版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领
文档简介
StateofAIfor
Decarbonisation2025
January2026
1
2
Contents
3Introduction
10UnlockingDomesticDecarbonisation
18EnablingNetZeroInfrastructure
27MaximisingFlexibilityinEnergyNetworks
36DecarbonisingManufacturingInputs
44ImprovingManufacturingProcessEfficiency
53OptimisingSoilManagement
62MinimisingMethaneinAgriculture
70OptimisingEVInfrastructureandCharging
78DecarbonisingFreightandFleets
3
2025:lotsofhype,sometangibleprogress
Thehype
AIhasbeenconstantlyintheheadlinesthisyear.Jaw-
droppinglylargeinvestmentsinAIdatacentres
underpinnedUSeconomicgrowth.Securingenough
energyforthosedatacentresbecameamajorconcern
andtechcompaniesstartedfundingnuclearpowerplantsandgridreinforcement.Businessesaroundtheworld
startedintegratinggenerativeAIintotheirsoftwareandprocesses,withmixedresults.Debatesaboutthefutureofworkintensified,withpeopleeitherworriedorexcitedaboutAI’spotentialtoautomatejobs.Evennational
carbonbudgetsstartedincludingassumptionsthatAIcouldsavemillionsoftonnesofCO2emissions.
Butamidstthattorrentofnewsandhypeitcanbehardtofindtheanswertoanimportantquestion:howeffectivelyisAIbeingappliedtokeysocietalchallengeslike
decarbonisation?
Thereality
ThisannualreportanswersthatquestionwithareviewofhowUKapplicationsofAIfordecarbonisationhavematuredoverthelastyear.Thisrangesfrom
applicationsthatarereachingscaleandmeaningfullycontributingtonationaldecarbonisation,throughto
earlierstageresearchthathasmadenotableprogress.
Therehavebeentangiblestepsforwardinsomeareas
thisyear.AI-poweredEVchargingisalreadyplayingasignificantroleinmanagingourlowcarbonelectricitygrid.HeatpumpinstallationsarequickerandcheaperbecauseofAI.SteelfurnacesandcementplantsarereducingemissionsthroughAI-optimisedoperations.
ButinotherareasprogresshasbeensloworhinderedbygenerativeAIhype.Lotsmoreworkisrequiredto
fullyrealisethebenefitsofAIfordecarbonisation.
4
MeasuringprogressonthenineGrandChallenges
Inour
originalreport
ADViCEidentified
sevendecarbonisationGrandChallengeswhereAIcouldmosthaveimpact.
Thisyearwehavealsoaddedtwotransport-relatedGrandChallenges.
EachGrandChallengeisbrokendownintomorespecificchallengesforAItotackleinthe
DecarbonisationChallengeCards
.
ToidentifyprogressthisyeartheADViCE
teamofdomainexpertsreviewedpotentialexamplesofprogressineachspecific
challenge(sourcedusingAI-basedresearchtoolsandconversationswithstakeholders).ForeachGrandChallengetheexpertsthenidentifiedkeythemesandexamplesof
whereAIhasmadeprogressin2025.
5
Overallprogressin2025
In2025AIhadameasurableimpactontheUK’sabilitytooperatealowcarbonelectricitygrid.AI-poweredsolarnowcastingreducedemissionsbyanestimated300,000tonnes,smartEVchargingloweredpeakelectricityusagefromEVsby42%andvirtualpowerplantshelpedbalancethegrid.
ThisyearsawanotablestepforwardinAIapplicationstodomesticdecarbonisationwiththenationalrolloutoftwoAItoolslookingtostreamlinedifferentpartsoftheheatpumpinstallationjourney.
Inmanufacturingandtransport,adoptionofAIforprocessefficiencyandoptimisationcontinuedtogrow.Largerorganisationsstartedtomovefromdemonstrationtodeployment,thoughsmaller
organisationstypicallylaggedbehind.
AIforsoilmanagementprogressedfromtheoreticalpilotingtoearlycommercialisation,whileapplicationstoagriculturalmethanereductionmadesomeprogressattheresearchstage.
LessprogresswasmadeonapplyingAItodecarbonisingmanufacturinginputsorelectrifyingfreight,partlyduetothehighcapitalcostsinvolved.ThesemayneedadditionalinterventionstoaccelerateAIadoption.
Visualsummaryofprogress
ShownbelowisanassessmentofhowfarsolutionsforeachGrandChallengeprogressedbetween2024and2025,alongwitharoughestimateofpotentialprogressin2026basedontheenablersandblockersidentifiedinthisreport.
NosignificantEarlyexploratoryMajorresearchProofofconceptPilotsinsomeMeaningfulimpactforGrowingimpactNational-scaleMaximumdecarbonisation
decarbonisationimpactimpact
workresearchprogrammesdemonstratorsorganisationsindividualorgsacrosssector
I
Unlockingdomesticdecarbonisation
20242025⃞2026
EnablingNetZeroinfrastructure2024一20252026
20242025
2026
-----⃞
Decarbonisingmanufacturinginputs
Improvingmanufacturingprocessefficiency
Optimisingsoilmanagement
Minimisingmethaneinagriculture
OptimisingEVinfrastructureandcharging
Decarbonisingfreightandfleets
2026
2025
2024
2026
2024
2025
------------⃞
2026
2024—一2025
--------------⃞
2026*
20242025
2026
2024
2025
Maximisingflexibilityinenergynetworks202420252026*
2024
Key
Maturityattheendof2024
Maturityattheendof2025
Potential
maturityattheendof2026
(estimate)
2025
2026
6
*Note:shorterarrowsfor2026in‘National-scaledecarbonisationimpact’arenotanindicatorprogressisexpectedtoslow,butbecausethefigureisfocusedonthegrandchallengesinearlierstages.AI’sdecarbonisationimpactforthemorematurechallengesisexpectedtogrowsteadilyin2026andfuturereportsmaystarttotrackthatmoreclearly.
7
Somekeynumbersfrom2025
79%
ofEVowners
haveasmart
charger
300,000
tonnesofCO2
avoidedeach
yearusingsolar
nowcasting
1500
farmsusing
autonomous
droneflightsto
inspectcrops
2%
reductioninCO2
fromcement
productionusing
AI
50%
reductioninheat
pump
installationtime
usingAItools
Thestartupecosystemcontinuestogrow
TheUKAIfordecarbonisationecosystemhasshownsteadygrowthinrecentyears.Mostcompaniesareintheseedandventurestage,whichbroadlymirrorstheoverallAIecosystemintheUK.
ActiveCompanies
379*
366
7%
343
307
12%
268
Seed
236
45%
Venture
Growth
Established
36%
20202021202220232024Q3*2025
8*DatasourcedfromBeauhurstcompanydatabaseonaselectionof380companiesidentifiedthroughacustomquery.DatacollectedupuntilQ3of2025.Categoriesofevolutionstagesofcompaniescanbefound
here
Isthisreportmissingsomething?
Youcancontactusat
ADViCE@turing.ac.uk
,or
signupto
ourmailinglist
tobenotifiedofthewebinars,workshopsandinperson
eventsthatwehost.MoredetailsaboutthelaunchofthedatabaseofAIfor
decarbonisationsolutionswillbeannouncedshortly.Wealsohavea
knowledge
base
withkeyresources.
AboutADViCE
AIforDecarbonisation’s
VirtualCentreofExcellence(ADViCE)isaprogrammefundedbytheDepartmentforEnergySecurityandNetZero.Itisapartnership
betweenDigitalCatapult,EnergySystemsCatapultandTheAlanTuring
Institute.
ADViCE
existstojoinuptheAIfor
decarbonisationecosystem.We’realwayskeentohearaboutwhatyou’redoinginthisspace.
Thisisanannualreportandforfuture
editionswewillbeworkingwiththe
ecosystemtocurateacontinuallyupdateddatabaseofAIfordecarbonisation
solutionsandtheirprogress.
We’dlovetohearfromyouwhetheryou:
•haveadecarbonisationchallengeyouthinkismissing
•areworkingonanAIsolutionthatisdeliveringdecarbonisation
•orjustknowaboutsomethinginterestingwehaven’tcovered.
9
Grand
Challenge1
Unlocking
Domestic
Decarbonisation
10
11
Overview:UnlockingDomesticDecarbonisation
TherearesomewellstudiedAIapplication
areaswhichhavebothacademicresearchandahandfulofearly-stagecommercialofferings,butfewexamplesofsuccessatscaleyet.
BothdataavailabilityandthepotentialmarketforAIaregrowingasdomesticdecarbonisationpicksupspeed.
Residentialheatingisresponsibleformorethan13%ofgreenhousegasemissionseachyear,
andsoisessentialindecarbonisingtheUK
economy.However,decarbonisinghomes
requireschangestobothheatingsystemsandconsumerbehavioursineveryhomeintheUK.Engagingconsumersinthatprocess,financingit,anddeliveringitatpaceareallmajor
challenges.
Therehavebeenanumberoflargepublic
sectorinnovationfundingprogrammesinthisareaoverthelastfewyears,includingtheNetZeroHeatprogramme.
ThisyearsawanotablestepforwardinAI
applicationstothisareawiththenationalrolloutoftwoAItoolslookingtostreamlinedifferent
partsoftheheatpumpinstallationjourney.
12
AIapplicationsthatcouldaddressthischallenge
13
ThemesinAIadoptionfordomesticdecarbonisation
1
2
3
4
5
UseofAIinacceleratingheatingsystemdesignhasmovedfrompilots(e.g.
Geo
’sAISmartHeatPathwayin
2022-24)tonational-scaleproductlaunches(
HeatGeek’sZeroDisruptAI
)thatarehelpingtosignificantlyreduceheatpumpinstallationcosts.
Frictionintheadminprocessforlowcarbontechnology(LCT)installshasbeenreducednationwide,withtheabilitytoprovidesame-dayauto-approvalsforLCTinstallsusing
ENA’sConnectDirect
.
Specialistchatbotsaimedatsupporting
installers
and
consumers
withheatpumpinstallationshavebeendevelopedandtrialled–includingapublic-facingrollout–buthavenotyetseensignificanttraction.
AdoptionofAItoimproveidentificationofvulnerableenergyconsumersisnowwidespread,from
callanalysisby
ScottishPower
,to
fuelpovertyriskmapping
tobettertargetgrantsupport,toSSEN
forecastingfuturevulnerability
atalocallevel.
AItoautomaticallybreaksmartmeterusageintodifferentappliances(knownasnon-intrusiveloadmonitoring,NILM)remainsanactiveareaofacademicresearch.Itisalreadyemployedinanumberofconsumer-facingappsincluding
Loop
,whichhashelped>150kusersreduceenergyusagebyanaverageof15%.
Heatpumpdesign:HeatGeekZeroDisrupt
HeatGeek
haveusedAIextensivelyinautomatingpartsoftheheatingsystemdesignprocess.Thisincludes:
•LiDAR-basedautomatedinternalsurveyingofhomes
•Computervision-basedheatpumpsitingassessment
•Automationofformpopulationandcommunications
•AI-basedselectionofoptimaldesignparameterstoachievetargetefficiencyatminimumcost
Intrialsthishas
reducedcosttocustomers
*by~75%andinstallationtimeby~50%,andhasnowbeenrolledout
nationallyandisinusebyallHeatGeekinstallers.
Thisdirectlyaddressesthecostanddisruptionchallengesthatareslowingheatpumpadoptionandislikelyto
significantlyacceleratetherateofheatpumpinstallations.
14*afterboiler
upgradeschemegrantsof£7.5karetakenintoaccount
15
Connectionapproval:ENAConnectDirect
The
EnergyNetworkAssociation
haveintegratedAIintothenationalconnectionapplicationservicefordomesticlow
carbontechnologies(LCTs)likeheatpumpsandEVchargers.
Itutilisescomputervisionto
reviewphotosofcut
-
outs
(essentiallythefusebetweenahomeandthegrid)to
removetheneedforahumantorevieweveryphotoandenableinstantapprovalofapplicationswhereitwasclearnocut-outupgradewasrequired.
Ithasreducedthetimeandcostofcompletingpaperworkforinstallersandhasbeenusedtospeedupover
185kLCT
approvals
forconsumers.
16
Enablersforthenext12months
Heatpumpinstallationsincreasing
HeatpumpinstallationratesareincreasingandtheUKnowhasover
300,000heatpumps
installed.
ThismeansthereisbothmoredataavailablefortrainingAI(including
freedatafromtrials
)andalargermarkettodriverevenuesforAIinnovators.
Smartmeterdatastartstobecomemoreavailable
Severalongoinginitiatives
tomakeaccesstosmartmeterdataeasierareunlikelytohaveanimpactin2026,but
syntheticsmartmeterdata
isavailablenow.
GenerativeAItoolingmatures
AsgenerativeAImodels(andassociatedtooling)mature,itbecomesincreasinglyfeasibleforthemtobeusedatscaletosupportconsumersindecarbonisingtheirhomes.
17
Remaininggapsandbarriers
Electricityremainsmoreexpensivethangas
Despiterecentgovernmentmovestoreducepolicycostsonelectricity,electricityremainsmorethan4xas
expensiveasgas.Thismakesitextremelydifficultfor
heatpumpstobecostcompetitivewithgasboilers(eventhoughtheyare3-4xmoreefficient).
ThisincreasestheimportanceofusingAItoimproveheatpumpoperation–bothimprovingefficiencyandshiftingusagesohouseholdscanbenefitfromtime-of-usetariffs.
Interoperabilityisacontinuingchallenge
LackofopenAPIsandinteroperablestandardsfordataandcontrolsremainsasignificantbarriertoapplyingAItodomesticheating.
Grand
Challenge2
EnablingNetZero
Infrastructure
18
19
Overview:EnablingNetZeroInfrastructure
SomeestablishedareasofAIusage,particularlyinoptimisingroutesandlayoutsfornewassets,havecontinuedtoseesteadygrowthinadoption.
Continuedpoliticalfocusonthisareameansitis
likelywewillcontinuetoseestrongfundingforAI
applicationsinthisareaoverthenexttwelvemonths(includingviatheStrategicInnovationFundand
GreenIndustriesGrowthAccelerator),butthe
potentialforbreakthroughremainsconstrainedby
challengesintegratingwithhard-to-change
bureaucraticprocesses,aswellasinsufficientdata.
Electrificationofheatingandtransportation,
combinedwithincreasedrenewables,meanswe
needbothsignificantexpansionofourelectricity
networksandwaystomanagenetworkconstraints.Deliveringattherequiredscale–andpace–isa
realchallenge,withnewrenewableprojectsheldupbydelaysoruncertaintyinnetworkconnections.
Despitesignificantattentionfromgovernmentandindustry,therehasbeenlittleprogressonapplyingAItothegridconnectionqueuethisyear.
ApplicationsofAIforreal-timeoptimisationandcontrolhavereachedreal-worldpilotstagein
offshorewindanddistributionnetworks.
20
AIapplicationsthatcouldaddressthischallenge
21
ThemesinAIadoptionforNetZeroinfrastructure
1
2
3
4
5
BigpromisesaboutAIbeingabletostreamlinegridconnectionsandaccelerateapprovalshaveyettodeliver,withthegovernment-announced
Connect
tool(matchingcapacitytodemand)havingbeenpaused.
AItoolstooptimiseinfrastructureplacementarerelativelymatureandwidelyadopted(e.g.
ContinuumOptioneer
,KinewellEnergy’s
cable
and
turbine
layoutoptimisation),reducingprojectdevelopmenttimelinesandcosts.
AIforreal-timecontrolandoptimisationofassetshasseennoticeableprogress,includingoperationalpilotsof
reinforcementlearningforwindfarmcontrol(
AIOLUS
)anddistributedcoordinationandcontrolofnetworkassets(
Constellation
).TheNationalEnergySystemOperator’s
VOLTA
programmeisscopingoutadoptionofAIwithinthenationalcontrolroom.
Cross-sectorandmulti-scaleplanningremainsakeychallengethatisstartingtobeaddressed.Therearenotabledata-sharinganddigitaltwininitiativesinthisspace(
ENSIGN
&
CReDo
+),buttheenablingconditionsforAItohavealargeimpactarenotyetinplace.
GenerativeAIhaslargelybeenlimitedtodataqualityenablers(e.g.
publicsentimentanalysis
and
datadiscovery
)andisnotcurrentlydisplacingcoreengineering/optimisationworkflows.
RouteOptimisation:ContinuumIndustries
Deployingnewenergyinfrastructureisslowandcostlybecauseplannersmustmanuallyevaluatethousandsofrouteoptionsagainstengineering,environmental,andpermittingconstraints.
ContinuumIndustries
isworkingwithmajornetworkoperators,includingNationalGrid,SSENTransmission,SGN,andNationalGasTransmission,tosolvethiswithOptioneer.
TheplatformusesAI-drivengeospatialoptimisationandconstraint-basedsearchevaluatemillionsofrouting
scenariosinminutes,balancingcost,environmentalimpact,andengineeringfeasibility.
Itdelivered
~60%reductioninprogrammetime
forSSEN’s132kVextensionand
~93%reduction
for
NGG’spipelinestudy.Thisreducescostsandacceleratesinfrastructureneededforelectrificationandhydrogentransition.In2025thetoolhasalsobeenextended
tocoverthe
screeningphase
ofrenewabledevelopment.
ContinuumarecurrentlyscalingwithliveUKdeploymentsand£8.2mSeriesAfunding.
22
23
IntelligentWindFarmControl:AIOLUS
Windfarmslose
10
-
20%oftheirpotentialoutputto"wakeeffects"
whereupstreamturbinesslowthewindfordownstreamturbines,butcurrentcontroltechnologiescan'toptimisethewholefarm.
UniversityofWarwickhasdeveloped
AIOLUS
,thefirstEuropeandeepreinforcementlearningsystemforwholewindfarmcontrol.Itusesreinforcementlearningtooptimiseturbinesettingsinreal-timetominimisewakeeffectsandmaximisefarm-wideoutput.
Inthelastyearthishasmovedfromlate-stagedevelopment
(ManchesterPrizefinalistinMay2024and£415kEPSRCgrant)intoareal-worldpilotwithoperationalcontrolofawindfarm.
Thiscoulddelivera3-5%increaseinannualenergyoutput-
equivalenttopowering1millionUKhouseholdsfromexisting
windcapacitywithoutnewinfrastructure.Byoptimisingexistingassets,itreducesneedfornewlandandoffshoredevelopments.
Localgridoptimisation:Constellation
UKPowerNetworkspartneredwithABB,
GeneralElectric/GE
Vernova
,UniversityofStrathclydePNDC,andmoretodevelopthe
world’sfirstsmartsubstations
capableofanalysingmillionsof
datapointsandreconfiguringnetworksettingsinreal-time.
Smartsubstationsforecastandanalyselocalpowerflowsandcommunicatewitheachother(ratherthanrelyingoncentral
control)tofreeupcapacityandincreaseresilience.
MLmodelsaretrainedcentrallythendistributedtosubstationsforautonomousoperation,providingresiliencewhen
communicationsfail.
Thefirstsmartsubstationwasinstalled
Jan2025inMaidstone
,withfivemoretobeinstalledbySeptember2026.ThesolutionstrialledaspartofConstellationcould
savecustomersinGB
£132m
by2030.Constellationestimatestheycanalsosave
17m
tCO2by2050
iffullyrolledout.
24
25
Enablersforthenext12months
Politicalappetiteishigh
Pressuretobothkeepenergybillsdownandspeedup
connections,particularlyfordatacentres,meanspoliticalsupportforAIapplicationsinthisspaceisextremely
strong.The
AIEnergyCouncil
focusesonspeedingup
gridconnectionsandissupportedbyworkatDESNZandNESO.The
CleanPower2030
targetcreatesahard
deadline,incentivisingexperimentationwithAIsolutions.
AIcompaniesneedenergyinfrastructurenow
AccesstoenergyfordatacentresisbecomingabindingconstraintonlargeAIcompanies.Theywillinvestboth
cashandtalentinunlockingthat,andwillbepredisposedtoAI-basedsolutions.Thisislikelytoincludecreationofnewrevenuestreams-e.g.Piclo’sdatacentre
connection
accelerationprogramme
intheUSwhichisexchanging
energyflexibilityservicesforfasterconnection.
Remaininggapsandbarriers
Changingprocessesrequiresmorethantechnology
Manyoftheprocessesinvolvedininfrastructure
developmentareformalisedunderlegislationor
regulation,whichrequiresfocusedpoliticalwilltochangequickly.AImayhelpspeedupcertainelementsbut
cannotstreamlineentireprocesses(orchangecultures)inhighlyregulatedareas.
Planningdataremainsfrustratinglypatchyandopaque
Thecomplexity(andhistoricallymanualnature)ofthe
planningprocessmeansconsistent,goodqualitydataisrarelyavailable.ThismakesitchallengingtobuildAI
toolsinthisarea(see
Yottar’sdevelopmentdiary
).
Therehasbeengradualprogressonthis,andOfgem’s
latestreview
proposesimportantactionsfornetworksthatwouldfurtherclosethisgap.
Grand
Challenge3
Maximising
Flexibilityin
EnergyNetworks
27
28
Overview:MaximisingFlexibilityinEnergyNetworks
thesystemoperator,savinghundredsofthousandsoftonnesofemissionsandtensofmillionsofpounds.
AIadoptionforforecastingandoptimisationremainsmixed.Manyorganisationcontinuetorelysolelyon
statisticalforecastingandmathematicaloptimisationtechniques,butanincreasingnumber(particularly
batteryoptimisers)areutilisingMLandReinforcementLearningforcompetitiveadvantage.
Inthenext12monthsweexpecttoseeAIadoptionatscaledeliveringincreasinglylargeimpacts(both
environmentalandfinancial)acrossthisGrandChallengeduetotherapidlygrowingmarket,andstrongfitforAIcapabilities.
Ahighrenewablesfuturerequiresenergydemandtoflexsoweconsumeandstoreenergywhenthewindisblowingandthesunisshining.Thisisaradical
changeinnetworkandmarketoperation,andis
fundamentallydependentonusingAItoforecastandoptimisedemandandsupplyatmuchmoregranularlevelsthaneverbefore.
2025hasseenanaccelerationinthedeploymentofflexibilityonthenetwork,withAIplayinganessentialroleinthat.Mostoftheimpacthascomefroma
growingmarketusingexistingMLtools,ratherthannewAI-driventechnologicalbreakthroughs.
However,akeydevelopmentisthatdeeplearningbasedsolarforecastswerefullyoperationalisedby
AIapplicationsthatcouldaddressthischallenge
30
ThemesinAIadoptionforenergyflexibility
1
2
3
4
5
AI-basedvirtualpowerplantorchestrationishelpingbalancethegridatscale.Therearenowmultiple
organisations(including
Kraken
,
Kaluza
,
Flexitricity
,and
Arenko
)usingAItoaggregatedistributedenergyresourceslikeEVchargers,batteries,andindustrialloadsintovirtualpowerplantsatuptoGWscale.
AIisalsobeingusedtomatchrenewablegeneratorstodemand
atalocallevel
,increasingmarginsforrenewablesandmakingthemmoreeconomicallyviable.
Solarnowcasting
(forecastingforthenextfewhours)ismateriallyimprovingcontrol-roomdecisions,savingtensofmillionsongridoperatingcostsand
hundredsofthousandsoftonnesofCO2peryear
.
Forecastingremainsthefoundationforflexibility.ManyorganisationsrelyonMLforforecastsofdemand,generationandprice.Thelastyearhasseen
someprogress
in
research
on
foundationmodels
for
forecastingbutthereal-worldimpactoftheseremainstobeproven.
AIisstartingtomakebuildingsflexiblegridassets.Companieslike
GridEdge
and
CarbonLaces
are
automatingreal-timeloadshiftinganddemandresponsewithAI-ledoptimisationofbuildingenergydeliver15–34%reductionsinpeakdemandintrialsduring2025.
Solarforecasting:OpenClimateFix
Solargeneration'sunpredictabilityforcesgridoperatorstomaintainexpensivefossilfuelbackupcapacity,drivingupcosts.OpenClimateFixpartneredNESOtodeploy
QuartzSolar
Thetechnologyachieved40%improvementoverpreviousforecastsandisfully
.
operationalinNESO'scontrolroom,saving
Adeepneuralnetworkcombinesmeasuredsolar
generation,numericalweatherpredictionandsatelliteimagerytopredictcloudmomentsandsolargenerationacrosstheUKoverthenextfewhours.
inbalancingcosts,withpotentialtoreach£150mby2035.
300,000tonnesofCO2and£30millionperyear
31
32
Localenergymarkets:UrbanChain
Directpeer-to-peerdomesticenergysupplycanmake
energycheaperbydecouplingrenewablesfromvolatile
wholesalegasmarkets,butisnotyetpermittedintheUK.
Instead,UrbanChainactsasaregulatedenergysupplierbutusesAIandblockchaintocreateprivatelocalenergymarketsthatmatchrenewablegeneratorswith
consumers.ThisAI-drivenoptimisationpairsrenewablegeneratorswithconsumersonahalf-hourlybasis,
deliveringupto
upto25%bettermarginsforgenerators
andsavingsforconsumers.
Thebusinessisscalingrapidly,havingrecorded
10x
revenuegrowthto£25m
on
over200GWhofassets
in2024,andispursuinga
£50mSeriesB
in2025.
33
Powersystemsfoundationmodel:GridFM
Managingthegridisdependentonsolvingaseriesof
optimal
powerflow
problems,whichisslowandcomputeintensive.
Researchersfrom
IBM,ImperialCollegeandothers
areworkingonafoundationmodel
pre
-
trainedon300,000+optimalpowerflow
problemstocreateafoundationmodelforgridoperations.
Itusesgraphneuralnetworkstolearnfundamentalpowerflowphysicsandgriddynamicsfromrealandsimulateddata.
Stillintheresearchstate,theopen-sourcecollaborationaimstodeliveranewAI-basedapproachthatprovidesordersof
magnitudefasterpowerflowsimulations.
Ifsuccessful,thishasthepotentialtomakeourexisting
optimisationofgridoperationsmoreeffective,butalsoto
potentiallyunlocknewwaysofoptimisingthegridinnear-realtime.
34
Enablersforthenext12months
TheNationalEnergySystemOperatorcontinuestodigitalise
DigitalisationofNESO’soperationsthroughtheOpenBalancingPlatformisbroadeningaccesstogridservicesfornewentrants,includingthoseusingAIoptimisation.Ithashelpedgeneratorsunderstandthereasonsforoperatingdecisions,andhascontributedtoa
five
-
foldincreaseinbatterydispatch
.The
REVEAL
platformalsoenablestestingofnovel
温馨提示
- 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
- 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
- 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
- 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
- 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
- 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
- 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
最新文档
- 2026届吉林省油田第十一中学生物高一下期末综合测试模拟试题含解析
- 鸡兔同笼课件创作说明
- 空气污染复习题及答案
- 高中历史专题2近代中国维护国家主权的斗争3伟大的抗日战争课件人民版
- 2026年冷链温控系统项目公司成立分析报告
- 2026年医疗废弃物智能处理项目公司成立分析报告
- 2026年城市领航辅助驾驶(NOA)项目公司成立分析报告
- 2026年凝聚态电池项目公司成立分析报告
- 2026年临床不可替代创新药项目公司成立分析报告
- 2026年健康保险产品创新项目可行性研究报告
- 2025跨境电商购销合同范本(中英文对照)
- 《骆驼祥子》知识点24章分章内容详述(按原著)
- 2025年人教版九年级物理知识点全面梳理与总结
- DB33T 2256-2020 大棚草莓生产技术规程
- 《建设工程造价咨询服务工时标准(房屋建筑工程)》
- 工程(项目)投资合作协议书样本
- 半导体技术合作开发合同样式
- 制程PQE述职报告
- 小广告清理服务投标方案
- 细胞治疗行业商业计划书
- 护士慎独精神的培养
评论
0/150
提交评论