侯俊雄,李琦,林绍福,冯逍,朱亚杰

• 1:北京大学遥感与地理信息系统研究所
• 2:北京大学智慧城市研究中心
• 3:北京未来网络科技高精尖创新中心

摘要(Abstract)：

针对当前我国重污染天气实时的空气质量预报问题,该文提出了一种基于长短期记忆神经网络的PM2.5浓度实时预报方法。此方法结合了北京市地面空气质量监测数据、天气预报模式的气象预报数据及东亚地区污染物排放清单进行分析,在将高层大气状态及排放状况融入了预报模型的同时,利用LSTM模型模拟区域PM2.5浓度的时序连续变化特征,建立了0~72h的区域PM2.5浓度实时预报模型。实验证明,该方法可以有效表征大气污染物变化的时序特征,从而进行更为精准的长时PM2.5浓度预报。同时,使用门限重复单元作为LSTM神经网络的核心,在保障模型精度的同时,进一步减少了模型训练时间,提高了模型的计算效率。

关键词(KeyWords)： PM2.5实时预报;门限重复单元;WRF;深度学习

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 侯俊雄,李琦,林绍福,冯逍,朱亚杰

DOI: 10.16251/j.cnki.1009-2307.2018.07.013

参考文献(References)：

• [1]ZHANG Y,BOCQUET M,MALLET V,et al.Realtime air quality forecasting,part I:history,techniques,and current status[J].Atmospheric Environment,2012,60(32):632-655.
• [2]曹春艳.环境空气质量管理模式研究进展[J].工业安全与环保,2007,33(12):32-35.(CAO Chunyan.Research progress of atmospheric environmental management models[J].Industrial Safety and Environmental Protection,2007,33(12):32-35.)
• [3]谭成好,陈昕,赵天良,等.空气质量数值模型的构建及应用研究进展[J].环境监控与预警,2014(6):1-7.(TAN Chenghao,CHEN Xin,ZHAO Tianliang,et al.Research progress on the development and application of air quality models[J].Environmental Monitoring and Forewarning,2014(6):1-7.)
• [4]NISKA H,RANTAMKI M,HILTUNEN T,et al.Evaluation of an integrated modelling system containing a multi-layer perceptron model and the numerical weather prediction model HIRLAM for the forecasting of urban airborne pollutant concentrations[J].Atmospheric Environment,2005,39(35):6524-6536.
• [5]朱亚杰,李琦,侯俊雄,等.基于支持向量回归的PM2.5浓度实时预报[J].测绘科学,2016,41(1):12-17.(ZHU Yajie,LI Qi,HOU Junxiong et al.Real time prediction of PM2.5concentration based on support vector regression algorithms[J].Science of Surveying and Mapping,2016,41(1):12-17.)
• [6]侯俊雄,李琦,朱亚杰,等.基于随机森林的PM2.5实时预报系统[J].测绘科学,2017,42(1):1-6.(HOU Junxiong,LI Qi,ZHU Yajie,et al.Real-time forecasting system of PM2.5concentration based on spark framework and random forest model[J].Science of Surveying&Mapping,2017,42(1):1-6.)
• [7]FENG X,LI Q,ZHU Y,et al.Artificial neural networks forecasting of PM2.5pollution using air mass trajectory based geographic model and wavelet transformation[J].Atmospheric Environment,2015,107:118-128.
• [8]范竣翔,李琦,朱亚杰,等.基于RNN的空气污染时空预报模型研究[J].测绘科学,2017,42(7):76-83.(FAN Junxiang,LI Qi,ZHU Yajie,et al.Aspatio-temporal prediction framework for air pollution based on deep RNN[J].Science of Surveying and Mapping,2017,42(7):76-83.)
• [9]侯俊雄,李琦,朱亚杰,等.融机器学习与WRF大气模式的PM2.5预报方法[J].测绘科学,2018,42(2):114-120.(HOU Junxiong,LI Qi,ZHU Yajie,et al.PM2.5forecasting method based on machine learning and WRF hybrid model[J].Science of Surveying and Mapping,2018,42(2):114-120.)
• [10]CHAN C K,YAO X.Air pollution in mega cities in China[J].Atmospheric Environment,2008,42(1):1-42.
• [11]关大博,刘竹.雾霾真相:京津冀地区PM2.5污染解析及减排策略研究[M].北京:中国环境出版社,2014:1-4.(GUAN Dabo,LIU Zhu.Tracing back the smog:source analysis and control strategies for PM2.5pollution in Beijing-Tianjin-Hebei[M].Beijing:China Environmental Science Press,2014:1-4.)
• [12]HUANG R J,ZHANG Y,BOZZETTI C,et al.High secondary aerosol contribution to particulate pollution during haze events in China[J].Nature,2014,514(7521):218-222.
• [13]EUROPE W.Health aspects of air pollution,results from the who project systematic review of health aspects of air pollution in Europe[J].Heat Vent,2004,46(8):1953-1956.
• [14]MIKOLOV T,KARAFIT M,BURGET L,et al.Recurrent neural network based language model[C]∥Conference of the International Speech Communication Association.Chiba,Japan:DBLP,2010:1045-1048.
• [15]HOCHREITER S,SCHMIDHUBER J.Long shortterm memory[J].Neural Computation,1997,9(8):1735-1780.
• [16]CHUNG,J Y,CAGLAR G L,CHO K C,et al.Empirical evaluation of gated recurrent neural networks on sequence modeling[C]∥Presented in NIPS 2014Deep Learning and Representation Learning Workshop.[S.l.]:[s.n.],2014.