CN116702926A - Air quality mode forecasting machine learning integrated correction method - Google Patents
Air quality mode forecasting machine learning integrated correction method Download PDFInfo
- Publication number
- CN116702926A CN116702926A CN202310540962.5A CN202310540962A CN116702926A CN 116702926 A CN116702926 A CN 116702926A CN 202310540962 A CN202310540962 A CN 202310540962A CN 116702926 A CN116702926 A CN 116702926A
- Authority
- CN
- China
- Prior art keywords
- data
- air quality
- forecast
- historical
- live
- 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.)
- Pending
Links
- 238000012937 correction Methods 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000010801 machine learning Methods 0.000 title claims abstract description 18
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000007781 pre-processing Methods 0.000 claims abstract description 10
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 238000010276 construction Methods 0.000 claims abstract description 5
- 238000012795 verification Methods 0.000 claims abstract description 4
- 238000011156 evaluation Methods 0.000 claims description 14
- 239000003344 environmental pollutant Substances 0.000 claims description 8
- 231100000719 pollutant Toxicity 0.000 claims description 8
- 230000010354 integration Effects 0.000 claims description 7
- 238000005516 engineering process Methods 0.000 claims description 5
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 238000009795 derivation Methods 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 2
- 230000000875 corresponding effect Effects 0.000 description 9
- 238000013507 mapping Methods 0.000 description 3
- 238000003915 air pollution Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000003203 everyday effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- 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)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an air quality mode forecasting machine learning integrated correction method, which comprises the following steps of S1, acquiring historical air quality forecasting data, historical weather forecasting data, historical air quality live data and historical weather live data; s2, preprocessing and characteristic construction are carried out on historical air quality forecast data, historical weather forecast data, historical air quality live data and historical weather live data to obtain preprocessing data and characteristic values, and the processed data are divided into a training set, a testing machine and a verification set according to the proportion of 7:2:1; s3, constructing a correction model based on Catboost by utilizing the historical air quality forecast data and the historical air quality live data and the historical weather forecast data and the historical weather live data; s4, based on the Catboost prediction model, inputting the predicted air quality characteristics and the predicted weather quality characteristics, and obtaining a corrected prediction result. The invention has high forecasting precision.
Description
Technical Field
The invention belongs to the technical field of intersection of computer science and environmental science, and particularly relates to an air quality mode forecasting machine learning integration correction method.
Background
In recent years, due to rapid urban and industrialized development, the problem of air pollution is increasingly serious, and environmental protection departments put higher demands on the forecasting and management of air pollution, so that more accurate and fine forecasting of future air quality is expected. With the development and popularization of technology, more and more numerical mode technologies are beginning to be applied to air quality forecasting, and WRF-Chem, CMAQ, CAMx and the like are typical modes.
When the air quality numerical mode is practically applied, the numerical mode is found to be based on the physical and chemical principles of movement and change of the atmospheric components, and the change trend of the concentration of the atmospheric pollutants in a future period of time can be predicted to a certain extent. However, due to the complexity and error accumulation of the atmospheric system, there is a large gap between the model forecast result and the corresponding actual atmospheric pollution concentration. Therefore, aiming at the forecasting result of the numerical mode, the method is combined with various effective methods to correct so as to improve the forecasting level, and has important display significance.
Aiming at the correction problem of air quality mode forecast, a plurality of researchers apply the front technology to conduct quantitative research in recent years, and important progress is made. Chen Lei and the like evaluate and correct the analysis of air quality forecast in Ningbo region based on CUACE mode; zhang and the like correct the air quality numerical value prediction result by using a set deep learning method, and perform PM2.5 error correction on the Xinjiang Wu Chang Shicheng city group; the Chinese zodiac and the like are coupled based on a machine learning algorithm for air quality numerical prediction correction, and the air quality model prediction results of four conventional pollutants are corrected; the air quality prediction PM2.5 was corrected in the Chongqing region based on machine learning.
The problem of correction of air quality mode forecast is still fresh at present, and the situation of forecasting six conventional pollutants based on the long-term air quality numerical mode forecast result is still fresh, or the relatively stable forecast quality improvement is obtained in medium-and-long-term forecast.
Disclosure of Invention
The invention aims to provide an air quality model forecasting machine learning integrated correction method which can improve the forecasting precision of a numerical forecasting model.
In order to achieve the above object, the technical scheme of the present invention is as follows:
an air quality model forecasting machine learning integrated correction method comprises the following steps:
s1, acquiring historical air quality forecast data, historical weather forecast data, historical air quality live data and historical weather live data;
s2, preprocessing and characteristic construction are carried out on historical air quality forecast data, historical weather forecast data, historical air quality live data and historical weather live data to obtain preprocessing data and characteristic values, and the processed data are divided into a training set, a testing machine and a verification set according to the proportion of 7:2:1;
s3, constructing a correction model based on Catboost by utilizing the historical air quality forecast data and the historical air quality live data and the historical weather forecast data and the historical weather live data;
s4, based on the Catboost prediction model, inputting the predicted air quality characteristics and the predicted weather quality characteristics, and obtaining a corrected prediction result.
As an improvement to the technical proposal, the elements of the air quality are PM2.5, PM10, O3, NO2, CO and SO2; the elements of weather are air temperature, air pressure, humidity, wind direction and wind speed.
As an improvement to the above technical solution, the air quality prediction data includes a future 7-day prediction result reported by the air quality mode at the Beijing time of 20, and outputs six types of pollutant concentration data in total, namely PM2.5, PM10, O3, NO2, CO, SO2; the weather forecast data comprises the forecast result of 7 days in future, which is reported by a weather mode at the Beijing time of 20, and comprises air temperature, air pressure, humidity, wind direction and wind speed.
As an improvement to the above solution, the air quality live data and the weather live data include quality-controlled site monitoring data and comprehensive live data; wherein the air quality live data output six types of pollutant concentration data in total, namely PM2.5, PM10, O3, NO2, CO and SO2; the weather live data includes five types of data, namely air temperature, air pressure, humidity, wind direction and wind speed.
As an improvement of the above technical solution, in the step S2, the method for preprocessing the history data is:
s201, firstly, dividing forecast data into 1-24, 25-48, 49-72, 73-96, 97-120, 121-144, 145-168, 169-192, 193-216, 217-240 according to forecast time, respectively corresponding to each forecast day, combining all data of the same forecast day into a data set, and then combining air quality forecast data and air image forecast data according to a time consistency principle;
s201, forecasting data of stations, splitting each station into a data set of each forecasting day according to forecasting time, and then merging all station data according to time to form a wide table, wherein fields of the wide table are forecasting elements of each station and longitude and latitude of the station;
s203, combining the forecast data and the site forecast data according to a time consistency principle, and sorting each field name without repeated data;
s204, after the operation is finished, forming a complete data set for each forecast day, and creating 10 data sets in total;
s205, merging 10 data sets with live data according to time respectively.
As an improvement to the above technical solution, in the step S2, the feature mining method is as follows:
s206, firstly, carrying out correlation analysis on the data set, removing the characteristic with weak correlation, and removing the characteristic with the absolute value of correlation smaller than or equal to 0.1;
s207, creating new features by using a feature derivation technology; the new features include combinations of the original features, intersections of the original features; the feature combination is to perform arithmetic operation on the features; the feature intersection is to perform intersection combination on a plurality of features, and perform intersection and Cartesian product compensation operation;
s208, performing time sliding operation on the data, performing hysteresis operation on the forecast data, and obtaining time corresponding to the live time to obtain the hysteresis characteristics of the live elements 。
As an improvement of the above technical solution, in the step S4, the method of prediction and evaluation is:
using the test set data, and adopting the root mean square error as an evaluation index, wherein the formula of the root mean square error is as follows:
wherein ,representing predicted value, y i Represents an observed value, n represents the number of observations required for evaluation;
at the same time use R 2 As an auxiliary evaluation index, the formula is as follows:
wherein Representing predicted value, y i Representing observations->Represents the arithmetic mean of the observations, n represents the number of observations required for evaluation.
Compared with the prior art, the invention has the advantages and positive effects that:
1. according to the invention, on the correction model, the historical air quality forecast data, the historical weather forecast data, the historical air quality live data and the historical weather live data are fused, the characteristic engineering is carried out, the historical live data is fully utilized, and reasonable prediction can be made for future trends. 2. The invention can output the time-by-time prediction correction result which is up to 240 hours, and the prediction precision is higher than the original air quality numerical mode and the original meteorological numerical mode.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of an algorithm of the present invention;
FIG. 2 is a plot of numerical pattern forecast data versus live data;
fig. 3 is a graph comparing the forecast data after correction with the live data.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention needs to correct the air quality and weather forecast data, so that a data set required by the forecast correction is firstly created. In general, it is necessary to take air quality forecast data and air quality live data, weather forecast data and weather live data for a long period of time as training data. The historical live and historical forecast data is then correlated in time. Since the forecast data is usually reported at 20 hours in the evening, and the air quality condition of a future period is forecasted, different forecast timepieces of different reporting time can correspond to the same time, and the confusion of variable and label mapping can be caused. In order to solve the problem, the forecast data is generally divided into separate data sets according to the number of forecast days, each data set only comprises data of a certain forecast day, according to the method, the forecast data set which is the same as the number of forecast days is manufactured, and then the live data and the forecast data are combined together according to the corresponding live time of the forecast time to form a complete live forecast mapping data set. And then, carrying out feature engineering on the live forecast mapping data set, adding space-time influencing features such as distinguishing between workdays and weekends, holiday features, important activity features and the like, and screening out features suitable for being put into a machine learning model. And then, establishing a machine learning model based on the characteristic data set, selecting a certain air quality element (such as O3) as a model target, and training the model to obtain an optimal predicted value.
As shown in fig. 1, the specific steps are as follows:
step one, acquiring historical air quality forecast data and historical weather forecast data, wherein the historical air quality forecast data and the historical weather forecast data comprise hour-by-hour PM2.5, PM10, NO2, CO, O3, SO2, air pressure, air temperature, wind direction, wind speed, humidity and the like which are reported at 20 points every day. Wherein, the air quality element is ug/m3 except the CO of mg/m 3; wherein the air pressure unit: hundred pascals (hPa), humidity unit: percent (%), wind direction unit: starting clockwise angle in north, wind speed unit: m/s.
Step two, preprocessing the historical data
And 2.1, dividing forecast data. Firstly, dividing forecast data into 1-24, 25-48, 49-72, 73-96, 97-120, 121-144, 145-168, 169-192, 193-216 and 217-240 according to forecast time, respectively corresponding to each forecast day, combining all data of the same forecast day into a data set, and then combining urban air quality forecast and urban weather forecast data according to a time consistency principle. And dividing the station forecast data into a data set of each forecast day according to forecast time for each station, and then merging all station data according to time to form a wide table, wherein the fields of the wide table are forecast elements of each station, longitude and latitude of the station and the like. And combining the city forecast data and the site forecast data according to the time consistency principle, and sorting each field name without repeated data. After the above operations, one complete data set is constructed for each forecast day, creating a total of 10 data sets. The 10 data sets were then each combined with the live data over time.
2.2, feature construction. Firstly, carrying out correlation analysis on a data set, removing the characteristic of weak correlation, and removing the characteristic that the absolute value of the correlation is smaller than or equal to 0.1. Next, new features are created using some feature derivation techniques. If the characteristics are combined, carrying out arithmetic operation on the characteristics; and (3) feature crossing, namely performing crossing combination on a plurality of features, such as performing operations of cross-union and Cartesian product compensation.
Besides the method for creating the characteristics, the time sliding operation can be performed on the data, the hysteresis operation is performed on the forecast data, and the obtained time corresponds to the live time, so that the live factor hysteresis characteristics are obtained.
