基于LSTM网络的大雾临近预报模型及应用
|
摘要
长短期记忆网络(LSTM)是一种时间递归神经网络,适合于预测时间序列延续性相对较长的事件.本文基于LSTM网络构建了一个全新的大雾临近预报框架,首先将地面气象要素观测资料转化成时间序列数据,并基于此序列进行建模.为了验证提出的模型的准确性,将安徽省81个国家站近2年地面气象要素数据转换为序列数据,基于该数据集对未来1–4小时进行逐小时大雾预报实验,实验结果显示本文提出的模型其TS-Score分别为61%、55%、36%和31%,明显优于卷积神经网络(CNN)以及传统机器学习算法如支持向量机(SVM)和K-近邻算法(KNN)的预测结果,是大雾临近预报的一种有效预报方法.
Long-Term and Short-Term Memory(LSTM) network is a time recursive neural network, which is suitable for predicting events with relatively long delay in time series. In this study, a new fog proximity prediction framework based on LSTM network is constructed, which can transform meteorological observation data into time series data and model them based on time series data. In order to validate the proposed model effectively, this study transforms the surface meteorological data of 81 national stations in Anhui Province from October 2015 to June 2017 into sequence data and constructs a validation data set. Based on this data set, the future 1–4 hourly fog forecasting experiments are carried out.The experimental results show that the proposed model's TS-Scores are 61%, 55%, 36%, and 31%, respectively, which are obviously better than CNN and those of traditional machine learning algorithms such as SVM and KNN. It is an effective method for fog prediction.
Long-Term and Short-Term Memory(LSTM) network is a time recursive neural network, which is suitable for predicting events with relatively long delay in time series. In this study, a new fog proximity prediction framework based on LSTM network is constructed, which can transform meteorological observation data into time series data and model them based on time series data. In order to validate the proposed model effectively, this study transforms the surface meteorological data of 81 national stations in Anhui Province from October 2015 to June 2017 into sequence data and constructs a validation data set. Based on this data set, the future 1–4 hourly fog forecasting experiments are carried out.The experimental results show that the proposed model's TS-Scores are 61%, 55%, 36%, and 31%, respectively, which are obviously better than CNN and those of traditional machine learning algorithms such as SVM and KNN. It is an effective method for fog prediction.
引文
1吴兑,邓雪娇,毛节泰,等.南岭大瑶山高速公路浓雾的宏微观结构与能见度研究.气象学报, 2007, 65(3):406–415.[doi:10.3321/j.issn:0577-6619.2007.03.009]
2李秀连,陈克军,王科,等.首都机场大雾的分类特征和统计分析.气象科技, 2008, 36(6):717–723.[doi:10.3969/j.issn.1671-6345.2008.06.008]
3 崔新强,周小兰,付佳,等.高速铁路安全运行高影响天气条件等级标准研究.灾害学,2016,31(3):26–30.[doi:10 .3969/j.issn.1000-811X.2016.03.005]
4黄健,吴兑,黄敏辉,等.1954-2004年珠江三角洲大气能见度变化趋势.应用气象学报,2008,19(1):61-70.[doi:10.3969/j.issn.1001-7313.2008.01.009]
5范引琪,李春强.1980-2003年京、津、冀地区大气能见度变化趋势研究.高原气象,2008,27(6):1392-1400.
6刘晓舟,许潇锋,杨军.华东三市能见度、气溶胶和太阳辐射变化特征.气象科技,2013,41(2):352-359.[doi:10.3969/j.issn.1671-6345.2013.02.027]
7刘骞,盛立芳,王园香,等.气象要素对中国大气能见度长期变化影响的定量研究.气候与环境研究,2016,21(1):47-55.
8中华人民共和国国家统计局.中华人民共和国2017年国民经济和社会发展统计公报.人民日报,2018-03-01(10).
9周须文,时青格,贾俊妹,等.低能见度雾的分级预报方法研究.热带气象学报,2014,30(1):161-166.[doi:10.3969/j.issn.1004-4965.2014.01.018]
10吴彬贵,张建春,李英华,等.天津港秋冬季低能见度数值释用预报研究.气象,2017,43(7):863-871.
11黄政,袁成松,包云轩,等.基于不同参数化方案的高速公路大雾过程的数值模拟试验.气象,2016,42(8):944-953.
12许爱华,陈翔翔,肖安,等.江西省区域性平流雾气象要素特征分析及预报思路.气象,2016,42(3):372-381.
13崔广新,李殿奎.基于自编码算法的深度学习综述.计算机系统应用,2018,27(9):47-51.[doi:10.15888/j.cnki.csa.006542]
14童基均,常晓龙,赵英杰,等.基于深度学习的运动目标实时识别与定位.计算机系统应用,2018,27(8):28-34.[doi:10.15888/j.cnki.csa.006525]
15张骥,余娟,汪金礼,等.基于深度学习的输电线路外破图像识别技术.计算机系统应用,2018,27(8):176-179.[doi:10.15888/j.cnki.csa.006458]
16许小峰.从物理模型到智能分析--降低天气预报不确定性的新探索.气象,2018,44(3):341-350.
17Graves A.Generating sequences with recurrent neural networks.arXiv:1308.0850,2013.
18Graves A,Jaitly N.Towards end-to-end speech recognition with recurrent neural networks.Proceedings of the 31st International Conference on Machine Learning.Beijing,China.2014.1764-1772.
19Sundermeyer M,Ney H,Schlüter R.From feedforward to recurrent LSTM neural networks for language modeling. IEEE/ACM Transactions on Audio,Speech,and Language Processing,2015,23(3):517-529.[doi:10.1109/TASLP.2015.2400218]
20Graves A.Long short-term memory.Graves A.Supervised Sequence Labelling with Recurrent Neural Networks.Berlin,Heidelberg:Springer,2012.37-45.
21於雯,周武能.基于LSTM的商品评论情感分析.计算机系统应用,2018,27(8):159-163.[doi:10.15888/j.cnki.csa.006483]
22史梦飞,杨燕,贺樑,等.基于Bi-LSTM和CNN并包含注意力机制的社区问答问句分类方法.计算机系统应用,2018,27(9):157-162.[doi:10.15888/j.cnki.csa.006536]
23曹国清,张晓明,陈亚峰.基于PCA-LSTM的多变量矿山排土场滑坡预警研究.计算机系统应用,2018,27(11):252-258.[doi:10.15888/j.cnki.csa.006646]
24李胜宇,高俊波,许莉莉.面向酒店评论的情感分析模型.计算机系统应用,2017,26(1):227-231.[doi:10.15888/j.cnki.csa.005511]
25赵熙,李京萌,童红梅.济南机场低能见度和低跑道视程对比分析.干旱气象,2017,35(5):847-856.
26Abadi M,Agarwal A,Barham P,et al.TensorFlow:Largescale machine learning on heterogeneous distributed systems.arXiv:1603.04467,2016.
2李秀连,陈克军,王科,等.首都机场大雾的分类特征和统计分析.气象科技, 2008, 36(6):717–723.[doi:10.3969/j.issn.1671-6345.2008.06.008]
3 崔新强,周小兰,付佳,等.高速铁路安全运行高影响天气条件等级标准研究.灾害学,2016,31(3):26–30.[doi:10 .3969/j.issn.1000-811X.2016.03.005]
4黄健,吴兑,黄敏辉,等.1954-2004年珠江三角洲大气能见度变化趋势.应用气象学报,2008,19(1):61-70.[doi:10.3969/j.issn.1001-7313.2008.01.009]
5范引琪,李春强.1980-2003年京、津、冀地区大气能见度变化趋势研究.高原气象,2008,27(6):1392-1400.
6刘晓舟,许潇锋,杨军.华东三市能见度、气溶胶和太阳辐射变化特征.气象科技,2013,41(2):352-359.[doi:10.3969/j.issn.1671-6345.2013.02.027]
7刘骞,盛立芳,王园香,等.气象要素对中国大气能见度长期变化影响的定量研究.气候与环境研究,2016,21(1):47-55.
8中华人民共和国国家统计局.中华人民共和国2017年国民经济和社会发展统计公报.人民日报,2018-03-01(10).
9周须文,时青格,贾俊妹,等.低能见度雾的分级预报方法研究.热带气象学报,2014,30(1):161-166.[doi:10.3969/j.issn.1004-4965.2014.01.018]
10吴彬贵,张建春,李英华,等.天津港秋冬季低能见度数值释用预报研究.气象,2017,43(7):863-871.
11黄政,袁成松,包云轩,等.基于不同参数化方案的高速公路大雾过程的数值模拟试验.气象,2016,42(8):944-953.
12许爱华,陈翔翔,肖安,等.江西省区域性平流雾气象要素特征分析及预报思路.气象,2016,42(3):372-381.
13崔广新,李殿奎.基于自编码算法的深度学习综述.计算机系统应用,2018,27(9):47-51.[doi:10.15888/j.cnki.csa.006542]
14童基均,常晓龙,赵英杰,等.基于深度学习的运动目标实时识别与定位.计算机系统应用,2018,27(8):28-34.[doi:10.15888/j.cnki.csa.006525]
15张骥,余娟,汪金礼,等.基于深度学习的输电线路外破图像识别技术.计算机系统应用,2018,27(8):176-179.[doi:10.15888/j.cnki.csa.006458]
16许小峰.从物理模型到智能分析--降低天气预报不确定性的新探索.气象,2018,44(3):341-350.
17Graves A.Generating sequences with recurrent neural networks.arXiv:1308.0850,2013.
18Graves A,Jaitly N.Towards end-to-end speech recognition with recurrent neural networks.Proceedings of the 31st International Conference on Machine Learning.Beijing,China.2014.1764-1772.
19Sundermeyer M,Ney H,Schlüter R.From feedforward to recurrent LSTM neural networks for language modeling. IEEE/ACM Transactions on Audio,Speech,and Language Processing,2015,23(3):517-529.[doi:10.1109/TASLP.2015.2400218]
20Graves A.Long short-term memory.Graves A.Supervised Sequence Labelling with Recurrent Neural Networks.Berlin,Heidelberg:Springer,2012.37-45.
21於雯,周武能.基于LSTM的商品评论情感分析.计算机系统应用,2018,27(8):159-163.[doi:10.15888/j.cnki.csa.006483]
22史梦飞,杨燕,贺樑,等.基于Bi-LSTM和CNN并包含注意力机制的社区问答问句分类方法.计算机系统应用,2018,27(9):157-162.[doi:10.15888/j.cnki.csa.006536]
23曹国清,张晓明,陈亚峰.基于PCA-LSTM的多变量矿山排土场滑坡预警研究.计算机系统应用,2018,27(11):252-258.[doi:10.15888/j.cnki.csa.006646]
24李胜宇,高俊波,许莉莉.面向酒店评论的情感分析模型.计算机系统应用,2017,26(1):227-231.[doi:10.15888/j.cnki.csa.005511]
25赵熙,李京萌,童红梅.济南机场低能见度和低跑道视程对比分析.干旱气象,2017,35(5):847-856.
26Abadi M,Agarwal A,Barham P,et al.TensorFlow:Largescale machine learning on heterogeneous distributed systems.arXiv:1603.04467,2016.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |