基于主动学习和克里金插值的空气质量推测

doi:10.11959/j.issn.2096-0271.2018061

摘要/Abstract

摘要：

空气质量监测站仅能在少数位置部署，故而无法获取城市中每个位置的空气质量信息。提出了一种基于主动学习和克里金插值的空气质量推测算法。该算法首先选用克里金插值作为基础的空气质量推测算法，然后结合主动学习的思想，对置信度最大的位置进行优先采样，最终建立基于主动学习的插值模型，通过最少的监测点对空气质量进行采样，最大限度地提升推测其他位置空气质量的准确度。研究结果表明，所提算法能够有效地提高空气质量推测精度，同时减少监测站采样数量，降低部署成本。

关键词: 克里金插值, 空气质量指数, 主动学习, 空间插值, 空气质量推测

Abstract:

For the air quality monitoring station,it can only be deployed in a few locations and cannot obtain the air quality information of each location in the city.An air quality estimation algorithm based on active learning and Kriging interpolation was proposed.Firstly,Kriging interpolation was used as the basic air quality estimation algorithm.Secondly,combined with the idea of active learning,the position-first sampling with the highest confidence of the model was searched.Finally,an interpolation model based on active learning was established to select the least position-to-air quality.Sampling was performed to maximize the accuracy of air quality at other locations.The results show that the proposed algorithm can effectively improve the accuracy of air quality estimation,reduce the number of sampling stations and reduce the deployment cost.

Key words: Kriging interpolation, air quality index, active learning, spatial interpolation, air quality estimation

常慧娟, 於志文, 於志勇, 安琦, 郭斌. 基于主动学习和克里金插值的空气质量推测[J]. 大数据, 2018, 4(6): 54-64.

Huijuan CHANG, Zhiwen YU, Zhiyong YU, Qi AN, Bin GUO. Air quality estimation based on active learning and Kriging interpolation[J]. Big Data Research, 2018, 4(6): 54-64.

