基于深度学习模型的广州市大气PM<sub>2.5</sub>和PM<sub>10</sub>浓度预测

黄春桃; 范东平; 卢集富; 廖启丰

doi:10.13205/j.hjgc.202112020

基于深度学习模型的广州市大气PM_2.5和PM₁₀浓度预测

doi: 10.13205/j.hjgc.202112020

1. 广州华立学院, 广州 511300;
2. 生态环境部华南环境科学研究所, 广州 510530;
3. 广东科迪隆科技有限公司, 广州 510000

基金项目:

广东省重点学科培育项目(粤教研函[2017]1号)

广东省教育厅重点科研平台青年创新人才项目(自然科学)(2018KQNCX35)。

详细信息

通讯作者:
黄春桃(1985-),女,硕士,主要研究方向为大气污染治理、水处理技术。610140538@qq.com

计量
- 文章访问数: 543
- HTML全文浏览量: 115
- PDF下载量: 25
- 被引次数: 0
出版历程
- 收稿日期: 2021-05-12
- 网络出版日期: 2022-03-30
- 刊出日期: 2022-03-30

PREDICTION OF PM_2.5 AND PM₁₀ CONCENTRATION IN GUANGZHOU BASED ON DEEP LEARNING MODEL

1. Guangzhou Huali College, Guangzhou 511300, China;
2. South China Institute of Environmental Sciences, Ministry of Ecology and Environment, Guangzhou 510530, China;
3. Guangdong Kedilong Technology Co., Ltd, Guangzhou 510000, China

摘要

摘要: 精准预测大气污染颗粒物PM_2.5、PM₁₀浓度能为大气污染防治提供科学依据,但目前较多PM_2.5和PM₁₀浓度预测在缺少污染源排放清单和能见度数据时,预测精度不高。而目前深度学习模型应用于PM_2.5和PM₁₀浓度预测的研究还鲜见报道。基于广州市2015年6月1日—2018年1月10日的空气质量和气象监测历史数据,分别构建了随机森林模型(RF)、XGBoost模型2种传统的机器学习模型和长短时记忆网络(LSTM)、门控循环单元网络(GRU)2种深度学习模型,并对广州市的PM_2.5、PM₁₀日均浓度值进行预测。结果表明:在缺少污染源排放清单和能见度数据时,4种模型也能较好地预测PM_2.5、PM₁₀日均浓度。根据MSE、RMSE、MAPE、MAE和R²等评价指标,对4个模型的PM_2.5、PM₁₀预测效果进行测评,得出深度学习GRU模型预测效果均为最佳,RF模型的预测结果均为最差。相比目前研究及应用较多的RF模型、XGBoost模型、LSTM模型,基于深度学习的GRU模型能更好地预测PM_2.5、PM₁₀浓度。
- PM_2.5 /
- PM₁₀ /
- 深度学习模型 /
- 浓度预测 /
- 影响因素
Abstract: Precisely predicting the concentration of PM_2.5 and PM₁₀ in air pollution can provide a scientific basis for the prevention and control of air pollution. However, in the absence of pollution source emission inventory and visibility data, the prediction accuracy of the existing PM_2.5 and PM₁₀ concentration prediction methods are not high. In addition, it is rarely reported that the current deep learning models have been applied successively to PM_2.5 and PM₁₀ concentration prediction research. Based on the historical air quality monitoring data and weather monitoring historical data in Guangzhou from June 1, 2015 to January 10, 2018, two traditional machine learning models(random forest model(RF) and XGBoost model) and two deep learning models(short-long-term memory network(LSTM) and gated recurrent unit network(GRU) were constructed respectively, to predict the daily average concentration of PM_2.5 and PM₁₀ in Guangzhou. The results showed that the four models could also well predict the daily average concentration of PM_2.5 and PM₁₀ in the absence of pollution source emission inventory and visibility data. According to the evaluation metrics, i.e., MSE, RMSE, MAPE, MAE, and R², the PM_2.5 and PM₁₀ prediction effects of the four models were evaluated. The results indicated that the prediction effect of the deep GRU model was the best and the prediction results of the RF model were the worst. Compared with the commonly used RF model, XGBoost model, and LSTM model, the GRU model based on deep learning could better predict PM_2.5 and PM₁₀ concentration.
- PM_2.5 /
- PM₁₀ /
- deep learning /
- concentration prediction /
- influencing factors

HTML全文

参考文献(26)

[1]	刘佳澍,顾远,马帅帅,等.常州夏冬季PM_2.5中无机组分昼夜变化特征与来源解析[J].环境科学,2018,39(3):980-989.
[2]	段文娇,郎建垒,程水源,等.京津冀地区钢铁行业污染物排放清单及对PM_2.5影响[J].环境科学,2018,39 (4):1445-1454.
[3]	余钟奇,瞿元昊,周广强,等.2018年秋冬季长江三角洲区域PM_2.5污染来源数值研究[J].中国环境科学,2020,40(10):4237-4246.
[4]	黄小刚,邵天杰,赵景波,等.汾渭平原PM_2.5浓度的影响因素及空间溢出效应[J].中国环境科学,2019,39(8):3539-3548.
[5]	黄小刚,邵天杰,赵景波,等.长江经济带空气质量的时空分布特征及影响因素[J].中国环境科学,2020,40(2):874-884.
[6]	陈波,李少宁,杨新兵,等.北京春季PM_2.5和PM₁₀污染水平及影响因素研究[J].内蒙古农业大学学报(自然科学版),2018,39(5):36-44.
[7]	肖悦,田永中,许文轩,等.近10年中国空气质量时空分布特征[J].生态环境学报,2017,26(2):243-252.
[8]	黄含含,王羽琴,李升苹,等.西安市PM_2.5中水溶性离子的季节变化特征[J].环境科学,2020,41(6):2528-2535.
[9]	杨文涛,姚诗琪,邓敏,等.北京市PM_2.5时空分布特征及其与PM₁₀关系的时空变异特征[J].环境科学,2018,39(2):684-690.
[10]	赵雪,侯丽丽,王鑫龙,等.基于LUR模型的2019年北京地区PM_2.5与PM₁₀浓度空间分异模拟[J].环境科学学报,2020,40(11):4060-4069.
[11]	郭庆元,杨晓春,吴其重,等.基于数值模拟对西安地区冬季一次重污染过程PM_2.5区域来源解析[J].环境科学学报,2020,40(9):3103-3111.
[12]	YU M F,ZHU Y,LIN C J,et al.Effects of air pollution control measures on air quality improvement in Guangzhou,China[J].Journal of Environmental Management,2019,244:127-137.
[13]	SHEN L,JACOB D J,MICKLEY L J,et al.Insignificant effect of climate change on winter haze pollution in Beijing [J].Atmospheric Chemistry and Physics,2018,18 (23):17489-17496.
[14]	李岚淼,李龙国,李乃稳.城市雾霾成因及危害研究进展[J].环境工程,2017,35(12):92-97 ,104.
[15]	雷钦.大气细颗粒物PM_2.5的危害及其治理策略[J].低碳世界,2020,10(11):23-24.
[16]	ERIC L,ISAC L,MARIANNE H,et al.Ambient ultrafine particle concentration and incidence of childhood cancers[J].Environment International,2020,145:106135.
[17]	SOURANGSU C,SAGNIK D,SMITH K R.Ambient PM_2.5 exposure and expected premature mortality to 2100 in India under climate change scenarios[J].Nature Communications,2018,9(1).
[18]	周广强,谢英,吴剑斌,等.基于WRF-Chem模式的华东区域PM_2.5预报及偏差原因[J].中国环境科学,2016,36(8):2251-2259.
[19]	CHEN J J,LU J,AVISE J C,et al.Seasonal modeling of PM_2.5 in California’s San Joaquin Valley[J].Atmospheric Environment,2014,92:182-190.
[20]	康俊锋,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.
[21]	DEUKWOO L,SOOWON L.Hourly prediction of particulate matter (PM_2.5) concentration using time series data and random forest[J].KIPS Transactions on Software and Data Engineering,2020,9(4).
[22]	黄赫,周勇,刘宇杰,等.基于多源环境变量和随机森林的农用地土壤重金属源解析:以襄阳市襄州区为例[J].环境科学学报,2020,40(12):4548-4558.
[23]	史飞飞,周秉荣,颜亮东,等.近32年隆宝高寒湿地时空变化特征及其气候驱动力分析[J].高原气象,2020,39(6):1282-1294.
[24]	刘志壮,吕谋,周国升.基于小波组合模型的短期城市用水量预测[J].给水排水,2020,56(10):110-114 ,131.
[25]	郭宇龙,李岚涛,陈伟强,等.基于红边光谱特征和XGBoost算法的冬小麦叶绿素浓度估算研究[J].红外,2020,41(11):33-43.
[26]	唐科,秦敏,赵星,等.基于Stacking集成学习模型的气态亚硝酸预测[J].中国环境科学,2020,40(2):582-590.