Journal of Computer Applications ›› 2024, Vol. 44 ›› Issue (8): 2643-2650.DOI: 10.11772/j.issn.1001-9081.2023081169

• Frontier and comprehensive applications • Previous Articles    

Multi-granularity abrupt change fitting network for air quality prediction

Qianhong SHI1, Yan YANG1(), Yongquan JIANG1, Xiaocao OUYANG1, Wubo FAN2, Qiang CHEN2, Tao JIANG2, Yuan LI2   

  1. 1.School of Computing and Artificial Intelligence,Southwest Jiaotong University,Chengdu Sichuan 611756,China
    2.Air Environmental Research Institute,Sichuan Academy of Eco?Environmental Sciences,Chengdu Sichuan 610046,China
  • Received:2023-08-31 Revised:2023-09-11 Accepted:2023-10-09 Online:2024-08-22 Published:2024-08-10
  • Contact: Yan YANG
  • About author:SHI Qianhong, born in 1999, M. S. candidate. His research interests include data mining, sequential data analysis.
    JIANG Yongquan, born in 1981, Ph. D., assistant research fellow. His research interests include scientific intelligence, computer vision.
    OUYANG Xiaocao, born in 1995, Ph. D. candidate. Her research interests include deep learning, spatio-temporal data mining.
    FAN Wubo, born in 1988, Ph. D., engineer. His research interests include air pollution prevention and control.
    CHEN Qiang, born in 1981, Ph. D., senior engineer. His research interests include air pollution prevention and control.
    JIANG Tao, born in 1981, M. S., senior engineer. His research interests include air pollution prevention and control.
    LI Yuan, born in 1988, M. S., senior engineer. Her research interests include air pollution prevention and control.
  • Supported by:
    National Natural Science Foundation of China(61976247)

面向空气质量预测的多粒度突变拟合网络

石乾宏1, 杨燕1(), 江永全1, 欧阳小草1, 范武波2, 陈强2, 姜涛2, 李媛2   

  1. 1.西南交通大学 计算机与人工智能学院,成都 611756
    2.四川省生态环境科学研究院 大气环境研究所,成都 610046
  • 通讯作者: 杨燕
  • 作者简介:石乾宏(1999—),男,河南荥阳人,硕士研究生,主要研究方向:数据挖掘、序列数据分析
    杨燕(1964—),女,安徽合肥人,教授,博士,CCF会员,主要研究方向:人工智能、大数据分析与挖掘 yyang@swjtu.edu.cn
    江永全(1981—),男,四川泸州人,助理研究员,博士,主要研究方向:科学智能、计算机视觉
    欧阳小草(1995—),女,四川广元人,博士研究生,主要研究方向:深度学习、时空数据挖掘
    范武波(1988—),男,四川广元人,工程师,博士,主要研究方向:大气污染防治
    陈强(1981—),男,四川泸州人,高级工程师,博士,主要研究方向:大气污染防治
    姜涛(1981—),男,四川宜宾人,高级工程师,硕士,主要研究方向:大气污染防治
    李媛(1988—),女,四川南部人,高级工程师,硕士,主要研究方向:大气污染防治。
  • 基金资助:
    国家自然科学基金资助项目(61976247)

Abstract:

Air quality data, as a typical spatio-temporal data, exhibits complex multi-scale intrinsic characteristics and has abrupt change problem. Concerning the problem that existing air quality prediction methods perform poorly when dealing with air quality prediction tasks containing large amount of abrupt change, a Multi-Granularity abrupt Change Fitting Network (MACFN) for air quality prediction was proposed. Firstly, multi-granularity feature extraction was first performed on the input data according to the periodicity of air quality data in time. Then, a graph convolution network and a temporal convolution network were used to extract the spatial correlation and temporal dependence of the air quality data, respectively. Finally, to reduce the prediction error, an abrupt change fitting network was designed to adaptively learn the abrupt change part of the data. The proposed network was experimentally evaluated on three real air quality datasets, and the Root Mean Square Error (RMSE) decreased by about 11.6%, 6.3%, and 2.2% respectively, when compared to the Multi-Scale Spatial Temporal Network (MSSTN). The experimental results show that MACFN can efficiently capture complex spatio-temporal relationships and performs better in the task of predicting air quality that is prone to abrupt change with a large magnitude of variability.

Key words: air quality prediction, deep learning, spatio-temporal feature, multi-granularity, abrupt change

摘要:

空气质量数据作为一种典型的时空数据,具有复杂的多尺度内在特性并存在突变的问题。针对现有空气质量预测方法在处理包含大量突变数据的空气质量预测任务时表现不佳的问题,提出一种面向空气质量预测的多粒度突变拟合网络(MACFN)。首先,针对空气质量数据在时间上的周期性,对输入数据进行了多粒度的特征提取。然后,采用图卷积网络与时间卷积网络分别提取空气质量数据的空间关联性与时间依赖性。最后,设计一个突变拟合网络自适应地学习数据中的突变部分,从而减小预测误差。所提网络在3个真实的空气质量数据集上进行了实验评估,与多尺度时空网络(MSSTN)相比,均方根误差(RMSE)分别下降约11.6%、6.3%和2.2%。实验结果表明,MACFN能有效捕捉复杂的时空关系,并在变化幅度较大、易发生突变的空气质量预测任务中有更好表现。

关键词: 空气质量预测, 深度学习, 时空特征, 多粒度, 突变

CLC Number: