融合机器学习与SHAP值算法的居民需求响应个体异质性因素挖掘与应用研究

王兆华, 刘杰, 王博, 邓娜娜, 聂富华

系统工程理论与实践 ›› 2024, Vol. 44 ›› Issue (7) : 2247-2259.

PDF(2208 KB)
中国科学院数学与系统科学研究院期刊网
PDF(2208 KB)
系统工程理论与实践 ›› 2024, Vol. 44 ›› Issue (7) : 2247-2259. DOI: 10.12011/SETP2023-0677
论文

融合机器学习与SHAP值算法的居民需求响应个体异质性因素挖掘与应用研究

    王兆华1,2,3, 刘杰1,2,3, 王博1,2,3, 邓娜娜1,2,3, 聂富华1,2,3
作者信息 +

Research on mining and applications of individual heterogeneity factors in resident demand response by integrating machine learning and SHAP value algorithm

    WANG Zhaohua1,2,3, LIU Jie1,2,3, WANG Bo1,2,3, DENG Nana1,2,3, NIE Fuhua1,2,3
Author information +
文章历史 +

摘要

本研究基于大规模居民电力需求响应(EDR: electricity demand response)实验以及家庭用电调查数据, 利用机器学习和SHAP (Shapley additive explanatory)值算法从全局和个体两个层面对影响居民参与需求响应的影响因素进行了识别和异质性分析. 研究发现, 居民是否参与需求响应活动是外部激励, 家庭结构, 用电规律与习惯倾向, 用电知识等因素共同作用的结果, 其效应的大小和极性存在着丰富的异质性.其中, 电话营销等外部激励对用户参与需求响应影响最大, 其效果在年龄较大以及受教育程度较高的群体较为明显; 响应时段基准用电量在1度左右的用户参与倾向较大; 节能环保意识较强且具有较高节电条件的家庭参与概率更高. 同时, 依据SHAP值的交互以及分解性质, 在后续需求响应活动中对用户进行分类营销, 可以节省93.9%的营销成本, 并提高46.4%的参与人数. 本研究对不同群体的异质性进行了更为细致的分析研究, 为未来新型电力系统下进行更为精确和智能的需求响应提供了重要支撑.

Abstract

Based on the large-scale residential electricity demand response (EDR) experiment and household survey data, this study uses machine learning and SHAP (Shapley additive explanatory) value algorithm to identify and analyze the influencing factors of residents' participation in demand response from the whole and individual levels. Our study found that whether residents participate in EDR activities is the result of the joint action of external incentives, family structure, electricity use habits, electricity use knowledge, and the value and polarity of the effect is varied in heterogeneity. Among them, external incentives such as telemarketing have the greatest impact on customers' participation in demand response, and the effect is more obvious among older and more educated groups; customers with a baseline electricity consumption of about 1 kW⋅h during the response period have a higher tendency to participate; households with a stronger awareness of energy conservation and higher conditions for saving electricity have a higher probability of participation. At the same time, according to the interaction and decomposition properties of SHAP value, classified marketing for users in subsequent EDR activities can save 93.9% of marketing costs and increase the number of participants by 46.4%. This study has carried out a more detailed analysis and research on the heterogeneity of different groups, providing an important support for more accurate and intelligent EDR to China's new power system.

关键词

需求响应 / 因素分析 / 机器学习 / SHAP值

Key words

demand response / factor analysis / machine learning / SHAP value

引用本文

EndNote

Ris (Procite)

Bibtex

导出引用
王兆华 , 刘杰 , 王博 , 邓娜娜 , 聂富华. 融合机器学习与SHAP值算法的居民需求响应个体异质性因素挖掘与应用研究. 系统工程理论与实践, 2024, 44(7): 2247-2259 https://doi.org/10.12011/SETP2023-0677
WANG Zhaohua , LIU Jie , WANG Bo , DENG Nana , NIE Fuhua. Research on mining and applications of individual heterogeneity factors in resident demand response by integrating machine learning and SHAP value algorithm. Systems Engineering - Theory & Practice, 2024, 44(7): 2247-2259 https://doi.org/10.12011/SETP2023-0677
中图分类号： F423.3   

参考文献

[1] 秦大河. 气候变化科学与人类可持续发展[J]. 地理科学进展, 2014, 33(7): 874-883. Qin D H. Climate change science and sustainable development[J]. Progress in Geography, 2014, 33(7): 874-883.
[2] IPCC. Climate change 2022: Impacts, adaptation and vulnerability[R]. Cambridge: Cambridge University Press, 2022.
[3] 秦云, 徐新武, 王蕾, 等. IPCC AR6报告关于气候变化适应措施的解读[J]. 气候变化研究进展, 2022, 18(4): 452-459. Qin Y, Xu X W, Wang L, et al. Interpretation of the IPCC AR6 on adaptation options of climate change[J]. Climate Change Research, 2022, 18(4): 452-459.
[4] 胡鞍钢. 中国实现2030年前碳达峰目标及主要途径[J]. 北京工业大学学报(社会科学版), 2021, 21(3): 1-15. Hu A G. China’s goal of achieving carbon peak by 2030 and its main approaches[J]. Journal of Beijing University of Technology (Social Sciences Edition), 2021, 21(3): 1-15.
[5] 赵冉. 构建以新能源为主体的新型电力系统[N]. 中国电力报, 2021-04-01(002). doi: 10.28061/n.cnki.ncdlb.2021.000594.2021-04-01. Zhao R. Building a new type of power system with new energy as the main body[N]. China Electric Power News, 2021-04-01(002). doi: 10.28061/n.cnki.ncdlb.2021.000594.2021-04-01.
[6] 国家发改委. 国家能源局有关负责同志就《关于完善能源绿色低碳转型体制机制和政策措施的意见》答记者问[J]. 财经界, 2022, 612(5): 1-3. National Energy Administration. Reply to reporters’ questions on “Opinions on improving the system, mechanism, and policy measures for green and low carbon energy transformation” by relevant offcials of the National Energy Administration[J]. Money China, 2022, 612(5): 1-3.
[7] 韩肖清, 李廷钧, 张东霞, 等. 双碳目标下的新型电力系统规划新问题及关键技术[J]. 高电压技术, 2021, 47(9): 3036-3046. Han X Q, Li T J, Zhang D X, et al. New issues and key technologies of new power system planning under double carbon goals[J]. High Voltage Engineering, 2021, 47(9): 3036-3046.
[8] 许博, 岳欣明, 关艳, 等. 针对用电负荷“峰谷倒挂” 现象的混合型电力需求响应策略[J]. 系统工程理论与实践, 2022, 42(8): 2129-2138. Xu B, Yue X M, Guan Y, et al. A hybrid response rate to solve the problem of “peak valley inversion”[J]. Systems Engineering—Theory & Practice, 2022, 42(8): 2129-2138.
[9] Agrawal V V Y, Şafak Y. Design of electricity demand-response programs[J]. Management Science, 2022, 68(10): 7441-7456.
[10] 代业明, 高亚丽, 尹慧, 等. 考虑售电商可再生能源补贴的电力市场综合需求响应策略研究[J/OL]. 中国管理科学, 1-21.[2024-05-02]. https://doi.org/10.16381/j.cnki.issn1003-207x.2022.2441. Dai Y M, Gao Y L, Yin H, et al. Research on integrated demand response strategy of electricity market considering renewable energy subsidies[J/OL]. Chinese Journal of Management Science, 1-21.[2024-05-02]. https://doi.org/10.16381/j.cnki.issn1003-207x.2022.2441.
[11] Tao L, Gao Y. Real-time pricing for smart grid with distributed energy and storage: A noncooperative game method considering spatially and temporally coupled constraints[J]. International Journal of Electrical Power & Energy Systems, 2020, 115: 105487.
[12] Dong J, Xue G, Li R. Demand response in China: Regulations, pilot projects and recommendations—A review[J]. Renewable and Sustainable Energy Reviews, 2016, 59: 13-27.
[13] 卓振宇, 张宁, 谢小荣, 等. 高比例可再生能源电力系统关键技术及发展挑战[J]. 电力系统自动化, 2021, 45(9): 171-191. Zhuo Z Y, Zhang N, Xie X R, et al. Key technologies and developing challenges of power system with high proportion of renewable energy[J]. Automation of Electric Power Systems, 2021, 45(9): 171-191.
[14] 陶小马, 周雯. 电力需求响应的研究进展及文献述评[J]. 北京理工大学学报(社会科学版), 2014, 16(1): 32-40.Tao X M, Zhou W. A review of the research on electricity demand response[J]. Journal of Beijing Institute of Technology (Social Sciences Edition), 2014, 16(1): 32-40.
[15] 赵晓东, 王娟, 周伏秋, 等. 构建新型电力系统亟待全面推行电力需求响应——基于11省市电力需求响应实践的调研 [J]. 宏观经济管理, 2022(6): 52-60. Zhao X D, Wang J, Zhou F Q, et al. The construction of a new type of power system urgently requires the comprehensive implementation of power demand response—A survey based on the practice of power demand response in 11 provinces and cities[J]. Macroeconomic Management, 2022(6): 52-60.
