人工智能趨勢與公用事業中的應用.pdf

《人工智能趨勢與公用事業中的應用.pdf》由會員分享，可在線閱讀，更多相關《人工智能趨勢與公用事業中的應用.pdf（70頁珍藏版）》請在三個皮匠報告上搜索。

1、1CONFIDENTIALDistributed AI for Grid Optimization-DER Connect and ManageEPRI AI Summit|Jan 7,2025Dr.Yingchen“YC”Zhang,VP of Product,Utilidata2CONFIDENTIAL2015Deployed the utility industrys first real-time machine-learning grid optimization solution 2019Developed first applications for meter company

2、software platforms2021Partnered with NVIDIA to develop first distributed AI platform for the electric grid 2023Early adopters secure$100M funding from Department of Energy to scale Karman2024Hubbell becomes first hardware provider to embed KarmanUtilidata has operated real-time machine learning soft

3、ware on the grid for over a decade3CONFIDENTIALMeet Karman The first native distributed AI platform for the grid Built on an NVIDIA GPU with a combination of on-chip and cloud-based software,Karman:Enhances resiliencyMakes distributed energy resources more valuableRevolutionizes customer interaction

4、s with energy data Offsets the need for infrastructure investments4CONFIDENTIALA modified version of the NVIDIA Jetson Orin Nano:Smaller size to easily embed in edge devicesRemoved USB features not required for grid operations,including circuitry for cameras,HDMI and audioAllows a broader temperatur

5、e range for devices in rugged conditionsEnhanced security features for grid operations,including board-to-board connectorIncreased storage to accommodate third-party softwareProprietary on-chip softwareCustom Karman module Nvidia Jetson Orin Nano(Left)and Karman custom NVIDIA mod(Right)5Karman is a

6、different kind of edge solution Open,modern architectureSoftware-defined data processing and communications to support unlimited applications Robust communicationsCommunicates to a centralized ADMS,other Karman units,DERs,and additional devicesAdvanced computationOperates 100 x faster than market so

7、lutions to enable decision-making locally at every endpointEasy access to data Core services analyze millions of data points and deliver actionable insights via APIs and a user interface 6CONFIDENTIALKarman detects EV charging and distinguishes it from other loads at the siteKarman uses on-chip algo

8、rithms to detect the start time of EV charging within seconds and instantly makes those events known to other applications,such as DER schedulingTraditional and inefficient methods look at energy use trends over time and retrospectively select the times that EV charging is likely to occur7CONFIDENTI

9、ALEV detection and power quality correlation 8CONFIDENTIALWe partnered with the University of Michigan to study EV charging behavior and its impact on the gridKarman collected real-time waveform resolution voltage,current,and power data at the edge of the grid,allowing researchers to analyze and det

10、ect EV charging patterns and better understand how to manage EV demand9CONFIDENTIALEV charging caused large,rapid swings in current drawInconsistent power draw results in inefficient energy consumption,which could lead to overheating lines,power loss,and outages10CONFIDENTIALEV charging lowers local

11、 voltage significantly Areas with high concentrations of unmanaged EV charging are at risk of more-frequent outages11CONFIDENTIALEV charging causes variability in local voltage The ability to measure these changes with edge computing allows utilities to understand which locations can accommodate EV

12、charging 12CONFIDENTIALEV charging lowers power quality by introducing current harmonics generated from the conversion of AC to DC power Low power quality causes equipment degradation and failure for both utilities and consumers(flickering lights,excessive motor wear and tear,and premature failure o

13、f home appliances)13CONFIDENTIALEV charging in the presence of a grid fault showed varied conditions to support voltage recoveryWith this level of data from Karman,utilities can leverage EVs to support grid reliability and mitigate the impact of grid faults14CONFIDENTIALLocal computing,analysis,and

14、predictions will allow utilities to manage and coordinate EV charging demandKarman analyzes large amounts of granular data in real-time,which reduces latency and allows utilities to mitigate negative impacts of EVs,improve grid planning,and leverage EVs for reliability15CONFIDENTIALEPRI IEL Lab Demo

15、16CONFIDENTIALEPRI Lab demo configurationForm 2S meter socketNetmeterSCPIOpen DSSHardcodedMQTTMQTTSimulatorsModbusModbusUtilityUtilidata(meter collar)Load bank(NHR)Native loadsESPVPhysical hardwareSoftware modulesElectrical connectionWi-Fi connection17CONFIDENTIALDistributed DERMS demonstrationDER d

16、ispatch optimization algorithm automates solar+storage controls to minimize customer billEPRI lab demonstration(12/4/24)Thank you18 2024 Blue Wave AI LabsJ.Thomas GruenwaldExecutive Vice President,Co-FounderJanuary 7,2025Nuclear Industry Use of AIFor Online Thermal Limit Predictability19Machine Lear

17、ning,AI Experts,Real-World NuclearExtensive,real-world nuclear experienceMore than half of US BWRs use our software and services.Have solved several long-standing reactor issuesHave intimate knowledge of modern nuclear core design across utilities,fuel brands,and product linesPeopleEngineers and phy

18、sicists with broad,nuclear,AI and simulation experienceCollaborationLocated near Purdue University Close relationships with Purdue Nuclear Engineering DepartmentSponsored PhD student working to integrate AI into modern core simulators Sponsored Purdue Data Mine project on Generative AINRC/DOE and Na

19、tional Lab joint projectsFounded in 2016,and already trusted by over half the boiling water reactors in the U.S.domestic fleet.We are an AI-centric,industry-focused innovation company serving the nuclear energy industry.We combine the insight of exceptional scientific technical talent with the lates

20、t advancements in AI and Machine Learning to transform data into solutions for the worlds most difficult problems.BW LocationsBW Locations20 2024 Blue Wave AI Labs.All Rights Reserved.ABOUT BLUE WAVEProduct Portfolio and Pipeline21Nuclear News,February 2022 2024 Blue Wave AI Labs.All Rights Reserved

21、.PROVEN AI/ML CAPABILITIESIncreasing Model ComplexityBlue wave the first to apply ai for saving nuclear fuelMaking Nuclear Safer and More CompetitiveMaking Nuclear Safer and More CompetitiveBlue Wave Expertise Nuclear Energy;BWR,PWR,ComponentsCommercial nuclear power engineering and managementApplic

22、ation of AI/ML to nuclear power operationsNuclear Regulatory ProcessGenerative AI applied to document preparation,review,and automationSaaS(Software as a Service),AI/ML,Amazon Web Service Integration(AWS)Customer Benefits TodayMinimize new fuel purchases for each cycleAvoid generation losses from un

23、planned power deratesMitigate Cobalt-60 Carryover dose rates with high moisture carry-over(MCO)-a common BWR concernMitigate costly reactivity management challenges in operation and in the fuel management/core design phase$3 million of Savings per Unit per Year at typical plantBWnuclear.ai:Currently

24、 Deployed in Half of US BWR FleetThermalLimit.aiMachine Learning can achieve more accurate online parameter predictions than Core Simulators aloneMCO.aieigenvalue.aiParameterTyp.Avg.Bias of BW PredictionTyp.Industry Bias without BW toolsMCO Current Cycle.%No other way to predict MCO exists.MCO Futur

25、e Cycle.%No other way to predict MCO exists.Keff50 pcm200-300 pcmMFLCPR0.3010-21.510-2-310-2MAPRAT0.7510-2410-2-810-2MFLPD0.7510-2410-2-810-222 2024 Blue Wave AI Labs.All Rights Reserved.Confidential and Proprietary Information.Began with a Pilot Study at Limerick Clean Energy Center in April of 202

26、3 Deployed to our BWnuclear.ai suite of tools on AWS GovCloudTHERMAL LIMIT MODEL DEVELOPMENT23A Brief HistoryAfter successful demonstration of Methodology for Thermal Limit Bias Predictability,ThermalLimits.ai was developed and expanded across Constellations BWR Fleet in Spring of 2024.More Stations

27、 to follow!2024 Blue Wave AI Labs.All Rights Reserved.Compliance with Technical Specification and Thermal Operating Limits are essential to the safe operation of an NPP.In a BWR,three major limits are tracked(MFLPD,MFLCPR,and MAPRAT).Example of typical limits areChallenge:A large and inconsistent bi

28、as between offline and online limits makes it difficult to engineer in appropriate levels of margin to these limits.Excess margin=Over-fueling the core(excess direct fuel costs)Insufficient margin=Operation challenges resulting in power derates and decreases energy capability of the core to avoid ex

29、ceeding a limit(power generation losses).Blue Wave Product:A method for consistent and accurate estimation of online thermal limits from training data coming from earlier cycles.24OfflineCORE SIMULATOR MODELINGOnlineLPRM results+BW correctionsBACKGROUND ON THERMAL LIMITS 2024 Blue Wave AI Labs.All R

30、ights Reserved.ObjectiveCompliance with Tech Specs and thermal operating limits,i.e.,maintaining safetyReduce the bias between offline(black line)and online(blue line)by more consistently and accurately estimating thermal limits(purple line)ImplementationAI enabled tool supporting reload core design

31、Ability to adjust design goals for scenario planning during fuel cycleBenefitsReduces excess margin or over-fueling the core resulting in direct fuel cost savingsRestores margin and lessens operational challenges when approaching administrative limits,thereby preventing power generation lossesResult

32、sOnline to offline bias consistently reduced by a factor of 3 to 5,on average,across all generating stations.Accurate in-cycle thermal limit predictions along with LPRM.ai has already prevented costly actionsTHERMAL LIMITS.AIThermal Limit bias reduction through an encoder-decoder convolutional neura

33、l networkBias between Bias between online and online and offline limitoffline limitIssue with in-core instrumentation25 2024 Blue Wave AI Labs.All Rights Reserved.THERMAL LIMIT MONITORING ARCHITECTURECore Monitoring SystemqOnline Power DistributionqOnline Thermal LimitsqCompliance with Tech SpecsqRe

34、al Plant OperationLPRMsTIPsCore SimulatorqOffline Power DistributionqOffline Thermal LimitsqUsed for reload core designqIntended Plant OperationFeedback from Nuclear InstrumentationDetector CalibrationOperationDesignqGaps in modeling&simulationqError&uncertainty in physical measurementqLeads to over

35、ly conservative design and/or operational challenges;costs millions per reloadBlue Wave Provides AI systems for TIP Alignment,LPRM Analysis and Thermal Limit Predictions26Gaps in key plant parameters limit design and operation 2024 Blue Wave AI Labs.All Rights Reserved.THERMAL LIMIT MODELING APPROAC

36、H27 Direct-Bias Methodology Error correction network that takes offline thermal limit as input and adjusts the power distribution to more accurately predict the expected online thermal limit Convolutional Neural Network(CNN)encoder-decoder networkInputs:Offline MFLPD array and other cycle parameters

37、Output:Online MFLPD arrayConvolutional Neural Networks are Computationally Efficient in Dealing with Large ArraysExample Input to BWnuclear.ai BWR Core Prediction Software Suite 2024 Blue Wave AI Labs.All Rights Reserved.The Bias requires use of larger thermal limit design margins,which increases fu

38、el costs.The Bias forces deeper than planned use of control rods that can lead to power derates and generation revenue losses.Unplanned rod patterns lead to MFLPD management challenges for Operations and re-work for Nuclear Analysis engineers.Online values depend upon accurate LPRM and TIP operation

39、.Scale of BIAS is 10-2 and scale of MSE is 10-4MFLPD is the maximum over a 30 x30 x25 array.MSEarray is across the whole array.MSEMFLPD is the squared error of the max values(=MFLPD values).BIAS is calculated for the MFLPD values.OfflineMSEarray:4.18MSEMFLPD:14.77BIASmean:3.44BIASmax:7.3728TYPICAL B

