CN113807432A - Air temperature forecast data correction method based on deep learning - Google Patents
Air temperature forecast data correction method based on deep learning Download PDFInfo
- Publication number
- CN113807432A CN113807432A CN202111084626.1A CN202111084626A CN113807432A CN 113807432 A CN113807432 A CN 113807432A CN 202111084626 A CN202111084626 A CN 202111084626A CN 113807432 A CN113807432 A CN 113807432A
- Authority
- CN
- China
- Prior art keywords
- data
- layer
- model
- air temperature
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012937 correction Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000013135 deep learning Methods 0.000 title claims abstract description 22
- 238000000605 extraction Methods 0.000 claims abstract description 22
- 230000004927 fusion Effects 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 238000013507 mapping Methods 0.000 claims abstract description 6
- 238000009499 grossing Methods 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 19
- 230000004913 activation Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 10
- 239000013598 vector Substances 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 4
- 238000004458 analytical method Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 6
- 238000007689 inspection Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention belongs to the technical field of weather forecast, and particularly relates to an air temperature forecast data correction method based on deep learning. In the data preprocessing stage, the spatial resolution is improved while the temperature forecast data is converted into lattice point data by using a nearest neighbor interpolation method, and Gaussian noise is removed by smoothing the temperature data by adopting Gaussian filtering; in the stage of constructing a deep learning network structure, the time resolution is improved by utilizing the upsampling process, meanwhile, the time characteristic is extracted by adopting the LSTM, the time characteristic and the numerical prediction characteristic extracted by the UNet network are subjected to weighted fusion, and the temperature prediction precision is improved by utilizing the nonlinear mapping capability of the deep learning network and the information extraction capability of lattice data. In conclusion, the air temperature forecast data correction model can calculate a more accurate correction value, can improve the space-time resolution of air temperature forecast, reduces the manpower consumption, and provides correction service with high resolution and accurate analysis for future refined grid point forecast.
Description
Technical Field
The invention belongs to the technical field of weather forecast, and particularly relates to an air temperature forecast data correction method based on deep learning.
Background
In recent years, the meteorological department vigorously develops the intelligent grid forecasting service, and the dominant position of a high-resolution forecasting mode is more prominent. However, the high resolution prediction has many limitations, and the main reason is that the initial value error and the mode error caused by the initial condition, the boundary condition, the physical process, etc. cannot be eliminated, so the development of the correction technology cannot be ignored. A reasonable, objective and quantitative correction method is a bridge for connecting numerical model prediction and refined weather prediction and is also a key of high-resolution prediction in a period of time in the future.
Temperature, one of meteorological factors that have important influences on livelihood and economy, is limited by mode resolution and calculation capability, and may have a large prediction error. At present, most of forecast products of the air temperature forecast service in a numerical mode adopt a statistical correction method, namely, after linear regression and sliding average processing are carried out on historical temperature data in a medium-short period, a change trend in a near period of the air temperature is found as a correction basis, and then the forecast result is subjected to experience correction by combining with artificial experience. And some methods also adopt machine learning algorithms such as a support vector machine and a random forest to carry out forecast correction. However, in the existing air temperature forecasting technology research, few technologies for comprehensively solving the problems of spatial-temporal resolution improvement and accuracy improvement exist, and the problems are generally solved by adopting independent technical schemes such as scale reduction, forecasting correction and the like, which is one of the important factors leading to slow development of refined weather forecasting.
In the existing weather forecast correction technology, a statistical correction method depends on artificial subjective experience too much, the false alarm rate is greatly improved due to misjudgment, and the statistical correction method cannot adapt to the development of a future high-resolution forecast mode; the machine learning correction method is difficult to train samples with huge data quantity, and noise errors are easily generated for the chaos effect of weather.
Disclosure of Invention
The invention provides an air temperature forecast data correction method based on deep learning, which solves the problems of dependence on manpower and incapability of high-resolution forecast and poor processing capability on a large amount of data in a machine learning process by utilizing the nonlinear mapping capability of a deep learning network and the information extraction capability of lattice data; removing Gaussian noise by adopting a Gaussian filtering smoothing technology in a data preprocessing stage; meanwhile, the time attribute is subjected to feature extraction by adopting the LSTM, and the time attribute is used as the weight of the deep neural network, so that the RMSE (root mean square error) of the temperature forecast can be reduced, and the temperature forecast accuracy can be effectively improved.
In order to achieve the purpose, the invention adopts the following technical scheme:
a deep learning-based air temperature forecast data correction method, as shown in fig. 1, includes the following steps:
s1, acquiring original weather forecast data and historical weather observation data, wherein weather elements included in the original weather forecast data and the historical weather observation data need to have air temperature elements and at least one other weather element characteristic related to the air temperature elements;
s2, preprocessing the data acquired in the step S1, including:
s21, performing missing value processing and abnormal value processing on the historical meteorological observation data;
s22, converting the original weather forecast data into lattice point data with the same spatial resolution as the historical weather observation data by adopting a nearest neighbor interpolation method;
s23, extracting an hour label, and constructing a time data set T ═ h0,h1,h2,…,hnIn which h isiThe hour data of the ith sample is shown, and n is the total amount of the samples;
s24, fusing the characteristics of other meteorological elements except the temperature element in the original meteorological forecast data and the historical meteorological observation data to construct a model characteristic data set X ═ X0,x1,x2,…,xnIn which xiIs a vector, and the dimension is consistent with the quantity of the characteristics of other meteorological elements except the gas temperature element;
s25, fusing the air temperature elements in the original weather forecast data obtained in the step S22 after lattice point formation and the air temperature elements in the historical weather observation data obtained in the step S21, and constructing a model label data set Y ═ Y0,y1,y2,…,ynIn which y isiIs the ith temperature element data;
s26, smoothing the model label data set Y by adopting Gaussian filtering to remove Gaussian noise;
s3, constructing a deep learning network model, which comprises a time feature extraction module, a numerical prediction feature extraction module and a feature fusion module;
the time feature extraction module comprises 1 embedded layer, 3 LSTM layers and 1 reconstruction layer which are sequentially connected and is used for carrying out feature extraction on a time data set T and outputting a group of time feature vectors FT(ii) a Simultaneously performing upsampling on the time dimension of the model characteristic data set X;
the numerical prediction feature extraction module is a UNet network and is used for carrying out numerical prediction feature extraction on the model feature data set X and the model label data set Y; one side of the UNet network is a coding module, the other side of the UNet network is a decoding module, wherein the coding module is used for extracting deep-level features and comprises 4 coding layers, the coding layer network structures are completely the same, but the coding layer network structures are different in parameters and not shared, and each coding layer comprises 3 convolutional layers, 1 normalization layer, 1 activation function layer and 1 maximum pooling layer; the decoding module is used for restoring the high-level features acquired by the encoding module to the initial resolution, and comprises 4 decoding layers, wherein the network structures of the decoding layers are completely the same, but the parameters of the decoding layers are different and are not shared, each decoding layer comprises 1 upsampling layer, one connecting layer, 3 convolutional layers, 2 normalization layers, 2 activation function layers and a full connecting layer, the connecting layer in each decoding layer is connected with the activation function layer in the corresponding encoding layer, and the UNet network outputs a group of numerical prediction feature vectors FP；
The feature fusion module takes the time features as weights and is used for weighting numerical prediction features, namely FT×FPThen extracting the correlation among all the characteristics through 3 full connection layers, and mapping the correlation to an output space;
s4, dividing the time data set, the model characteristic data set and the model label data set into a training set and a testing set, and training the deep learning network model constructed in the step S3: presetting a model training loss function as MSE, a result evaluation index as RMSE, an activation function as relu function, an optimizer as Adam gradient descent method, and optimizing parameters by a Bayesian optimization method according to the number of convolution kernels, the size of the convolution kernels, the learning rate and the embedding dimension; obtaining a trained deep learning network model, namely an air temperature correction model, after training;
and S5, inputting the acquired real-time weather forecast data as input data into an air temperature correction model, wherein the output data of the model is corrected data.
