CN109886496A - A kind of agriculture production prediction method based on weather information - Google Patents
A kind of agriculture production prediction method based on weather information Download PDFInfo
- Publication number
- CN109886496A CN109886496A CN201910147418.8A CN201910147418A CN109886496A CN 109886496 A CN109886496 A CN 109886496A CN 201910147418 A CN201910147418 A CN 201910147418A CN 109886496 A CN109886496 A CN 109886496A
- Authority
- CN
- China
- Prior art keywords
- weather information
- moment
- yield
- coding
- information
- 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
Classifications
-
- 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
Abstract
The invention belongs to agriculture production forecast fields, disclose a kind of agriculture production prediction method based on weather information, model towards meteorological data be mainly to be made of three parts by the model of sequence form tissue, first part is encoder, the meteorology in each stage is encoded to yield coding by it, can construct encoder using LSTM structure comprising historical information model to encode the yield in each stage;Second part is attention calculator, this part is the influence power for measuring the meteorology of different phase to ultimate output, and attention calculator is made of the neural network of a shallow-layer;Part III is decoder, and the weight and coding and decoding that decoder obtains preceding two parts are ultimate output.Interim prediction of the invention is to first pass through LSTM structure weather information is projected as yield coding, then obtain by decoding, and gained precision is much higher than conventional linear regression model.
Description
Technical field
The present invention relates to a kind of agriculture production prediction method based on weather information.
Background technique
The accumulated time effect for the adaptability and environmental resource that crop changes environmental resource determines the life of crop
Long situation, and then influence crop yield.Crop is different in demand of the different growth and development stages for varying environment resource, and
And plant growth is a continuous long-term accumulated process, has irreversibility.Most Statistic analysis models using accumulated temperature and
Rainfall is accumulated to assess crop for the adaptive of environmental resource, but majority is studied the environmental response of each growing stage of crop
It is simplified to same level, thus crop growth process can not be embodied, the dynamic need of environmental resource is changed.Plant growth
Journey mechanism model have through the mechanism such as crop photosynthesis, respiration process emulate high time resolution based on carbon at
The crop growth conditions divided, but could not also have the mechanism that can be analyzed the Short Term Anomalous such as high temperature or arid temperature or water stress
Model.Current process of crop growth mechanism model and statistical regression analysis model, can not quantitative analysis for different phase
Anomalous weather has ignored the timing spy of crop modeling input feature vector for the cumulative effect of plant growth in the model analysis stage
Property.
The traditional prediction technique of farm output is mainly based upon statistics calculating and extracts with manual features, by different regions
History annual output finishing analysis, and the relationship of the attributes such as hand digging yield and meteorology, soil does yield according to this relationship
Prediction.Since the relationship of farm output and each influence factor is complicated, and influence factor is more.This mode calculates complexity, workload
Greatly, and precision of prediction is lower.In recent years with the fast development of computer technology and the algorithm for relying on computer, artificial intelligence
The method of energy starts gradually to be applied to agriculture production forecast.Traditional multilinear fitting, bayes method are all under this progress
Further applied.But these methods or from parameter fitting or from the angle of probability distribution, all do not consider data when
Sequence characteristics.And not only the growth of crops itself has temporal aspect in agriculture scene, the attribute for influencing crop growth is same
Sample has temporal aspect.Therefore the temporal aspect in agriculture scene how is grasped, and keeps the advantage of parameter fitting in agricultural production
It is an important problem in prediction.
Summary of the invention
The object of the present invention is to provide a kind of agriculture production prediction method based on weather information, can have by weather information
Effect ground prediction agricultural production amount, the process for influencing crop growth for weather information have the characteristics that timing and sensibility, and
Crops prediction need to have in-advance demand, devise a kind of agricultural production based on attention mechanism and shot and long term memory models
Measure prediction technique.
