基于时空优化长短期记忆网络与烟花算法的AQI预测
|
摘要
针对传统深度学习模型在预测空气质量指数(air quality index,AQI)时,难以从时间角度建模、网络超参数选取困难等问题,提出一种基于长短期记忆(long short-term memory,LSTM)网络和烟花算法(fireworks algorithm,FWA)的AQI预测模型LSTM-FWA。首先,以武汉市历史空气质量和气象监测数据为研究对象,利用LSTM网络中隐含层节点之间相互连接的结构特点,对空气质量的时间变化特征进行建模;接下来,考虑到种群多样性和并发性,将烟花算法应用到超参数组合优化问题中;最后,对模型输入分别进行时间、空间、时空角度的优化,实验结果表明基于时空优化的LSTM-FWA模型预测性能提升最为明显。将LSTM-FWA与其他预测模型进行比较,并全面分析不同模型在各种优化策略下的性能。实验结果显示,本文提出的时空优化LSTM-FWA模型对于AQI预测具有最优的性能。
Since the traditional deep learning models have difficulties in time modeling for air quality index(AQI) prediction and hyper-parameters selecting for network models, we propose an AQI prediction model LSTM-FWA based on long short-term memory(LSTM) network and fireworks algorithm(FWA). Firstly, the data of historical air quality and meteorological monitoring data from Wuhan city are taken as the research object, and the time variation characteristics of air quality are modeled by the structure of interconnections among hidden layer nodes in the LSTM network. Then, the swarm intelligent algorithm is used to optimize the selection of hyper-parameters and fireworks algorithm is applied considering the population diversity and concurrency. Finally, the model input is optimized with temporal, spatial, spatial-temporal techniques respectively, and experimental results show that spatial-temporal optimization based LSTM-FWA has the most obvious improvement for predictive performance. Our LSTM-FWA model was compared with other predictive models, and the performance of different models under various optimization strategies was comprehensively analyzed. The experimental results show that the proposed spatial-temporal LSTM-FWA model has the optimal prediction performance for AQI prediction.
Since the traditional deep learning models have difficulties in time modeling for air quality index(AQI) prediction and hyper-parameters selecting for network models, we propose an AQI prediction model LSTM-FWA based on long short-term memory(LSTM) network and fireworks algorithm(FWA). Firstly, the data of historical air quality and meteorological monitoring data from Wuhan city are taken as the research object, and the time variation characteristics of air quality are modeled by the structure of interconnections among hidden layer nodes in the LSTM network. Then, the swarm intelligent algorithm is used to optimize the selection of hyper-parameters and fireworks algorithm is applied considering the population diversity and concurrency. Finally, the model input is optimized with temporal, spatial, spatial-temporal techniques respectively, and experimental results show that spatial-temporal optimization based LSTM-FWA has the most obvious improvement for predictive performance. Our LSTM-FWA model was compared with other predictive models, and the performance of different models under various optimization strategies was comprehensively analyzed. The experimental results show that the proposed spatial-temporal LSTM-FWA model has the optimal prediction performance for AQI prediction.
引文
[1]KELLY F J,FUSSELL J C.Air pollution and public health:Emerging hazards and improved understanding of risk[J].Environmental Geochemistry and Health,2015,37(4):631-649.DOI:10.1007/s10653-015-9720-1.
[2]SALMOND J,SABEL C E,VARDOULAKIS S.Towards the integrated study of urban climate,air pollution,and public health[J].Climate,2018,6(1):1-4.DOI:10.3390/cli6010014.
[3]PéREZ V A,ARASA R,CODINA B,et al.Enhancing air quality forecasts over Catalonia(Spain)using model output statics[J].Journal of Geoscience&Environment Protection,2015,3(8):9-22.DOI:10.4236/gep.2015.38002.
[4]AZID A,JUAHIR H,TORIMAN M E,et al.Prediction of the level of air pollution using principal component analysis and artificial neural network techniques:A case study in Malaysia[J].Water,Air,&Soil Pollution,2014,225(8):2063-1-2063-14.DOI:10.1007/s11270-014-2063-1.
[5]MISHRA D,GOYAL P,UPADHYAY A.Artificial intelligence based approach to forecast PM2.5during haze episodes:A case study of Delhi,India[J].Atmospheric Environment,2015,102:239-248.DOI:10.1016/j.atmosenv.2014.11.050.
[6]HE K,ZHANG X Y,REN S Q,et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE Press,2016:770-778.DOI:10.1109/CVPR.2016.90.
[7]LITJENS G,KOOI T,BEJNORDI B E,et al.A survey on deep learning in medical image analysis[J].Medical Image Analysis,2017,42(9):60-88.DOI:10.1016/j.media.2017.07.005.