Step three, constructing a correction model based on Catboost
CatBOOST is a Yandex open-source machine learning algorithm. It can be easily integrated with a deep learning framework. It can handle a variety of data types. The CatBOOST is a GBDT framework which is based on a symmetrical decision tree (oblivious tree) and is realized by a learner, fewer in parameters, supports category variables and high accuracy, and solves the problems of Gradient Bias (Gradient Bias) and Prediction Bias (Prediction shift), so that the occurrence of overfitting is reduced, and the accuracy and generalization capability of an algorithm are improved. The training set, the test set and the validation set are divided according to a ratio of 7:2:1 by taking the forecast characteristic as an independent variable and taking the live data as the dependent variable. And then putting the data into model training, and adjusting model parameters including the maximum decision number, the learning rate, the maximum tree depth, the maximum leaf tree and the like, and training until the model effect reaches the optimal.
Step four, prediction and assessment
In order to evaluate the prediction result, the root mean square error is used as an evaluation index by using the test set data, and the formula of the root mean square error is as follows:
wherein ,representing predicted value, y i Represents the observed value, and n represents the number of observations required for evaluation.
At the same time use R 2 As an auxiliary evaluation index, the formula is as follows:
wherein Representing predicted value, y i Representing observations->Represents the arithmetic mean of the observations, n represents the number of observations required for evaluation.
Example 1
In order to better explain the technical scheme of the invention, the invention selects 6 air quality monitoring station data, WRF mode forecast meteorological data and CAMx mode forecast air quality data in the Yibin city from 2021 month 1 to 2022, and the specific implementation steps are as follows:
step 1: historical data is obtained.
Taking WRF mode data and CAMx mode data which are continuously operated for more than one year, wherein the WRF mode data and the CAMx mode data comprise future 240-hour weather forecast data and air quality factor forecast data which are reported at 20 points every day, namely temperature, air pressure, humidity, wind speed, wind direction and the like every hour, and the temperature units are as follows: DEG C, barometric unit: hPa, humidity unit: % wind direction unit: the wind speed unit is manufactured by the angle of clockwise right north: m/s; PM2.5 (ug/m 3), PM10 (ug/m 3), NO2 (ug/m 3), CO (mg/m 3), O3 (ug/m 3), SO2 (ug/m 3), AQI, etc. per hour. And taking air quality forecast data corresponding to the air quality monitoring stations in 6 areas of Yibin city.
The history live data of Yibin city corresponding to the mode forecast time is taken, and the history live data comprises PM2.5 (ug/m 3), PM10 (ug/m 3), NO2 (ug/m 3), CO (mg/m 3), O3 (ug/m 3), SO2 (ug/m 3), AQI, primary pollutants and the like.
Step 2: data preprocessing
Dividing the forecast data into 10 parts according to the number of forecast days, transversely combining each part of data according to the forecast time corresponding principle, and combining the data after the combination with the historical live data in a time corresponding way. And then, the data are arranged, and feature construction is carried out.
1. Arithmetic operation features. PM2.5 is part of PM10, PM2.5/PM10 can be calculated for city data and site data; and carrying out logarithmic operation on the AQI.
2. And (5) spatial feature operation. And taking the site position, and calculating the distance between the site position and the position of the center of Yibin city as a space influence characteristic.
3. Feature normalization. And (3) normalizing each feature by using a mean value normalization method, and removing dimension influence. The normalized formula is:where μ represents the sample mean and s represents the sample standard deviation.
4. The processed data are divided into a training set, a testing machine and a verification set, and the ratio is 7:2:1.
Step 3, constructing a single-element correction model based on Catboost
The invention selects Catboost as a training frame to construct a single-element correction model. The internal structure is a gradient rise based on a symmetrical tree. The training parameter settings for the model are shown in table 1:
table 1 model training parameters
Training the training set by taking PM2.5 as a target variable according to the above super parameters, wherein the root mean square error obtained on the testing set is 14.2ug/m < 3 >. And the prediction error of the original mode is 35.6ug/m < 3 >. The correction models of the other five elements were trained using the same operations.