Compared with the prior art, the method has the advantages that in the data preprocessing stage, the spatial resolution is improved while the temperature forecast data is converted into the lattice point data by using the nearest neighbor interpolation method, and the temperature data is smoothed by adopting Gaussian filtering to remove Gaussian noise; in the stage of constructing a deep learning network structure, the time resolution is improved by utilizing the upsampling process, meanwhile, the time characteristic is extracted by adopting the LSTM, the time characteristic and the numerical prediction characteristic extracted by the UNet network are subjected to weighted fusion, and the temperature prediction precision is improved by utilizing the nonlinear mapping capability of the deep learning network and the information extraction capability of lattice data. In conclusion, the air temperature forecast data correction model can calculate a more accurate correction value, can improve the space-time resolution of air temperature forecast, reduces the manpower consumption, and provides a correction service with high resolution and accurate analysis for future refined grid point forecast.
Drawings
FIG. 1 is a schematic diagram of the logic sequence of the present invention;
fig. 2 is a schematic diagram of the UNet network structure of the present invention.
Detailed Description
The scheme of the invention is further described below:
the detailed steps of the invention are as follows:
s1, acquiring original weather forecast data and historical weather observation data, wherein weather elements included in the original weather forecast data and the historical weather observation data need to have air temperature elements and at least one other weather element characteristic related to the air temperature elements;
s2, preprocessing the data acquired in the step S1, including:
s21, performing missing value processing and abnormal value processing on the historical meteorological observation data;
missing value processing: judging whether missing values exist in the atmospheric temperature meteorological elements and other related meteorological features within a preset time period or not, counting the number of the missing values, filling by the average value of the previous and next data when the number of the missing values is smaller than a set value and the previous and next data of the missing values are not empty, and filling by the average value of all data of the corresponding meteorological elements within the preset time period when at least one of the previous and next data of the missing values is empty; when the number of the missing values is larger than or equal to a set value, the meteorological elements are removed;
abnormal value processing: a) checking the climate limit value, and removing the data violating the climate rule; b) extreme value inspection, which eliminates data with extremely low or impossible probability in a certain time period in a certain region; c)3 sigma principle inspection, calculating the average value of all data of each meteorological element in the same space-time range
And standard deviation of
Wherein IiFor the meteorological data at time i, the interval range
The external data is regarded as an abnormal value and is removed; d) and time-varying inspection, which judges according to the time-space resolution of the received meteorological data, wherein common time-varying inspection comprises 1-hour time-varying inspection and 3-hour time-varying inspection, and data exceeding the time-varying upper and lower limit values are eliminated.
S22, converting the original weather forecast data into lattice point data with the same spatial resolution as the historical weather observation data by adopting a nearest neighbor interpolation method;
s23, extracting an hour label, and constructing a time data set T ═ h0，h1，h2，…，hnIn which h isiThe hour data of the ith sample is shown, and n is the total amount of the samples;
s24, fusing the characteristics of other meteorological elements except the temperature element in the original meteorological forecast data and the historical meteorological observation data to construct a model characteristic data set X ═ X0,x1,x2，…，xnIn which xiIs a vector, and the dimension is consistent with the quantity of the characteristics of other meteorological elements except the gas temperature element;
s25, fusing the air temperature elements in the original weather forecast data obtained in the step S22 after lattice point formation and the air temperature elements in the historical weather observation data obtained in the step S21, and constructing a model label data set Y ═ Y0，y1，y2，…，ynIn which y isiIs the ith temperature element data;
s26, smoothing the model label data set Y by adopting Gaussian filtering to remove Gaussian noise;
s3, constructing a deep learning network model, which comprises a time feature extraction module, a numerical prediction feature extraction module and a feature fusion module;
the time feature extraction module comprises 1 embedded layer, 3 LSTM layers and 1 reconstruction layer which are sequentially connected and is used for carrying out feature extraction on a time data set T and outputting a group of time feature vectors FT(ii) a Simultaneously performing upsampling on the time dimension of the model characteristic data set X;
the numerical prediction feature extraction module is a UNet network and is used for carrying out numerical prediction feature extraction on the model feature data set X and the model label data set Y; as shown in fig. 2, one side of the UNet network is an encoding module, and the other side is a decoding module, where the encoding module is used for extracting deep level features, and includes 4 encoding layers, each encoding layer has the same network structure, but different parameters and is not shared, and each encoding layer includes 3 convolutional layers, 1 normalization layer, 1 activation function layer, and 1 activation function layerA maximum pooling layer; the decoding module is used for restoring the high-level features acquired by the encoding module to the initial resolution, and comprises 4 decoding layers, wherein the network structures of the decoding layers are completely the same, but the parameters of the decoding layers are different and are not shared, each decoding layer comprises 1 upsampling layer, one connecting layer, 3 convolutional layers, 2 normalization layers, 2 activation function layers and a full connecting layer, the connecting layer in each decoding layer is connected with the activation function layer in the corresponding encoding layer, and the UNet network outputs a group of numerical prediction feature vectors FP；
The feature fusion module takes the time features as weights and is used for weighting numerical prediction features, namely FT×FPThen extracting the correlation among all the characteristics through 3 full connection layers, and mapping the correlation to an output space;
s4, dividing the time data set, the model characteristic data set and the model label data set into a training set and a testing set, and training the deep learning network model constructed in the step S3: presetting a model training loss function as MSE, a result evaluation index as RMSE, an activation function as relu function, an optimizer as Adam gradient descent method, and optimizing parameters by a Bayesian optimization method according to the number of convolution kernels, the size of the convolution kernels, the learning rate and the embedding dimension; obtaining a trained deep learning network model, namely an air temperature correction model, after training;
and S5, extracting weather forecast data of the real-time numerical weather forecast product as input data and inputting the input data into the air temperature correction model, and taking the corrected air temperature data segment output by the model as a correction result.
Most of the existing air temperature forecast correction technologies are statistical methods, and after the correction is carried out through methods such as linear regression and sliding average, the acquired air temperature change trend is used as a correction basis, and finally, the correction is carried out manually; the invention adopts a deep learning model, thereby reducing the manpower consumption. Machine learning and deep learning are partially applied to correction service, but the machine learning is difficult to train samples with huge data size, noise errors are easy to generate for meteorological chaos effect, the currently used ConvLSTM, MeLC-GRU and other deep models transmit time information mainly through the transformation and combination of state features on the time dimension, although the transmission of the time information can reduce the meteorological chaos effect to a certain extent, the feature mining is insufficient, and the space-time resolution is not improved; according to the invention, the spatial resolution is improved in the data preprocessing stage, and numerical prediction characteristics and time characteristics are respectively extracted by using UNet and LSTM in the deep learning model construction stage and are weighted and fused, so that the prediction precision can be improved, and the prediction space-time resolution can be improved.
Claims (1)
1. A deep learning-based air temperature forecast data correction method is characterized by comprising the following steps:
s1, acquiring original weather forecast data and historical weather observation data, wherein weather elements included in the original weather forecast data and the historical weather observation data need to have air temperature elements and at least one other weather element characteristic related to the air temperature elements;
s2, preprocessing the data acquired in the step S1, including:
s21, performing missing value processing and abnormal value processing on the historical meteorological observation data;
s22, converting the original weather forecast data into lattice point data with the same spatial resolution as the historical weather observation data by adopting a nearest neighbor interpolation method;
s23, extracting an hour label, and constructing a time data set T ═ h0,h1,h2,…,hnIn which h isiThe hour data of the ith sample is shown, and n is the total amount of the samples;
s24, fusing the characteristics of other meteorological elements except the temperature element in the original meteorological forecast data and the historical meteorological observation data to construct a model characteristic data set X ═ X0,x1,x2,…,xnIn which xiIs a vector, and the dimension is consistent with the quantity of the characteristics of other meteorological elements except the gas temperature element;
s25, fusing the air temperature elements in the original weather forecast data obtained by the lattice point step S22 and the air temperature elements in the historical weather observation data obtained by the step S21Building model tag data set Y ═ Y0,y1,y2,…,ynIn which y isiIs the ith temperature element data;
s26, smoothing the model label data set Y by adopting Gaussian filtering to remove Gaussian noise;
s3, constructing a deep learning network model, which comprises a time feature extraction module, a numerical prediction feature extraction module and a feature fusion module;
the time feature extraction module comprises 1 embedded layer, 3 LSTM layers and 1 reconstruction layer which are sequentially connected and is used for carrying out feature extraction on a time data set T and outputting a group of time feature vectors FT(ii) a Simultaneously performing upsampling on the time dimension of the model characteristic data set X;
