基于EMD-GWO-LSTM模型的新疆标准化降水蒸散指数预测方法研究

doi:10.13866/j.azr.2024.04.01

摘要/Abstract

摘要：

干旱预测一直是干旱研究领域的重大挑战，提高干旱预测的准确性是解决干旱问题的关键。基于1961—2019年新疆34个气象站点月降水和月平均气温数据，计算得到标准化降水蒸散指数（Standardized Precipitation Evapotranspiration Index，SPEI），对新疆气象干湿变化进行分析，提出一种经验模态分解方法（empirical mode decomposition, EMD）-灰狼优化算法（grey wolf optimizer，GWO）-长短期神经网络（long short-term memory network，LSTM）的数据分解型干旱组合预测模型进行预测，并进行模型性能评价。结果表明：（1）干旱周期性变化整体呈现平稳且周期长的特点；（2） EMD能够有效优化数据的平稳性，GWO优化预测模型参数，组合模型的预测精度相较于单一预测模型有明显提高；（3） 4个预测模型结果精度由高到低的排序为：EMD-GWO-LSTM、GWO-LSTM、GWO-支持向量回归（Support Vactor Regression，SVR）、LSTM，拟合优度分别为0.972、0.939、0.862、0.830，EMD-GWO-LSTM组合预测模型的预测精度优于其余3个预测模型。EMD-GWO-LSTM组合模型可有效提高气象干旱的预测精度，为新疆地区气象干旱预报及抗旱减灾工作提供了新的方法手段。

关键词: EMD-GWO-LSTM模型, 标准化降水蒸散指数, 干旱预测, 新疆

Abstract:

Drought prediction has always been a major challenge in the field of drought research. Improving the accuracy of drought prediction is the key to solving the drought problem. The standardized precipitation evapotranspiration index (SPEI) was calculated on the basis of the monthly precipitation and average temperature data from 34 meteorological stations in Xinjiang from 1961 to 2019. Dry and wet changes in the Xinjiang region were analyzed. An empirical mode decomposition (EMD)-Gray Wolf Optimizer (GWO)-long short-term memory network is proposed. A combination prediction model based on the data decomposition of LSTM was used to forecast the drought, and the performance of the model was evaluated. The results were as follows: (1) the drought periodicity was stable and the periodicity was long. (2) EMD can effectively optimize the stationarity of data, GWO can optimize the parameters of the prediction model, and the prediction accuracy of the combination model is significantly higher than that of the single prediction model. (3) The accuracy of the results of the four prediction models in descending order was as follows: EMD-GWO-LSTM, GWO-LSTM, GWO-support vector regression (SVR), and LSTM (goodness of fit: 0.972, 0.939, 0.862, 0.830, respectively). The prediction accuracy of the EMD-GWO-LSTM combination prediction model was higher than that of the other three prediction models. The EMD-GWO-LSTM combination prediction model can effectively improve the accuracy of meteorological drought prediction and provide a novel approach for meteorological drought forecasting and drought mitigation in Xinjiang.

Key words: EMD-GWO-LSTM model, standardized precipitation evapotranspiration index, drought prediction, Xinjiang

许超杰, 窦燕, 孟琪琳. 基于EMD-GWO-LSTM模型的新疆标准化降水蒸散指数预测方法研究[J]. 干旱区研究, 2024, 41(4): 527-539.

XU Chaojie, DOU Yan, MENG Qilin. Prediction of the standardized precipitation evapotranspiration index in the Xinjiang region using the EMD-GWO-LSTM model[J]. Arid Zone Research, 2024, 41(4): 527-539.