As can be seen from the comparison of fig. 2 and fig. 3, 1, the invention fuses the historical air quality forecast data, the historical weather forecast data, the historical air quality live data and the historical weather live data on the correction model, performs characteristic engineering, fully utilizes the historical live data, and can reasonably predict future trends. 2. The invention can output the time-by-time prediction correction result which is up to 240 hours, and the prediction precision is higher than the original air quality numerical mode and the original meteorological numerical mode.
All other embodiments, modifications, equivalents, improvements, etc., which are apparent to those skilled in the art without the benefit of this disclosure, are intended to be included within the scope of this invention.
Claims (7)
1. An air quality model forecast machine learning integrated correction method is characterized in that: the method comprises the following steps:
s1, acquiring historical air quality forecast data, historical weather forecast data, historical air quality live data and historical weather live data;
s2, preprocessing and characteristic construction are carried out on historical air quality forecast data, historical weather forecast data, historical air quality live data and historical weather live data to obtain preprocessing data and characteristic values, and the processed data are divided into a training set, a testing machine and a verification set according to the proportion of 7:2:1;
s3, constructing a correction model based on Catboost by utilizing the historical air quality forecast data and the historical air quality live data and the historical weather forecast data and the historical weather live data;
s4, based on the Catboost prediction model, inputting the predicted air quality characteristics and the predicted weather quality characteristics, and obtaining a corrected prediction result.
2. The air quality model predictive machine learning integration correction method of claim 1, wherein: the air quality factors are PM2.5, PM10, O3, NO2, CO and SO2; the elements of weather are air temperature, air pressure, humidity, wind direction and wind speed.
3. The air quality model predictive machine learning integration correction method of claim 1, wherein: the air quality forecast data comprises a future 7-day forecast result which is reported by the air quality mode at the Beijing time of 20, and outputs six types of pollutant concentration data in total, namely PM2.5, PM10, O3, NO2, CO and SO2; the weather forecast data comprises the forecast result of 7 days in future, which is reported by a weather mode at the Beijing time of 20, and comprises air temperature, air pressure, humidity, wind direction and wind speed.
4. The air quality model predictive machine learning integration correction method of claim 1, wherein: the air quality live data and the weather live data comprise quality-controlled site monitoring data and comprehensive live data; wherein the air quality live data output six types of pollutant concentration data in total, namely PM2.5, PM10, O3, NO2, CO and SO2; the weather live data includes five types of data, namely air temperature, air pressure, humidity, wind direction and wind speed.
5. The air quality model predictive machine learning integration correction method of claim 1, wherein: in the step S2, the method for preprocessing the history data is as follows:
s201, firstly, dividing forecast data into 1-24, 25-48, 49-72, 73-96, 97-120, 121-144, 145-168, 169-192, 193-216, 217-240 according to forecast time, respectively corresponding to each forecast day, combining all data of the same forecast day into a data set, and then combining air quality forecast data and air image forecast data according to a time consistency principle;
s201, forecasting data of stations, splitting each station into a data set of each forecasting day according to forecasting time, and then merging all station data according to time to form a wide table, wherein fields of the wide table are forecasting elements of each station and longitude and latitude of the station;
s203, combining the forecast data and the site forecast data according to a time consistency principle, and sorting each field name without repeated data;
s204, after the operation is finished, forming a complete data set for each forecast day, and creating 10 data sets in total;
s205, merging 10 data sets with live data according to time respectively.
6. The air quality model predictive machine learning integration correction method of claim 1, wherein: in the step S2, the feature mining method is as follows:
s206, firstly, carrying out correlation analysis on the data set, removing the characteristic with weak correlation, and removing the characteristic with the absolute value of correlation smaller than or equal to 0.1;
s207, creating new features by using a feature derivation technology; the new features include combinations of the original features, intersections of the original features; the feature combination is to perform arithmetic operation on the features; the feature intersection is to perform intersection combination on a plurality of features, and perform intersection and Cartesian product compensation operation;
s208, performing time sliding operation on the data, performing hysteresis operation on the forecast data, and obtaining time corresponding to the live time to obtain the hysteresis characteristics of the live elements 。
7. The air quality model predictive machine learning integration correction method of claim 1, wherein: in the step S4, the method of prediction and evaluation is as follows:
using the test set data, and adopting the root mean square error as an evaluation index, wherein the formula of the root mean square error is as follows:
wherein ,representing predicted value, y i Represents an observed value, n represents the number of observations required for evaluation;
at the same time use R 2 As an auxiliary evaluation index, the formula is as follows:
wherein Representing predicted value, y i Representing observations->Represents the arithmetic mean of the observations, n represents the number of observations required for evaluation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310540962.5A CN116702926A (en) | 2023-05-15 | 2023-05-15 | Air quality mode forecasting machine learning integrated correction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310540962.5A CN116702926A (en) | 2023-05-15 | 2023-05-15 | Air quality mode forecasting machine learning integrated correction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116702926A true CN116702926A (en) | 2023-09-05 |
Family
ID=87836465
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310540962.5A Pending CN116702926A (en) | 2023-05-15 | 2023-05-15 | Air quality mode forecasting machine learning integrated correction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116702926A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117290792A (en) * | 2023-11-14 | 2023-12-26 | 广东省气象服务中心(广东气象影视宣传中心) | Air pressure forecasting system and method based on machine learning |
-
2023
- 2023-05-15 CN CN202310540962.5A patent/CN116702926A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117290792A (en) * | 2023-11-14 | 2023-12-26 | 广东省气象服务中心(广东气象影视宣传中心) | Air pressure forecasting system and method based on machine learning |
CN117290792B (en) * | 2023-11-14 | 2024-05-28 | 广东省气象服务中心(广东气象影视宣传中心) | Air pressure forecasting system and method based on machine learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108491970B (en) | Atmospheric pollutant concentration prediction method based on RBF neural network | |
CN112465243B (en) | Air quality forecasting method and system | |
CN109116444B (en) | PCA-kNN-based air quality model PM2.5Forecasting method | |
CN112288164B (en) | Wind power combined prediction method considering spatial correlation and correcting numerical weather forecast | |
CN116630122B (en) | Lake ecological hydraulic regulation and control method and system based on hydrologic-ecological response relation | |
CN109508818B (en) | Online NOx prediction method based on LSSVM | |
CN101853290A (en) | Meteorological service performance evaluation method based on geographical information system (GIS) | |
CN114254802B (en) | Prediction method for vegetation coverage space-time change under climate change drive | |
CN116702926A (en) | Air quality mode forecasting machine learning integrated correction method | |
CN113743013A (en) | XGboost-based temperature prediction data correction method | |
CN112287299A (en) | River health change quantitative attribution method, device and system | |
JP5110891B2 (en) | Statistical prediction method and apparatus for influent water quality in water treatment facilities | |
CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
CN117078114B (en) | Water quality evaluation method and system for water-bearing lakes under influence of diversion engineering | |
CN110648023A (en) | Method for establishing data prediction model based on quadratic exponential smoothing improved GM (1,1) | |
CN112101612A (en) | Prediction method of town gas usage, storage medium and terminal equipment | |
CN114741972A (en) | Construction method of seasonal prediction model of air pollutant concentration | |
CN116050475A (en) | Training method and device for key pollutant concentration prediction model and computer equipment | |
CN115759469A (en) | Method and device for predicting peak load and occurrence time thereof | |
CN115526330A (en) | Organic matter navigation data calibration method and device, computer equipment and storage medium | |
CN114154686A (en) | Dam deformation prediction method based on ensemble learning | |
CN117909888B (en) | Intelligent artificial intelligence climate prediction method | |
CN114811858B (en) | Online learning method for air conditioner load | |
CN116128049B (en) | XGBoost model-based migration condition selection method for water quality prediction model | |
CN113408770B (en) | Equipment maintenance time prediction method based on deep learning |
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 |