To achieve the goals above, the present invention provides a kind of agriculture production prediction method based on weather information, including as follows
Step:
(1) sequential coding: being encoded to production information for accumulative weather information, and encoder is made of LSTM, defines the time
Sequence X={ x1, x2..., xT, wherein xTWeather information vector is represented, the characteristics of according to weather information to Yield of Corn,
Time series X is sequentially input to LSTM network;When LSTM model is calculated to moment t, x is inputtedt=[T1t,T2t,…,
Tnt], wherein T represents feature, and n represents the number of feature, and the state of memory unit is ct, the hiding feature h at this momentK=
ottanh(ct), otIt is the out gate of LSTM network neural member, calculates such as following formula:
ot=σ (Wo[T1t,T2t,…,Tnt]+Uoht-1);
Wherein σ is logistic sigmoid function, memory unit state ctUpdate be c by forgetting upper a momentt-1
In partial information, and add the information at this momentIt completes, calculates such as following formula:
Forget memory unit state c of upper a momentt-1Degree be by forgetting a fkControl, and new information is added to note
The degree of unit is recalled by input gate ikControl, these control doors calculate as follows:
ft=σ (Wf[T1t,T2t,…,Tnt]+Ufht-1+Vfct-1);
it=σ (Wi[T1t,T2t,…,Tnt]+Uiht-1+Vict-1);
It walks and calculates by T, last production forecast result is Ypre=Wpre×ht+biaspre, hTIt is hidden for the last one moment
Hide feature, in this structure the hidden unit h at each moment be at current state c to input x feature extraction as a result,
The time series data of corresponding input, h are the yield coding at corresponding each moment;
(2) calculate weight by attention mechanism: attention mechanism is made of one three layers of feedforward network FFN, in order to
The capability of fitting of more preferable balance network and the relationship of complexity, use relu as activation primitive first two layers, make network with non-thread
Property capability of fitting and guarantee parameter between lower degree of dependence, third layer be full articulamentum without activation primitive, in order to guarantee to pay attention to
The numerical value of power mechanism output has good discrimination, and is distributed in the section 0-1, and softmax is arranged after the output of third layer
Weight is handled, it is as follows to the influence degree or weight Att, calculating process of yield to calculate different phase weather information:
FFN (H)=max (0, max (0, W1H+b1)W2+b2)W3+b3;
Att=softmax (FFN (H));
(3) yield decodes: the production in each stage that the weighted value Att calculated first according to attention mechanism obtains coding
Amount information h be weighted summation obtain ultimate output coding, then by ultimate output coding projection be production forecast value P, calculate such as
Under:
P=(AttH) W+b;
Wherein H is the matrix of production information h composition, W and b be respectively model learning weight with it is bigoted, all by trained
It arrives.
Further, the parameter value of the model various pieces is obtained by back-propagation algorithm training.
Further, the training uses Adam optimization algorithm.
The present invention constructs by introducing LSTM model and attention mechanism and makees produce by meteorological data sequence estimation
The prediction model of amount.There is following three points advantage relative to conventional method:
(1) agricultural tradition is to use the mode of linear regression based on meteorological agriculture production forecast, and the present invention is based on depth
The model of feature learning when learning method establishes non-linear.
(2) present invention be fitted by attention mechanism corn growth different phase to meteorological sensibility, experiment
The result shows that the weight of attention mechanism generation is related to meteorological sensibility to crop and attention mechanism is for anomalous weather
Also there is certain sensibility.
(3) interim prediction of the invention is to first pass through lstm structure weather information is projected as yield coding, then pass through
Decoding obtains, and gained precision is much higher than conventional linear regression model.
The other feature and advantage of the embodiment of the present invention will the following detailed description will be given in the detailed implementation section.
Detailed description of the invention
Attached drawing is to further understand for providing to the embodiment of the present invention, and constitute part of specification, under
The specific embodiment in face is used to explain the present invention embodiment together, but does not constitute the limitation to the embodiment of the present invention.Attached
In figure:
Fig. 1 is the agriculture Production Forecast Models structural schematic diagram in one embodiment of the invention based on weather information;
Fig. 2 is that distribution map of the force value at the meteorological anomaly time is paid attention in one embodiment of the invention;
Fig. 3 is interim production forecast experimental result schematic diagram in one embodiment of the invention.