the numerical prediction feature extraction module is a UNet network and is used for carrying out numerical prediction feature extraction on the model feature data set X and the model label data set Y; one side of the UNet network is a coding module, the other side of the UNet network is a decoding module, wherein the coding module is used for extracting deep-level features and comprises 4 coding layers, the coding layer network structures are completely the same, but the coding layer network structures are different in parameters and not shared, and each coding layer comprises 3 convolutional layers, 1 normalization layer, 1 activation function layer and 1 maximum pooling layer; the decoding module is used for restoring the high-level features acquired by the encoding module to the initial resolution, and comprises 4 decoding layers, wherein the network structures of the decoding layers are completely the same, but the parameters of the decoding layers are different and are not shared, each decoding layer comprises 1 upsampling layer, one connecting layer, 3 convolutional layers, 2 normalization layers, 2 activation function layers and a full connecting layer, the connecting layer in each decoding layer is connected with the activation function layer in the corresponding encoding layer, and the UNet network outputs a group of numerical prediction feature vectors FP；
The feature fusion module takes the time features as weights and is used for weighting numerical prediction features, namely FT×FPThen extracting the correlation among all the characteristics through 3 full connection layers, and mapping the correlation to an output space;
s4, dividing the time data set, the model characteristic data set and the model label data set into a training set and a testing set, and training the deep learning network model constructed in the step S3: presetting a model training loss function as MSE, a result evaluation index as RMSE, an activation function as relu function, an optimizer as Adam gradient descent method, and optimizing parameters by a Bayesian optimization method according to the number of convolution kernels, the size of the convolution kernels, the learning rate and the embedding dimension; obtaining a trained deep learning network model, namely an air temperature correction model, after training;
and S5, inputting the acquired real-time weather forecast data as input data into an air temperature correction model, wherein the output data of the model is corrected data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111084626.1A CN113807432B (en) | 2021-09-16 | 2021-09-16 | Air temperature forecast data correction method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111084626.1A CN113807432B (en) | 2021-09-16 | 2021-09-16 | Air temperature forecast data correction method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113807432A true CN113807432A (en) | 2021-12-17 |
| CN113807432B CN113807432B (en) | 2024-04-30 |
Family
ID=78895499
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111084626.1A Active CN113807432B (en) | 2021-09-16 | 2021-09-16 | Air temperature forecast data correction method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113807432B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114385600A (en) * | 2022-03-23 | 2022-04-22 | 北京弘象科技有限公司 | Downscaling correction method and device for ECMWF temperature data |
| CN114463947A (en) * | 2022-01-17 | 2022-05-10 | 南京气象科技创新研究院 | Convective disaster-causing strong wind early warning and forecasting method based on space-time network convolution model |
| CN114611608A (en) * | 2022-03-14 | 2022-06-10 | 中国人民解放军国防科技大学 | Sea surface height numerical value prediction deviation correction method based on deep learning model |
| CN114742179A (en) * | 2022-06-13 | 2022-07-12 | 南京信息工程大学 | Grid point forecast deviation correction method based on ECMWF |
| CN114936620A (en) * | 2022-03-04 | 2022-08-23 | 中国人民解放军国防科技大学 | Sea surface temperature numerical value forecast deviation correction method based on attention mechanism |
| CN115796000A (en) * | 2022-10-26 | 2023-03-14 | 中国水利水电科学研究院 | Short-term air temperature forecast set correction method based on stacked machine learning algorithm |
| CN116070676A (en) * | 2023-03-28 | 2023-05-05 | 南京气象科技创新研究院 | Expressway road surface temperature forecasting method based on attention mechanism and self-encoder |
| CN116432702A (en) * | 2023-06-09 | 2023-07-14 | 南京信息工程大学 | Deep learning NFC-Net network model for ECMWF forecast product correction |
| CN116484995A (en) * | 2022-09-19 | 2023-07-25 | 北京城市气象研究院 | Method and system for rapidly forecasting concentration of atmospheric aerosol in area |
| CN116992249A (en) * | 2023-09-28 | 2023-11-03 | 南京信息工程大学 | Grid point forecast deviation correction method based on FMCNN-LSTM |
| CN118586896A (en) * | 2024-08-06 | 2024-09-03 | 中铁桥隧技术有限公司 | Bridge maintenance management system and method based on BIM technology |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180204111A1 (en) * | 2013-02-28 | 2018-07-19 | Z Advanced Computing, Inc. | System and Method for Extremely Efficient Image and Pattern Recognition and Artificial Intelligence Platform |
| CN109325624A (en) * | 2018-09-28 | 2019-02-12 | 国网福建省电力有限公司 | A kind of monthly electric power demand forecasting method based on deep learning |
| CN110309603A (en) * | 2019-07-05 | 2019-10-08 | 华北电力大学（保定） | A kind of short-term wind speed forecasting method and system based on wind speed characteristics |
| CN111339092A (en) * | 2020-02-24 | 2020-06-26 | 江苏省南通环境监测中心 | Deep learning-based multi-scale air quality forecasting method |
| CN112232561A (en) * | 2020-10-13 | 2021-01-15 | 三峡大学 | Power load probability prediction method based on constrained parallel LSTM quantile regression |
| CN112748483A (en) * | 2020-12-24 | 2021-05-04 | 北京思湃德信息技术有限公司 | Air temperature forecast deviation correction method and device based on deep learning |
-
2021
- 2021-09-16 CN CN202111084626.1A patent/CN113807432B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180204111A1 (en) * | 2013-02-28 | 2018-07-19 | Z Advanced Computing, Inc. | System and Method for Extremely Efficient Image and Pattern Recognition and Artificial Intelligence Platform |
| CN109325624A (en) * | 2018-09-28 | 2019-02-12 | 国网福建省电力有限公司 | A kind of monthly electric power demand forecasting method based on deep learning |
| CN110309603A (en) * | 2019-07-05 | 2019-10-08 | 华北电力大学（保定） | A kind of short-term wind speed forecasting method and system based on wind speed characteristics |
| CN111339092A (en) * | 2020-02-24 | 2020-06-26 | 江苏省南通环境监测中心 | Deep learning-based multi-scale air quality forecasting method |
| CN112232561A (en) * | 2020-10-13 | 2021-01-15 | 三峡大学 | Power load probability prediction method based on constrained parallel LSTM quantile regression |
| CN112748483A (en) * | 2020-12-24 | 2021-05-04 | 北京思湃德信息技术有限公司 | Air temperature forecast deviation correction method and device based on deep learning |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114463947A (en) * | 2022-01-17 | 2022-05-10 | 南京气象科技创新研究院 | Convective disaster-causing strong wind early warning and forecasting method based on space-time network convolution model |
| CN114463947B (en) * | 2022-01-17 | 2022-11-18 | 南京气象科技创新研究院 | Convective disaster-causing strong wind early warning and forecasting method based on space-time network convolution model |
| CN114936620A (en) * | 2022-03-04 | 2022-08-23 | 中国人民解放军国防科技大学 | Sea surface temperature numerical value forecast deviation correction method based on attention mechanism |
| CN114936620B (en) * | 2022-03-04 | 2024-05-03 | 中国人民解放军国防科技大学 | Sea surface temperature numerical forecasting deviation correcting method based on attention mechanism |
| CN114611608A (en) * | 2022-03-14 | 2022-06-10 | 中国人民解放军国防科技大学 | Sea surface height numerical value prediction deviation correction method based on deep learning model |
| CN114611608B (en) * | 2022-03-14 | 2024-04-09 | 中国人民解放军国防科技大学 | Sea surface height numerical value forecast deviation correction method based on deep learning model |
| CN114385600A (en) * | 2022-03-23 | 2022-04-22 | 北京弘象科技有限公司 | Downscaling correction method and device for ECMWF temperature data |
| CN114385600B (en) * | 2022-03-23 | 2022-07-19 | 北京弘象科技有限公司 | Downscaling correction method and device for ECMWF temperature data |
| CN114742179A (en) * | 2022-06-13 | 2022-07-12 | 南京信息工程大学 | Grid point forecast deviation correction method based on ECMWF |
| CN114742179B (en) * | 2022-06-13 | 2022-09-02 | 南京信息工程大学 | Grid point forecast deviation correction method based on ECMWF |
| CN116484995B (en) * | 2022-09-19 | 2023-10-27 | 北京城市气象研究院 | Method and system for rapidly forecasting concentration of atmospheric aerosol in area |
| CN116484995A (en) * | 2022-09-19 | 2023-07-25 | 北京城市气象研究院 | Method and system for rapidly forecasting concentration of atmospheric aerosol in area |
| CN115796000B (en) * | 2022-10-26 | 2023-06-20 | 中国水利水电科学研究院 | Short-term air temperature forecast set correction method based on stacked machine learning algorithm |
| CN115796000A (en) * | 2022-10-26 | 2023-03-14 | 中国水利水电科学研究院 | Short-term air temperature forecast set correction method based on stacked machine learning algorithm |
| CN116070676B (en) * | 2023-03-28 | 2023-06-30 | 南京气象科技创新研究院 | Expressway road surface temperature forecasting method based on attention mechanism and self-encoder |
| CN116070676A (en) * | 2023-03-28 | 2023-05-05 | 南京气象科技创新研究院 | Expressway road surface temperature forecasting method based on attention mechanism and self-encoder |
| CN116432702A (en) * | 2023-06-09 | 2023-07-14 | 南京信息工程大学 | Deep learning NFC-Net network model for ECMWF forecast product correction |
| CN116432702B (en) * | 2023-06-09 | 2023-08-29 | 南京信息工程大学 | Deep learning NFC-Net network model for ECMWF forecast product correction |
| CN116992249A (en) * | 2023-09-28 | 2023-11-03 | 南京信息工程大学 | Grid point forecast deviation correction method based on FMCNN-LSTM |
| CN116992249B (en) * | 2023-09-28 | 2024-01-23 | 南京信息工程大学 | Grid point forecast deviation correction method based on FMCNN-LSTM |
| CN118586896A (en) * | 2024-08-06 | 2024-09-03 | 中铁桥隧技术有限公司 | Bridge maintenance management system and method based on BIM technology |
| CN118586896B (en) * | 2024-08-06 | 2024-10-18 | 中铁桥隧技术有限公司 | Bridge maintenance management system and method based on BIM technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113807432B (en) | 2024-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113807432B (en) | Air temperature forecast data correction method based on deep learning | |
| CN116757534B (en) | Intelligent refrigerator reliability analysis method based on neural training network | |
| CN113723010B (en) | Bridge damage early warning method based on LSTM temperature-displacement correlation model | |
| CN108009674A (en) | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks | |
| CN111160628A (en) | Air pollutant concentration prediction method based on CNN and double-attention seq2seq | |
| CN111160620B (en) | Short-term wind power prediction method based on end-to-end memory network | |
| CN112712209B (en) | Reservoir warehousing flow prediction method and device, computer equipment and storage medium | |
| CN111401599B (en) | Water level prediction method based on similarity search and LSTM neural network | |
| CN111523706B (en) | Section lane-level short-term traffic flow prediction method based on deep learning combination model | |
| CN111242351A (en) | Tropical cyclone track prediction method based on self-encoder and GRU neural network | |
| CN111178585A (en) | Fault reporting amount prediction method based on multi-algorithm model fusion | |
| CN117494034A (en) | Air quality prediction method based on traffic congestion index and multi-source data fusion | |
| CN116128141B (en) | Storm surge prediction method and device, storage medium and electronic equipment | |
| CN112183897A (en) | Long-time prediction method for icing thickness of overhead transmission line based on deep learning | |
| CN116401516A (en) | Power load abnormal data detection and correction method based on deep learning | |
| CN113033903A (en) | Fruit price prediction method, medium and equipment of LSTM model and seq2seq model | |
| CN114357670A (en) | Power distribution network power consumption data abnormity early warning method based on BLS and self-encoder | |
| CN116702090A (en) | Multi-mode data fusion and uncertain estimation water level prediction method and system | |
| CN110260914B (en) | Engineering safety monitoring system region division method based on measuring point space-time characteristics | |
| CN113283546B (en) | Furnace condition abnormity alarm method and system of heating furnace integrity management centralized control device | |
| CN114692981A (en) | Medium-and-long-term runoff forecasting method and system based on Seq2Seq model | |
| CN117909888A (en) | Intelligent artificial intelligence climate prediction method | |
| CN117578441A (en) | Method for improving power grid load prediction precision based on neural network | |
| CN116525135B (en) | Method for predicting epidemic situation development situation by space-time model based on meteorological factors | |
| CN117290673A (en) | Ship energy consumption high-precision prediction system based on multi-model fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |