您的位置：山东大学 -> 科技期刊社 -> 《山东大学学报（工学版）》

山东大学学报(工学版) ›› 2014, Vol. 44 ›› Issue (6): 19-25.doi: 10.6040/j.issn.1672-3961.1.2014.180

• 机器学习与数据挖掘 • 上一篇    下一篇

基于深度信念网络的PM2.5预测

郑毅, 朱成璋   

  1. 国防科学技术大学计算机学院, 湖南 长沙 410073
  • 收稿日期:2014-01-23 修回日期:2014-10-27 出版日期:2014-12-20 发布日期:2014-01-23
  • 通讯作者: 朱成璋(1990-),男,湖南湘潭人,硕士研究生,主要研究方向为机器学习.E-mail:kevin.zhu.china@gmail.com E-mail:kevin.zhu.china@gmail.com
  • 作者简介:郑毅(1989-),男,重庆梁平人,硕士研究生,主要研究方向为数据挖掘.E-mail:justice131@163.com
  • 基金资助:
    国家高技术研究发展计划(863计划)资助项目(2011AA010702);湖南省科技厅计划资助项目(2012FJ4269)

A prediction method of atmospheric PM2.5 based on DBNs

ZHENG Yi, ZHU Chengzhang   

  1. College of Computer Science, National University of Defense Technology, Changsha 410073, Hunan, China
  • Received:2014-01-23 Revised:2014-10-27 Online:2014-12-20 Published:2014-01-23

PDF (PC)

2830

摘要： 提出一种基于深度信念网络(deep belief networks, DBNs)的区域PM2.5日均值预测方法,讨论了训练数据选择方式,并优化了DBNs参数设置。通过相关实验并与基于径向基神经网络(radial basis function, RBF)和反向传播神经网络(back propagation, BP)方法比较,验证了基于DBNs方法的可行性和预测精度。实验结果表明:基于DBNs的方法,区域(西安市)预测PM2.5日均值与观测日均值之间均方差(mean square error, MSE)为8.47×10-4mg2/m6;而采用相同数据集,基于RBF和BP的方法均方差为1.30×10-3mg2/m6和1.96×10-3mg2/m6。比较分析表明:基于DBNs的方法能较好预测区域整体PM2.5的日均值变化趋势,显著优于基于神经网络和径向基网络方法的预测结果。

关键词: 深度学习, 限制玻尔兹曼机, PM2.5预测, 机器学习, 深度信念网络

Abstract: A DBNs-based (deep belief networks) method for forecasting the daily average concentrations of PM2.5 in Xian was proposed. Besides, the way to select training data set as well as the DBNs parameters was optimized. Then relative experiments and comparison with methods based on BP (back propagation) and RBF (radial basis function) artificial neural network confirmed the feasibility and precision of DBNs. The results showed that the MSE (mean square error) between DBNs simulated PM2.5 daily average concentrations and observed ones was 8.47×10-4 mg2/m6, while the MSE of RBF and BP was 1.30×10-3 mg2/m6 and 1.96×10-3 mg2/m6 respectively. Therefore the DBNs-based method was fit for prediction of PM2.5 concentrations and it predicted more accurately than those methods based on RBF and BP artificial neural network.

Key words: restricted boltzmann machine, deep belief networks, PM2.5 prediction, deep learning, machine learning

中图分类号: 

