1、The use of AI for improving energy securityQuantitative exploration of the opportunities of the deployment of AI applications in the electricity systemIsmael Arciniegas Rueda,Hye Min Park,Joshua Steier,Henri van Soest,Harper Fine,Mlusine LebretFor more information on this publication,visit www.rand.

2、org/t/RRA2907-2About RAND Europe RAND Europe is a not-for-profit research organisation that helps improve policy and decision making through research and analysis.To learn more about RAND Europe,visit www.randeurope.org.Research Integrity Our mission to help improve policy and decision making throug

3、h research and analysis is enabled through our core values of quality and objectivity and our unwavering commitment to the highest level of integrity and ethical behaviour.To help ensure our research and analysis are rigorous,objective,and nonpartisan,we subject our research publications to a robust

4、 and exacting quality-assurance process;avoid both the appearance and reality of financial and other conflicts of interest through staff training,project screening,and a policy of mandatory disclosure;and pursue transparency in our research engagements through our commitment to the open publication

5、of our research findings and recommendations,disclosure of the source of funding of published research,and policies to ensure intellectual independence.For more information,visit www.rand.org/about/principles.RANDs publications do not necessarily reflect the opinions of its research clients and spon

6、sors.Published by the RAND Corporation,Santa Monica,Calif.,and Cambridge,UK 2024 RAND CorporationR is a registered trademark.Cover:Adobe StockLimited Print and Electronic Distribution Rights This publication and trademark(s)contained herein are protected by law.This representation of RAND intellectu

7、al property is provided for noncommercial use only.Unauthorised posting of this publication online is prohibited;linking directly to its webpage on rand.org is encouraged.Permission is required from RAND to reproduce,or reuse in another form,any of its research products for commercial purposes.For i

8、nformation on reprint and reuse permissions,please visit www.rand.org/pubs/permissions.iiiModern civilisation is critically dependent on access to electricity.Electricity systems underpin practically all the essential life functions we take for granted.Most economic activity would be impossible with

9、out access to electricity.However,existing grid systems were not designed to handle present let alone future electricity demand.At the same time,climate change urgently demands drastic changes to our energy systems.Developing and modernising grid infrastructure requires substantial investment,and se

10、curing necessary funding for these projects is complicated by finite resources and competing budget priorities.Artificial intelligence(AI)applications have potential to solve many of the challenges facing the grid.They can fulfil a range of functions throughout the electricity system,making electric

11、ity cheaper and more reliable.In many cases,the deployment of AI simply extends existing methods and approaches.For example,AI applications that help increase electricity market clearing can build on many existing data applications.AI tools can also open up new ways of interacting within the electri

12、city grid,such as dynamic charging and discharging of electric vehicle batteries to provide flexible storage.These opportunities could help improve the overall energy security of the electricity system.However,the deployment of AI applications could also give rise to cybersecurity risks,the risk of

13、unexplained or unexpected actions,or supplier dependency and vendor lock-in.The speed at which AI is developing means many of these risks are not yet well understood.This technical report accompanies the main policy report titled The Use of AI for Improving Energy Security(RR-A2907-1).It describes t

14、he methods,tools and results of our quantitative exploration of the impact of AI-based applications on the energy security of the European power grid.This study was made possible by internal RAND funding through the RAND Initiated Research(RIR)programme.We would like to thank participants in the bac

15、kcasting workshop held in Brussels on 9 September 2023 for making time for this project and sharing their insights.We would also like to thank our quality assurance reviewers Lucia Retter and Andy Skelton for their useful feedback.Finally,we would like to thank Nancy Staudt,Lisa Jaycox and the membe

16、rs of the RIR administrative team for their support over the course of this project.RAND is a not-for-profit research organisation that aims to improve policy and decision-making in the public interest,through research and analysis.Our clients include governments,institutions,non-governmental organi

17、sations and firms with a need for rigorous,independent and multidisciplinary analysis.For more information about this document,please contact:Ismael Arciniegas RuedaSenior Economist,RAND Corporation1200 S Hayes St22202 Arlington,VA,USAEmail:iruedarand.orgHenri van SoestSenior Analyst,RAND EuropeRue