图/表 10

图1

表1

表2

图2

图3

表3

表4

表5

表6

图4

参考文献 55

[1]	吴燕锋, 巴特尔·巴克, 罗那那. 1961—2012年北疆干旱时空变化[J]. 中国沙漠, 2017, 37(1): 158-166.
	[Wu Yanfeng, Bake Batur, Luo Nana. Spatiotemporal pattern of drought in North Xinjiang, China in 1961-2012[J]. Journal of Desert Research, 2017, 37(1): 158-166.] doi: 10.7522/j.issn.1000-694X.2015.00212
[2]	邹旭恺, 张强. 近半个世纪我国干旱变化的初步研究[J]. 应用气象学报, 2008, 19(6): 679-687.
	[Zou Xukai, Zhang Qiang. Preliminary study on variation in drought over China during past 50 years[J]. Journal of Applied Meteorology, 2008, 19(6): 679-687.]
[3]	赵晓妮, 黄萌田, 庞博. IPCC最新发布《气候变化2021:公众摘要》[N]. 中国气象报, 2022-11-11(001).
	[Zhao Xiaoni, Huang Mengtian, Pang Bo. IPCC recently released Climate Change 2021: Public Summary[N]. China Meteorological Science, 2022-11-11(001). ]
[4]	《中国21世纪议程》编制组编. 中国21世纪议程[M]. 北京: 中国环境科学出版社, 1994.
	[Compilation Team of China’s Agenda 21. China’s Agenda 21[M]. Beijing: China Environmental Science Press, 1994.]
[5]	马荣, 张明军, 王圣杰. 近50 a西北干旱区冬季积雪日数变化特征[J]. 自然资源学报, 2018, 33(1): 129-140.
	[Ma Rong, Zhang Mingjun, Wang Shengjie. Variation characteristics of snow cover days in arid region of Northwest China in last 50 years[J]. Journal of Natural Resources, 2018, 33(1): 129-140.]
[6]	吴秀兰, 段春锋, 玛依拉·买买提艾力, 等. 基于MCI的新疆近60 a干旱时空特征分析[J]. 干旱区研究, 2022, 39(1): 75-83.
	[Wu Xiulan, Duan Chunfeng, Mayila Maimaitiaili, et al. Analysis of the temporal-spatial variation characteristics of drought in the Xinjiang based on the meteorological drought comprehensive index[J]. Arid Zone Research, 2022, 39(1): 75-83.]
[7]	Yu Meixiu, Li Qiongfang, Hayes Michael, et al. Are droughts becoming more frequent or severe in China based on the Standardized Precipitation Evapotranspiration Index: 1951-2010?[J]. International Journal of Climatology, 2014, 34(34): 545-558. doi: 10.1002/joc.2014.34.issue-3
[8]	Mishraa K, Singh V P. A review of drought concepts[J]. Journal of Hydrology, 2010, 391: 202-216. doi: 10.1016/j.jhydrol.2010.07.012
[9]	宋玉鑫. 新疆极端气候及干湿变化特征研究[D]. 郑州: 郑州大学, 2021.
	[Song Yuxin. Study on Variation Characteristics of Extreme Climate and Drought-Humid in Xinjiang[D]. Zhengzhou: Zhengzhou University, 2021.]
[10]	米前川, 高西宁, 李玥, 等. 深度学习方法在干旱预测中的应用[J]. 应用气象学报, 2022, 33(1): 104-114.
	[Mi Qianchuan, Gao Xining, Li Yue, et al. Application of deep learning method to drought prediction[J]. Journal of Applied Meteorological Science, 2022, 33(1): 104-114.]
[11]	孙荣强. 干旱定义及其指标评述[J]. 灾害学, 1994, 9(1): 17-21.
	[Sun Rongqiang. Definition of drought and review of its indicators[J]. Journal of Catastrophology, 1994, 9(1): 17-21.]
[12]	王舒, 肖高翔. 4种气象干旱指数在新疆的适用性分析[J]. 人民长江, 2021, 52(9): 86-100.
	[Wang Shu, Xiao Gaoxiang. Applicability analysis of four meteorological drought indices in Xinjiang[J]. Yangtze River, 2021, 52(9): 86-100.]
[13]	程军, 李云祯, 邹渝. 新疆干旱时空动态及其对气候变化的响应[J]. 自然资源遥感, 2022, 34(4): 216-224.
	[Cheng Jun, Li Yunzhen, Zou Yu. Spatial and temporal dynamics of drought in Xinjiang and its response to climate change[J]. Remote Sensing for Natural Resources, 2022, 34(4): 216-224.]
[14]	粟晓玲, 张更喜, 冯凯. 干旱指数研究进展与展望[J]. 水利与建筑工程学报, 2019, 17(5): 9-18.
	[Su Xiaoling, Zhang Gengxi, Feng Kai. Progress and perspective of drought index[J]. Journal of Water Resources and Architectural Engineering, 2019, 17(5): 9-18.]
[15]	王素萍, 王劲松, 张强, 等. 几种干旱指标对西南和华南区域月尺度干旱监测的适用性评价[J]. 高原气象, 2015, 34(6): 1616-1624. doi: 10.7522/j.issn.1000-0534.2014.00089