Specific embodiment
It is described in detail below in conjunction with specific embodiment of the attached drawing to the embodiment of the present invention.It should be understood that this
Locate described specific embodiment and be merely to illustrate and explain the present invention embodiment, is not intended to restrict the invention embodiment.
In one embodiment of the invention, the corn yield variation of 12, U.S. state 1971-2016 is chosen to implement
The research object of case constructs model, specific structure by mode as described in Figure 1 characterized by the climatic informations such as temperature, precipitation
Build that steps are as follows:
(1) sequential coding: being encoded to production information for accumulative weather information, and encoder is made of LSTM, defines the time
Sequence X={ x1, x2..., xT, wherein xTWeather information vector is represented, the characteristics of according to weather information to Yield of Corn,
Time series X is sequentially input to LSTM network;When LSTM model is calculated to moment t, x is inputtedt=[T1t,T2t,…,
Tnt], wherein T represents feature, and n represents the number of feature, and the state of memory unit is ct, the hiding feature h at this momentK=
ottanh(ct), otIt is the out gate of LSTM network neural member, calculates such as following formula:
ot=σ (Wo[T1t,T2t,…,Tnt]+Uoht-1);
Wherein σ is logistic sigmoid function, memory unit state ctUpdate be c by forgetting upper a momentt-1
In partial information, and add the information at this momentIt completes, calculates such as following formula:
Forget memory unit state c of upper a momentt-1Degree be by forgetting a fkControl, and new information is added to note
The degree of unit is recalled by input gate ikControl, these control doors calculate as follows:
ft=σ (Wf[T1t,T2t,…,Tnt]+Ufht-1+Vfct-1);
it=σ (Wi[T1t,T2t,…,Tnt]+Uiht-1+Vict-1);
It walks and calculates by T, last production forecast result is Ypre=Wpre×ht+biaspre, hTIt is hidden for the last one moment
Hide feature, in this structure the hidden unit h at each moment be at current state c to input x feature extraction as a result,
The time series data of corresponding input, h are the yield coding at corresponding each moment;
(2) calculate weight by attention mechanism: attention mechanism is made of one three layers of feedforward network FFN, in order to
The capability of fitting of more preferable balance network and the relationship of complexity, use relu as activation primitive first two layers, make network with non-thread
Property capability of fitting and guarantee parameter between lower degree of dependence, third layer be full articulamentum without activation primitive, in order to guarantee to pay attention to
The numerical value of power mechanism output has good discrimination, and is distributed in the section 0-1, and softmax is arranged after the output of third layer
Weight is handled, it is as follows to the influence degree or weight Att, calculating process of yield to calculate different phase weather information:
FFN (H)=max (0, max (0, W1H+b1)W2+b2)W3+b3;
Att=softmax (FFN (H));
(3) yield decodes: the production in each stage that the weighted value Att calculated first according to attention mechanism obtains coding
Amount information h be weighted summation obtain ultimate output coding, then by ultimate output coding projection be production forecast value P, calculate such as
Under:
P=(AttH) W+b;
Wherein H is the matrix of production information h composition, W and b be respectively model learning weight with it is bigoted, all by trained
It arrives;The parameter value of the model various pieces is obtained by back-propagation algorithm training;The training uses Adam optimization algorithm.
It chooses root-mean-square error (RMSE) and is used as evaluation index, oppose with conventional method and other machine learning methods
Than, the result is as follows:
The method AT-LSTM proposed by the present invention of table 1 and other methods performance comparison
In view of crops are influenced situation difference, present invention structure on the frame of LSTM by meteorology in different stages of growth
Attention mechanism has been built, in order to examine the operative condition of attention mechanism, has been tested as follows:
The first step chooses annual 30th week attention value here according to attention value garbled data as screening
Variable (because in the 30th week corn to the sensitivity of temperature and precipitation), it is greater than 0.7 and less than 0.1 that screening conditions, which are respectively set,
The high value part of attention value is obtained, with low magnitude portion.
Second step counts attention high value part, with each time institute accounting of 2000-2016 in low magnitude portion respectively
Example.
Third step, distribution situation of the comparative analysis attention Distribution value between a time.
As shown in Fig. 2, top half be attention > 0.7 when attention value distribution situation, lower half portion is
The distribution situation abscissa of attention value represents the time when attention < 0.1, and it includes this section that color, which represents,
The quantity of attention.It can be found that the high value part of attention in 2012 is significantly more than other times, and without low value portion
Point attention value, and the U.S. in 2012 receives serious natural calamity, the influence for the arid that crops receive and subtract
It produces.This shows that the attention mechanism of model of the present invention is sensitive to abnormal weather.
In face of predicting task, effective result can be obtained before event actually occurs and is important.In order to which testing model mentions
The ability of preceding prediction, the present invention devise interim production forecast experiment, and experimental result is indulged as shown in figure 3, abscissa is cycle
Coordinate is the RMSE between true value and predicted value.In conjunction with table 1 it can be found that AT-LSTM model at 23 weeks, prediction result was
Prediction result better than Lasso method at 39 weeks, prediction result has been better than prediction result of the RF method at 39 weeks at 28 weeks, and
Optimal result was obtained at 35 weeks, this can illustrate that AT-LSTM model is advantageous in stage prediction.
The optional embodiment of the embodiment of the present invention is described in detail in conjunction with attached drawing above, still, the embodiment of the present invention is simultaneously
The detail being not limited in above embodiment can be to of the invention real in the range of the technology design of the embodiment of the present invention
The technical solution for applying example carries out a variety of simple variants, these simple variants belong to the protection scope of the embodiment of the present invention.
It is further to note that specific technical features described in the above specific embodiments, in not lance
In the case where shield, it can be combined in any appropriate way.In order to avoid unnecessary repetition, the embodiment of the present invention pair
No further explanation will be given for various combinations of possible ways.
In addition, any combination can also be carried out between a variety of different embodiments of the embodiment of the present invention, as long as it is not
The thought of the embodiment of the present invention is violated, equally should be considered as disclosure of that of the embodiment of the present invention.
Claims (3)
1. a kind of agriculture production prediction method based on weather information, which comprises the steps of:
(1) sequential coding: being encoded to production information for accumulative weather information, and encoder is made of LSTM, defines time series X
={ x1, x2..., xT, wherein xTWeather information vector is represented, the characteristics of according to weather information to Yield of Corn, by the time
Sequence X is sequentially input to LSTM network;When LSTM model is calculated to moment t, x is inputtedt=[T1t, T2t..., Tnt], wherein
T represents feature, and n represents the number of feature, and the state of memory unit is ct, the hiding feature h at this momentk=ottanh(ct),
otIt is the out gate of LSTM network neural member, calculates such as following formula:
ot=σ (Wo[T1t, T2t..., Tnt]+Uoht-1);
Wherein σ is logistic sigmoid function, memory unit state ctUpdate be c by forgetting upper a momentt-1In
Partial information, and add the information at this momentIt completes, calculates such as following formula:
Forget memory unit state c of upper a momentt-1Degree be by forgetting a fkControl, and new information is added to memory unit
Degree by input gate ikControl, these control doors calculate as follows:
ft=σ (Wf[T1t, T2t..., Tnt]+Ufht-1+Vfct-1);
it=σ (Wi[T1t, T2t..., Tnt]+Uiht-1+Vict-1);
It walks and calculates by T, last production forecast result is Ypre=Wpre×ht+biaspre, hTFor the hiding spy at the last one moment
Sign, the hidden unit h at each moment is to the feature extraction for inputting x at current state c as a result, corresponding in this structure
The time series data of input, h are the yield coding at corresponding each moment;
(2) calculate weight by attention mechanism attention: attention mechanism is made of one three layers of feedforward network FFN,
It in order to more preferably balance the capability of fitting of network and the relationship of complexity, uses relu as activation primitive first two layers, there is network
Lower degree of dependence between nonlinear fitting ability and guarantee parameter, third layer is full articulamentum without activation primitive, in order to guarantee
The numerical value of attention mechanism output has good discrimination, and is distributed in the section 0-1, is arranged after the output of third layer
Softmax handles weight, and it is as follows to the influence degree or weight Att, calculating process of yield to calculate different phase weather information:
FFN (H)=max (0, max (0, W1H+b1)W2+b2)W3+b3;
Att=softmax (FFN (H));
(3) yield decodes: the weighted value Att calculated first according to attention mechanism believes the yield in each stage that coding obtains
Breath h be weighted summation obtain ultimate output coding, then by ultimate output coding projection be production forecast value P, calculate it is as follows:
P=(AttH) W+b;
Wherein H be production information h composition matrix, W and b be respectively model learning weight with it is bigoted, all by train obtain.
2. the agriculture production prediction method according to claim 1 based on weather information, which is characterized in that the model is each
Partial parameter value is obtained by back-propagation algorithm training.
3. the agriculture production prediction method according to claim 2 based on weather information, which is characterized in that the training uses
Adam optimization algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910147418.8A CN109886496B (en) | 2019-02-27 | 2019-02-27 | Agricultural yield prediction method based on meteorological information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910147418.8A CN109886496B (en) | 2019-02-27 | 2019-02-27 | Agricultural yield prediction method based on meteorological information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109886496A true CN109886496A (en) | 2019-06-14 |
CN109886496B CN109886496B (en) | 2023-02-03 |
Family
ID=66929747
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910147418.8A Active CN109886496B (en) | 2019-02-27 | 2019-02-27 | Agricultural yield prediction method based on meteorological information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109886496B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110880127A (en) * | 2019-10-24 | 2020-03-13 | 北京三快在线科技有限公司 | Consumption level prediction method and device, electronic equipment and storage medium |
CN111027752A (en) * | 2019-11-19 | 2020-04-17 | 浙江大学 | Crop yield estimation method based on deep spatiotemporal feature joint learning |
WO2021262279A1 (en) * | 2020-06-24 | 2021-12-30 | X Development Llc | Predicting climate conditions based on teleconnections |
CN113919587A (en) * | 2021-11-03 | 2022-01-11 | 中大检测(湖南)股份有限公司 | Geological disaster prediction method and system based on neural network |
CN114925947A (en) * | 2022-03-04 | 2022-08-19 | 武汉大学 | Phenological adaptive crop physiological index deep learning estimation method and system |
CN115860269A (en) * | 2023-02-20 | 2023-03-28 | 南京信息工程大学 | Crop yield prediction method based on triple attention mechanism |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160350653A1 (en) * | 2015-06-01 | 2016-12-01 | Salesforce.Com, Inc. | Dynamic Memory Network |
CN108024158A (en) * | 2017-11-30 | 2018-05-11 | 天津大学 | There is supervision video abstraction extraction method using visual attention mechanism |
CN108388900A (en) * | 2018-02-05 | 2018-08-10 | 华南理工大学 | The video presentation method being combined based on multiple features fusion and space-time attention mechanism |
-
2019
- 2019-02-27 CN CN201910147418.8A patent/CN109886496B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160350653A1 (en) * | 2015-06-01 | 2016-12-01 | Salesforce.Com, Inc. | Dynamic Memory Network |
CN108024158A (en) * | 2017-11-30 | 2018-05-11 | 天津大学 | There is supervision video abstraction extraction method using visual attention mechanism |
CN108388900A (en) * | 2018-02-05 | 2018-08-10 | 华南理工大学 | The video presentation method being combined based on multiple features fusion and space-time attention mechanism |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110880127B (en) * | 2019-10-24 | 2022-04-12 | 北京三快在线科技有限公司 | Consumption level prediction method and device, electronic equipment and storage medium |
CN110880127A (en) * | 2019-10-24 | 2020-03-13 | 北京三快在线科技有限公司 | Consumption level prediction method and device, electronic equipment and storage medium |
CN111027752A (en) * | 2019-11-19 | 2020-04-17 | 浙江大学 | Crop yield estimation method based on deep spatiotemporal feature joint learning |
WO2021098472A1 (en) * | 2019-11-19 | 2021-05-27 | 浙江大学 | Crop yield estimation method based on deep temporal and spatial feature combined learning |
US11741555B2 (en) | 2019-11-19 | 2023-08-29 | Zhejiang University | Crop yield estimation method based on deep temporal and spatial feature combined learning |