40、IAS FOR MFLPD 2024 Blue Wave AI Labs.All Rights Reserved.Training Set:The eight previous Fuel CyclesPerformance Improvement:Mean bias is reduced by a factor of 3.62 Max Bias reduced by a factor of 2.03Potential Savings Revise the core design to have fewer fuel bundles Avoid generation losses due to

41、derates.Note:Scale of BIAS is 10-2.Scale of MSE is 10-4MFLPD is the maximum over a 30 x30 x25 array.MSEarray is across the whole array.MSEMFLPD is the squared error of the max values(=MFLPD values)BIAS is calculated for the MFLPD valuesOfflineMSEarray:4.18MSEMFLPD:14.77BIASmean:3.44BIASmax:7.37Blue

42、Wave ModelMSEarray:1.65MSEMFLPD:1.34BIASmean:0.95BIASmax:3.6329 2024 Blue Wave AI Labs.All Rights Reserved.BLUE WAVE MODEL PERFORMANCEMFLPD NODAL ACCURACY30The Blue Wave model prediction is significantly more accurate than the physics model predictionmax!|#!$#!%|max!|#!$#!&|Cycle 17Online vs.Offline

43、Online vs.BW Model 2024 Blue Wave AI Labs.All Rights Reserved.SUMMARY OF RESULTS31Observed Modeling Performance:Online-offline mean“bias”reduced by a factor of 3 to 5.5 across all cycles and stations modeledMax bias reduced by a factor of 2 to 4 at every plant where models have been developedEarly R

44、ealized Benefits:Elimination of TL design setdowns Most recently at PB 0.865 up to 0.900 for MFPLDCustomers are working to qualify methodology to design directly to BW predictionsAvoidance of remedial actions(and lost generation revenue)at multiple customer BWR sitesPrivileged and ConfidentialCASE S

45、TUDY 132Notable Examples of BenefitTL Models had not been developed at the time of Core Design 2024 Blue Wave AI Labs.All Rights Reserved.CASE STUDY 133Notable Examples of BenefitModel was available to troubleshoot high MFLPD during the cycle.Result was to avoid inserting shallow shaper rods for an

46、extended interval.CEG estimated this avoided 1.2 EFPDs of lost generation capability of the core 2024 Blue Wave AI Labs.All Rights Reserved.CASE STUDY 134Actual Event:Approaching thermal limits limit within a matter of daysRunaway MFLPD at 0.96 on path to reach procedural limit of 0.98 within days,B

47、lue Wave model predicted.918Operators thought some LPRMs may need to be placed out of service due to miscalibration,but couldnt efficiently verify this claim.Without Intervention,a short-term derate would be eventual course of action,followed by insertion of shaper control bladesMLFPD getting worse,

48、up to 0.975(model predicted 0.92)Operator requested Blue Wave to analyze ALL LPRMswe did,rank ordering them from most-to-least problematicWe identified 7 LPRMs with issues,recommending bypass reducing MFLPD from 0.975 to 0.955 Blue Wave predictions proven true and accurateBlue Wave recommended perfo

49、rming recalibration with TIP,based on model predictions(still 0.92)After TIP the MFLPD went to 0.92!Blue Wave tools helped address and closeout three related IRsOperator estimates that this support avoided generation losses of approximately$1.23MThis event was part of the application of the NEI Top

50、Innovative Practice award 2024 Blue Wave AI Labs.All Rights Reserved.USER INTERFACE35ThermalLimits.aiThe models are packaged and deployed to the AWS GovCloudEnsures compliance Ensures compliance with 10 CFR 810 Export Control RequirementsSafeguard sensitive data,protect data files with server-side e

51、ncryptionStrengthen identity managementRestrict the API calls users are able to make with identity federation,easy key rotation,and other powerful access control testing tools.Protect accounts and workloadsEnables pushing model updates without disruptionEnables pushing model updates without disrupti

52、onPlatform updates are seamless and customer requests Platform updates are seamless and customer requests can be rapidly deployed.can be rapidly deployed.2024 Blue Wave AI Labs.All Rights Reserved.Making a Material Impact on the Nuclear Fleetwith BWnuclear.ai(mco+eigenvalue+thermal limit)(Fresh Fuel

53、=$500K/assembly,Spent Fuel=$100K/assembly)36 2024 Blue Wave AI Labs.All Rights Reserved.Confidential and Proprietary Information.Unit#BW CyclesBatch Size before BWnuclearBatch Size2 CyclesAgoSavings2 CyclesAgoBatch Size Previous CycleSavings Previous CycleBatch Size Current CycleSavings Current Cycl

54、eTotal Saved BundlesFresh Fuel SavingsSpent Fuel Savings1227226842601216$8M$1.6M23272272026012264820$10M$2M3227226842621014$7M$1.4M43276272426016268828$14M$2.8M5227226842561620$10M$2M63200196419281881224$12M$2.4M7231630412312416$8M$1.6M823083008304412$6M$1.2M92144140414048$4M$0.8M1013082921616$8M$1.

55、6M1122082044200812$6M$1.2M1222082008200816$8M$1.6M13122822088$4M$0.8MTotal27210$105M$21M27 Cycles Planned with BWnuclear.ai 210 Fuel Bundles Saved$105M Fresh Fuel Cost Savings$21M Spent Fuel Cost Savings$81M Avoided Operational Costs$207M Total Cost Savings for13 UnitsAverage savings per unit per cy

56、cle:$7.7MAverage fresh fuel savings per unit per cycle:3.8MCycles planned with BWnuclear.ai 2024 Blue Wave AI LabsThank You37 2025 Electric Power Research Institute,Inc.All rights Grace SmithEPRI Data ScientistJanuary 7,2025EPRI Product ID:3002029854Data Handling and Its Impacts to AI ModelingAI&DX

57、in Electric Power Summit 2025 Electric Power Research Institute,Inc.All rights reserved.39Data GovernanceEPRI Product ID:3002027488Data Governance is the management of data availability,usability,integrity,and security within an organization.Involves establishing policies,procedures,and standards to

58、 ensure data is managed responsibly.Data Governance should be integrated into every area that involves data.FrameworkTeam and RolesPolicies and PracticesToolsKey Components 2025 Electric Power Research Institute,Inc.All rights reserved.40MetadataExample Temperature Sensor Metadatasensor_id:“TS-1001”

59、location:“Building A,Room 101”sensor_type:“Thermocouple”model:“ST-200”manufacturer:“Sensor Manufacturer Inc.”status:“active”hierarchy:“Boiler Room”,“Section B”,“Temperature Sensors”,“TS-1001”installation_date:“2023-05-16”last_callibration_date:“2024-11-20”accuracy:“0.5C”range:“-50C to 150C”compressi

60、on_rate:“1 reading per minute”data_format:“Digital output”calibration:“Using a standard reference thermometer”compliance:“ISO 9001 certified”Metadata is data about data.It provides information that helps to describe,organize,and manage other data.Metadata TypesDescriptive MetadataStructural Metadata

61、Administrative MetadataTechnical MetadataReference MetadataLegal Metadata 2025 Electric Power Research Institute,Inc.All rights reserved.41Faulty Metadata ExampleThis range is inaccurate due to poor metadata quality!2025 Electric Power Research Institute,Inc.All rights reserved.42Data Handling Proce

62、ss 2025 Electric Power Research Institute,Inc.All rights reserved.43Data Preparation for Advanced Monitoring SystemsData Origin&LineageData DestinationExpert Knowledge of the SystemData ModelingThermal Performance Monitoring(TPM)Advanced Pattern Recognition(APR)Vibration MonitoringTypes of Plant Onl

63、ine-MonitoringKey ConsiderationsEPRI Product ID:3002030671 2025 Electric Power Research Institute,Inc.All rights reserved.44Common Learning TechniquesSupervised LearningDuckLizardDogDogSupervised LearningPrediction ModelPrediction ModelDogUnsupervised LearningUnsupervised Learning 2025 Electric Powe

64、r Research Institute,Inc.All rights reserved.45Why Proper Data Handling andAccurate Metadata MatterEnsures Data QualityEnhances EfficiencySupports Real-time MonitoringScalabilityCompliance and SecurityIntegration with Systems 2025 Electric Power Research Institute,Inc.All rights reserved.46 2025 Ele

65、ctric Power Research Institute,Inc.All rights TOGETHERSHAPING THE FUTURE OF ENERGYNPP Realistic Facility Search systemfor Nuclear Power Plants in Korea2025.1KHNPalrightsreserved,plzProhibitunauthorizedcopieswithoutpermissionPresentation contents01NPP RealisticFacility Searchsystem02Development03Main

66、contentsperformance04Future planIntroductionInnovationImprove work productivity by providing and sharing NPP necessary informationRather than relying on Peoples memory to search for itGOALInformation searchInformation Production(Action)Information searchInformation ProductionFocusingon information p

67、roductionConceptGoogle MAPMobileGPSPaper based map bookWebprotalRestaurant menu Vistors ratingDesignated Vistors reviewDriver Number of Vistors RelatedCall Taxi Special notesTrafficVariousBig DataChatbotSearch engine1.Introductionshortening50u(Exelon)Explore and enter inaccessible space at any time(

68、2018 NRC meeting,VR initiative)Summary u(Thermal power/LNG)Taking the pictures of the inside,Reprocess the system flow on the picturesu(KHNP)Check the location of equipment in the CV,and mark the number of equipmentExelonThermalpowerplantLNGDevelopmentStatusKHNPDevelopmentDirectionItemAs-IsTo-BeAs-I

69、sTo-BePicturesintegration412Pieces4pieces/1 SpotAccessPCPC/MobileDataManagementPersonalPCServer(Web based)UsersFilebase(1Person/1PC)Web-based(Server)ExecutionPersonalPCWeb-BasedinformationLimited(PCDB)WithLegacysystemInputmethodN/AInputfunctionSearchmethodMouseclick+numberinput2.Industry trends51Met

70、hod Improvement ForFinding equipmentlocationShorteningFacility related information SearchtimeRelated document search of target facilityfield confirmationInformation search by employees in the LDMAction Periodic Test Valve Line-up Maintenance work System operation Abnormal etcTarget FacilityP&ID equi

71、pment position display manuallyDifficultyinreadabilityon GA drawingsproceduresP&IDDrawingsOne-click material search,Web based pre-checkuRelatedinformation interconnection searchwithoneclickuInformationsearchbypeopletextinput Periodic Test Valve Line-up Maintenance work Systemoperation Abnormal etcfi

72、eld confirmationActionWeb based realistic pre-checkimproving Field equipment search methodShorteninginformationsearch timeHumanerrorreductionInformation Search method InnovationThe effect of improvingwork productivity:54,750,000hrfor 60yearsReproducecircuitbreakerpositiondisplaydrawingsTablemanageme

73、ntof spatialinformationsuchashazardousspaces,enclosedareasMust remember Drawing or procedures No.Use*,takes longer than you expected3.Background524.Deriving functional requirements535.System Design&contentsReference Dwg.Panorama ViewFacility positioninformation SearchFacilities related information D

74、BFP AreaRADClosed SpaceCVTB IslandAB BDGDescriptionFPInputInputDBDSelectionNPPPlantViewAREARoom DoorFire escape PathwayPathH/SwaySearchEQCircuitInfo.BreakerBDGMoveLine-up InfoWorkPCM0bileSAP(Procedures,ISO)RtPMS(PI)E-P&ID(Equipmentlocation in DWG)Facilitymaster(Quality,safety etc)Legacy system conne

75、ction searchReference drawingNPPViewerDevelopmentControltarget facilitiesDBDev.Information search&display Facilitylocation,fireescape FP AREAdisplay DOOR,Room location Radiation,Closed space Shooting all areas&viewerFieldfacility locationdisplay(Labeling)Field facility location check Start/stopover/

76、Arrivalpathway Valves,Equipment list Derive Development list E-P&ID position search Equipment Master Plant facility location check Breaker,Hand Switch search ISO DWG etc Sap Operation status displayv Development Contents54CODINGDRAWINGDIGITALIZAITONNPP FACILITY SEARCH SYSTEMSHOOTING MERGINGTAGGINGDB

77、DESIGNTake4shots at90degreeintervalsMergethecapturespictures6.Development method557.DevelopmentInitiatives20261Plant(2unit)DevelopmentAll Plants3PhaseWorkprocessinnovation2030 3 KHNPS Initiatives2P H A S ESystemintegration(2unit)&expansion2023Demonstration Development1Phase1unitincludingconstruction

78、NPPPhase 1Phase 2Phase3OperationNPPdevelopment(han-bitUnit6)PilotDevelopmentofconstructionNPP(Shinhanul#2)1powerplantdevelopment(2unit)Developmentofothertypesofmodels CombiningAi,Bigdata,Chatbot IntegratingbusinessprocessR&D(21.423.3)Advanced&incrementalexpansionApply to all power plantsWork56Portal

79、ContentsOne-ClickGA Plant viewer Facility Related information(40,000/1unit)DBWorkportal(PC)Mobile application with Wireless InfraRelated information Linked searchLDM(Procedures,ISO,Equipment master,ETC),Legacy system(E-P&ID,PI,ETC)Reference Drawing DigitalizationNPP Plant ViewerFacility location sea

80、rchClosed SpaceFire zoneRadiationEmergencyevacuationReactorTBNAuxiliaryCPB BDGBDGBDGBDGCPBBDGDescriptionTag inputTree1.System Configuration2.Key FunctionsLocate Facility position with tag Number&descriptionAREA,Zone Information DigitalizationE-P&ID Drawing link SearchLegacy system(Procedures,ISO)Lin

81、k SearchApplying various fields&safety ImprovementPre-job Briefing&3-Way CommunicationPredictable Plant operationEstablishing a realistic training base for fire brigadeEstablishing work planAbnormal3.Applicationcontents(1)Various Training3-Way CommunicationWorkProcessHuman errorField workDigitalizin

82、g checklists attached equipmentEnhance Workers safetyProvide on-site event informationAdd pipe flow formationInformation DisplayInformation displayWear protective gearAlarm informationDynamic information3.Applicationcontents(2)Personal Information ManagementFacility Tagging ScalablityAutomated uploa

83、dITC/Power outage workFlexible SystemConfiguration ManagementInformation storage/SharingExpand search target equipment3.Applicationcontents(3)SimulationMain performance23Q123Q223Q323Q424Q124Q2Users accumulated trend(23Q1 24 Q2)9,25010,11811,35611,93615,08916,134Users70,00073,883(250/1day)60,00050,00

84、057,749Quarter40,00042,660Accumulated30,00030,72420,00019,36810,0001.Users TrendVirtual SimilarityEffectivenessExcellent(45%)Good(45%)Excellent(82%)Training EffectGood(18%)Product FeasibilityPatent ProtectionuContinuous Patent registration(4pieces)using development method of this patentuSME technolo

85、gy transfer contracts(2 cases),Promoting overseas exportsuUtilizing this technology from Construction to O&M including TraininguUsing Education practical courses(Over 2,000 Employees a year)S/4HANA17092.Application cases&main performanceuPatent developed by KHNP itself achieved 4 winsat the Indonesi

86、a invention competition-Best award(1),Gold award(1),Special award(2)3.International invention day 4 wins(2024)Future planSYSTEmPROCESsMETAVERSEPre-jobbriefing,informationsharingFaultnotificationinputSAPLinkworkorderinput SAP link1.Future planothersysteminterconnectionMetaverseestablishment possibleaddedonconferencecal Multi participation among workers of head office,Plant office MCR,Field operatorsnotificationWorkorderOverseasConsulting Global access such as UAE,Czech etcAdvancementfireChatbotsearch,BigdataanalysisMCRalarmCRTinterconnection(alarmequipment-field equipment22QT&EXTATheTEXeTnd