	[Wang Suping, Wang Jinsong, Zhang Qiang, et al. Applicability evaluation of drought indices in monthly scale drought monitoring in southwestern and southern China[J]. Plateau Meteorology, 2015, 34(6): 1616-1624.] doi: 10.7522/j.issn.1000-0534.2014.00089
[16]	Palmer W C. Meteorological Drought[M]. Washington, DC: US Department of Commerce, Weather Bureau, 1965.
[17]	McKee T B, Doesken N J, Kleist J. The relationship of drought frequency and duration to time scales[C]// Proceedings of the 8^th Conference on Applied Climatology, Boston, MA: American Meteorological Society, 1993: 179-183.
[18]	Gocic M, Trajkovic S. Drought characterisation based on water surplus variability index[J]. Water Resour Manage, 2014, 28: 3179-3191. doi: 10.1007/s11269-014-0665-4
[19]	任建成, 巩在武, 郑宝枝, 等. 基于SPI的近30年黄河三角洲地区旱涝时空特征[J]. 水土保持研究, 2016, 23(1): 46-51.
	[Ren Jiancheng, Gong Zaiwu, Zheng Baozhi, et al. Spatiotemporal characteristics of drought/flood in the Yellow River Delta based on SPI in recent 30 years[J]. Research of Soil and Water Conservation, 2016, 23(1): 46-51.]
[20]	Vicente-Serrano S M, Beguería S, López-Moreno J I. A multiscalar drought index sensitive to global warming: The Standardized Precipitation Evapotranspiration Index[J]. Journal of Climate, 2010, 23(7): 1696-1718. doi: 10.1175/2009JCLI2909.1
[21]	轩俊伟, 郑江华, 刘志辉. 基于SPEI的新疆干旱时空变化特征[J]. 干旱区研究, 2016, 33(2): 338-344.
	[Xuan Junwei, Zheng Jianghua, Liu Zhihui. SPEI-based spatiotemporal variation of drought in Xinjiang[J]. Arid Zone Research, 2016, 33(2): 338-344.]
[22]	杨星星, 杨云川, 邓思敏, 等. 基于SPEI的广西干旱综合特征及农业旱灾风险研究[J]. 水土保持研究, 2020, 27(4): 113-121.
	[Yang Xingxing, Yang Yunchuan, Deng Simin, et al. Research on drought characteristics and agricultural drought risk assessment in Guangxi based on SPEI[J]. Research of Soil and Water Conservation, 2020, 27(4): 113-121.]
[23]	马尚谦, 张勃, 张佳琦, 等. 标准化降水蒸散指数在华北平原的适用性分析[J]. 中国农业资源与区划, 2020, 41(2): 235-245.
	[Ma Shangqian, Zhang Bo, Zhang Jiaqi, et al. Applicability analysis of Standardized Precipitation Evapotranspiration Index in North China Plain[J]. China Agricultural Resources and Regional Planning, 2020, 41(2): 235-245.]
[24]	张露, 朱仲元, 席小康, 等. 基于SPEI的锡林河流域干旱演化特征分析[J]. 干旱区研究, 2020, 37(4): 819-829.
	[Zhang Lu, Zhu Zhongyuan, Xi Xiaokang, et al. Analysis of drought evolution in the Xilin River Basin based on Standardized Precipitation Evapotranspiration Index[J]. Arid Zone Research, 2020, 37(4): 819-829.]
[25]	任贤月, 穆振侠, 周育琳. 基于不同蒸散方法的SPEI在天山南北坡气象干旱的差异性分析[J]. 南水北调与水利科技, 2019, 17(3): 48-55.
	[Ren Xianyue, Mu Zhenxia, Zhou Yulin. The difference meteorological drought of SPEI on the north and south slopes of Tianshan Mountains with different evapotranspiration methods[J]. South-to-North Water Transfers and Water Science & Technology, 2019, 17(3): 48-55.]
[26]	赵慧, 姚俊强, 李新国, 等. 新疆气候干湿变化特征分析[J]. 中山大学学报(自然科学版), 2020, 59(4): 126-133.
	[Zhao Hui, Yao Junqiang, Li Xinguo, et al. The characteristics of climate change in Xinjiang during 1961-2015[J]. Acta Scientiarum Naturalium Universitatis Sunyatseni, 2020, 59(4): 126-133.]
[27]	Hao Z, Singh V P. Drought characterization from a multivariate perspective: A review[J]. Journal of Hydrology, 2015, 527: 668-678. doi: 10.1016/j.jhydrol.2015.05.031
[28]	Abbe A B, Inoubli R, Rhif M, et al. Combining deep learning methods and multi-resolution analysis for drought forecasting modeling[J]. Earth Science Informatics, 2023, 16: 1811-1820. doi: 10.1007/s12145-023-01009-4
[29]	张建海, 张棋, 许德合, 等. ARIMA-LSTM组合模型在基于SPI干旱预测中的应用——以青海省为例[J]. 干旱区地理, 2020, 43(4): 1004-1013.
	[Zhang Jianhai, Zhang Qi, Xu Dehe, et al. Application of a combined ARIMA-LSTM model based on SPI for the forecast of drought: A case study in Qinghai Province[J]. Arid Land Geography, 2020, 43(4): 1004-1013.]
[30]	牛芳鹏, 白云岗, 王新涛, 等. 新疆干旱灾害风险区划特征研究[J]. 水利规划与设计, 2023(9): 9-13.
	[Niu Fangpeng, Bai Yungang, Wang Xintao, et al. Research on characteristics of drought disaster risk regionalization in Xinjiang[J]. Water Conservancy Planning and Design, 2023(9): 9-13.]
[31]	Poornima S, Pushpalatha M. Drought prediction based on SPI and SPEI with varying timescales using LSTM recurrent neural network[J]. Soft Comput, 2019, 23: 8399-8412. doi: 10.1007/s00500-019-04120-1
[32]	许德合, 张棋, 黄会平. ARIMA-SVR组合模型在基于标准化降水指数干旱预测中的应用[J]. 干旱地区农业研究, 2020, 38(2): 276-282.
	[Xu Dehe, Zhang Qi, Huang Huiping. Application of the combined ARIMA-SVR model in drought prediction based on the Standardized Precipitation Index[J]. Agricultural Research in the Arid Areas, 2020, 38(2): 276-282.]
[33]	李子阳, 王肖鑫, 张恩典, 等. 基于VMD-GRU的大型灌区干旱预测模型研究[J]. 中国农村水利水电, 2023(3): 130-137.
	[Li Ziyang, Wang Xiaoxin, Zhang Endian, et al. Research on the drought prediction model of large irrigation areas based on VMD-GRU[J]. China Rural Water and Hydropower, 2023(3): 130-137.]
[34]	刘雪梅, 宋文辉, 钱峰, 等. 基于VMD-CQPSO-GRU模型的气象干旱预测方法[J]. 华北水利水电大学学报(自然科学版), 2021, 42(4): 31-40.
	[Liu Xuemei, Song Wenhui, Qian Feng, et al. Prediction method of meteorological drought based on VMD-CQPSO-GRU method[J]. Journal of North China University of Water Resources and Electric Power, 2021, 42(4): 31-40.]
[35]	Malik A, Tikhamarine Y, Souag-Gamane D, et al. Support vector regression integrated with novel meta-heuristic algorithms for meteorological drought prediction[J]. Meteorology and Atmospheric Physics, 2021, 133: 891-909. doi: 10.1007/s00703-021-00787-0
[36]	董翰林, 王文婷, 谢云, 等. 新疆气候干湿变化特征及其影响因素[J]. 干旱区研究, 2023, 40(12): 1875-1884.
	[Dong Hanlin, Wang Wenting, Xie Yun, et al. The characteristics and influencing factors of dry and wet climate changes in Xinjiang[J]. Arid Zone Research, 2023, 40(12): 1875-1884.]
[37]	阿迪拉·阿布都热合曼, 昝梅. 基于MCI的新疆干旱变化特征分析[J]. 水文, 2023, 43(1): 96-101.
	[Adila Abudureheman, Zan Mei. Change characteristics of drought in Xinjiang Based on MCI[J]. Journal of China Hydrology, 2023, 43(1): 96-101.]
[38]	秦大辉, 杨灵, 谌伦超, 等. 基于多源数据的新疆干旱特征及干旱模型研究[J]. 自然资源遥感, 2022, 34(1): 151-157.
	[Qin Dahui, Yang Ling, Chen Lunchao, et al. A study on the characteristics and model of drought in Xinjiang based on multi-source data[J]. Remote Sensing of Natural Resources, 2022, 34(1): 151-157.]
[39]	刘武. 借助ArcGIS软件采用泰森多边形法计算2020年黑龙江省降雨量[J]. 吉林水利, 2021(5): 8-11, 20.
	[Liu Wu. Calculating the rainfall of Heilongjiang Province in 2020 by Tyson polygon method with the help of ArcGIS software[J]. Jilin Water Resources, 2021(5):8-11, 20. ]
[40]	Mirjalili S, Mirjalili S M, Lewis A. Grey wolf optimizer[J]. Advances in Engineering Software, 2014, 69: 46-61. doi: 10.1016/j.advengsoft.2013.12.007
[41]	丁严, 许德合, 曹连海, 等. 基于CEEMD的LSTM和ARIMA模型干旱预测适用性研究——以新疆为例[J]. 干旱区研究, 2022, 39(3): 734-744.
	[Ding Yan, Xu Dehe, Cao Lianhai, et al. Applicability of the LSTM and ARIMA model in drought prediction based on CEEMD: A case study in Xinjiang[J]. Arid Zone Research, 2022, 39(3): 734-744.]
[42]	王林, 陈文. 标准化降水蒸散指数在中国干旱监测的适用性分析[J]. 高原气象, 2014, 33(2): 423-431. doi: 10.7522/j.issn.1000-0534.2013.00048
	[Wang Lin, Chen Wen. Applicability analysis of Standardized Precipitation Evapotranspiration Index in drought monitoring in China[J]. Plateau Meteorology, 2014, 33(2): 423-431.] doi: 10.7522/j.issn.1000-0534.2013.00048
[43]	郭燕云, 胡琦, 傅玮东, 等. 基于SPEI指数的新疆天山草地近55 a干旱特征[J]. 干旱区研究, 2019, 36(3): 670-676.
	[Guo Yanyun, Hu Qi, Fu Weidong, et al. Drought trend over the grasslands in the Tianshan Mountains, Xinjiang in recent 55 years based on SPEI[J]. Arid Zone Research, 2019, 36(3): 670-676.]
[44]	郭冬, 吐尔逊·哈斯木, 吴秀兰, 等. 四种气象干旱指数在新疆区域适用性研究[J]. 沙漠与绿洲气象, 2022, 16(3): 90-101.
	[Guo Dong, Turson Hasmu, Wu Xiulan, et al. Applicability of four meteorological drought indices in Xinjiang[J]. Desert and Oasis Meteorology, 2022, 16(3): 90-101.]
[45]	吴志勇, 程丹丹, 何海, 等. 综合干旱指数研究进展[J]. 水资源保护, 2021, 37(1): 36-45.
	[Wu Zhiyong, Cheng Dandan, He Hai, et al. Research progress of composite drought index[J]. Water Resources Protection, 2021, 37(1): 36-45.]
[46]	黄静, 张运, 汪明秀, 等. 近17年新疆干旱时空分布特征及影响因素[J]. 生态学报, 2020, 40(3): 1077-1088.
	[Huang Jing, Zhang Yun, Wang Mingxiu, et al. Spatial and temporal distribution characteristics of drought and its relationship with meteorological factors in Xinjiang in last 17 years[J]. Acta Ecologica Sinica, 2020, 40(3): 1077-1088.]
[47]	赵紫竹, 张宝林, 潘丽杰, 等. 基于SPEI的内蒙古东部干旱诊断与预测[J]. 环境生态学, 2023, 5(7): 39-48.
	[Zhao Zizhu, Zhang Baolin, Pan Lijie, et al. Drought diagnosis and prediction using SPEI in eastern Inner Mongolia[J]. Environmental Ecology, 2023, 5(7): 39-48.]
[48]	王冉, 后麒麟, 石如玉, 等. 基于变分模态分解与集成深度模型的锂电池剩余寿命预测方法[J]. 仪器仪表学报, 2021, 42(4): 111-120.
	[Wang Ran, Hou Qilin, Shi Ruyu, et al. Remaining useful life prediction method of lithium battery based on variational mode decomposition and integrated deep model[J]. Chinese Journal of Scientific Instrument, 2021, 42(4): 111-120.]
[49]	刘振男, 杜尧, 韩幸烨, 等. 基于遗传算法优化极限学习机模型的干旱预测: 以云贵高原为例[J]. 人民长江, 2020, 51(8): 13-18.
	[Liu Zhennan, Du Yao, Han Xingye, et al. Drought prediction based on genetic algorithm-optimized extreme learning machine model: Case of Yunnan: Guizhou Plateau[J]. Yangtze River, 2019, 51(8): 13-18.]
[50]	魏腾飞, 潘庭龙. 基于改进PSO优化LSTM网络的短期电力负荷预测[J]. 系统仿真学报, 2021, 33(8): 1866-1874. doi: 10.16182/j.issn1004731x.joss.20-0297
	[Wei Tengfei, Pan Tinglong. Short-term power load forecasting based on LSTM neural network Optimized by Improved PSO[J]. Journal of System Simulation, 2021, 33(8): 1866-1874.] doi: 10.16182/j.issn1004731x.joss.20-0297
[51]	刘扬, 王立虎, 杜帅兵, 等. 改进VMD-QR-ELM混合模型在径流预报中的应用[J]. 水文, 2022, 42(6): 31-34, 51.
	[Liu Yang, Wang Lihu, Du Shuaibing, et al. Application of improved VMDQR-ELM hybrid model in runoff forecast[J]. Journal of China Hydrology, 2022, 42(6): 31-34, 51.]
[52]	Taesam Lee. EMD and LSTM hybrid deep learning model for predicting sunspot number time series with a cyclic pattern[J]. Solar physics, 2020, 295(6): 1-23. doi: 10.1007/s11207-019-1566-8
[53]	Tan Qiaofeng, Lei Xiaohui, Wang Xu, et al. An adaptive middle and long-term runoff forecast model using EE-MD-ANN hybrid approach[J]. Journal of Hydrology, 2018, 567: 767-780. doi: 10.1016/j.jhydrol.2018.01.015
[54]	黄睿茜, 赵俊芳, 霍治国, 等. 深度学习技术在农业干旱监测预测及风险评估中的应用[J]. 中国农业气象, 2023, 44(10): 943-952.
	[Huang Ruixi, Zhao Junfang, Huo Zhiguo, et al. Application of deep learning technology in monitoring, forecasting and risk assessment of agricultural drought[J]. Chinese Journal of Agrometeorology, 2023, 44(10): 943-952.] doi: 10.3969/j.issn.1000-6362.2023.10.007
[55]	高秉丽, 巩杰, 李焱, 等. 基于SPEI的黄河流域多尺度干湿特征分析[J]. 干旱区研究, 2022, 39(3): 723-733.
	[Gao Bingli, Gong Jie, Li Yan, et al. Analysis of multi-scalar characteristics of dry and wet conditions in the Yellow River Basin based on SPEI[J]. Arid Zone Research, 2022, 39(3): 723-733.]

时间尺度	Sen’s斜率
月	-4.4752×10^-4
春季	-0.0139
夏季	-0.0103
秋季	-0.0120
冬季	0.0010
年	-0.0169

时间尺度	主周期	旱涝交替尺度/a	周期个数/个	时间尺度	主周期	旱涝交替尺度/a	周期个数/个
年	第一主周期	16	2.5	秋	第一主周期	16	2.5
	第二主周期	3	13		第二主周期	9	4.5
	第三主周期	5	7.5		第三主周期	5	8.5
	第四主周期	7	5.5		第四主周期	2	22
	第五主周期	4	10		第五主周期	3	13
春	第一主周期	6.5	6	冬	第一主周期	16	2.5
	第二主周期	13	3		第二主周期	6	6.5
	第三主周期	3	13		第三主周期	4	10.5
	第四主周期	8	5		第四主周期	3	15
夏	第一主周期	20	2		第五主周期	9	4.5
	第二主周期	3	13
	第三主周期	8	5

名称	IMF1	IMF2	IMF3	IMF4	IMF5	IMF6	Res
学习率	0.0039	0.0148	0.0479	0.0816	0.0592	0.0454	0.0607
隐藏层节点	53	31	55	10	12	28	18
L2正则化系数	0.0815	0.0311	0.0045	0.0830	0.0012	0.0142	0.0312
网络结构	12-53-1	12-31-1	12-55-1	12-10-1	12-12-1	12-28-1	12-18-1

	RMSE	MAE	SMAPE
IMF1	0.089	0.069	1.310
IMF2	0.028	0.022	0.450
IMF3	0.032	0.023	0.258
IMF4	0.038	0.026	0.267
IMF5	0.006	0.005	0.075
IMF6	0.004	0.002	0.033
Res	0.040	0.017	0.036

模型	RMSE	MAE	SMAPE	R²
LSTM	0.287	0.201	0.398	0.830
GWO-SVR	0.247	0.171	0.505	0.862
GWO-LSTM	0.165	0.114	0.384	0.939
EMD-GWO-LSTM	0.112	0.087	0.315	0.972