CN111027752B (en) * | 2019-11-19 | 2022-06-21 | 浙江大学 | Crop yield estimation method based on deep spatiotemporal feature joint learning |
WO2021262279A1 (en) * | 2020-06-24 | 2021-12-30 | X Development Llc | Predicting climate conditions based on teleconnections |
US11243332B2 (en) | 2020-06-24 | 2022-02-08 | X Development Llc | Predicting climate conditions based on teleconnections |
US11668856B2 (en) | 2020-06-24 | 2023-06-06 | Mineral Earth Sciences Llc | Predicting climate conditions based on teleconnections |
CN113919587A (en) * | 2021-11-03 | 2022-01-11 | 中大检测(湖南)股份有限公司 | Geological disaster prediction method and system based on neural network |
CN114925947A (en) * | 2022-03-04 | 2022-08-19 | 武汉大学 | Phenological adaptive crop physiological index deep learning estimation method and system |
CN114925947B (en) * | 2022-03-04 | 2023-04-07 | 武汉大学 | Phenological adaptive crop physiological index deep learning estimation method and system |
CN115860269A (en) * | 2023-02-20 | 2023-03-28 | 南京信息工程大学 | Crop yield prediction method based on triple attention mechanism |
Also Published As
Publication number | Publication date |
---|---|
CN109886496B (en) | 2023-02-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109886496A (en) | A kind of agriculture production prediction method based on weather information | |
Nasr et al. | Cross entropy error function in neural networks: Forecasting gasoline demand. | |
Han et al. | Coupling a bat algorithm with xgboost to estimate reference evapotranspiration in the arid and semiarid regions of china | |
CN111027175B (en) | Method for evaluating social and economic influences of flood based on coupling model integrated simulation | |
CN116451879B (en) | Drought risk prediction method and system and electronic equipment | |
CN106716479A (en) | A computer implemented method of deriving performance from a local model | |
CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
CN110119086A (en) | A kind of tomato greenhouse environmental parameter intelligent monitoring device based on ANFIS neural network | |
CN114662742A (en) | Dynamic decision method and device for field irrigation and fertilization system | |
CN111831966B (en) | Combined river channel water level forecasting method based on high-dimensional probability distribution function | |
CN105184400A (en) | Tobacco field soil moisture prediction method | |
CN114611608A (en) | Sea surface height numerical value prediction deviation correction method based on deep learning model | |
Hosseini et al. | Evaluation of data-driven models to downscale rainfall parameters from global climate models outputs: the case study of Latyan watershed | |
Robati et al. | Inflation rate modeling: Adaptive neuro-fuzzy inference system approach and particle swarm optimization algorithm (ANFIS-PSO) | |
Li et al. | An Integrated Artificial Neural Network-based Precipitation Revision Model. | |
CN116629453B (en) | Remote sensing yield estimation method suitable for whole crop growth period | |
Lin et al. | Stochastic conservative contextual linear bandits | |
Huang et al. | Hybrid neural network models for hydrologic time series forecasting based on genetic algorithm | |
Kosamkar et al. | Agriculture crop simulation models using computational intelligence | |
CN113537645A (en) | Soybean yield prediction method based on machine learning fusion satellite and weather data | |
CN113300884A (en) | GWO-SVR-based step-by-step network flow prediction method | |
Arya et al. | Effective LSTM Neural Network with Adam Optimizer for Improving Frost Prediction in Agriculture Data Stream | |
CN117423003B (en) | Winter wheat seedling condition grading remote sensing monitoring method in overwintering period | |
CN117272813B (en) | Soil humidity prediction method based on water balance constraint deep learning | |
Zhang et al. | Bayesian calibration of aquacrop 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |