CN116663608A - Extreme drought accurate prediction method and system - Google Patents

Extreme drought accurate prediction method and system Download PDF

Info

Publication number
CN116663608A
CN116663608A CN202310666885.8A CN202310666885A CN116663608A CN 116663608 A CN116663608 A CN 116663608A CN 202310666885 A CN202310666885 A CN 202310666885A CN 116663608 A CN116663608 A CN 116663608A
Authority
CN
China
Prior art keywords
drought
spi
training
model
prediction
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
Application number
CN202310666885.8A
Other languages
Chinese (zh)
Inventor
喻国强
田冉
丁严
俞斌
连丞龙
胡振泽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo Yinzhou District Planning And Design Institute
Original Assignee
Ningbo Yinzhou District Planning And Design Institute
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ningbo Yinzhou District Planning And Design Institute filed Critical Ningbo Yinzhou District Planning And Design Institute
Priority to CN202310666885.8A priority Critical patent/CN116663608A/en
Publication of CN116663608A publication Critical patent/CN116663608A/en
Pending legal-status Critical Current

Links

Classifications

    • 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
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • 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
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2123/00—Data types
    • G06F2123/02—Data types in the time domain, e.g. time-series data
    • 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)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses an extreme drought accurate prediction method and system. Training the LSTM network model by utilizing the decomposition result, and taking the trained LSTM network model as a drought prediction model, so as to realize the prediction of drought; the system comprises a data processing module, a model training module and a model analysis module; the method solves the problems that in the prior art, local characteristics are not extracted from the drought index SPI time sequence obtained according to precipitation data, and single model is directly used for prediction, so that the data is low in predictability and easy to generate local optimum, and the effect of accurately predicting drought is difficult to achieve.

Description

Extreme drought accurate prediction method and system
Technical Field
The invention relates to the field of urban meteorological disaster early warning, in particular to an extreme drought accurate prediction method and system.
Background
Precipitation is one of the main factors affecting drought occurrence, and quantitative prediction of drought based on precipitation is one of the ways to achieve accurate prediction. However, since precipitation data has nonlinear characteristics and varies in different areas and time, it is difficult to directly use precipitation for comparison and feature learning between different scales. Therefore, in related studies of drought prediction, the extent, duration and extent of influence of drought are generally quantitatively assessed using drought index.
On the basis of quantitative evaluation of drought based on drought index, a model is usually required to be constructed and trained to select an optimal model for predicting drought occurrence, and in the prior art, an ARIMA model or an ANN model is used for predicting drought conditions. However, the drought index SPI time sequence obtained by the two models according to the precipitation data has non-stable and nonlinear characteristics, and the partial characteristics of the sequence are not extracted to enable the sequence to be stable, so that the data has low predictability, the problem of partial optimal easily occurs, and the effect of accurately predicting drought is difficult to achieve.
Disclosure of Invention
The invention aims to: the invention aims to provide an extreme drought accurate prediction method and an extreme drought accurate prediction system which can predict drought conditions with high accuracy.
The technical scheme is as follows: in order to achieve the above purpose, the method for accurately predicting extreme drought according to the present invention comprises the following steps:
step S1: acquiring daily precipitation data of a target area;
step S2: for precipitation data, calculating SPI time sequences B of different time scales based on a drought index SPI, and dividing the SPI time sequences B into a training set and a testing set;
step S3: decomposing the SPI time sequence B in the training set and the testing set by using a complementary integrated empirical mode CEEMD decomposition method to respectively obtain IMF components and trend items;
step S4: training the LSTM network model by using IMF components and trend items in the training set to obtain a drought prediction model;
step S5: predicting IMF components and trend items in the test set by using a drought prediction model;
step S6: and analyzing the prediction result.
The step S1 of acquiring the daily precipitation data of the target area refers to downloading a daily precipitation data set of the target area weather station from a national weather science data center, wherein the daily precipitation data set includes date, temperature and precipitation data, and extracting date data and precipitation data of the date from the data set.
The step S2 is to calculate SPI time series B of different time scales based on the drought index SPI, and divide the SPI time series B into a training set and a test set, namely, calculating and obtaining SPI values of 1, 3, 6 and 12 month time scales of daily rainfall data according to a calculation formula of the drought index SPI to form four-scale SPI time series B, and dividing the SPI time series B into the training set and the test set according to a ratio of 4:1;
the calculation formula of the drought index SPI is as follows:
where a is the positive and negative coefficient of probability density, when b= -1,when b=1, _a-> G (x) is the cumulative probability; constant c 0 ＝2.515517，c 1 ＝0.802853，c 2 ＝0.010328，d 1 ＝1.432788，d 2 ＝0.189269，d 3 ＝0.001308。
The complementary integrated empirical mode CEEMD decomposition method in step S3 decomposes the training set and the SPI time series B in the test set to obtain IMF components and trend terms, respectively, specifically:
adding n groups of auxiliary white noise including positive noise and negative noise to the SPI time sequence B to obtain a positive noise sequence H 1 And negative noise sequence H 2 The sequences obtained at this time are as follows, with a total number of 2n:
wherein N is an auxiliary sequence;
the resulting sequence is then used to determineRespectively decomposing to obtain m IMF components, each component being marked as +.>And->Where i=1, …, n, j=1, …, m, t refer to a time series;
for each set of IMF componentsAnd->Averaging to obtain the value of the j-th IMF:
taking the obtained IMF value as a final decomposition result, namely decomposing the SPI time sequence B into:
wherein r is (t) Is a residual trend term.
The training of the LSTM network model is performed by using IMF components and trend terms of the training set in step S4, where the activating function of the LSTM network model selects ReLU, and the weight is updated once every training sample, and the loss function uses mean square error MSE:
wherein X is J Is the observed value, X is the training result value, N is X J J is the number of updates, j=1, 2, N;
in the training process, as the iteration times are increased, the MSE gradually decreases, the model precision is gradually improved, and when the MSE value is increased, the training of the LSTM network model is stopped.
The trained judgment standard of the LSTM network model is as follows: evaluation index R 2 A value of greater than 0.8, wherein R 2 The method comprises the following steps:
wherein: x is X J Is an observation value of the current,is X J X is the training result value, N is X J J is the number of updates, j=1, 2,..n.
The predicting IMF components and trend items in the test set by using the drought prediction model in step S5 specifically includes: pre-prediction using drought prediction modelThe IMF component and the trend term obtained by decomposing the SPI time sequence B of the test set are measured, wherein the SPI time sequence B comprises SPI values of four months of time scales of 1, 3, 6 and 12, and the IMF component and the trend term are respectively marked as IMF 1 、IMF 2 、…、IMF m And Res, the prediction results are respectively marked as P1 and P 2 、…、P m+1 ；
Four sets of predictions comprising four month time scales of 1, 3, 6, 12 were summed to obtain the complementary integrated empirical mode CEEMD-LSTM predicted SPI values by the following formula, where m is the number of IMF components, k=1, 2,..:
the step S6 is to analyze the predicted result, namely, divide the drought grade according to the grade standard of the predicted SPI value, and when the SPI value is smaller than-0.5, drought occurs.
The invention also provides an extreme drought accurate prediction system, which comprises a data processing module, a model training module and a model analysis module;
the data processing module is used for calculating SPI time sequences B of different time scales for the acquired precipitation data and dividing the SPI time sequences B into a training set and a testing set;
the model training module comprises the step of training an LSTM network model;
the model analysis module is used for performing performance analysis on the trained LSTM network model.
The beneficial effects are that: the invention has the following advantages: 1. the invention provides a method and a system for predicting drought occurrence, wherein the method comprises the steps of obtaining precipitation data, calculating SPI time sequences of different time scales, decomposing the SPI time sequences to extract local features of decomposed precipitation components on different scales, training an LSTM network model by utilizing a decomposition result, and taking the trained LSTM network model as a drought prediction model, so that high-precision drought prediction is realized;
2. the invention adopts the CEEMD signal decomposition method to extract the local characteristics of the SPI time sequence and stabilize the sequence, thereby reducing the complexity of the sequence and improving the predictability of the sequence in drought;
3. compared with the traditional drought prediction model, the LSTM network model is higher in modeling timeliness and better in learning effect as one of the deep learning methods; in the LSTM network model training process, the invention adopts the strategy of 'early stop method', so that the model stops iterating when the accuracy is highest, thereby improving the prediction drought efficiency and accuracy of the model under each time scale.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention;
FIG. 2 is a schematic representation of the results of predicting drought conditions.
Detailed Description
The technical scheme of the present invention will be described in detail with reference to the following examples and the accompanying drawings.
As shown in fig. 1, the method for accurately predicting extreme drought according to the invention comprises the following steps:
step S1: acquiring daily precipitation data of a target area; the daily precipitation data set of the target area weather station is downloaded from the national weather science data center, and comprises date, temperature and precipitation data, and date data and precipitation data of the date are extracted from the data set.
Step S2: for precipitation data, calculating SPI time sequences B of different time scales based on a drought index SPI, and dividing the SPI time sequences B into a training set and a testing set; the method comprises the steps of calculating and obtaining SPI values of 1, 3, 6 and 12 month time scales according to a calculation formula of a drought index SPI by using daily precipitation data to form a four-scale SPI time sequence B, and dividing the SPI time sequence B into a training set and a testing set according to a ratio of 4:1;
the calculation formula of the drought index SPI is as follows:
where a is the positive and negative coefficient of probability density, when b= -1,when b=1, _a-> G (x) is the cumulative probability; constant c 0 ＝2.515517，c 1 ＝0.802853，c 2 ＝0.010328，d 1 ＝1.432788，d 2 ＝0.189269，d 3 ＝0.001308。
Step S3: decomposing the SPI time sequence B in the training set and the testing set by using a complementary integrated empirical mode CEEMD decomposition method to respectively obtain IMF components and trend items; the method comprises the following steps:
adding n groups of auxiliary white noise including positive noise and negative noise to the SPI time sequence B to obtain a positive noise sequence H 1 And negative noise sequence H 2 The sequences obtained at this time are as follows, with a total number of 2n:
wherein N is an auxiliary sequence;
the resulting sequence is then used to determineRespectively decomposing to obtain m IMF components, each component being marked as +.>And->Where i=1, …, n, j=1, …, m, t refer to a time series;
for each set of IMF componentsAnd->Averaging to obtain the value of the j-th IMF:
taking the obtained IMF value as a final decomposition result, namely decomposing the SPI time sequence B into:
wherein r is (t) Is a residual trend term.
Step S4: training the LSTM network model by adopting IMF components and trend items of a training set; wherein, the activating function of the LSTM network model selects ReLU, and the weight is updated once every training sample, and the loss function adopts MSE:
wherein X is J Is the observed value, X is the training result value, N is X J J is the number of updates, j=1, 2, N;
in the training process, as the iteration times are increased, the MSE gradually decreases, the model precision is gradually improved, and when the MSE value is increased, the training of the LSTM network model is stopped.
The trained judgment standard of the LSTM network model is as follows: evaluation index R 2 The value is more than 0.8, and the evaluation index R 2 The value is an index for measuring the model fitting goodness, the closer the value is to 1, the better the fitting effect of the model on the data is, R 2 The calculation formula is as follows:
wherein: x is X J Is an observation value of the current,is X J X is the training result value, N is X J J is the number of updates, j=1, 2,..n.
Step S5: predicting IMF components and trend items in the test set by using a drought prediction model; the method comprises the following steps: using a drought prediction model to predict IMF components and trend terms obtained by decomposing a test set SPI time sequence B, wherein the SPI time sequence B comprises SPI values of four month time scales of 1, 3, 6 and 12, and the IMF components and the trend terms are respectively marked as IMFs 1 、IMF 2 、…、IMF m And Res, the prediction results are respectively marked as P1 and P 2 、…、P m+1 ；
Four sets of predictions comprising four month time scales of 1, 3, 6, 12 were summed to obtain the complementary integrated empirical mode CEEMD-LSTM predicted SPI values by the following formula, where m is the number of IMF components, k=1, 2,..:
step S6: the prediction result is analyzed, namely, the predicted SPI value is divided into drought grades according to the drought grade grading standard, and the drought grade grading standard table is shown in the table 1, and when the SPI value is smaller than-0.5, drought occurs.
Table 1: drought grade grading standard table
As shown in fig. 2, a schematic diagram of the result of the prediction of drought conditions is shown. From the graph, it can be seen that fig. 2 (b) shows the prediction result of the drought prediction model of the present invention on the drought spatial distribution, which is very close to the actual situation shown in fig. 2 (a), and illustrates the advantages of CEEMD-based on non-stationary signal processing, and the model has better performance on SPI prediction.
The invention also provides an extreme drought accurate prediction system, which comprises a data processing module, a model training module and a model analysis module;
the data processing module is used for calculating SPI time sequences B of different time scales for the acquired precipitation data and dividing the SPI time sequences B into a training set and a testing set; the model training module comprises the step of training an LSTM network model; the model analysis module is used for performing performance analysis on the trained LSTM network model.
The invention provides a method and a system for predicting drought occurrence. According to the method, SPI time sequences of different time scales are calculated by acquiring precipitation data, and are decomposed, so that local features of decomposed precipitation components on different scales are extracted, an LSTM network model is trained by utilizing a decomposition result, and the trained LSTM network model is used as a drought prediction model, so that high-accuracy prediction of drought conditions is realized.
The invention adopts CEEMD signal decomposition method to extract local characteristics of SPI time sequence and make the sequence stable, thereby reducing complexity of the sequence and improving predictability of the sequence in drought.
The LSTM network model is used as one of the deep learning methods, and can effectively solve the problem that long-time sequences are difficult to learn in RNNs. Compared with the traditional drought prediction model, the LSTM modeling timeliness is higher, and the learning effect is better. In the LSTM network model training process, the invention adopts the strategy of 'early stop method', so that the model stops iterating when the accuracy is highest, thereby improving the prediction drought efficiency and accuracy of the model under each time scale.

Claims (10)

1. An extreme drought accurate prediction method is characterized by comprising the following steps:
step S1: acquiring daily precipitation data of a target area;
step S2: for precipitation data, calculating SPI time sequences B of different time scales based on a drought index SPI, and dividing the SPI time sequences B into a training set and a testing set;
step S3: decomposing the SPI time sequence B in the training set and the testing set by using a complementary integrated empirical mode CEEMD decomposition method to respectively obtain IMF components and trend items;
step S4: training the LSTM network model by using IMF components and trend items in the training set to obtain a drought prediction model;
step S5: predicting IMF components and trend items in the test set by using a drought prediction model;
step S6: and analyzing the prediction result.
2. The method according to claim 1, wherein the step S1 of obtaining daily precipitation data of the target area refers to downloading a daily precipitation data set of a weather station of the target area from a national weather science data center, wherein the daily precipitation data set includes date, temperature and precipitation data, and the date data and the precipitation data of the date are extracted from the data set.
3. The method for accurately predicting extreme drought according to claim 1, wherein in step S2, for precipitation data, calculating SPI time series B of different time scales based on a drought index SPI, and dividing the SPI time series B into a training set and a test set, namely calculating daily precipitation data according to a calculation formula of the drought index SPI to obtain SPI values of 1, 3, 6, 12 month time scales, forming four-scale SPI time series B, and dividing the SPI time series B into the training set and the test set according to a ratio of 4:1.
4. The method for accurately predicting extreme drought according to claim 3, wherein the calculation formula of the drought index SPI is as follows:
where a is the positive and negative coefficient of probability density, when b= -1,when b=1, _a-> Gx is the cumulative probability; constant c 0 ＝2.515517，c 1 ＝0.802853，c 2 ＝0.010328，d 1 ＝1.432788，d 2 ＝0.189269，d 3 ＝0.001308。
5. The method for accurately predicting extreme drought according to claim 1, wherein the decomposing method for complementary integrated empirical mode CEEMD in step S3 decomposes the SPI time series B in the training set and the test set to obtain IMF components and trend terms, respectively, specifically:
adding n groups of auxiliary white noise including positive noise and negative noise to the SPI time sequence B to obtain a positive noise sequence H 1 And negative noise sequence H 2 The sequences obtained at this time are as follows, with a total number of 2n:
wherein N is an auxiliary sequence;
the resulting sequence is then used to determineRespectively decomposing to obtain m IMF components, each component being marked as +.>And->Where i=1, …, n, j=1, …, m, t refer to a time series;
for each set of IMF componentsAnd->Averaging to obtain the value of the j-th IMF:
taking the obtained IMF value as a final decomposition result, namely decomposing the SPI time sequence B into:
wherein r is t Is a residual trend term.
6. The method for extreme drought accurate prediction according to claim 1, wherein the training of the LSTM network model using IMF components and trend terms of the training set in step S4, wherein the activation function of the LSTM network model uses ReLU, and the weight is updated once every training sample, and the loss function uses the mean square error MSE:
wherein X is J Is the observed value, X is the training result value, N is X J J is the number of updates, j=1, 2, N;
in the training process, as the iteration times are increased, the MSE gradually decreases, the model precision is gradually improved, and when the MSE value is increased, the training of the LSTM network model is stopped.
7. The method for accurately predicting extreme drought according to claim 6, wherein the trained decision criteria of the LSTM network model are: evaluation index R for measuring model fitting goodness 2 A value of greater than 0.8, wherein R 2 The method comprises the following steps:
wherein: x is X J Is an observation value of the current,is X J X is the training result value, N is X J J is the number of updates, j=1, 2,..n.
8. The method for accurately predicting extreme drought according to claim 1, wherein the predicting IMF components and trend terms in the test set using the drought prediction model in step S5 is specifically as follows: using a drought prediction model to predict IMF components and trend terms obtained by decomposing a test set SPI time sequence B, wherein the SPI time sequence B comprises SPI values of four month time scales of 1, 3, 6 and 12, and the IMF components and the trend terms are respectively marked as IMFs 1 、IMF 2 、…、IMF m And Res, the prediction results are respectively marked as P1 and P 2 、…、P m+1 ；
Four sets of predictions comprising four month time scales of 1, 3, 6, 12 were summed to obtain the complementary integrated empirical mode CEEMD-LSTM predicted SPI values by the following formula, where m is the number of IMF components, k=1, 2,..:
9. the method for accurately predicting extreme drought according to claim 1, wherein the step S6 of analyzing the prediction result means that the predicted SPI value is classified according to a drought classification standard, and drought occurs when the SPI value is less than-0.5.
10. The extreme drought accurate prediction system is characterized by comprising a data processing module, a model training module and a model analysis module;
the data processing module is used for acquiring precipitation data, calculating SPI time sequences B of different time scales and dividing the SPI time sequences B into a training set and a testing set;
the model training module is used for training the LSTM network model;
the model analysis module is used for performing performance analysis on the trained LSTM network model.
CN202310666885.8A 2023-06-07 2023-06-07 Extreme drought accurate prediction method and system Pending CN116663608A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310666885.8A CN116663608A (en) 2023-06-07 2023-06-07 Extreme drought accurate prediction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310666885.8A CN116663608A (en) 2023-06-07 2023-06-07 Extreme drought accurate prediction method and system

Publications (1)

Publication Number Publication Date
CN116663608A true CN116663608A (en) 2023-08-29

Family

ID=87727583

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310666885.8A Pending CN116663608A (en) 2023-06-07 2023-06-07 Extreme drought accurate prediction method and system

Country Status (1)

Country Link
CN (1) CN116663608A (en)

Similar Documents

Publication Publication Date Title
CN111104981A (en) Hydrological prediction precision evaluation method and system based on machine learning
CN111665575B (en) Medium-and-long-term rainfall grading coupling forecasting method and system based on statistical power
CN106600037B (en) Multi-parameter auxiliary load prediction method based on principal component analysis
CN114168906B (en) Mapping geographic information data acquisition system based on cloud computing
CN104820873A (en) Fresh water acute standard prediction method based on metal quantitative structure-activity relationship
CN108920427B (en) Dynamic estimation method for time-varying parameters of monthly scale hydrological model
CN111260111A (en) Runoff forecast improvement method based on meteorological big data
CN109190839B (en) Wind direction-fused intelligent rolling prediction method for wind speed along railway
CN115453064B (en) Fine particulate matter air pollution cause analysis method and system
CN110533007A (en) A kind of vehicle-mounted strain of bridge influences the identification of line feature intelligent and extracting method
CN116910041B (en) Daily correction method for remote sensing precipitation product based on scale analysis
CN112418476A (en) Ultra-short-term power load prediction method
CN110765418A (en) Intelligent set evaluation method and system for basin water and sand research model
Chen et al. Proactive quality control: Observing system simulation experiments with the Lorenz’96 model
CN115096357A (en) Indoor environment quality prediction method based on CEEMDAN-PCA-LSTM
CN110533249B (en) Metallurgical enterprise energy consumption prediction method based on integrated long-term and short-term memory network
CN108830405B (en) Real-time power load prediction system and method based on multi-index dynamic matching
CN113281229B (en) Multi-model self-adaptive atmosphere PM based on small samples 2.5 Concentration prediction method
CN116663608A (en) Extreme drought accurate prediction method and system
CN115510945B (en) Geological disaster probability forecasting method based on principal component and Logistic analysis
CN109697630B (en) Sparse regression-based merchant passenger flow volume multi-factor analysis and prediction method
CN115728290A (en) Method, system, equipment and storage medium for detecting chromium element in soil
CN110648023A (en) Method for establishing data prediction model based on quadratic exponential smoothing improved GM (1,1)
CN107066786A (en) Aerosol optical depth inversion algorithm based on neutral net
CN112528566A (en) Real-time air quality data calibration method and system based on AdaBoost training model

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