Crop yield prediction method for applying biological bacterial fertilizer in multi-climate mode

Abstract

The invention relates to a crop yield prediction method for applying a biological bacterial fertilizer in a multi-climate mode, and belongs to the field of agricultural environmental science. On the basis of a biological bacterial fertilizer database, dimension reduction analysis is carried out on multiple types of biological bacterial fertilizers, clustering and data set division are carried out on data after dimension reduction, data amplification is carried out on training set data, yield prediction model construction is further carried out in combination with test set data, and finally climate feature classification is carried out on a research area data set. And crop yield prediction of specific biological bacterial fertilizer application in different climate modes is completed. The method comprises six links of biological bacterial fertilizer database construction, dimensionality reduction analysis, clustering analysis, data amplification, yield prediction model construction and specific biological bacterial fertilizer yield prediction, and the links supplement each other, so that effective screening and application of the biological bacterial fertilizer under different climate characteristics are guaranteed, and the crop yield is predicted; harmonious symbiosis of soil health and crop growth is realized, and remarkable technical effects and application prospects are brought to agricultural production.

Description

Technical Field

[0001] The invention relates to a method for predicting crop yields corresponding to the application of different biological fertilizers under multiple climate modes, and belongs to the field of agricultural environmental science. Background Art

[0002] With the continuous improvement of people's living standards, especially the demand for improved quality of life, both at home and abroad are actively developing green agriculture to produce safe and pollution-free green food. In the process of producing green food, it is required to use no or as little chemical fertilizers, chemical pesticides and other chemical substances as possible. Therefore, biofertilizers have begun to play a role in agricultural production, alleviating or reducing agricultural product pollution while improving the quality of agricultural products (Zhang Beier. Effects of biofertilizers on farmland soil conservation based on field scale [D]. Shanghai: Donghua University, 2023). However, due to the characteristics of its own microorganisms, there are certain restrictions on application conditions, and it mainly has a positive impact on agricultural product yields, but the impact on different crops is different (Wang Zhe, Tu Chunbao, Ma Kun, et al., Research progress on the effects of biofertilizers on soil and crops [J]. Zhejiang Agricultural Sciences, 2024 (10): 2523-2528; Qu Cheng, Liu Fen, Fu Aibin, et al., Effects of biofertilizers and chemical fertilizers on rice growth characteristics and yield [J]. Hybrid Rice, 2023, 38 (4): 134-139).

[0003] The main problems with the above technical methods are: (1) There are many types of biofertilizers, and my country's soil characteristics are complex, so its targeted fertility improvement effect on soil is unclear; (2) There are limited cases of biofertilizer implementation, and its impact on crop yields is unclear. Under different climate models, a method is needed to predict crop yields for the application of biofertilizers. Summary of the invention

[0004] Based on the biological fertilizer database, the present invention performs dimensionality reduction analysis on multiple types of biological fertilizers, clusters the data after dimensionality reduction, divides the data set, performs data amplification on the training set data, further combines the test set data to construct a yield prediction model, and finally classifies the climate characteristics of the study area data set to complete the crop yield prediction of specific biological fertilizer application under different climate modes. The present invention includes six links: biological fertilizer database construction, dimensionality reduction analysis, cluster analysis, data amplification, yield prediction model construction and specific biological fertilizer yield prediction. Each link complements each other, ensures the effective screening and application of biological fertilizers under different climate characteristics, and predicts crop yields, realizes the harmonious symbiosis of soil health and crop growth, and brings significant economic and social benefits to agricultural production. To achieve the above purpose, the technical scheme of the present invention is as follows.

[0005] A method for predicting crop yield by applying biofertilizer under multiple climate models comprises the following steps:

[0006] (1) Construction of the biofertilizer database, including the collection of case information datasets, research area datasets, and research area historical climate datasets;

[0007] (2) Dimensionality reduction analysis

[0008] (2.1) Data standardization

[0009] The original data x of all indicator data except yield in the case information dataset and all indicator data in the study area dataset are standardized to obtain the standardized data x'. The calculation formula is as follows:

[0010] x'=[x-ave(x)] / std(x)

[0011] Where ave(x) is the mean of the original data, and std(x) is the standard deviation of the original data;

[0012] The indicators and types of all indicator data except output in the case information dataset and the original data of all indicator data in the research area dataset are consistent, and are all digital;

[0013] (2.2) Data Dimensionality Reduction

[0014] (2.2.1) Use PCA method to perform dimensionality reduction analysis on the standardized data x' to obtain the component matrix;

[0015] (2.2.2) Based on the standardized data x' information, find the corresponding original data x of the case information dataset, perform matrix multiplication operation on the original data and the component matrix, and obtain the dimension-reduced dataset;

[0016] (2.2.3) Extract the dimension reduction data corresponding to the study area data set to obtain the study area dimension reduction data set, and the remaining data is the case dimension reduction data set;

[0017] (3) Cluster analysis

[0018] (3.1) Weight ratio

[0019] (3.1.1) Calculate the difference g in the applicable range of the same type of biological fertilizer and obtain regional information. The specific formula is as follows:

[0020] g=max[max(lon)-min(lon),max(lat)-min(lat)]*110

[0021] Where lon is the longitude of biofertilizer application, lat is the latitude of biofertilizer application; max(lon) is the maximum longitude of biofertilizer application, min(lon) is the minimum longitude of biofertilizer application; max(lat) is the maximum latitude of biofertilizer application, min(lat) is the minimum latitude of biofertilizer application,

[0022] (3.1.2) Select regional production data, perform spatial interpolation, and obtain the base value c;

[0023] (3.1.3) Weight ratio calculation

[0024] Calculate the weighted longitude and latitude w(lon) and w(lat) of the same type of biological fertilizer as follows:

[0025] w(lon)=g / c*lon w(lat)=g / c*lat

[0026] The weighted longitude and latitude of all biological fertilizers are calculated using the same method;

[0027] (3.2) Cluster analysis

[0028] Using the clustering model, cluster analysis is performed on all indicators and weight latitude and longitude in the case dimension reduction data set to obtain n cluster data sets;

[0029] Said n≥3;

[0030] (3.3) Data partitioning

[0031] (3.3.1) Based on the clustering data set, search for the production data in the corresponding case information data set, merge the production data with the clustering data set, and obtain n new clustering data sets;

[0032] (3.3.2) All indicators in the n new clustering data sets are randomly divided into two data sets, namely a training set and a test set; the random division ratio and method are consistent in all the clustering data sets;

[0033] (4) Data augmentation

[0034] The same data augmentation model is used to augment the n training set data respectively, and the augmented data and the training set data are merged into a new training set;

[0035] (5) Construction of production prediction model

[0036] (5.1) Model construction

[0037] Combine the data of the new training set as input, select the prediction model, adjust the parameters, and perform repeated training until R 2 ≥0.7;

[0038] (5.2) Model verification

[0039] (5.2.1) merging the data of the test set;

[0040] (5.2.2) Extract all indicators except yield from the test set data and input them into the trained model to obtain predicted yield data;

[0041] (5.2.3) Linearly fit the test set yield index data with the model predicted yield data to obtain cR 2 ; When cR 2 ≥0.7, the yield prediction model training is completed and the next step is entered. 2 <0.7, return to the data amplification step and repeat the subsequent steps until cR 2 ≥0.7;

[0042] (6) Prediction of specific biological fertilizer production

[0043] (6.1) Production forecast

[0044] The climate data in the historical climate data set of the research area is used to replace the climate data in the dimension reduction data set of the research area, and matrix multiplication operation is performed with the component matrix to obtain the climate dimension reduction data set; the yield prediction model is input to obtain the yield prediction data of all biofertilizers;

[0045] (6.2) Curve fitting

[0046] (6.2.1) The climate data in the historical climate dataset of the study area are classified from small to large to obtain the amount of climate classification data r;

[0047] The number of classifications r≥5;

[0048] (6.2.2) Establish a fitting curve based on the biofertilizer production data and the corresponding climate classification data. Obtain a fitting curve graph for each type of biofertilizer, and calculate the fitting curve area s of all fitting curve graphs;

[0049] (6.3) Yield forecast distribution map

[0050] Based on the fitting curve area and meteorological classification data, the yield m of each category of biological fertilizer is calculated, and the predicted distribution map of the biological fertilizer yield under different climatic conditions of all categories is drawn. The calculation formula is as follows:

[0051] m=s / r.

[0052] Furthermore, the fitting curve adopts the species sensitivity distribution method;

[0053] Furthermore, the case information of the case information data set includes, but is not limited to: one or more of: biological fertilizer characteristics, implementation location, implementation year, soil properties, climate information, and crop seed characteristics;

[0054] Furthermore, the biological fertilizer characteristics include: type, effective live bacteria abundance, effective live bacteria count, N content, P content, K content, organic matter content, etc. The implementation location includes longitude and latitude, soil properties, including one or more of N, P, K, pH, Ca, Mg, S, Fe, Cu, B, Zn, Mo, Se, Mn, I, and Cl; the climate information includes temperature, humidity, rainfall, etc.; the crop seed characteristics, such as crop type, growth period, protein content, yield, etc.;

[0055] Furthermore, the data collection methods include but are not limited to consulting literature, books, experimental reports, project summary reports, etc.;

[0056] Furthermore, the method for obtaining the implementation location is as follows: when the biological fertilizer is implemented by region, if the area of ​​the region is less than 500 square meters, the longitude and latitude of the four boundaries and the center point of the region are collected; if the area of ​​the region is ≥ 500 square meters, in addition to the longitude and longitude of the four boundaries and the center point, the longitude and latitude information in the region is evenly collected according to demand;

[0057] Furthermore, relevant information about the study area is collected, and the indicators correspond to the database cases;

[0058] Furthermore, the units of the yield data are unified, including but not limited to kg / mu, t / mu, kg / ha, t / ha, etc.;

[0059] Furthermore, the interpolation methods include but are not limited to: ordinary kriging, simple kriging, co-kriging, pan-kriging, etc.;

[0060] Furthermore, the clustering model includes but is not limited to GMM, K-means, hierarchical clustering, etc.;

[0061] Furthermore, the ratio of the data volume of the training set and the test set includes but is not limited to: 1-9:9-1;

[0062] Furthermore, the ratio of the data volume of the training set to the test set is 5 / 5, 6 / 4, 7 / 3, 8 / 2 or 9 / 1;

[0063] Furthermore, the data augmentation model includes but is not limited to SMOT, GNN, etc.;

[0064] Furthermore, the prediction model includes but is not limited to random forest, support vector machine, Markov chain, decision tree, etc.;

[0065] Furthermore, the data volume of the climate dimension reduction data set of the study area is 5-15 times the data volume of the dimension reduction data set of the study area, that is, one climate data is replaced by 5-15 historical climate data;

[0066] The advantages and effects of this application are as follows:

[0067] (1) The present invention performs data amplification on existing biofertilizer cases, uses multiple models in combination with model verification, ensures the accuracy of amplified data, solves the problem of insufficient existing data, and lays a data foundation for later yield prediction;

[0068] (2) The present invention takes dimension reduction, clustering, data amplification and other methods as the premise, combines the double-layer model verification, and cyclically trains the yield prediction model, which improves the accuracy of the existing prediction model by 20-30%, ensuring the accuracy of the prediction results;

[0069] (3) The present invention fully considers the impact of climate on crop yields, selects the best applicable bio-fertilizer type, and uses a prediction model to predict crop yields under different climate conditions in the future, clarifies regional yield data in advance, and provides guarantees for orderly market development and social stability.

[0070] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application so that it can be implemented in accordance with the contents of the specification, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following is a detailed description of the preferred embodiments of the present application in conjunction with the accompanying drawings as follows.

[0071] Based on the detailed description of the specific embodiments of the present application in combination with the accompanying drawings below, those skilled in the art will become more aware of the above and other objects, advantages and features of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings without creative work. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual scale.

[0073] Figure 1 A technical roadmap for crop yield prediction methods using biofertilizers for a multi-climate model;

[0074] Figure 2 is the base value distribution diagram;

[0075] Figure 3 This is the clustering result diagram;

[0076] Figure 4 This is the yield prediction distribution map. Specific embodiments

[0077] To make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme in the embodiment of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiment of the present application. Obviously, the described embodiment is a part of the embodiment of the present application, rather than all of the embodiments. In the following description, specific details such as specific configuration and components are provided only to help fully understand the embodiments of the present application. Therefore, it should be clear to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, for clarity and brevity, the description of known functions and structures is omitted in the embodiment.

[0078] It should be understood that the references to "one embodiment" or "this embodiment" throughout the specification mean that the specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "one embodiment" or "this embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0079] In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplicity and clarity, and does not in itself indicate the relationship between the various embodiments and / or settings discussed.

[0080] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" in this article describes another type of association object relationship, indicating that there can be two relationships. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the previous and next associated objects are in an "or" relationship.

[0081] The term "at least one" in this article is merely a description of the association relationship of associated objects, indicating that there may be three relationships. For example, at least one of A and B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0082] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusions.

[0083] Example 1

[0084] 1. Construction of biological fertilizer database

[0085] (1) By consulting literature, experimental reports, and project reports, case information on biofertilizers was collected and a case information data set was constructed. Information indicators included the number of effective living bacteria in biofertilizers, organic matter content, longitude and latitude, implementation year, N, P, K, pH, Ni, Cu, Zn, average annual temperature, average annual rainfall, crop growth period, and yield;

[0086] (2) The same method was used to collect the data set of County T (except for the yield), and the annual average temperature data of County T for the past ten years was collected from the website of the regional meteorological bureau, ranging from 15 to 18 degrees. The biological fertilizers applied in County T include two types, namely nitrogen-fixing bacteria and compound bacteria.

