基于LASSO-GWO-KELM的工业碳排放预测方法研究

张新生; 魏志臻; 陈章政; 韩轶伟

doi:10.13205/j.hjgc.202310018

基于LASSO-GWO-KELM的工业碳排放预测方法研究

doi: 10.13205/j.hjgc.202310018

1. 西安建筑科技大学管理学院, 西安 710055;
2. 陕西省新型城镇化和人居环境研究院, 西安 710055

基金项目:

陕西省社科界重大理论与现实问题研究联合项目"城镇化进程的陕西市域碳减排重点路径研究"（2022HZ1522）

详细信息

作者简介:
张新生,教授,博士研究生导师,主要研究方向为城镇化与碳排放、机器学习与应用。zhangxs@xauat.edu.cn

计量
- 文章访问数: 168
- HTML全文浏览量: 21
- PDF下载量: 4
- 被引次数: 0
出版历程
- 收稿日期: 2023-07-08
- 网络出版日期: 2023-12-26

PREDICTION OF INDUSTRIAL CARBON EMISSIONS IN SHAANXI PROVINCE BASED ON LASSO-GWO-KELM MODEL

1. School of Management, Xi'an University of Architecture and Technology, Xi'an 710055, China;
2. The Research Institute of New Urbanization and Human Settlement in Shaanxi Province, Xi'an 710055, China

摘要

摘要: 针对工业碳排放系统的总量预测问题，建立基于套索回归（LASSO）、灰狼优化算法（GWO）和核极限学习机（KELM）相结合的模型提高碳排放量预测精度。首先根据IPCC公式法与电热分摊法核算2000—2020年工业直接与间接碳排放量，运用STIRPAT模型选取国内生产总值、能源结构、固定资产投资等指标；然后通过灰色关联分析、LASSO回归模型筛选出7个显著影响因素；再接着对工业碳排放系统的参数数据进行预处理并输入至KELM模型，使用GWO算法优化KELM正则化系数（C）和核函数参数（γ）；最后将预测结果集成汇总，并对比分析LASSO-GWO-KELM、LASSO-SSA-KELM、LASSO-SFO-KELM、LASSO-KELM和LASSO-ELM预测结果。结果显示：LASSO-GWO-KELM模型预测值与实际值拟合，其均方误差、平均绝对误差、均方根误差、平均绝对百分比误差分别为0.02%、1.09%、1.33%和1.17%，均优于其他模型，证明该模型能够更为准确地预测工业碳排放量，为我国尽早实现“双碳”目标提供参考。
- 工业 /
- 套索回归 /
- 核极限学习机 /
- 灰狼优化算法 /
- 碳排放预测
Abstract: A model based on LASSO regression (LASSO), Grey Wolf Optimization algorithm (GWO) and nuclear Extreme Learning Machine (KELM) was established to improve the prediction accuracy of industrial carbon emissions. Firstly, the direct and indirect carbon emissions of the industry during 2000 to 2020 were calculated according to the IPCC formula method and electrothermal allocation method respectively, and the gross domestic product (GDP), energy structure, and fixed asset investment were selected by the STIRPAT model. Then seven significant influencing factors were selected by grey correlation analysis and the LASSO regression model. Secondly, the parameter data of the industrial carbon emission system were preprocessed and input into the KELM model, and the KELM regularization coefficient (C) and kernel function parameter (γ) were optimized using the GWO algorithm. Finally, the forecast results were integrated and summarized, and the forecast results of LASSO-GGO-KELM, LASSO-SSA-KELM, LASSO-SFO-KELM, LASSO-KELM, and LASSO-ELM were compared and analyzed. The results showed that the predicted value of the LASSO-GGO-KELM model fits well with the actual value, and its mean square error, mean absolute error, root mean square error, and mean absolute percentage error was 0.02%, 1.09%, 1.33%, and 1.17% respectively, which was superior to other models, proving that this model can predict industrial carbon emissions more accurately. This study can provide a reference for China to realize the Double Carbon Goal as soon as possible.
- industry /
- LASSO regression /
- kernel extreme learning machine (KELM) /
- grey wolf optimization algorithm (GWO) /
- carbon emission prediction

HTML全文

参考文献(27)

[1]	杨冕, 卢昕, 段宏波.中国高耗能行业碳排放因素分解与达峰路径研究[J].系统工程理论与实践, 2018, 38(10):2501-2511.
[2]	张巍.基于STIRPAT模型的陕西省工业碳排放量预测和情景分析[J].可再生能源, 2017(5):771-777.
[3]	王向前, 夏丹.工业煤炭生产-消费两侧碳排放及影响因素研究:基于STIRPAT-EKC的皖豫两省对比[J].软科学, 2020, 34(8):84-89.
[4]	张国兴, 苏钊贤.黄河流域交通运输碳排放的影响因素分解与情景预测[J].管理评论, 2020, 32(12):283-294.
[5]	刘炳春, 符川川, 李健.基于PCA-SVR模型的中国CO₂排放量预测研究[J].干旱区资源与环境, 2018, 32(4):51-61.
[6]	赵金辉, 李景顺, 王潘乐, 等.基于Lasso-BP神经网络模型的河南省碳达峰路径研究[J].环境工程, 2022, 40(12):151-164.
[7]	徐丽, 曲建升, 李恒吉, 等.西北地区居民生活碳排放现状分析及预测[J].干旱区地理, 2019, 42(5):1166-1175.
[8]	WANG Z X, YE D J.Forecasting Chinese carbon emissions from fossil energy consumption using non-linear grey multivariable models[J].Journal of Cleaner Production, 2017, 142(8):600-612.
[9]	王勇, 毕莹, 王恩东.中国工业碳排放达峰的情景预测与减排潜力评估[J].中国人口·资源与环境, 2017, 27(10):131-140.
[10]	邵帅, 张曦, 赵兴荣.中国制造业碳排放的经验分解与达峰路径:广义迪氏指数分解和动态情景分析[J].中国工业经济, 2017(3):44-63.
[11]	SUN W, LIU M H.Prediction and analysis of the three major industries and residential consumption CO₂ Emissions based on least squares support vector machine in China[J].Journal of Cleaner Production, 2016, 122(2):144-153.
[12]	胡剑波, 罗志鹏, 李峰."碳达峰"目标下中国碳排放强度预测:基于LSTM和ARIMA-BP模型的分析[J].财经科学, 2022(2):89-101.
[13]	胡振, 龚薛, 刘华.基于BP模型的西部城市家庭消费碳排放预测研究:以西安市为例[J].干旱区资源与环境, 2020, 34(7):82-89.
[14]	潘思羽, 张美玲.基于BP神经网络的甘肃省二氧化碳排放预测及影响因素研究[J].环境工程, 2023, 41(7):1-12.
[15]	朱盛恺.基于极限学习机模型的空气质量二次预报[J].软件工程, 2022, 25(8):39-42.
[16]	HUANG G B.An insight into extreme learning machines:random neurons, random features and kernels[J].Cognitive Computation, 2014, 6(3):376-390.
[17]	CHENG C, TAY W P, HUANG G B.Extreme learning machines for intrusion detection[C]//Proceedings of the 2012 International Joint Conference on Neural Networks (IJCNN).Brisbane, QLD:IEEE, 2012:1-8.
[18]	高金贺, 郑宝珠, 周伟昊, 等.基于GA-SVR的城市交通运输碳排放预测研究[J].东华理工大学学报(自然科学版), 2022, 45(3):269-274.
[19]	王珂珂, 牛东晓, 甄皓, 等.基于WOA-ELM模型的中国碳排放预测研究[J].生态经济, 2020, 36(8):20-27.
[20]	何洋, 李丽敏, 温宗周, 等.基于GWO-ELM算法模型的水体含沙量预测[J].科学技术与工程, 2022, 22(3):910-917.
[21]	PAUSTIAN K, RAVINDRANATH N H, AMSTEL A V.2006 IPCC guidelines for national greenhouse gas inventories[J].International Panel on Climate Change, 2006.DOI: http://dx.doi.org/10.1016/S1462-9011(99)00023-4.
[22]	国家发展和改革委员会应对气候变化司.中国2008年温室气体清单研究[M].北京:中国计划出版社, 2014.
[23]	张新生, 张玥.基于Lasso-PSO-BP神经网络的腐蚀管道失效压力的预测[J].材料保护, 2020, 53(4):46-52.
[24]	SUN W, HUANG C C.Predictions of carbon emission intensity based on factor analysis and an improved extreme learning machine from the perspective of carbon emission efficiency[J].Journal of Cleaner Production, 2022, 338:130414.
[25]	汪涛.2035年远景目标和"十四五"规划要点[J].企业观察家, 2021(2):24.
[26]	马高权, 周娜, 谢蒙飞, 等.基于GWO-KELM模型的变压器油纸套管典型绝缘故障辨识方法[J].电网与清洁能源, 2023, 39(5):38-48.
[27]	骆正山, 于瑶如, 骆济豪, 等.基于IAOA-KELM的储气库注采管柱内腐蚀速率预测[J].安全与环境学报, 2023:1-9.doi: 10.13637 /j.issn.1009-6094.2023.0267.