[8]TELLO G,AL-JARRAH O Y,YOO P D,et al.Deepstructured machine learning model for the recognition of mixed-defect patterns in semiconductor fabrication processes[J].IEEE Transactions on Semiconductor Manufacturing,2018,31(2):315-322.DOI:10.1109/TSM.2018.2825482.
[9]尹文君,张大伟,闫京海.基于深度学习的大数据空气污染预报[J].中国环境管理,2015,7(6):46-52.DOI:10.3969/j.issn.1674-6252.2015.06.011.YI W J,ZHANG D W,YAN J H.Deep learning based air pollutant forecasting with big data[J].Environmental Conformity Assessment,2015,7(6):46-52.DOI:10.3969/j.issn.1674-6252.2015.06.011(Ch).
[10]董婷,赵俭辉,胡勇.基于时空优化深度神经网络的AQI等级预测[J].计算机工程与应用,2017,53(21):7-23.DOI:10.3778/j.issn.1002-8331.1705-0420.DONG T,ZHAO J H,HU Y.AQI levels prediction based on deep neural network with spatial and temporal optimizations[J].Computer Engineering and Applications,2017,53(21):7-23.DOI:10.3778/j.issn.1002-8331.1705-0420(Ch).
[11]AL-MOLEGI A,JABREEL M,GHALEB B.STF-RNN:Space time features-based recurrent neural network for predicting people next location[C]//IEEE Symposium Series on Computational Intelligence(SSCI).New York:IEEE Press,2017:1-7.DOI:10.1109/SSCI.2016.7849919.
[12]CHANDRA R,ZHANG M J.Competition and collaboration in cooperative coevolution of Elman recurrent neural networks for time-series prediction[J].IEEE Transactions on Neural Networks&Learning Systems,2015,26(12):3123-3136.DOI:10.1109/tnnls.2015.2404823.
[13]ZHAO J L,QU H,ZHAO J H,et al.Towards traffic matrix prediction with LSTM recurrent neural networks[J].Electronics Letters,2018,54(9):566-568.DOI:10.1049/el.2018.0336.
[14]ZHANG Q,WANG H,DONG J Y,et al.Prediction of sea surface temperature using long short-term memory[J].IEEE Geoscience and Remote Sensing Letters,2017,14(10):1745-1749.DOI:10.1109/LGRS.2017.2733548.
[15]TAN Y,ZHU Y C.Fireworks algorithm for optimization[C]//International Conference in Swarm Intelligence.Heidelberg:Springer,2010:355-364.DOI:10.1007/978-3-642-13495-1_44.
[16]ZHANG B,ZHENG Y J,ZHANG M X,et al.Fireworks algorithm with enhanced fireworks interaction[J].IEEE/ACM Transactions on Computational Biology&Bioinformatics,2017,14(1):42-45.DOI:10.1109/TCBB.2015.2446487.
[17]HUANG X B,LI H B,ZHU Y C.Short-term ice accretion forecasting model for transmission lines with modified timeseries analysis by fireworks algorithm[J].IET Generation Transmission&Distribution,2018,12(5):1074-1080.DOI:10.1049/iet-gtd.2017.0619.
[18]AHMAD I,BASHERI M,IQBAL M J,et al.Performance comparison of support vector machine,random forest,and extreme learning machine for intrusion detection[J].IEEEAccess,2018,6:33789-33795.DOI:10.1109/ACCESS.2018.2841987.
[19]LIU Y D,LIU S T,WANG Y Z,et al.A stochastic computational multi-layer perceptron with backward propagation[J].IEEE Transactions on Computers,2018,67(9):1273-1286.DOI:10.1109/TC.2018.2817237.
[20]TRAN S N,GARCEZ A S A.Deep logic networks:Inserting and extracting knowledge from deep belief networks[J].IEEE Transactions on Neural Networks and Learning Systems,2018,29(2):246-258.DOI:10.1109/TNNLS.2016.2603784.
[21]JIA C C,SHAO M,LI S,et al.Stacked denoising tensor auto-encoder for action recognition with spatiotemporal corruptions[J].IEEE Transactions on Image Processing,2018,27(4):1878-1887.DOI:10.1109/TIP.2017.2781299.
[2]SALMOND J,SABEL C E,VARDOULAKIS S.Towards the integrated study of urban climate,air pollution,and public health[J].Climate,2018,6(1):1-4.DOI:10.3390/cli6010014.
[3]PéREZ V A,ARASA R,CODINA B,et al.Enhancing air quality forecasts over Catalonia(Spain)using model output statics[J].Journal of Geoscience&Environment Protection,2015,3(8):9-22.DOI:10.4236/gep.2015.38002.
[4]AZID A,JUAHIR H,TORIMAN M E,et al.Prediction of the level of air pollution using principal component analysis and artificial neural network techniques:A case study in Malaysia[J].Water,Air,&Soil Pollution,2014,225(8):2063-1-2063-14.DOI:10.1007/s11270-014-2063-1.
