Dangane da rahoton Kasuwancin AI da AI da aka fitar kwanan nan 2021-2026, ƙimar karɓar AI a cikin Saitunan masana'antu ya karu daga kashi 19 zuwa kashi 31 cikin sama da shekaru biyu kawai. Baya ga kashi 31 cikin 100 na masu amsawa waɗanda suka gama fitar da AI a cikin ayyukansu, wani kashi 39 cikin ɗari a halin yanzu suna gwaji ko gwajin fasahar.
AI yana fitowa a matsayin babbar fasaha ga masana'antun da kamfanonin makamashi a duk duniya, kuma IoT bincike ya annabta cewa kasuwar mafitacin AI na masana'antu zai nuna haɓakar haɓakar ƙwayar cuta ta shekara-shekara (CAGR) na 35% don kaiwa dala biliyan 102.17 nan da 2026.
Zamanin dijital ya haifar da Intanet na Abubuwa. Ana iya ganin cewa bullowar bayanan sirri na wucin gadi ya kara saurin ci gaban Intanet na Abubuwa.
Bari mu kalli wasu abubuwan da ke haifar da haɓakar AI da AIoT masana'antu.
Factor 1: Ƙari da ƙari kayan aikin software don AIoT masana'antu
A cikin 2019, lokacin da ƙididdigar Iot ta fara rufe AI masana'antu, akwai fewan samfuran software na AI da aka sadaukar daga masu siyar da fasahar aiki (OT). Tun daga wannan lokacin, yawancin dillalai na OT sun shiga kasuwar AI ta hanyar haɓakawa da samar da mafita na software na AI a cikin nau'ikan dandamali na AI don ginin masana'anta.
Dangane da bayanai, kusan dillalai 400 suna ba da software na AIoT. Adadin masu siyar da software da ke shiga kasuwar AI masana'antu ya karu sosai a cikin shekaru biyu da suka gabata. Yayin binciken, IoT Analytics ya gano 634 masu samar da fasahar AI ga masana'antun / abokan ciniki na masana'antu. Daga cikin waɗannan kamfanoni, 389 (61.4%) suna ba da software na AI.
Sabuwar dandamalin software na AI yana mai da hankali kan yanayin masana'antu. Bayan Uptake, Braincube, ko C3 AI, ɗimbin dillalai masu haɓaka fasahar aiki (OT) suna ba da dandamali na software na AI. Misalai sun haɗa da ABB's Genix Industrial Analytics da AI suite, Rockwell Automation's FactoryTalk Innovation suite, Schneider Electric na kansa masana'antun tuntuɓar dandamali, kuma mafi kwanan nan, takamaiman add-ons. Wasu daga cikin waɗannan dandamali suna yin niyya ga lokuta masu yawa na amfani. Misali, dandali na Genix na ABB yana samar da nazarce-nazarce na ci gaba, gami da aikace-aikacen da aka riga aka gina da kuma ayyuka don gudanar da ayyukan aiki, mutuncin kadara, dorewa da ingantaccen sarkar samarwa.
Manyan kamfanoni suna sanya kayan aikinsu na software na ai a filin shago.
Samuwar kayan aikin software kuma ana samun su ta sabbin kayan aikin software na musamman na AWS, manyan kamfanoni kamar Microsoft da Google. Misali, a cikin Disamba 2020, AWS ta saki Amazon SageMaker JumpStart, fasalin Amazon SageMaker wanda ke ba da saiti na da aka riga aka gina da kuma hanyoyin da za a iya daidaita su don yawancin abubuwan amfani da masana'antu na yau da kullun, kamar PdM, hangen nesa na kwamfuta, da tuƙi mai cin gashin kansa. dannawa kadan kawai.
Abubuwan ƙayyadaddun yanayin amfani da software suna haifar da haɓaka amfani.
Takamaiman kayan aikin software, kamar waɗanda aka mayar da hankali kan kiyaye tsinkaya, suna zama gama gari. IoT Analytics ya lura cewa adadin masu samarwa da ke amfani da hanyoyin sarrafa bayanan samfuran AI na tushen AI (PdM) mafita software ya karu zuwa 73 a farkon 2021 saboda karuwar nau'ikan hanyoyin bayanai da kuma amfani da samfuran horarwa, da kuma tartsatsi. amfani da fasahar haɓaka bayanai.
Factor 2: Ana sauƙaƙe haɓakawa da kiyaye hanyoyin AI
Koyon inji mai sarrafa kansa (AutoML) yana zama daidaitaccen samfur.
Saboda daɗaɗɗen ayyukan da ke da alaƙa da koyon injin (ML), saurin haɓakar aikace-aikacen koyon injin ya haifar da buƙatuwar hanyoyin koyo na injin da za a iya amfani da su ba tare da gwaninta ba. Sakamakon filin bincike, ci gaba da sarrafa injina, ana kiransa AutoML. Kamfanoni iri-iri suna yin amfani da wannan fasaha a matsayin wani ɓangare na sadaukarwar AI don taimakawa abokan ciniki haɓaka ƙirar ML da aiwatar da shari'ar amfani da masana'antu cikin sauri. A cikin Nuwamba 2020, alal misali, SKF ta ba da sanarwar samfur na tushen atomatik wanda ke haɗa bayanan sarrafa injin tare da rawar jiki da bayanan zafin jiki don rage farashi da ba da damar sabbin samfuran kasuwanci don abokan ciniki.
Ayyukan koyo na inji (ML Ops) suna sauƙaƙe sarrafa samfuri da kiyayewa.
Sabon horo na ayyukan koyo na inji yana nufin sauƙaƙe kiyaye samfuran AI a cikin mahallin masana'antu. Ayyukan ƙirar AI yawanci suna ƙasƙantar da lokaci yayin da abubuwa da yawa ke shafar shuka (misali, canje-canjen rarraba bayanai da ƙa'idodin inganci). Sakamakon haka, kulawar ƙirar ƙira da ayyukan koyo na inji sun zama dole don saduwa da ƙayyadaddun buƙatun yanayin masana'antu (misali, ƙirar da ke ƙasa da 99% na iya kasa gano halayen da ke haifar da lafiyar ma'aikaci).
A cikin 'yan shekarun nan, yawancin farawa sun shiga sararin ML Ops, ciki har da DataRobot, Grid.AI, Pinecone/Zilliz, Seldon, da Weights & Biases. Kamfanoni da aka kafa sun kara ayyukan koyan inji zuwa abubuwan da suke bayarwa na software na AI, gami da Microsoft, wanda ya gabatar da gano faɗuwar bayanai a cikin Azure ML Studio. Wannan sabon fasalin yana bawa masu amfani damar gano canje-canje a cikin rarraba bayanan shigar da ke lalata aikin ƙirar.
Factor 3: Sirrin wucin gadi da aka yi amfani da shi ga aikace-aikacen da ake da su da kuma lokuta masu amfani
Masu samar da software na gargajiya suna ƙara ƙarfin AI.
Bugu da ƙari ga manyan kayan aikin software na AI a kwance kamar MS Azure ML, AWS SageMaker, da Google Cloud Vertex AI, kayan aikin software na gargajiya kamar Tsarin Gudanar da Kulawa da Kwamfuta (CAMMS), Tsarin aiwatar da Kisa (MES) ko tsarin samar da albarkatun kasuwanci (ERP) yanzu ana iya ingantawa sosai ta hanyar allurar damar AI. Misali, ERP mai ba da Epicor Software yana ƙara ƙarfin AI ga samfuran da ke akwai ta hanyar Epicor Virtual Assistant (EVA). Ana amfani da wakilan EVA masu hankali don sarrafa ayyukan ERP, kamar sake tsara ayyukan masana'antu ko yin tambayoyi masu sauƙi (misali, samun cikakkun bayanai game da farashin samfur ko adadin abubuwan da ake samu).
Ana haɓaka shari'o'in amfani da masana'antu ta amfani da AIoT.
Ana haɓaka shari'o'in amfani da masana'antu da yawa ta ƙara ƙarfin AI zuwa kayan aikin kayan masarufi/software. Kyakkyawan misali shine hangen nesa na inji a aikace-aikacen sarrafa inganci. Tsarin hangen nesa na inji na al'ada suna sarrafa hotuna ta hanyar haɗe-haɗe ko kwamfutoci masu hankali sanye da software na musamman waɗanda ke kimanta ƙayyadaddun sigogi da ƙofa (misali, babban bambanci) don tantance ko abubuwa suna nuna lahani. A lokuta da yawa (misali, kayan aikin lantarki tare da nau'ikan wayoyi daban-daban), adadin ƙimar ƙarya yana da yawa sosai.
Duk da haka, ana sake farfado da waɗannan tsarin ta hanyar basirar wucin gadi. Misali, mai samar da injin masana'antu Vision Cognex ya fito da sabon kayan aikin zurfafa koyo (Vision Pro Deep Learning 2.0) a cikin Yuli 2021. Sabbin kayan aikin sun haɗa tare da tsarin hangen nesa na al'ada, yana ba masu amfani da ƙarshen damar haɗa zurfin koyo tare da kayan aikin hangen nesa na gargajiya a cikin aikace-aikacen iri ɗaya zuwa saduwa da buƙatun likita da muhallin lantarki waɗanda ke buƙatar ingantacciyar ma'auni na karce, gurɓatawa da sauran lahani.
Factor 4: Ana haɓaka kayan aikin AIoT na masana'antu
Kwakwalwar AI tana haɓaka cikin sauri.
Abubuwan da aka haɗa AI kwakwalwan kwamfuta suna girma cikin sauri, tare da zaɓuɓɓuka iri-iri don tallafawa haɓakawa da tura samfuran AI. Misalai sun haɗa da sabbin sassan sarrafa hoto na NVIDIA (Gpus), A30 da A10, waɗanda aka gabatar a cikin Maris 2021 kuma sun dace da yanayin amfani da AI kamar tsarin shawarwari da tsarin hangen nesa na kwamfuta. Wani misali shine Rukunin Gudanar da Tensors na ƙarni na huɗu na Google (TPus), waɗanda ke da ƙarfi na musamman-manufa hadedde da'irori (ASics) waɗanda za su iya cimma har sau 1,000 mafi inganci da sauri cikin haɓaka ƙirar ƙira da tura don takamaiman aikin AI (misali, gano abu). , Rarraba hoto, da ma'auni na shawarwari). Yin amfani da kayan aikin AI da aka keɓe yana rage lokacin ƙididdige samfurin daga kwanaki zuwa mintuna, kuma ya tabbatar da cewa ya zama mai canza wasa a lokuta da yawa.
Ana samun kayan aikin AI mai ƙarfi nan da nan ta hanyar ƙirar biyan kuɗi-da-amfani.
Kamfanonin Superscale suna haɓaka sabobin su koyaushe don samar da albarkatun ƙididdiga a cikin gajimare ta yadda masu amfani da ƙarshen za su iya aiwatar da aikace-aikacen AI na masana'antu. A cikin Nuwamba 2021, alal misali, AWS ta sanar da sakin hukuma na sabbin misalan tushen GPU, Amazon EC2 G5, wanda NVIDIA A10G Tensor Core GPU ke bayarwa, don aikace-aikacen ML iri-iri, gami da hangen nesa na kwamfuta da injunan shawarwari. Misali, mai ba da tsarin ganowa Nanotronics yana amfani da misalan Amazon EC2 na ingantaccen tsarin kula da ingancin AI na tushen AI don haɓaka ƙoƙarin sarrafawa da cimma daidaiton ƙimar ganowa a cikin kera microchips da nanotubes.
Kammalawa da Hasashen
AI yana fitowa daga masana'anta, kuma zai kasance a ko'ina a cikin sabbin aikace-aikace, irin su PdM na tushen AI, da kuma haɓakawa ga software da ake amfani da su. Manyan kamfanoni suna fitar da shari'o'in amfani da AI da yawa da bayar da rahoton nasara, kuma yawancin ayyukan suna da babban koma baya kan saka hannun jari. Gabaɗaya, haɓakar girgije, dandamali na iot da kwakwalwan AI masu ƙarfi suna ba da dandamali don sabon ƙarni na software da haɓakawa.
Lokacin aikawa: Janairu-12-2022