18、de la Loi 82/Bte 31040 Brussels,BelgiumEmail:vansoestrandeurope.orgPrefaceivThe use of AI for improving energy security-Technical reportEXECUTIVE SUMMARYBackground and contextElectricity systems around the world are under pressure due to aging infrastructure,rising demand for electricity and the nee

19、d to decarbonise our energy supplies at pace.AI applications help address these pressures and increase overall energy security.For example,AI applications can reduce peak demand through demand response,improve the efficiency of wind farms and facilitate the integration of large numbers of electric v

20、ehicles into the power grid.However,the widespread deployment of AI applications could also come with heightened cybersecurity risks,the risk of unexplained or unexpected actions,or supplier dependency and vendor lock-in.The speed at which AI is developing means many of these opportunities and risks

21、 are not yet well understood.This technical report accompanies the main policy report titled The Use of AI for Improving Energy Security(RR-A2907-1).It describes the methods,tools and results of our quantitative exploration of the impact of AI-based applications on the energy security of the Europea

22、n power grid.Study objectives and research approachThe objective of this study was to assess the current state of AI applications for the power grid and their impact on various energy security metrics.To do this,we used PyPSA-Eur,a Python-based power system model,to explore the extent to which diffe

23、rent AI applications can improve energy security under different AI deployment scenarios.Our research focused on the European energy system and the short term(less than five years),assuming a static power grid topology and considering AI applications at deployment readiness level(TRL 8 and 9).We dev

24、eloped quantitative methods to evaluate the performance of AI on energy security on alternative scenarios of AI deployment in the European power grid.To measure energy security attributes for each scenario,we examined four key dimensions of energy security:availability(ensuring sufficient supply res

25、erve to meet demand),affordability(average and marginal cost of electricity),accessibility(determining whether electricity generation is relying on imported fossil fuel)and acceptability(evaluating the carbon emissions of the electricity).This technical report provides quantitative evidence for the

26、impact of AI on energy security and translates the evidence into policy insights on the role of AI in the power grid.Our research also provides observations and lessons learned on the advantages and challenges of using open-source modelling for research on power grids.Primary findingsWe evaluated th

27、e impact of three AI applications across four scenarios:AI-driven load reduction,AI-driven load shifting,AI-driven wind wake steering,and all three applications combined.Comparing these scenarios to a baseline without AI applications,we found that AI-driven load reduction provided benefits across al

28、l energy security metrics,ranging vfrom 3%to 22%.However,the distributional impact varied between countries,ranging from significant improvements to negligible changes.AI-driven load shifting improved two energy security metrics(availability and affordability),but left the others unchanged compared

29、to the benchmark scenario.The distributional impact was mixed,with some countries showing improvement and others remaining unchanged or even deteriorating compared to the benchmark.Wind wake steering had minimal impact due to the low penetration of wind energy in the current system but could become

30、more significant as Europe deploys more wind energy to achieve its clean energy goals.In the all-application combined scenario,the reserve margin metric performed best,while average and maximum locational marginal prices(LMPs)were worse than in Scenario 1:AI-driven load reduction.Other metrics were

31、on par with that scenario.These findings highlight that the impact of AI on energy security will depend on how it is rolled out,and that in the case of the European power grid trade-offs between member states are possible.Policy recommendationsOur analysis found that each AI scenario improved at lea

32、st one energy security metric,but the magnitude varied and trade-offs between countries and different metrics were observed.AI-supporting policies should align with the security metric of greater saliency.The impact of AI applications may depend on energy market rules and market incentives.Policymak

33、ers should consider AIs role in current policy discussions on market restructuring or design.Implementing different AI applications together could have effects on energy security.Policymakers should recognise that AIs impact on energy security is not necessarily additive and carefully consider inter

34、actions between applications to avoid adverse impacts.Load-reducing behind-the-meter(BTM)AI applications have significant positive impacts on all four energy security metrics.Policymakers should incentivise the use of AI heating,ventilation and air conditioning(HVAC)control technology for buildings,

35、smart metering,virtual power plants,and distributed energy resources to improve energy security.viThe use of AI for improving energy security-Technical reportPreface iiiExecutive summary ivFigures viiTables viiiAbbreviations and acronyms ixChapter 1.Introduction 1Methods 2Definition of energy securi

36、ty 3Chapter 2.Energy Security 3Potential of AI for energy systems 5AI applications for energy 5Chapter 3.AI in energy 5Modelling approach 7Chapter 4.Quantitative modelling 7Modelling limitations 8Power system model and PyPSA-Eur 9Scenarios 11Benchmark 13Scenario 1(S1):AI-driven load reduction 17Scen

37、ario 2(S2):AI-driven load shifting 21Scenario 3(S3):Wind wake steering 26Scenario 4(S4):All together 29Findings 31Chapter 5.Findings and discussion 31Discussion 35Policy insights 39Chapter 6.Policy implications and further research 39Future research opportunities 40Benefits and challenges 41Chapter

38、7.Open-source model discussion 41Lessons learned 42Chapter 8.Conclusions 43Glossary 44References 45Table of contentsviiFiguresFigure 4.1 Analysis framework 7Figure 4.2.Clustering of European power grid 10Figure 4.3.Our construction of the European grid network(37 nodes)13Figure 4.4.PyPSA output:Elec

39、tricity generation/discharge by source for benchmark 14Figure 4.5.PyPSA output:Hourly power generation by source for benchmark 15Figure 4.6.PyPSA output:Maximum LMP of electricity by node for benchmark 16Figure 4.7.Annual electricity consumption by sector in 2013 for EU countries 17Figure 4.8.Annual

40、 electricity consumption by sector in 2013 for EU countries 18Figure 4.9.PyPSA output:Electricity generation/discharge by source for benchmark versus S1 19Figure 4.10.PyPSA output:Hourly power generation by source for S1 20Figure 4.11.PyPSA output:Maximum LMP of electricity by node for benchmark and

41、 S1 21Figure 4.12.Comparison of hourly load shapes for the system:Adjusted shape to reflect 9%peak load reduction 22Figure 4.13.Comparison of hourly load shapes:Adjusted shape for centralised reduction versus decentralised reduction centralised peak reduction for Albania 23Figure 4.14.Comparison of

42、hourly load shapes:Adjusted shape for centralised reduction versus decentralised reduction decentralised peak reduction for Albania 23Figure 4.15.PyPSA output:Electricity generation/discharge by source for benchmark versus S2 24Figure 4.16.PyPSA output:Hourly power generation by source for S2 25Figu

43、re 4.17.PyPSA output:Maximum LMP of electricity by node for benchmark versus S2 26Figure 4.18.PyPSA output:Electricity generation/discharge by source for benchmark versus S3 27Figure 4.19.PyPSA output:Hourly power generation by source for S3 28Figure 4.20.PyPSA output:Maximum LMP of electricity by n

44、ode for benchmark versus S3 28viiiThe use of AI for improving energy security-Technical reportTablesTable 4.1.Recent AI applications in energy by system and goals 6Table 4.2.Summary of scenarios 12Table 5.1.Energy security metric comparison 31Table 5.2.Changes in accessibility by country and scenari

45、o 32Table 5.3.Changes in acceptability by countries and scenarios 33Table 5.4.Changes in affordability by country and scenario 34Table 5.5.Energy security metric comparison between centralised and decentralised load shifting 38Figure 4.21.PyPSA output:Electricity generation/discharge by source for b

46、enchmark versus S4 29Figure 4.22.PyPSA output:Hourly power generation by source for S4 30Figure 4.23.PyPSA output:Maximum LMP of electricity by node for benchmark versus S4 30Figure 5.1.Distributional impact for S1(load reduction)compared to benchmark 35Figure 5.2.Distributional impact for S2 load s

47、hifting compared to benchmark 37ixAbbreviations and acronymsACAlternating currentAIArtificial intelligenceBTMBehind-the-meterCCGTCombined cycle gas turbineDCDirect currentELCCEffective load carrying capacityENTSO-EThe European Network of Transmission System OperatorsEUEuropean UnionEVElectric vehicl

48、eFDFossil fuel dependencyGHGGreenhouse gasHVACHeating,ventilation and air conditioningIFOMIn-front-of-the-meterLMPLNGLocational marginal priceLiquified natural gasMLMachine learningOCGTOpen cycle gas turbineOFFWINDOffshore windONWINDOnshore windPHSPumped hydro storageRORTRLRun-of-riverTechnology rea

49、diness levelVREVariable renewable energy1Chapter 1.INTRODUCTION This research project investigates the opportunities and risks that may flow from the deployment of AI applications in the electricity system.In this introductory chapter,we set out the background for the research.Electricity systems ar

50、ound the world are changing rapidly due to rising demand,the need to replace aging infrastructure and the need to decarbonise the electricity system.These changes are increasing the electricity systems vulnerability to disruption.In this fast-changing environment,AI applications may help improve the

51、 resilience of the electricity systems,but they may also exacerbate vulnerabilities and increase new risks.This technical report accompanies the main policy report titled The use of AI for improving energy security(RR-A2907-1).It describes of the methods,tools and results of our quantitative explora

52、tion of the impact of 1 Kemp(2023).2 Eddy et al.(2023).AI-based applications on the energy security of the European power grid.ContextIn 2021,gas prices began to rise in Europe as the economy reopened following the COVID-19 pandemic,leading to a rise in electricity prices,with negative impacts on af

53、fordability.This was further exacerbated by the Russian war in Ukraine,which had a profound impact on Europes energy relationship with Russia.At the time of writing this report(summer 2023),the energy crisis in Europe seems to have eased somewhat,in part due to diversification of energy suppliers an

54、d a mild winter.However,energy concerns remain,driven by increasing energy demand from China,which will increase competition with Europe for liquefied natural gas(LNG)imports1 and the possibility of colder winters to come.2 12The use of AI for improving energy security-Technical reportAgainst this b

55、ackdrop,we wanted to understand quantitatively whether and how AI applications could help the constrained European electric system by asking the following questions:1.To what extent could AI applications at technology readiness level(TRL)3 8 or 9 help improve energy security,if such applications wer

56、e widely adopted in the short term(10%on accessibility).Given the technological readiness of this approach,our results support policies that incentivise the use of:a.AI HVAC control technology for buildingsb.Smart metering c.VPP and distributed energy resources.Chapter 6.POLICY IMPLICATIONS AND FURT

57、HER RESEARCH143240The use of AI for improving energy security-Technical reportFuture research opportunitiesWe identified several major opportunities that can be built upon the research conducted using PyPSA.The identified expansions are as follows:49 For an application in India see Spencer et al.(20

58、21).50 PyPSA meets Earth(2022).Expanding the time horizon to long-term future:Expanding the time horizon from short-term(5 years)will allow for the inclusion of AI technologies at a lower level of technological readiness than those considered in this research.Expanding the scope to include other ene

59、rgy sectors:Expanding the scope to include other energy sectors such as primary fuel sources and transportation would help to address a broad energy system question,such as navigating through climate policy and energy system transition.The PyPSA tool already has these features that can be adopted wi

60、thout extensive time or resources.Testing for a longer period time:We were only able to assess the energy system for a week given computational and solver constraints.Including a longer assessment period would allow researchers to examine the system more holistically as demand and supply vary season

61、ally,especially to account for a grid with high penetration of intermittent renewable energy whose generation may vary and be uncertain based on weather conditions.Allowing additional levels of granularity in the optimisation:Given computational constraints,the research was limited to a one-node-per

62、-country for most countries.This limitation was due to the use of a solver that,although available at no cost,has limited power.Additional funds can be allocated to allow for additional levels of granularity in the optimisation,which will help to assess more distributional impact and better represen

63、t the actual system.Exploring other geographic regions:There are many projects49 underway that are using open data to construct energy system in various regions.This expansion will open up the possibility to explore energy policy problems on a global scale.5041One of the ancillary objectives of this

64、 research was to gain experience with open-source power flow modelling.In this section,we share some of the insights the team gained from this exercise that can guide future deployment of AI in the electricity system.Benefits and challenges In recent years,open-source tools and open data have become

65、 increasingly popular in clean energy transition research,due to 51 Groissbck(2019).improved computing capabilities and the need for lower-cost entry solutions for non-industry use cases,such as for researchers,policy analysts and others.PyPSA is a mature open-source tools and has functions comparab

66、le with commercial and proprietary tools.51 It is intuitive and easy to learn.However,during our use of PyPSA,we encountered several issues that impeded our progress in conducting research.Operating system issues with installation:The first issue we encountered was related to the installation proces

67、s.The core function and its dependencies require a Linux system and do not work with Windows.This presented a challenge for us,as we were using Windows,and had to switch to Linux to use PyPSA.This resulted in additional time and effort being expended on the installation process.De-bugging and error

68、handling:We found that de-bugging and handling errors were not intuitive and required us to rely on open discussion forums for answers.This process was unreliable as we sometimes received answers in a matter of days,but at other times we did not receive any response at all.Having to wait to resolve

69、errors before we could continue our work slowed the progress of our research.Deviating from basic set-up:Another issue we encountered was that deviating from the basic set-up,such as choosing a different year to study,required additional processes and runtime.We often faced errors that prevented us

70、from completing the run.As a result,we had to change the plan and limit the project scope to the data and set-up available to us.Frequent updates:We found that PyPSA had frequent updates,with a new version release almost every month.When there was a major update to the tool,we had to configure the t

71、ool again,as a significant amount of the workflow was modified or merged with other functions.This resulted in additional time and effort spent re-doing work we had already completed.Chapter 7.OPEN-SOURCE MODEL DISCUSSION42The use of AI for improving energy security-Technical reportLessons learned O

72、ur experience using PyPSA revealed several issues,such as difficulties with installation,error handling and deviations from the basic set-up,as well as frequent updates that required us to redo work.Through our PyPSA experience,we learned the inherent trade-offs between using open-source and commerc

73、ial tools.While open-source tools like PyPSA offer extensive capabilities and cost-effectiveness,they can also present unique challenges such as system-specific requirements,less intuitive debugging and frequent updates that necessitate reconfiguration.Our experience highlighted the importance of fa

74、ctoring these considerations into the project timeline,especially when facing tight deadlines.Despite these challenges,the understanding of the tool we gained through the project will be valuable for future undertakings.The lessons learned from this research is testament to the evolving nature of po

75、wer system modelling,where the balance between functionality,ease of use and cost-effectiveness is key.43Our exploratory results suggest that AI applications tackling the efficiency and optimisations of demand(BTM)may have a significant positive impact on energy security metrics.The impact varies ac

76、cording to the attribute of demand targeted(e.g.average demand versus peak demand).In addition,our results also indicate significant variability of impact between countries,with some benefiting more than others.Our results should be caveated by the exploratory nature of our analysis,its short-term h

77、orizon as well as the limited AI applications/scenarios analysed.We conclude our report with a list of lessons learned on the use of open-source power flow modelling techniques,which we think are worth highlighting as a result of this research.Chapter 8.CONCLUSIONS4344The use of AI for improving ene

78、rgy security-Technical reportGLOSSARYBTMBehind-the-meter(BTM)refers to resources typically owned and operated by the customer and used to offset their energy consumption from the grid.Correction factorThis factor represents wake losses for wind energy production in PyPSA-Eur model.IFOMIn-front-of-th

79、e-meter(IFOM)refers to the part of the grid owned and operated by utilities or other grid operators,including power plants,transmission lines and distribution networks that deliver electricity to end users.Load scaling factorThis factor allows users to linearly scale the load in PyPSA-Eur model.Rese

80、rve marginReserve margin refers to the amount of excess capacity available in a power system to meet unexpected increases in demand or unexpected outages of power plants or other energy resources.45REFERENCESACER and CEER.Annual Report on the Results of Monitoring the Internal Electricity and Natura

81、l Gas Markets in 2021.2022.As of 24 April 2024:https:/www.ceer.eu/documents/104400/7517827/ACER-CEER+Annual+Report+on+the+Results+of+Monitoring+the+Internal+Electricity+and+Natural+Gas+Markets+in+2021/Adaramola Muyiwa S.andPer ge.Krogstad.2011.Experimental investigation of wake effects on wind turbi

82、ne performance.Renewable Energy 36(8):2078-2086.As of 24 April 2024:https:/doi.org/10.1016/j.renene.2011.01.024Ahlqvist,Victor,Pr Holmberg and Thomas Tangers.2022.A survey comparing centralized and decentralized electricity markets.Energy Strategy Reviews 40:100812.As of 24 April 2024:https:/doi.org

83、/10.1016/j.esr.2022.100812Ang,Beng Wah,W.L.Choong and Adam Tsan Sheng Ng.2015.Energy security:Definitions,dimensions and indexes.Renewable and Sustainable Energy Reviews 42:1077-1093.As of 24 April 2024:https:/doi.org/10.1016/j.rser.2014.10.064Bosh,Jonathan,Iain Staffell and Adam D.Hawkes.2018.Tempo

84、rally explicit and spatially resolved global offshore wind energy potentials.Energy 163:766-781.As of 24 April 2024:https:/doi.org/10.1016/j.energy.2018.08.153Bowler,Ben,Marcus Asprou,Balint Hartmann,Peyman Mazidi and Elias Kyriakides.2020.Enabling Flexibility Through Wholesale Market Changes A Euro

85、pean Case Study.In Flexitranstore:Special Session in the 21st International Symposium on High Voltage Engineering,edited by Blint Nmeth and Lambros Ekonomou,13-22.International:Springer International Publishing.CEER.2020.Recommendations on Dynamic Price Implementation.As of 24 April 2024:https:/www.

86、ceer.eu/documents/104400/-/-/2cc6dfac-8aa7-9460-ac19-4cdf96f8ccd0 Cherp,Aleh and Jessica Jewell.2014.The concept of energy security:Beyond the four As.Energy Policy 75(December):415-421.As of 24 April 2024:https:/doi.org/10.1016/j.enpol.2014.09.005Chinnakannan,Anubama,Oskar Nilsson,Karim Hussain and

87、 Victor Avelar.2022.Using AI to Optimize HVAC Systems in Buildings:A Real-world Example,White Paper 508,Version 1.Schneider Electric Energy Management Research Center,24 June,2.32 p.m.GMT.As of 24 April 2024:https:/ one of the worlds largest wind companies is using AI to capture more energy.Microsof

88、t,24 March,8.54 a.m.AEST.As of 24 April 2024:https:/ and Stanley Reed.2023.Europe Has Weathered an Energy Crisis,for Now.New York Times,February 24.As of 24 April 2024:https:/ Dioxide Emissions Coefficients.As of 24 April 2024:https:/www.eia.gov/environment/emissions/co2_vol_mass.phpEntezari,Ashkan,

89、Alireza Aslani,Rahim Zahedi and Younes Noorollahi.2023.Artificial intelligence and machine learning in energy systems:A bibliographic perspective.Energy Strategy Reviews 45(January):101017.As of 24 April 2024:https:/doi.org/10.1016/j.esr.2022.101017 46The use of AI for improving energy security-Tech

90、nical reportFairley,Peter.2019.Algorithms Help Turbines Share the Wind Field tests affirm the energy boost long promised by optimized“wake steering”of wind turbines.IEEE Spectrum,1 July,6.17 p.m.CEST.As of 24 April 2024:https:/spectrum.ieee.org/teaching-wind-turbines-wake-steering GridPath.GridPath,

91、Advanced Software for Power-System Planning.2023.As of 24 April 2024:https:/www.gridpath.io/Groissbck,Marcus.2019.Are open source energy system optimization tools mature enough for serious use?.Renewable and Sustainable Energy Reviews102:234-248.As of 24 April 2024:https:/doi.org/10.1016/j.rser.2018

92、.11.020Harrison-Atlas,Dylan,Ryan N.King and Andrew Glaws.2021.Machine learning enables national assessment of wind plant controls with implications for land use.Wind Energy 25(10):618-638.As of 24 April 2024:https:/doi.org/10.1002/we.2689 Hrsch,Jonas,Fabian Hofmann,David Schlachtberger and Tom Brown

93、.2018.PyPSA-Eur:An open optimisation model of the European transmission system.Energy Strategy Reviews 22(4):207-215.As of 24 April 2024:https:/arxiv.org/abs/1806.01613 Intharak,Narumon,Jupri Haji Julay,Endang Jati Mat Sahid,Tomoko Matsumoto,Satoshi Nakanishi and Alberto Gotardo Ormeno Aquino.2007.A

94、 Quest for Energy Security in the 21st Century:Resources and Constraints.Tokyo:Institute of Energy Economics.IRENA and KEEI.2021.Renewable Energy and Electricity Interconnections for a Sustainable Northeast Asia.As of 24 April 2024:https:/www.irena.org/publications/2021/May/Renewable-Energy-and-Elec

95、tricity-Interconnections-for-a-Sustainable-Northeast-Asia Kemp,John.2023.Chinas Reduced LNG Appetite Eased Europe Gas Crises.Reuters,Commentary,14 February,8.30 p.m.EST.As of 24 April 2024:https:/ Halhoul,Mohamed Essaaidi,Mohamed Ben Haddou,Basheer Qolomany,Junaid Qadir,Muhammad Anan,Ala Al-Fuqaha,M

96、ohamed Riduan Abid and Driss Benhaddou.2021.Intelligent building control systems for thermal comfort and energy efficiency:A systematic review of artificial intelligence-assisted techniques.Renewable and Sustainable Energy Reviews 144:110969.As of 24 April 2024:https:/doi.org/10.1016/j.rser.2021.110

97、969Moorhouse,David.2002.Detailed Definitions and Guidance for Application of Technology Readiness Levels.Journal of Aircraft 39(1):190-192.As of 24 April 2024:https:/doi.org/10.2514/2.2916Neumann,Fabian,Lisa Zeyen,Marta Victoria,Fabian Hofmann,Jonas Hrsch and Tom Brown.2020.PyPSA-Eur:An open optimis

98、ation model of the European transmission system.Karlsruher Institut fr Technologie.As of 28 May 2024:https:/pypsa-eur.readthedocs.io/en/latest/index.html PyPSA.2023.PyPSA-Eur:A Sector-Coupled Open Optimisation Model of the European Energy System.GitHub.As of 24 April 2024:https:/ PyPSA meets Earth.2

99、023.Welcome to the PyPSA-Earth documentation.As of 24 April 2024:https:/pypsa-earth.readthedocs.io/en/latest/47Roald Line A.,David Pozo,Anthony Papavasiliou,Daniel K.Molzahn,Jalal Kazempour and Antonio Conejo.2023.Power systems optimization under uncertainty:A review of methods and applications.Elec

100、tric Power Systems Research.As of September 2023:https:/ R.O.,Icaro H.Honorato,Rodrigo Fiorotti,Wanderley C.Celeste,Leonardo J.Silvestre and Jair A.Silva.2021.An Artificial FIntelligence based scheduling algorithm for demand-side energy management in Smart Homes.Applied Energy 282:116145.As of 24 Ap

101、ril 2024:https:/doi.org/10.1016/j.apenergy.2020.116145 Saroj,Ana.2022.How Vestas is using AI to Capture more Energy.Minds.ai,30 March,2.39 p.m.CEST.As of 24 April 2024:https:/minds.ai/post/how-vestas-is-using-ai-to-capture-more-energy/Scott,J.2019.Grid Edge:Artificial Intelligence for Energy Systems

102、,Presentation delivered at International Energy Agency Workshop on Modernising Energy Efficiency through Digitalisation.IEA.As of 24 April 2024:https:/ GridEdge.pdf Sder,Lennart,Peter D.Lund,Hardi Koduvere,Torjus Folsland Bolkesj,Geir Hyvik Rosseb,Emilie Rosenlund-Soysal,Klaus Skytte,Jonas Katz and

103、Dagnija Blumberga.2018.A review of demand side flexibility potential in Northern Europe.Renewable and Sustainable Energy Reviews 91:654-664.As of 24 April 2024:https:/doi.org/10.1016/j.rser.2018.03.104Spencer,Thomas,Neshwin Rodrigues,Raghav Pachouri and Shubham Thkre.2021.A model-based assessment of

104、 variable renewable grid integration costs in India.The Energy and Resources Institute.As of 24 April 2024:https:/www.teriin.org/sites/default/files/2021-02/A_Modal-Based_Assessment_Report_0.pdfSwitch.2023.Switch Power System Planning Model.GitHub.24 April 2024:https:/switch-model.org/Wahid,Fazli,Lo

105、kman Hakim Ismail,Rozaida Ghazali and Muhammad Aamir.2019.An Efficient Artificial Intelligence Hybrid Approach for Energy Management in Intelligent Buildings.KSII Transactions on Internet&Information Systems 13(12):5904-5927.As of 24 April 2024:https:/doi.org/10.3837/tiis.2019.12.007 Zettelmeyer,Jermomin,Simone Tagliapietra,Georg Zachmann and Conall Heussaf.2022.Beating the European energy crisis.IMF,1 December,4.19 p.m.CEST.As of 24 April 2024:https:/www.imf.org/en/Publications/fandd/issues/2022/12/beating-the-european-energy-crisis-Zettelmeyer