图/表 14

表1

图1

图2

表2

表3

图3

图4

图5

图6

表4

图7

图8

图9

图10

参考文献［16］

［1］	李杰, 瞿亮, 桑会勇 ,等. PM_(2.5)浓度插值中不同空间插值方法对比［J］. 测绘科学, 2016(4): 50-54,101. LI J , QU L , SANG H Y ,et al. Comparison of different spatial interpolation methods in PM_(2 . 5)concentration interpolation［J］. Surveying and Mapping Science, 2016(4): 50-54,101.
［2］	VARDOULAKIS S , FISHER B E A , PERICLEOUS K ,et al. Modelling air quality in street canyons:a review［J］. Atmospheric Environment, 2003,37(2): 155-182.
［3］	BURROWS W R , BENJAMIN M , BEAUCHAMP S ,et al. CART decisiontree statistical analysis and prediction of summer season maximum surface ozone for the vancouver,montreal,and atlantic regions of Canada［J］. Journal of Applied Meteorology, 1995,34(8): 1848-1862.
［4］	SHARMA M , AGGARWAL S , BOSE P ,et al. Meteorology-based forecasting of air quality index using neural network［C］// IEEE International Conference on Industrial Informatics,August 21-24,2003,Banff,Canada. Piscataway:IEEE Press, 2003.
［5］	宋宇辰, 甄莎 . BP神经网络和时间序列模型在包头市空气质量预测中的应用［J］. 干旱区资源与环境, 2013,27(7): 65-70. SONG Y C , ZHEN S . Application of BP neural network and time series model in air quality prediction of Baotou City［J］. Journal of Arid Land Resources and Environment, 2013,27(7): 65-70.
［6］	祝翠玲, 蒋志方, 王强 . 基于B-P神经网络的环境空气质量预测模型［J］. 计算机工程与应用, 2007(22): 223-227. ZHU C L , JIANG Z F , WAN G Q . A prediction model of ambient air quality based on B-P neural network［J］. Computer Engineering and Applications, 2007(22): 223-227.
［7］	ZHENG Y , YI X , LI M ,et al. Forecasting fine-grained air quality based on big data［C］// The 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Minin g,August 10-13,2015,Sydney,Australia. New York:ACM Press, 2015: 2267-2276.
［8］	林开春, 邵峰晶 . 基于随机森林和神经网络的空气质量预测研究［J］. 青岛大学学报(工程技术版), 2018,33(2): 32-36. LIN K C , SHAO F J . Study on air quality prediction based on random forest and neural network［J］. Journal of Qingdao University (Engineering ＆ Technology Edition), 2018,33(2): 32-36.
［9］	孟倩 . 基于随机森林视角的空气质量分类预测［J］. 重庆工商大学学报(自然科学版), 2018,35(3): 30-34. MENG Q . Classification and prediction of air quality based on random forest pers pective［J］. Journal of Chongqing Technology and Business University (Natural Science Edition), 2018,35(3): 30-34.
［10］	苏静, 吴海平 . 应用灰色理论模型预测环境空气质量变化趋势——以靖江市环境空气质量预测为例［J］. 污染防治技术, 2010,23(4): 10-12,22. SU J , WU H P . Predicting the change trend of ambient air quality by using grey theory model—— taking the prediction of ambient air quality in Jingjiang City as an example［J］. Pollution Control Technology, 2010,23(4): 10-12,22.
［11］	杨锦伟, 孙宝磊 . 基于灰色马尔科夫模型的平顶山市空气污染物浓度预测［J］. 数学的实践与认识, 2014,44(2): 64-70. YANG J W , SUN B L . The prediction of air pollutant concentration in Pingdingshan City based on grey Markov model［J］. Mathematics in Practice and Theory, 2014,44(2): 64-70.
［12］	闫妍, 张云鹏, 李铠月 ,等. 基于BP神经网络的西安环境空气质量的预测［J］. 电子设计工程, 2013,21(21): 54-57. YAN Y , ZHANG Y P , LI K Y ,et al. Prediction of ambient air quality in Xi’an based on BP neural network［J］. Electronic Design Engineering, 2013,21(21): 54-57.
［13］	段悦, 舒红, 胡泓达 ,等. 全球 MODIS 气温数据的修正秩克里金插值分析［J］. 武汉大学学报(信息科学版), 2015(8): 1036-1041. DUAN Y , SHU H , HU H D ,et al. Modified rank kriging interpolation analysis of global MODIS temperature data［J］. Journal of Wuhan University (Information Science Edition), 2015(8): 1036-1041.
［14］	ZHOU Z H , LI M . Semisupervised regression with cotraining-style algorithms［J］. IEEE Transactions on Knowledge ＆ Data Engineering, 2007,19(11): 908-913.
［15］	韩忠华 . Kriging模型及代理优化算法研究进展［J］. 航空学报, 2016,37(11): 3197-3225. HAN Z H . Research progress of Kriging model and agent optimization algorithm［J］. Acta Aeronautica ET Astronautica Sinica, 2016,37(11): 3197-3225.
［16］	CAI D , HE X . Manifold adaptive experimental design for text categorization［J］. IEEE Transactions on Knowledge ＆ Data Engineering, 2012,24(4): 707-719.

AQI	SO₂（日均值）（μg/m³）	NO₂（日均值）（μg/m³）	PM10（日均值）（μg/m³）	O₃（日均值）（μg/m³）	PM2.5（日均值）（μg/m³）
0	0	0	0	0	0
50	50	40	50	160	35
100	150	80	150	200	75
150	475	180	250	300	115
200	800	280	350	400	150
300	1 600	565	420	800	250
400	2 100	750	500	1 000	350
500	2 620	940	600	1 200	500

station_id	time	PM_2.5浓度（μg/m³）	PM₁₀浓度（μg/m³）	NO₂浓度（μg/m³）	CO浓度（mg/m³）	O₃浓度（μg/m³）	SO₂浓度（μg/m³）
1001	2014/5/10:00	138	159.4	56.3	0.9	50.8	17.2

station_id	name_chinese	name_english	latitude	longitude	district_id
1001	海淀北部新区	Haidian Beibu Xinqu	40.090 679	116.173 553	101

簇	MAPE	RMSE
簇1	6.35%	10.98
簇2	9.07%	2.96