HOURLY TEMPERATURE FORECAST METHOD OF SINGLE STATION BASED ON 1DCNN AND LSTM
-
摘要: 针对海量气象观测数据间存在大量的物理噪声、与气温无关的冗余特征以及时间相关性，提出了一种将一维卷积神经网络(1DCNN)和长短期记忆神经网络(LSTM)相结合的多信息融合气温预报方法。首先，运用差分法对气象观测数据进行预处理，得到平稳时间序列数据；其次，运用1DCNN提取与气温变化相关的特征变量作为神经网络模型的输入变量；最后，运用1DCNN和LSTM构建多信息融合气温预报模型1DCNN-LSTM，并以云南省昆明市历史气象观测数据为例，与传统的LSTM、1DCNN和反向传播神经网络(BP)对未来24小时的逐时气温预报进行了比较研究。研究结果表明，1DCNN-LSTM的均方根误差(RMSE)相较于LSTM、1DCNN和BP最大降低了5.221%、19.350% 和9.253%，平均绝对误差(MAE)最大降低了4.419%、17.520% 和8.089%。为气温的精准预报提供了参考依据。Abstract: To deal with the large amount of physical noise, redundant features unrelated to temperature, and time correlation between massive meteorological observational data, a multi-information fusion temperature forecast method combining one-dimensional convolutional neural network (1DCNN) and long short-term memory neural network (LSTM) is proposed. First, the difference method is used to preprocess the meteorological observational data to obtain stationary time series data. Second, 1DCNN is used to extract feature variables related to temperature changes as the input variables of the neural network model. Finally, 1DCNN and LSTM are used to establish a multi-information fusion temperature prediction model 1DCNN-LSTM. Taking the historical meteorological observational data of Kunming City in Yunnan Province as an example, the model is compared with the traditional LSTM, 1DCNN and Back Propagation Neural Network (BP) of the hourly temperature forecast in the next 24 hours. The results show that compared with those of LSTM, 1DCNN and BP, the root mean square error (RMSE) of 1DCNN-LSTM is reduced by 5.221%, 19.350%, and 9.253%, and the mean absolute error (MAE) is reduced by 4.419%, 17.520% and 8.089%, respectively. This research method provides a reference for the accurate prediction of air temperature.
-
表 1 1DCNN-LSTM相较于1DCNN、LSTM和BP模型在RMSE、MAE、r和p上改进的百分比
模型 评价指标 1小时 2小时 3小时 4小时 5小时 6小时 7小时 8小时 9小时 10小时 LSTM RMSE 5.221 4.216 3.651 3.184 2.493 2.103 1.703 1.219 0.854 0.519 MAE 4.419 3.757 3.415 3.095 2.509 2.129 1.760 1.370 1.141 0.676 r 0.119 0.121 0.123 0.125 0.126 0.128 0.129 0.130 0.130 0.131 p 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1DCNN RMSE 19.350 16.241 14.335 12.497 10.650 9.331 7.973 6.634 5.552 4.694 MAE 17.520 15.140 13.541 12.060 10.497 8.967 7.878 6.548 5.317 4.384 r 0.435 0.443 0.451 0.458 0.465 0.470 0.475 0.478 0.480 0.482 p 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 BP RMSE 9.253 7.685 6.798 5.908 4.955 4.264 3.545 2.753 2.110 1.527 MAE 8.089 7.034 6.332 5.757 4.951 4.289 3.650 2.917 2.358 1.713 r 0.225 0.228 0.232 0.235 0.238 0.240 0.242 0.244 0.245 0.246 p 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 
下载: 导出CSV
-
[1] CURCEAC S, TERNYNCK C, OUARDA T B, et al. Short-term air temperature forecasting using nonparametric functional data analysis and SARMA models[J]. Environmental Modelling & Software, 2019, 111(1): 394-408. [2] MORAL-CARCEDO J, PÉREZ-GARCÍA J. Temperature effects on firms'electricity demand: an analysis of sectorial differences in Spain [J]. Appl Energy, 2015, 142(3): 407-425. [3] 熊敏诠, 代刊, 唐健. 春季中国日最高气温延伸期预报误差分析及订正[J]. 热带气象学报, 2020, 36(6): 795-804. [4] SALCEDO-SANZ S, DEO R C, CARRO-CALVO L, et al. Monthly prediction of air temperature in Australia and New Zealand with machine learning algorithms[J]. Theor Appl Climatol, 2016, 125(1-2): 13-25. [5] WEBBER H, EWERT F, KIMBALL B A, et al. Simulating canopy temperature for modelling heat stress in cereals[J]. Environmental Modelling & Softwarel, 2016, 77(3): 143-155. [6] FUENTES M, CAMPOS C, GARCIA-LOYOLA S. Application of artificial neural networks to frost detection in central Chile using the next day minimum air temperature forecast[J]. Chilean Journal of Agricultural Research, 2018, 78(3): 327-338. [7] MADHIARASAN M, TIPALDI M, SIANO P. Analysis of artificial neural network performance based on influencing factors for temperature forecasting applications[J]. Journal of High Speed Networks, 2020, 26(3): 209-223. [8] ABDALLAH W, ABDALLAH N, MARION J M, et al. A hybrid methodology for short term temperature forecasting[J]. International Journal of Intelligence Science, 2020, 10(3): 65-81. [9] 赖绍钧, 何芬, 吴毅伟, 等. 地表利用城市化在一次极端高温过程数值模拟中的影响[J]. 热带气象学报, 2020, 36(3): 347-359. [10] 熊雄, 姚薇, 叶小岭, 等. 基于空间回归的B样条拟合在地面气温资料质量控制中的应用[J]. 热带气象学报, 2020, 36(4): 489-498. [11] 张金秀, 杨晓玲, 孙占峰, 等. 河西走廊东部强降温气候特征及预报方法研究[J]. 中国农学通报, 2020, 36(10): 114-121. [12] 张颖超, 姚润进, 熊雄. 一种基于PSO-MEF的地面气温观测资料的质量控制方法[J]. 热带气象学报, 2017, 33(2): 273-280. [13] 李江峰, 蔡晓军, 王文, 等. 偏最小二乘回归在水汽和地面气温多模式集成预报中的应用研究[J]. 地球科学进展, 2018, 33(4): 404-415. [14] DONGJIN C, CHEOLHEE Y, JUNGHO I, et al. Comparative assessment of various machine learning-based bias correction methods for numerical weather prediction model forecasts of extreme air temperatures in urban areas[J]. Earth and Space Science, 2020, 7(4): 1-18. [15] 贺圣平, 王会军, 李华, 等. 机器学习的原理及其在气候预测中的潜在应用[J]. 大气科学学报, 2021, 44(1): 26-38. [16] 张亚刚, 杨银, 张成军, 等. 基于径向基函数神经网络温度预报订正方法及评估[J]. 热带气象学报, 2021, 37(1): 136-144. [17] MAGDALENA G, JAROMIR K, PIOTR B, et al. Combined TBATS and SVM model of minimum and maximum air temperatures applied to wheat yield prediction at different locations in Europe[J]. Agricultural and Forest Meteorology, 2020, 281(2): 1-19. [18] 周满国, 黄艳国, 杨训根. 基于GRU神经网络与灰色模型集成的气温预报[J]. 热带气象学报, 2020, 36(6): 855-864. [19] 樊仲欣, 陈旭红, 谭桂容. 基于递归小波神经网络的江苏城镇夏季最高气温预报预警技术[J]. 自然灾害学报, 2019, 28(6): 56-69. [20] 施晓芬, 陈翔, 曹永勇, 等. 基于回归BP神经网络误差分级迭代法的逐时气温预测方法研究[J]. 数学的实践与认识, 2019, 49(1): 145-151. [21] 李文杰, 罗文斌, 李艺文, 等. 基于可分离卷积与LSTM的语音情感识别研究[J]. 信息技术, 2020, 44(10): 61-66. [22] 马玉环, 张瑞军, 武晨, 等. 深度残差网络和LSTM结合的图像序列表情识别[J]. 重庆邮电大学学报(自然科学版), 2020, 32(5): 874-883. [23] 倪铮, 梁萍. 基于LSTM深度神经网络的精细化气温预报初探[J]. 计算机应用与软件, 2018, 35(11): 233-237. [24] 陶晔, 杜景林. 基于随机森林的长短期记忆网络气温预测[J]. 计算机工程与设计, 2019, 40(3): 737-743. [25] 周南, 杜军平, 姚旭, 等. 基于卷积神经网络的微博话题内容搜索方法[J]. 计算机科学与探索, 2019, 13(5): 753-764. [26] DIAS L M, FATIH D, ADNAN Y. Speech emotion recognition with deep convolutional neural networks[J]. Biomedical Signal Processing and Control, 2020, 59(5): 1-11. [27] SEPP H, UUML J, RGEN S. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1 735-1 780. -
点击查看大图
图(10) / 表(1)
计量
- 文章访问数: 293
- HTML全文浏览量: 40
- PDF下载量: 44
- 被引次数: 0
粤公网安备 4401069904700003号