• 论文 • 上一篇    下一篇

融合数值模式预报数据的深度学习PM2.5浓度预测模型

王舒扬1, 姜金荣2, 迟学斌2, 唐晓3   

  1. 1. 中国科学院计算机网络信息中心, 中国科学院大学, 北京 100190;
    2. 中国科学院计算机网络信息中心, 北京 100190;
    3. 中国科学院大气物理研究所, 北京 100029
  • 收稿日期:2020-10-30 出版日期:2022-06-14 发布日期:2022-06-10
  • 基金资助:
    国家重点研发计划(2016YFB0200800)和中科院信息化专项课题(XXH13506-302)资助

王舒扬, 姜金荣, 迟学斌, 唐晓. 融合数值模式预报数据的深度学习PM2.5浓度预测模型[J]. 数值计算与计算机应用, 2022, 43(2): 142-153.

Wang Shuyang, Jiang Jinrong, Chi Xuebin, Tang Xiao. A DEEP LEARNING MODEL FOR FORECASTING PM2.5 COMBINED WITH NUMERICAL MODEL DATA[J]. Journal on Numerica Methods and Computer Applications, 2022, 43(2): 142-153.

A DEEP LEARNING MODEL FOR FORECASTING PM2.5 COMBINED WITH NUMERICAL MODEL DATA

Wang Shuyang1, Jiang Jinrong2, Chi Xuebin2, Tang Xiao3   

  1. 1. Computer Network Information Center, Chinese Academy of Sciences;University of Chinese Academy of Sciences, Beijing 100190, China;
    2. Computer Network Information Center, Chinese Academy of Sciences, Beijing 100190, China;
    3. Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China
  • Received:2020-10-30 Online:2022-06-14 Published:2022-06-10
PM2.5污染问题是中国近年来引起广泛关注的环境问题,对PM2.5浓度进行预报有重要意义.传统的预报方法是基于空气动力学理论的数值模式预报方法.最近几年深度学习方法被广泛应用于PM2.5浓度预报问题.之前的深度学习预报方法主要是使用观测站的观测数据建立单点式的预报模型.本文使用ConvLSTM深度神经网络建立模型,在中国及周边区域的PM2.5数据集上实现了网格化的序列到序列预报.模型通过卷积模块提取空间特征,通过LSTM模块提取时间特征,适合解决PM2.5网格化预报问题.同时,模型中使用了再分析数据和模式数据两种不同来源的数据结合起来进行预报,融合了深度学习方法和传统数值模式方法.实验表明,模型的均方根误差比数值模式预报下降30.2%,具有良好的预报效果.
Particulate Matter(PM2.5) pollution caused widespread concern recent years in China. It is very significant to predict PM2.5 concentration. Traditional prediction method is numerical model method, which is based on aerodynamics. Deep learning model is applied to forecast PM2.5 concentration these years. The data used by past research is mainly from monitoring station. In this paper, a sequential grid forecast model is proposed in PM2.5 dataset among China and adjacent regions, and the model is based on Convolutional Long Short-Term Memory(ConvLSTM) deep learning neural network. This model is suitable for PM2.5 prediction, as the convolutional module can extract spatial feature and the LSTM module can extract time feature. The model uses both re-analysis data and numerical model data, and combine deep learning method and numerical model method. The experimental results showed that the Root Mean Square Error(RMSE) decrease 30.2%, compared with numerical model method.

MR(2010)主题分类: 

()
[1] 崔相辉,谢剑锋,张丰,丁琳,李增顺,郝震寰,刘勇,赵起超.基于深度学习的PM2.5预测模型建立[J].北京测绘, 2017,(06):22-27.
[2] 曲悦,钱旭,宋洪庆,何杰,李剑辉,修昊.基于机器学习的北京市PM2.5浓度预测模型及模拟分析[J].工程科学学报, 2019, 41(03):401-407.
[3] Huang C, Kuo P. A Deep CNN-LSTM Model for Particulate Matter (PM2.5) Forecasting in Smart Cities[J]. Sensors, 2018, 18(7), 2220.
[4] Kong, L, Tang X, Zhu J, Wang Z, Li J, Wu H, Wu Q, Chen H, Zhu L, Wang W, Liu B, Wang Q, Chen D, Pan Y, Song T, Li F, Zheng H, Jia G, Lu M, Wu L and Carmichael G R. A Six-year long (2013-2018) High-resolution Air Quality Reanalysis Dataset over China base on the assimilation of surface observations from CNEMC, Earth Syst. Sci. Data Discuss., https://doi.org/10.5194/essd-2020-100, in review, 2020.
[5] Jin X, Yang N, Wang X, et al. Integrated Predictor Based on Decomposition Mechanism for PM2.5 Long-Term Prediction[J]. Applied Sciences, 2019, 9(21).
[6] Hochreiter S, Schmidhuber, J. Long Short-Term Memory. Neural Computation. 1997, 9:1735-1780.
[7] Shi X, Chen Z, Wang H, et al. Convolutional LSTM Network:a machine learning approach for precipitation nowcasting[C]. Neural information processing systems, 2015, 802-810.
[8] Shi X, Gao Z, Lausen L, et al. Deep Learning for Precipitation Nowcasting:A Benchmark and A New Model[C]. Neural information processing systems, 2017, 5617-5627.
[9] Bengio S, Vinyals O, Jaitly N, et al. Scheduled sampling for sequence prediction with recurrent Neural networks[C]. Neural information processing systems, 2015, 1171-1179.
[10] Wang Y, Long M, Wang J, et al. PredRNN:recurrent neural networks for predictive learning using spatiotemporal LSTMs[C]. Neural information processing systems, 2017:, 879-888.
[11] 许柏宁,姜金荣,郝卉群,林鹏飞,何丹丹.一种基于区域海表面温度异常预测的ENSO预报深度学习模型[J].科研信息化技术与应用, 2017, 8(6):65-76.
[12] 张伟,王自发,安俊岭,等.利用BP神经网络提高奥运会空气质量实时预报系统预报效果[J].气候与环境研究, 2009, 15(5):595-601
[13] 程念亮,李红霞,孟凡,柴发合,程兵芬.我国城市PM2.5数值预报简述[J].安徽农业科学, 2015, 43(07):243-246+271.
[14] 戴李杰,张长江,马雷鸣.基于机器学习的PM2.5短期浓度动态预报模型[J].计算机应用, 2017, 37(11):3057-3063.
[15] 潘锦秀,晏平仲,孙峰,李云婷,刘保献,王占山,董瑞.多元线性回归方法对北京地区PM2.5预报的改进应用[J].中国环境监测, 2019, 35(02):43-52.
No related articles found!
阅读次数
全文


摘要