  • TU457
[1] 郭晓泽,单思行.针对PM2.5的综述[J].能源与节能,2012(11):58-59. GUO Xiaoze, SHAN Sixing. Overview of PM2.5[J]. Energyand Energy Conservation, 2012(11):58-59.
[2] 周蕊,邵帅,王彦清.PM2.5的毒性及机制的进展[J].科技传播,2012(6):100. ZHOU Xin, SHAO Shuai, WANG Yanqing. Progress of PM2.5 toxicity and mechanisms[J]. Science & Technology Information, 2012(6):100.
[3] 李恩敬,艾春艳,赵飞.基于文献计量的PM2.5国内外研究情况分析[J].环境与可持续发展,2013,38(4):34-37. LI Enjing, AI Chunyan, ZHAO Fei. Study on PM2.5 baseon papers[J]. Environment and Sustainable Development, 2013, 38(4):34-37.
[4] WANG W, GUO Y. Air pollution PM2.5 data analysis in Los Angeles long beach with seasonal ARIMA model[C]//Proceedings of International Conference on Energy and Environment Technology. Guiling: IEEE, 2009, 3:7-10.
[5] MCKENDRY I G. Evaluation of artificial neural networks for fine particulate pollution (PM10 and PM2.5) forecasting[J]. Journal of the Air & Waste Management Association, 2002, 52(9):1096-1101.
[6] ABRAHAM A. Artificial neural networks[J].Handbook of Measuring System Design, 2005, 129(2):901-908.
[7] YAO L, LU N, JIANG S. Artificial neural network (ANN) for multi-source PM2.5 estimation using surface, MODIS, and meteorological data[C]//Proceedings of International Conference on Biomedical Engineering and Biotechnology. Macau: IEEE, 2012:1228-1231.
[8] ZHU C L, JIANG Z F, WANG Q. Forecastingmodel of environment air quality based on BP neural network[J]. JisuanjiGongcheng yu Yingyong (Computer Engineering and Applications), 2007, 42(22):223-227.
[9] JIANG Z, MENG X, YANG C, et al. A BP neural network prediction model of the urban air quality based on rough set[C]//Proceedings of Fourth International Conference on Natural Computation. Jinan: IEEE, 2008, 1:362-370.
[10] ZHENG H M, SHANG X X. Study on prediction of atmospheric PM2.5 based on RBF neural network[C]//Proceedings of Fourth International Conference on Digital Manufacturing and Automation. Hefei: IEEE, 2013:1287-1289.
[11] REHMAN M Z, NAWI N M. The effect of adaptive momentum in improving the accuracy of gradient descent back propagation algorithm on classification problems[C]//Proceedings of Software Engineering and Computer Systems. Berlin:Springer, 2011:380-390.
[12] 王敏,邹滨,郭宇,等.基于BP人工神经网络的城市PM2.5浓度空间预测[J]. 环境污染与防治, 2013,35(9):63-66. WANG Min, ZHOU Bin, GUO Yu, et al. BP artificialneural network-based analysis of spatial variability of urban PM2.5 concentration[J]. Environmental Pollution & Control, 2013, 35(9):63-66.
[13] 陈一萍,郑朝洪.BP和RBF网络在厦门市大气环境质量评价中的比较[J].环保科技,2008,14(4):8-11. CHEN Yiping, ZHENG Chaohong. Comparison of environmental quality evaluation based on BP and RBF network in Xiamen[J]. Environmental Protection and Technology, 2008, 14(4):8-11.
[14] 刘妹琴,廖晓昕.RBF神经网络的一种鲁棒学习算法[J].华中理工大学学报,2000,28(2):8-10. LIU Meiqin, LIAO Xiaoxi. A robust RBF neural network learning algorithm[J]. Journal of Huazhong University of Science and Technology, 2000, 28(2):8-10.
[15] 孙志军,薛磊,许阳明,等.深度学习研究综述[J].计算机应用研究,2012,29(8):2806-2810. SUN Zhijun, XUE Lei, XU Yangming, et al. Overview ofdeep learning[J]. Application Research of Computers, 2012, 29(8):2806-2810.
[16] DJURDJEVIC P D, HUBER M. Deep belief network for modeling hierarchical reinforcement learning policies[C]//Proceedings of 2013 International Conference on Systems, Man, and Cybernetics. Manchester: IEEE, 2013:2485-2491.
[17] FOUSEK P, RENNIE S, DOGNIN P, et al. Direct product based deep belief networks for automatic speech recognition[C]//Proceedings of 2013 International Conference on Acoustics, Speech and Signal Processing. Vancouver: IEEE, 2013:3148-3152.
[18] ZHOU S, CHEN Q, WANG X. Discriminative deep belief networks for image classification[C]//Proceedings of 17th International Conference on Image Processing. Las Vegas: IEEE, 2010:1561-1564.
[19] GHAHABI O, HERNANDO J. Deep belief networks for i-vector based speaker recognition[C]//Proceedings of 2014 International Conference on Acoustics,Speechand Signal Processing. Florence: IEEE, 2014:1700-1704.
[20] RIOUX L, GIGUERE P. Sign language fingerspelling classification from depth and color images using a deep belief network[C]//Proceedings of 2014 Canadian Conference on Computer and Robot Vision. Quebec: IEEE, 2014:92-97.
[21] ZHU C C, YIN J P, LI Q. A stock decision support system based on DBNs[J].Journal of Computational Information Systems, 2014, 10(2):883-893.
[1] 张冕,黄颖,梅海艺,郭毓. 基于Kinect的配电作业机器人智能人机交互方法[J]. 山东大学学报(工学版), 2018, 48(5): 103-108.
[2] 谢志峰,吴佳萍,马利庄. 基于卷积神经网络的中文财经新闻分类方法[J]. 山东大学学报(工学版), 2018, 48(3): 34-39.
[3] 唐乐爽,田国会,黄彬. 一种基于DSmT推理的物品融合识别算法[J]. 山东大学学报(工学版), 2018, 48(1): 50-56.
[4] 周福娜,高育林,王佳瑜,文成林. 基于深度学习的缓变故障早期诊断及寿命预测[J]. 山东大学学报(工学版), 2017, 47(5): 30-37.
[5] 刘洋,刘博,王峰. 基于Parameter Server框架的大数据挖掘优化算法[J]. 山东大学学报(工学版), 2017, 47(4): 1-6.
[6] 魏波,张文生,李元香,夏学文,吕敬钦. 一种选择特征的稀疏在线学习算法[J]. 山东大学学报(工学版), 2017, 47(1): 22-27.
[7] 何正义,曾宪华,曲省卫,吴治龙. 基于集成深度学习的时间序列预测模型[J]. 山东大学学报(工学版), 2016, 46(6): 40-47.
[8] 周旺,张晨麟,吴建鑫. 一种基于Hartigan-Wong和Lloyd的定性平衡聚类算法[J]. 山东大学学报(工学版), 2016, 46(5): 37-44.
[9] 孟令恒,丁世飞. 基于单静态图像的深度感知模型[J]. 山东大学学报(工学版), 2016, 46(3): 37-43.
[10] 刘杰, 杨鹏, 吕文生, 刘阿古达木, 刘俊秀. 基于气象因素的PM2.5质量浓度预测模型[J]. 山东大学学报(工学版), 2015, 45(6): 76-83.
[11] 谢琳1,殷熙尧2,李凡长3,吴佳3. 一种逆归结学习表示[J]. 山东大学学报(工学版), 2013, 43(4): 46-50.
[12] 何雪英1,2, 秦伟1, 尹义龙1*, 赵联征1,乔昊3. 基于机器学习的视频指纹识别[J]. 山东大学学报(工学版), 2011, 41(4): 29-33.
[13] 梁春林1,彭凌西2*. 基于免疫网络的无监督式分类算法[J]. 山东大学学报(工学版), 2010, 40(5): 82-86.
[14] 郭茂祖 邹权 李文滨 韩英鹏. 生物信息学中的学习问题[J]. 山东大学学报(工学版), 2009, 39(3): 1-6.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 2018 《山东大学学报(工学版)》编辑部　
地址: 济南市山大南路27号（250100）　 电话: 0531-88366735　 E-mail: xbgxb@sdu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发