[5]MISHRA D,GOYAL P,UPADHYAY A.Artificial intelligence based approach to forecast PM2.5during haze episodes:A case study of Delhi,India[J].Atmospheric Environment,2015,102:239-248.DOI:10.1016/j.atmosenv.2014.11.050.
[6]HE K,ZHANG X Y,REN S Q,et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).New York:IEEE Press,2016:770-778.DOI:10.1109/CVPR.2016.90.
[7]LITJENS G,KOOI T,BEJNORDI B E,et al.A survey on deep learning in medical image analysis[J].Medical Image Analysis,2017,42(9):60-88.DOI:10.1016/j.media.2017.07.005.
[8]TELLO G,AL-JARRAH O Y,YOO P D,et al.Deepstructured machine learning model for the recognition of mixed-defect patterns in semiconductor fabrication processes[J].IEEE Transactions on Semiconductor Manufacturing,2018,31(2):315-322.DOI:10.1109/TSM.2018.2825482.
[9]尹文君,张大伟,闫京海.基于深度学习的大数据空气污染预报[J].中国环境管理,2015,7(6):46-52.DOI:10.3969/j.issn.1674-6252.2015.06.011.YI W J,ZHANG D W,YAN J H.Deep learning based air pollutant forecasting with big data[J].Environmental Conformity Assessment,2015,7(6):46-52.DOI:10.3969/j.issn.1674-6252.2015.06.011(Ch).
[10]董婷,赵俭辉,胡勇.基于时空优化深度神经网络的AQI等级预测[J].计算机工程与应用,2017,53(21):7-23.DOI:10.3778/j.issn.1002-8331.1705-0420.DONG T,ZHAO J H,HU Y.AQI levels prediction based on deep neural network with spatial and temporal optimizations[J].Computer Engineering and Applications,2017,53(21):7-23.DOI:10.3778/j.issn.1002-8331.1705-0420(Ch).
[11]AL-MOLEGI A,JABREEL M,GHALEB B.STF-RNN:Space time features-based recurrent neural network for predicting people next location[C]//IEEE Symposium Series on Computational Intelligence(SSCI).New York:IEEE Press,2017:1-7.DOI:10.1109/SSCI.2016.7849919.
[12]CHANDRA R,ZHANG M J.Competition and collaboration in cooperative coevolution of Elman recurrent neural networks for time-series prediction[J].IEEE Transactions on Neural Networks&Learning Systems,2015,26(12):3123-3136.DOI:10.1109/tnnls.2015.2404823.
[13]ZHAO J L,QU H,ZHAO J H,et al.Towards traffic matrix prediction with LSTM recurrent neural networks[J].Electronics Letters,2018,54(9):566-568.DOI:10.1049/el.2018.0336.
[14]ZHANG Q,WANG H,DONG J Y,et al.Prediction of sea surface temperature using long short-term memory[J].IEEE Geoscience and Remote Sensing Letters,2017,14(10):1745-1749.DOI:10.1109/LGRS.2017.2733548.
[15]TAN Y,ZHU Y C.Fireworks algorithm for optimization[C]//International Conference in Swarm Intelligence.Heidelberg:Springer,2010:355-364.DOI:10.1007/978-3-642-13495-1_44.
[16]ZHANG B,ZHENG Y J,ZHANG M X,et al.Fireworks algorithm with enhanced fireworks interaction[J].IEEE/ACM Transactions on Computational Biology&Bioinformatics,2017,14(1):42-45.DOI:10.1109/TCBB.2015.2446487.
[17]HUANG X B,LI H B,ZHU Y C.Short-term ice accretion forecasting model for transmission lines with modified timeseries analysis by fireworks algorithm[J].IET Generation Transmission&Distribution,2018,12(5):1074-1080.DOI:10.1049/iet-gtd.2017.0619.
[18]AHMAD I,BASHERI M,IQBAL M J,et al.Performance comparison of support vector machine,random forest,and extreme learning machine for intrusion detection[J].IEEEAccess,2018,6:33789-33795.DOI:10.1109/ACCESS.2018.2841987.
[19]LIU Y D,LIU S T,WANG Y Z,et al.A stochastic computational multi-layer perceptron with backward propagation[J].IEEE Transactions on Computers,2018,67(9):1273-1286.DOI:10.1109/TC.2018.2817237.
[20]TRAN S N,GARCEZ A S A.Deep logic networks:Inserting and extracting knowledge from deep belief networks[J].IEEE Transactions on Neural Networks and Learning Systems,2018,29(2):246-258.DOI:10.1109/TNNLS.2016.2603784.
[21]JIA C C,SHAO M,LI S,et al.Stacked denoising tensor auto-encoder for action recognition with spatiotemporal corruptions[J].IEEE Transactions on Image Processing,2018,27(4):1878-1887.DOI:10.1109/TIP.2017.2781299.