namazinga ekhwalithi). Ngenxa yalokho, ukugcinwa kwamamodeli kanye nokusebenza kokufunda komshini kuye kwaba yinto edingekayo ukuze kuhlangatshezwane nezidingo zekhwalithi ephezulu zezindawo zezimboni (isibonelo, amamodeli anokusebenza okungaphansi kuka-99% Angase ahluleke ukuhlonza ukuziphatha okusongela ukuphepha kwabasebenzi).
Eminyakeni yamuva nje, izinkampani eziningi ezintsha zijoyine isikhala se-ML Ops, okuhlanganisa i-DataRobot, i-Grid.AI, i-Pinecone/Zilliz, i-Seldon, kanye ne-Weights & Biases. Izinkampani ezisunguliwe zengeze imisebenzi yokufunda komshini ezinhlinzekweni zazo zesofthiwe ye-AI ekhona, okuhlanganisa i-Microsoft, eyethula ukutholwa kwedatha ku-Azure ML Studio. Lesi sici esisha senza abasebenzisi bakwazi ukubona izinguquko ekusakazweni kwedatha yokufaka ehlisa ukusebenza kwemodeli.
Isici 3: Ubuhlakani bokwenziwa obusetshenziswa kuzinhlelo zokusebenza ezikhona kanye namacala okusetshenziswa
Abahlinzeki besofthiwe bendabuko banezela amakhono e-AI.
Ngaphezu kwamathuluzi esofthiwe ye-AI amakhulu akhona avundlile njenge-MS Azure ML, i-AWS SageMaker, kanye ne-Google Cloud Vertex AI, amasudi esofthiwe yendabuko njenge-Computerized Maintenance Management Systems (CAMMS), izinhlelo zokukhiqiza (MES) noma ukuhlela izinsiza zebhizinisi (ERP) manje kungathuthukiswa kakhulu ngokufaka amakhono e-AI. Isibonelo, umhlinzeki we-ERP i-Epicor Software ingeza amakhono e-AI emikhiqizweni yayo ekhona nge-Epicor Virtual Assistant yayo (EVA). Ama-ejenti e-EVA ahlakaniphile asetshenziselwa ukwenza izinqubo ze-ERP ngokuzenzakalela, njengokuhlela kabusha imisebenzi yokukhiqiza noma ukwenza imibuzo elula (isibonelo, ukuthola imininingwane mayelana nentengo yomkhiqizo noma inani lezingxenye ezitholakalayo).
Amacala okusetshenziswa kwezimboni ayathuthukiswa ngokusebenzisa i-AIoT.
Amacala amaningana okusetshenziswa kwezimboni ayathuthukiswa ngokungeza amakhono e-AI engqalasizinda yehadiwe/isofthiwe ekhona. Isibonelo esicacile umbono womshini ezinhlelweni zokulawula ikhwalithi. Izinhlelo zombono womshini zendabuko zicubungula izithombe ngamakhompyutha ahlanganisiwe noma ahlukene afakwe isofthiwe ekhethekile ehlola amapharamitha kanye nemingcele enqunyiwe (isb., ukugqama okuphezulu) ukuze kunqunywe ukuthi izinto zibonisa amaphutha yini. Ezimweni eziningi (isibonelo, izingxenye ze-elekthronikhi ezinezinhlobo ezahlukene zocingo), inani lemiphumela engamanga liphezulu kakhulu.
Kodwa-ke, lezi zinhlelo ziyavuselelwa ngobuhlakani bokwenziwa. Isibonelo, umhlinzeki we-industrial machine Vision uCognex ukhiphe ithuluzi elisha le-Deep Learning (Vision Pro Deep Learning 2.0) ngoJulayi 2021. Amathuluzi amasha ahlangana nezinhlelo zombono zendabuko, okwenza abasebenzisi bokugcina bakwazi ukuhlanganisa ukufunda okujulile namathuluzi ombono wendabuko kuhlelo lokusebenza olufanayo ukuze bahlangabezane nezimo zezokwelapha neze-elekthronikhi ezidinga ukulinganiswa okunembile kokuklwebheka, ukungcola kanye nezinye izinkinga.
Isici 4: Ihadiwe ye-AIoT yezimboni iyathuthukiswa
Ama-chip e-AI ayathuthuka ngokushesha.
Ama-chip e-AI ehadiwe afakiwe akhula ngokushesha, anezinketho eziningi ezitholakalayo zokusekela ukuthuthukiswa nokusetshenziswa kwamamodeli e-AI. Izibonelo zifaka phakathi amayunithi okucubungula ihluzo akamuva e-NVIDIA (i-Gpus), i-A30 kanye ne-A10, aqalwa ngoMashi 2021 futhi afaneleka ezimweni zokusetshenziswa kwe-AI njengezinhlelo zokuncoma kanye nezinhlelo zokubona ikhompyutha. Esinye isibonelo yi-Google's Tensors Processing Units (TPus) yesizukulwane sesine, okuyizisekethe ezihlanganisiwe ezinamandla (i-ASics) ezingasetshenziswa ngempumelelo kanye nesivinini esifinyelela ku-1,000 ekuthuthukisweni kwemodeli kanye nokusetshenziswa kwemithwalo yemisebenzi ethile ye-AI (isb., ukutholwa kwezinto, ukuhlukaniswa kwezithombe, kanye nezilinganiso zokuncoma). Ukusebenzisa ihadiwe ye-AI ezinikele kunciphisa isikhathi sokubalwa kwemodeli kusukela ezinsukwini kuya kumaminithi, futhi kuye kwabonakala kushintsha umdlalo ezimweni eziningi.
Ihadiwe enamandla ye-AI itholakala ngokushesha ngemodeli yokukhokha ngokusetshenziswa ngakunye.
Amabhizinisi e-Superscale ahlala ethuthukisa amaseva awo ukuze enze izinsizakusebenza zekhompyutha zitholakale efwini ukuze abasebenzisi bokugcina bakwazi ukusebenzisa izinhlelo zokusebenza ze-AI zezimboni. NgoNovemba 2021, isibonelo, i-AWS yamemezela ukukhishwa okusemthethweni kwezinhlelo zayo zakamuva ezisekelwe ku-GPU, i-Amazon EC2 G5, enikwe amandla yi-NVIDIA A10G Tensor Core GPU, yezinhlelo zokusebenza ezahlukahlukene ze-ML, kufaka phakathi izinjini zokubona zekhompyutha kanye nezincomo. Isibonelo, umhlinzeki wezinhlelo zokuthola i-Nanotronics usebenzisa izibonelo ze-Amazon EC2 zesisombululo sayo sokulawula ikhwalithi esisekelwe ku-AI ukusheshisa imizamo yokucubungula nokufeza amazinga okuthola anembe kakhulu ekukhiqizweni kwama-microchip nama-nanotubes.
Isiphetho kanye Nethemba
I-AI iphuma efektri, futhi izoba khona kuzo zonke izinhlelo zokusebenza ezintsha, njenge-PdM esekelwe ku-AI, kanye nokuthuthukiswa kwesofthiwe ekhona kanye nezimo zokusetshenziswa. Amabhizinisi amakhulu aqala izimo eziningana zokusetshenziswa kwe-AI futhi abika impumelelo, futhi amaphrojekthi amaningi anenzuzo ephezulu ekutshalweni kwezimali. Sekukonke, ukukhuphuka kwefu, amapulatifomu e-iot kanye nama-chip e-AI anamandla kunikeza ipulatifomu yesizukulwane esisha sesofthiwe kanye nokwenza ngcono.
Isikhathi sokuthunyelwe: Jan-12-2022