基于多维数据与深度学习的区域发电碳排放因子预测研究

doi:10.3969/j.issn.2097-0706.2023.08.002

综合智慧能源 ›› 2023, Vol. 45 ›› Issue (8): 11-17.doi: 10.3969/j.issn.2097-0706.2023.08.002

基于多维数据与深度学习的区域发电碳排放因子预测研究

李方一¹^,²(), 李楠¹^,², 周琰¹^,², 谢武¹^,²

1.合肥工业大学管理学院，合肥 230009
2.能源环境智慧管理与绿色低碳发展安徽省哲学社会科学重点实验室（合肥工业大学），合肥 230009

收稿日期:2023-05-31 修回日期:2023-06-23 接受日期:2023-07-26 出版日期:2023-08-25 发布日期:2023-08-22
作者简介:李方一（1985），男，副教授，博士，从事大数据与能源环境管理等方面的研究，fyli@hfut.edu.cn。
基金资助:
国家自然科学基金项目(71902051)

Prediction on the regional carbon emission factor for power generation based on multi-dimensional data and deep learning

LI Fangyi¹^,²(), LI Nan¹^,², ZHOU Yan¹^,², XIE Wu¹^,²

1. School of Management, Hefei University of Technology, Hefei 230009, China
2. Anhui Key Laboratory of Philosophy and Social Sciences of Energy and Environment Smart Management and Green Low Carbon Development, Hefei University of Technology, Hefei 230009, China

Received:2023-05-31 Revised:2023-06-23 Accepted:2023-07-26 Online:2023-08-25 Published:2023-08-22
Supported by:
National Natural Science Foundation of China(71902051)

摘要/Abstract

摘要：

在碳交易背景下，对电力企业进行实时、准确、全面的碳排放计量是开展发电结构调整、技术创新、供需联动、碳交易等工作的基础。受制于数据采集与传输系统的限制，动态碳排放因子的测算与预测目前仍难以完全实现。采用深度学习方法，将双重注意力机制与传统的门控循环单元（GRU）神经网络融合，构建了GRU-Attention预测模型。以合肥市2022年的电力数据为样本，结合合肥市平均气象数据，对GRU模型、长短时记忆（LSTM）模型、基于双重注意力机制的LSTM-Attention模型和GRU-Attention模型进行训练，以实现小时级别的碳排放因子预测。利用不同的评价指标对4种预测模型进行对比，与GRU，LSTM，LSTM-Attention模型相比，GRU-Attention模型预测精度更高，有助于实现发电碳排放因子的中长期预测。

关键词: 碳排放计量, 碳排放因子, 门控循环单元, 双重注意力机制, 深度学习

Abstract:

With the support of carbon trading policy， the real-time， accurate and comprehensive measurement on power enterprises' carbon emissions is the basis for structure adjustment， technological innovation， supply and demand side interaction and carbon trading of power generation industry. The calculation and prediction on dynamic carbon emission factors is still limited by the data collection and transmission system. By taking deep learning， a prediction model， called GRU-Attention model， was built by combining dual attention mechanism with traditional Gate Recurrent Unit （GRU） neural network. Then， a GRU model， a Long Short-Term Memory （LSTM） model， a LSTM model based on dual attention mechanism（LSTM-Attention） and a GRU-Attention model were constructed and trained by the power data of Hefei in 2022 and average meteorological data of Hefei， to achieve hourly prediction on carbon emission factor. Comparing the prediction results made by the four models above， it is found that the prediction made by the GRU-Attention model is more accurate than that of the other three models， which can advance the mid- and long-term prediction on carbon emission factor.

Key words: measurement of carbon emissions, carbon emission factor, Gate Recurrent Unit, dual attention mechanism, deep learning

中图分类号:

TK01：F206

李方一, 李楠, 周琰, 谢武. 基于多维数据与深度学习的区域发电碳排放因子预测研究[J]. 综合智慧能源, 2023, 45(8): 11-17.

LI Fangyi, LI Nan, ZHOU Yan, XIE Wu. Prediction on the regional carbon emission factor for power generation based on multi-dimensional data and deep learning[J]. Integrated Intelligent Energy, 2023, 45(8): 11-17.

图/表 11

图1

表1

图2

图3

图4

表2

图5

图6

图7

图8

表3

参考文献 20

[1]	新华社. 中共中央国务院关于完整准确全面贯彻新发展理念做好碳达峰碳中和工作的意见[EB/OL].(2021-10-24)[2023-05-25]. http://www.gov.cn/zhengce/2021-10/24/content_5644613.htm.
[2]	中华人民共和国生态环境部. 碳排放权交易管理办法(试行)[EB/OL].(2021-10-24)[2023-05-25]. http://www.gov.cn/zhengce/zhengceku/2021-01/06/content_5577360.htm.
[3]	ANG B W. Decomposition analysis for policy making in energy:Which is the preferred method?[J]. Energy Policy, 2004, 32(9): 1131-1139. doi: 10.1016/S0301-4215(03)00076-4
[4]	潘家华, 庄贵阳, 郑艳, 等. 低碳经济的概念辨识及核心要素分析[J]. 国际经济评论, 2010(4): 88-101.
	PAN Jiahua, ZHUANG Guiyang, ZHENG Yan, et al. Clarification of the concept of low-carbon economy and analysis of its core elements[J]. International Economic Review, 2010(4): 88-101.
[5]	李小胜, 张焕明. 中国碳排放效率与全要素生产率研究[J]. 数量经济技术经济研究, 2016, 33(8): 64-79,161.
	LI Xiaosheng, ZHANG Huanming. Study on the efficiency and the total factor productivity of carbon emissions in China[J]. The Journal of Quantitative & Technical Economics, 2016, 33(8): 64-79,161.
[6]	ZHANG Q, YANG J, SUN Z, et al. Analyzing the impact factors of energy-related CO₂ emissions in China: What can spatial panel regressions tell us?[J]. Journal of Cleaner Production, 2017, 161:1085-1093. doi: 10.1016/j.jclepro.2017.05.071
[7]	陈公达, 邹祥波, 卢锐, 等. 中外火电企业碳排放统计方法与质量控制现状分析[J]. 热力发电, 2022, 51(10):54-60.
	CHEN Gongda, ZOU Xiangbo, LU Rui, et al. Domestic and international statistical methods and quality control status for carbon emission from fossil-fired power plants[J]. Thermal Power Generation, 2022, 51(10):54-60.
[8]	林玥廷, 李世明, 卢建刚, 等. 燃煤电厂碳排放在线监测和管理系统设计[J]. 自动化技术与应用, 2018, 37(4):139-141.
	LIN Yueting, LI Shiming, LU Jiangang, et al. Design of on-line monitoring and management system for carbon emission of coal-fired power plants[J]. Techniques of Automation and Applications, 2018, 37(4):139-141.
[9]	陈咏城, 唐雯, 马旭涛. 火力发电厂碳排放测量及分析[J]. 电站辅机, 2021, 42(3):14-17.
	CHEN Yongcheng, TANG Wen, MA Xutao. Measurement and analysis of carbon emission in thermal power plant[J]. Power Station Auxiliary Equipment, 2021, 42(3):14-17.
[10]	陈剑, 马大卫, 王正风, 等. 煤电机组CO₂排放原位在线监测系统研发与应用[J]. 电力科技与环保, 2022, 38(5):423-431.
	CHEN Jian, MA Dawei, WANG Zhengfeng, et al. Development and application of on-line CO₂ emission monitoring system in coal-fired power plant[J]. Electric Power Technology and Environmental Protection, 2022, 38(5):423-431.
[11]	李姚旺, 张宁, 杜尔顺, 等. 基于碳排放流的电力系统低碳需求响应机制研究及效益分析[J]. 中国电机工程学报, 2022, 42(8): 2830-2842.
	LI Yaowang, ZHANG Ning, DU Ershun, et al. Mechanism study and benefit analysis on power system low carbon demand response based on carbon emission flow[J]. Proceedings of the CSEE, 2022, 42(8): 2830-2842.
[12]	李善寿, 马枭杰, 潘璐茜, 等. 基于VMD和LSTM-CNN的短期负荷预测模型研究[J]. 控制工程, 2023, 30(3):1-11.
	LI Shanshou, MA Xiaojie, PAN Luxi, et al. Research on short-term load forecasting model based on VMD and LSTM-CNN[J]. Control Engineering of China, 2023, 30(3):1-11.
[13]	毛清华, 张强. 融合柯西变异和反向学习的改进麻雀算法[J]. 计算机科学与探索, 2021, 15(6):1155-1164. doi: 10.3778/j.issn.1673-9418.2010032
	MAO Qinghua, ZHANG Qiang. Improved sparrow algorithm combining Cauchy mutation and opposition-based learning[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(6): 1155-1164.
[14]	YUAN N J, ZHENG Y, ZHANG L H, et al. T-Finder:A recommender system for finding passengers and vacant taxis[J]. IEEE Transactions on Knowledge and Data Engineering, 2013, 25(10):2390-2403. doi: 10.1109/TKDE.2012.153
[15]	VERMA T, VARAKANTHAM P, KRAUS S, et al. Augmenting decisions of taxi drivers through reinforcement learning for improving revenues[C]//Proceedings of 2017 International Conference on Automated Planning and Scheduling. Palo Alto: AAAI Press, 2017:409-417.
[16]	李雅丽, 王淑琴, 陈倩茹, 等. 若干新型群智能优化算法的对比研究[J]. 计算机工程与应用, 2020, 56(22):1-12. doi: 10.3778/j.issn.1002-8331.2006-0291
	LI Yali, WANG Shuqin, CHEN Qianru, et al. Comparative study of several new swarm intelligence optimization algorithms[J]. Computer Engineering and Applications, 2020, 56(22):1-12. doi: 10.3778/j.issn.1002-8331.2006-0291
[17]	LI X L, PAN G, WU Z H, et al. Prediction of urban human mobility using large-scale taxi traces and its applications[J]. Frontiers of Computer Science, 2012, 6(1): 111-121. doi: 10.1007/s11704-011-1192-6
[18]	HWANG R H, HSUEH Y L, CHEN Y T, et al. An effective taxi recommender system based on a spatio-temporal factor analysis model[J]. Information Sciences, 2015, 314(12):28-40. doi: 10.1016/j.ins.2015.03.068
[19]	LUO Z W, LYU H M, FANG F, et al. Dynamic taxi service planning by minimizing cruising distance without passengers[J]. IEEE Access, 2018, 6:70005-70016. doi: 10.1109/Access.6287639
[20]	孙友源, 郑张, 秦亚琦, 等. 火电机组碳排放特性研究及管理建议[J]. 中国电力, 2018, 51(3):144-149.
	SUN Youyuan, ZHENG Zhang, QIN Yaqi, et al. Study on carbon emission characteristics and suggestions on carbon emission management of coal-fired power plant[J]. Electric Power, 2018, 51(3):144-149.

指标		最小值	最大值	平均值	中位数	标准差	方差
输入指标	气温/℃	-8.52	14.52	1.78	1.08	4.73	22.36
	湿度/%	22.56	100.00	63.63	68.07	20.43	417.49
	气压/hPa	1 010.890	1 035.030	1 022.120	1 020.777	6.200	38.430
	降水量/(mm·h^-1)	0	0.450 0	0.130 0	0	0.060 0	0.003 6
	经向风速/(m·s^-1)	-4.490	6.960	0.690	0.920	3.070	9.430
	纬向风速/(m·s^-1)	-10.250	4.480	-0.980	-0.796	3.440	11.820
	地面风速/(m·s^-1)	1.067	11.210	4.360	4.010	3.560	1.890
	风向/(°)	1.05	358.59	199.18	213.25	106.27	11 293.87
输出指标	碳排放因子/[g·(kW·h)^-1]	529.000	810.000	714.000	733.000	62.200	3 868.142

神经网络层	类型	激活	可学习参数量
输入层	序列输入	9×1×1	0
全连接层	全连接	9×1×1	72
ReLU层	Relu	9×1×1	0
Sigmoid层	sigmoid	9×1×1	0
Multiplication层	按元素相乘	9×1×1	0
LSTM层	GRU	10×1×1	480
LSTM层	GRU	20×1	1 500
全连接层	全连接	1×1	20
输出层	回归输出	1×1	0

模型	R²	δ_MAE	δ_MBE	δ_RMSE
LSTM	0.530 04	1.934 1	-0.295 11	1.905 0
GRU	0.500 02	2.173 5	-0.367 62	1.992 3
LSTM-Attention	0.864 27	1.032 5	-0.397 35	1.497 2
GRU-Attention	0.899 72	1.014 4	-0.166 08	1.307 0

基于多维数据与深度学习的区域发电碳排放因子预测研究

Prediction on the regional carbon emission factor for power generation based on multi-dimensional data and deep learning

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 20

相关文章 6

编辑推荐

Metrics

本文评价

[1]	金立, 张力, 唐杨, 唐侨, 任炬光, 杨焜, 刘小兵. 考虑温湿指数与耦合特征的综合能源负荷短期预测[J]. 综合智慧能源, 2023, 45(7): 70-77.
[2]	房瑞, 段志勇, 刘在智, 王宇轩, 刘辰希, 李浩, 范垂钢. 海洋垃圾治理技术综述[J]. 综合智慧能源, 2023, 45(5): 70-79.
[3]	翟广松, 王鹏, 梁鹏勋, 谢智锋, 殷豪. 基于纵横交叉算法-门控循环单元的日前电价预测模型[J]. 综合智慧能源, 2022, 44(11): 36-42.
[4]	肖新帅, 田秀霞, 徐曼. 基于端到端算法的绝缘子检测技术研究[J]. 华电技术, 2021, 43(2): 28-33.
[5]	吕政权, 李朝阳, 王海峰, 陈怡君, 彭道刚. 基于GRU-CNN的综合能源网络安全攻击检测方法[J]. 华电技术, 2021, 43(2): 9-14.
[6]	李小斌, 张红娜, 曲凯阳, 李凤臣 . 核能集中供热系统优越性分析[J]. 华电技术, 2020, 42(11): 69-82.