[16] Mizobuchi K, Takeuchi K. The influences of financial and non-financial factors on energy-saving behaviour: A field experiment in Japan[J]. Energy Policy, 2013, 63: 775-787.
[17] Aalami H A, Khatibzadeh A. Regulation of market clearing price based on nonlinear models of demand bidding and emergency demand response programs[J]. International Transactions on Electrical Energy Systems, 2016, 26(11): 2463-2478
[18] 涂京, 周明, 宋旭帆, 等. 居民用户参与电网调峰激励机制及优化用电策略研究[J]. 电网技术, 2019, 43(2): 443-453. Tu J, Zhou M, Song X F, et al. Research on incentive mechanism and optimal power consumption strategy for residential users’ participation in peak shaving of power grid[J]. Power System Technology, 2019, 43(2): 443-453.
[19] Yamaguchi Y, Chen C F, Shimoda Y, et al. An integrated approach of estimating demand response flexibility of domestic laundry appliances based on household heterogeneity and activities[J]. Energy Policy, 2020, 142: 111467.
[20] Frondel M, Kussel G, Sommer S. Heterogeneity in the price response of residential electricity demand: A dynamic approach for Germany[J]. Resource and Energy Economics, 2019, 57: 119-134.
[21] 蒋锋, 张文雅. 机器学习方法在经济研究中的应用[J]. 统计与决策, 2022, 38(4): 43-49. Jiang F, Zhang W Y. Application of machine learning methods in economic research[J]. Statistics & Decision, 2022, 38(4): 43-49.
[22] 夏飞, 张洁, 张浩, 等. 基于BIC准则和加权皮尔逊距离的居民负荷模式精细识别及预测[J]. 电子测量与仪器学报, 2020, 34(11): 33-42. Xia F, Zhang J, Zhang H, et al. Fine recognition and prediction of resident load pattern based on BIC criterion and weighted Pearson distance[J]. Journal of Electronic Measurement and Instrumentation, 2020, 34(11): 33-42.
[23] 葛磊蛟, 刘航旭, 赵康, 等. 面向商业和居民混合的配电网短期负荷预测HGWOACOA-LSTMN方法[J]. 天津大学学报(自然科学与工程技术版), 2021, 54(12): 1269-1279. Ge L J, Liu H X, Zhao K, et al. An HGWOACOA-LSTMN method for short-term load forecasting of distribution network for commercial and residential users[J]. Journal of Tianjin University (Science and Technology), 2021, 54(12): 1269-1279.
[24] 史佳琪, 张建华. 基于多模型融合Stacking集成学习方式的负荷预测方法[J]. 中国电机工程学报, 2019, 39(14): 4032-4042. Shi J Q, Zhang J H. Load forecasting based on multi-model by stacking ensemble learning[J]. Proceedings of the CSEE, 2019, 39(14): 4032-4042.
[25] Wang B, Deng N, Zhao W, et al. Residential power demand side management optimization based on fine-grained mixed frequency data[J]. Annals of Operations Research, 2021, 316: 1-20.
[26] Wang Z, Zhao W, Deng N, et al. Mixed data-driven decision-making in demand response management: An empirical evidence from dynamic time-warping based nonparametric-matching DID[J]. Omega, 2021, 100: 102233.
[27] Grinsztajn L, Oyallon E, Varoquaux G. Why do tree-based models still outperform deep learning on tabular data?[J]. Arxiv Preprint Arxiv: 2207.08815, 2022.
[28] Lundberg S M, Lee S I. A unified approach to interpreting model predictions[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach, California, USA; Curran Associates Inc. 2017: 4768-4777.
[29] Karn R R, Kudva P, Huang H, et al. Cryptomining detection in container clouds using system calls and explainable machine learning[J]. IEEE Transactions on Parallel and Distributed Systems, 2021, 32(3): 674-691.
[30] Lundberg S M, Lee S I. Consistent feature attribution for tree ensembles[J]. Arxiv Preprint Arxiv: 1706.06060, 2017.

基金

国家自然科学基金(72243001,72074026,72141302,72321002)
PDF(2208 KB)

736

Accesses

0

Citation

Detail
段落导航
相关文章

/

〈 〉