基于外界影响及时序因素的PM2.5预测研究

doi:10.16180/j.cnki.issn1007-7820.2022.03.008

摘要/Abstract

摘要：

随着雾霾问题逐渐加重,对其主成分之一PM_2.5的预测已成为广泛关注的问题。PM_2.5日浓度变化受多种因素影响,且具有非线性、时变性的特征,难以被准确预测。针对此问题,文中提出一种基于外界影响及时序因素的PM_2.5日浓度预测方法。该方法分离出PM_2.5日浓度的外界主要影响因素与时间因素,建立了基于外界主要影响因素的BP神经网络初步预测模型以及基于时间因素的EEMD-LSTM组合残差修正模型。使用杭州市2014年~2019年间的PM_2.5日浓度和其他相关因素数据进行仿真实验。结果表明,相比其他模型,文中所提出预测模型的均方根误差为2.74,预测准确率更高。

关键词: 雾霾, PM_2.5, BP, EEMD, LSTM, 时间序列预测, 神经网络, 时间序列分解, 组合预测

Abstract:

As the haze problem gradually worsens, the prediction of one of its main component PM_2.5 has become a widespread concern. The daily concentration of PM_2.5 is affected by many factors, and it has the characteristics of non-linear and time-varying, which is difficult to accurately predict.To solve this problem, a prediction method of PM_2.5 daily concentration based on external influences and time-series factors is proposed. With this method, the main external factors and time factors of PM_2.5 daily concentration are separated, and the BP neural network preliminary prediction model based on the main external factors and the combined residual correction model of EEMD-LSTM neural network based on time factor are established. The daily PM_2.5 concentration and other related factors data of Hangzhou from 2014 to 2019 are used for simulation experiments. The results show that compared with other models, the root mean square error of the prediction model proposed in the study is 2.74, and the prediction accuracy is higher.

Key words: haze, PM_2.5, BP, EEMD, LSTM, time series prediction, neural network, time series decompose, combination prediction

中图分类号:

TP39

杨艳梅,程宗毛. 基于外界影响及时序因素的PM_2.5预测研究[J]. 电子科技, 2022, 35(3): 51-57.

Yanmei YANG,Zongmao CHENG. Prediction of PM_2.5 Based on External Influences and Time-Series Factors[J]. Electronic Science and Technology, 2022, 35(3): 51-57.

图/表 10

图1

图2

表1

图3

图4

图5

表2

图6

图7

表3

参考文献 19

[1]	胡彬, 陈瑞, 徐建勋, 等. 雾霾超细颗粒物的健康效应[J]. 科学通报, 2015,60(30):2808-2823.
	Hu Bin, Chen Rui, Xu Jianxun, et al. Health effects of ambient ultrafine(nano) particles in haze[J]. Chinese Science Bulletin, 2015,60(30):2808-2823.
[2]	权建农, 徐祥德, 贾星灿, 等. 影响我国霾天气的多尺度过程[J]. 科学通报, 2020,65(9):810-824.
	Quan Jiannong, Xu Xiangde, Jia Xingcan, et al. Multi-scale processes in severe haze events in China and their interactions with aerosols: mechanisms and progresses[J]. Chinese Science Bulletin, 2020,65(9):810-824.
[3]	李波, 朱恩文, 冯倩. 基于时间序列的长沙市PM_2.5的统计分析[J]. 经济数学, 2017,34(1):105-110.
	Li Bo, Zhu Enwen, Feng Qian. Statistical analysis of changsha PM2.5 based on time series[J]. Journal of Quantitative Economics, 2017,34(1):105-110.
[4]	戴邵武, 陈强强, 刘志豪, 等. 基于EMD-LSTM 的时间序列预测方法[J]. 深圳大学学报(理工版), 2020,37(3):265-270.
	Dai Shaowu, Chen Qiangqiang, Liu Zhihao, et al. Time series prediction based on EMD-LSTM model[J]. Journal of Shenzhen University(Science and Engineering), 2020,37(3):265-270.
[5]	康俊峰, 黄烈星, 张春艳, 等. 多机器学习模型下逐小时PM_2.5预测及对比分析[J]. 中国环境科学, 2020,40(5):1895-1905.
	Kang Junfeng, Huang Liexing, ZhangChunyan,et al.Hourly PM_2.5 prediction and its comparative analysis under multimachine learning model[J]. China Environmental Science, 2020,40(5):1895-1905.
[6]	蒋奇峰, 杜景林, 周芸. 一种基于QPSO-RBF模型预测PM_2.5浓度值的方法研究及应用[J]. 计算机应用与软件, 2020,37(9):202-208.
	Jiang Qifeng, Du Jinglin, Zhou Yun. A method for predicting PM_2.5 concentration based on QPSO-RBF model and its application[J]. Computer Applications and Software, 2020,37(9):202-208.
[7]	管胜, 阮方鸣, 周奎, 等. BP神经网络预测电极速度影响放电参数分析[J]. 电子科技, 2019,32(6):43-48.
	Guan Sheng, Ruan Fangming, Zhou Kui, et al. Analysis of discharge parameters affected by BP neural network predicting electrode velocity[J]. Electronic Science and Technology, 2019,32(6):43-48.
[8]	Wu Z H, Huang N E. Ensemble empirical mode decomposition: a noise assisted data analysis method[J]. Advances in Adaptive Data Analysis, 2009,1(1):1-41.
[9]	Liu Y M, Zhao C C, Xiong M Y, et al. Assessment of bearing performance degradation via extension and EEMD combined approach[J]. Journal of Central South University, 2017,24(5):1155-1163.
[10]	Xu P P, Chen C J, Lou H F. Research on the method of EEMD in pulse wave signals prcessing[J]. Chinese Journal of Biomedical Engineering, 2018,27(3):100-108.
[11]	Xiao Y, Tian X T. Tourism traffic demand prediction using Google trends based on EEMD-DBN[J]. Engineering, 2020,12(3):194-215.
[12]	曲悦, 钱旭, 宋洪庆, 等. 基于机器学习的北京市PM_2.5浓度预测模型及模拟分析[J]. 工程科学学报, 2019,41(3):401-407.
	Qu Yue, Qian Xu, Song Hongqing, et al. Machine-learning- based model and simulation analysis of PM_2.5concentration prediction in Beijing[J]. Chinese Journal of Engineering, 2019,41(3):401-407.
[13]	尹光花, 刘小明, 张露, 等. 基于LSTM特征模板的短文本情感要素分析与研究[J]. 电子科技, 2018,31(11):38-41.
	Yin Guanghua, Liu Xiaoming, Zhang Lu, et al. Sentiment elements of internet short texts for analysis and research based on LSTM network mode[J]. Electronic Science and Technology, 2018,31(11):38-41.
[14]	Zhang B N. Foreign exchange rates forecasting with an EMD-LSTM neural networks model[J]. Journal of Physics Conference, 2018,1053(1):5-12.
[15]	陈振宇, 刘金波, 李晨, 等. 基于LSTM与XGBoost组合模型的超短期电力负荷预测[J]. 电网技术, 2020,44(2):614-620.
	Chen Zhenyu, Liu Jinbo, Li Chen, et al. Ultra short-term power load forecasting based on combined LSTM-XGBoost model[J]. Power System Technology, 2020,44(2):614-620.
[16]	董健卫, 陈艳美, 孟盼, 等. 回归分析与基于MIV的RBF神经网络在PM_2.5的相关因素分析中的应用[J]. 数学的实践与认识, 2017,47(10):127-136.
	Dong Jianwei, Chen Yanmei, Meng Pan, et al. The application of regression analysis and RBF neural network based on MIV in the related factors analysis of PM_2.5[J]. Mathematics in Practice and Theory, 2017,47(10):127-136.
[17]	陈川, 陈冬林, 何李凯, 等. 基于BPNN-EMD-LSTM 组合模型的城市短期燃气负荷预测[J]. 安全与环境工程, 2019,26(1):149-154.
	Chen Chuan, Chen Donglin, He Likai, et al. Short-term forecast of urban natural gas load based on BPNN-EMD-LSTM combined model[J]. Safety and Environmental Engineering, 2019,26(1):149-154.
[18]	曾浩, 丁镭. 长江经济带城市雾霾污染PM_2.5时空格局演变及影响因素研究[J]. 华中师范大学学报(自然科学版), 2019,53(5):724-734.
	Zeng Hao, Ding Lei. Study on spatial-temporal pattern evolution and influencing factors of urban haze pollution PM_2.5 in the Yangtze river economic belt[J]. Journal of Central China Normal University(Natural Sciences), 2019,53(5):724-734.
[19]	石昊苏, 方丽娟, 台志强, 等. 一种基于Retinex的雾霾图像增强改进算法[J]. 电子设计工程, 2018,26(4):180-183.
	Shi Haosu, Fang Lijuan, Tai Zhiqiang, et al. An improved haze image enhancement algorithm based on Retinex[J]. Electronic Design Engineering, 2018,26(4):180-183.

参考指标	比较指标	相关性系数
PM₂_.₅	PM₁₀	0.946 0
	SO₂	0.638 8
	CO	0.758 9
	NO₂	0.695 5
	O₃	-0.106 0
	平均温度	-0.406 9
	最高温度	-0.347 3
	最低温度	-0.444 4
	平均相对湿度	-0.155 9
	总降雨量/总融雪量	-0.199 7
	平均风速	-0.297 0
	最大持续风速	-0.274 6

输入数据	输出数据
x₁,x₂,…,x_n	x_n+1
x₂,x₃,…,x_n+1	x_n+2
?	?
x_N-n,x_N-n+1,…,x_N-1	x_N

模型	RMSE	R	MAPE/%
BP	10.30	0.86	30.23
BP-LSTM	5.99	0.91	17.43
BP-EMD-LSTM	3.73	0.96	10.58
BP-EEMD-LSTM	2.74	0.99	8.01