[0087] 2. Dimensionality reduction analysis

[0088] (1) Based on the standardized calculation formula, all numerical data except output in the case information dataset and all numerical data in the t county dataset are standardized;

[0089] (2) SPSS software was used to perform PCA analysis on the standardized data to obtain a component matrix, and all indicators were divided into 7 principal components;

[0090] (3) Retrieve the corresponding original data based on the standardized data, perform matrix multiplication on the original data and the component matrix to obtain a reduced-dimensionality data set;

[0091] 3. Cluster analysis

[0092] (1) The case information dataset contains three types of biological fertilizers (nitrogen-fixing bacteria, silicate bacteria, and composite bacteria). The difference in the applicable range of each type of biological fertilizer is calculated, and the ordinary Kriging interpolation method is used to perform spatial interpolation on the yield data to obtain the base value ( Figure 2 );

[0093] (2) According to the formula, calculate the latitude and longitude weight ratio w(lon) and w(lat) of each type of biological fertilizer;

[0094] (3) The K-means clustering method was used to cluster the dimension reduction data sets and weighted longitude and latitude of all bacterial fertilizers into four categories, of which the number of points in category 2 was the largest ( Figure 3 );

[0095] 4. Data Augmentation

[0096] (1) Find the corresponding production data in the case information dataset based on the dimension reduction dataset and merge it with the clustering dataset to obtain a new clustering dataset;

[0097] (2) Using the random classification method, the four new clustering data sets are divided into training sets and test sets in a ratio of 7 / 3;

[0098] (3) The SMOT model was used to perform data augmentation on all the data in the four training sets;

[0099] (4) Merge the augmented data and the training set data to form a new training set;

[0100] 5. Construction of yield prediction model

[0101] (1) Input the new training set data into the random forest model, iterate repeatedly, and modify the relevant parameters (number of trees, minimum number of samples for splitting internal nodes, etc.) until R 2 =0.72;

[0102] (2) Merge the test set data from the four new clustering data sets, extract all indicators in the test set except yield, input them into the trained random forest model, and obtain the predicted yield data;

[0103] (3) Linearly fit the test set yield data with the model predicted yield data to obtain cR 2 =0.75, the model effect meets the requirements;

[0104] 6. Prediction of specific biological fertilizer production

[0105] (1) Replace the annual average temperature of the current year with the historical ten-year average temperature of county t, and perform matrix multiplication operation with the component matrix to obtain the climate dimension reduction data set;

[0106] (2) Input the yield prediction model to obtain all biofertilizer yield data;

[0107] (3) The annual average temperature was classified into 5 levels according to size, and the species sensitivity distribution method was used to establish a fitting curve with the biofertilizer yield data and the corresponding temperature classification data to obtain the two curve areas;

[0108] (4) Draw a predicted distribution map of biofertilizer yield ( Figure 4 ).

[0109] The above description is only the preferred embodiment of the present invention, which does not limit the protection scope of the present invention. For those skilled in the art, the present invention can be modified and varied in various ways. Within the spirit and principle of the present invention, any change, modification, replacement, integration and parameter change of these embodiments by conventional substitution or capable of achieving the same function without departing from the principle and spirit of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for predicting crop yields by applying biofertilizers under multiple climate models, characterized in that: The steps include: (1) Construction of the biofertilizer database, including the collection of case information datasets, research area datasets, and research area historical climate datasets; (2) Dimensionality reduction analysis (2.1) Data standardization The original data x of all indicator data except yield in the case information dataset and all indicator data in the study area dataset are standardized to obtain the standardized data x'. The calculation formula is as follows: x'=[x-ave(x)] / std(x) Where ave(x) is the mean of the original data, and std(x) is the standard deviation of the original data; The indicators and types of all indicator data except output in the case information dataset and the original data of all indicator data in the research area dataset are consistent, and are all digital; (2.2) Data Dimensionality Reduction (2.2.1) Use PCA method to perform dimensionality reduction analysis on the standardized data x' to obtain the component matrix; (2.2.2) Based on the standardized data x' information, find the corresponding original data x of the case information dataset, perform matrix multiplication operation on the original data and the component matrix, and obtain the dimension-reduced dataset; (2.2.3) Extract the dimension reduction data corresponding to the study area data set to obtain the study area dimension reduction data set, and the remaining data is the case dimension reduction data set; (3) Cluster analysis (3.1) Weight ratio (3.1.1) Calculate the difference g in the applicable range of the same type of biological fertilizer and obtain regional information. The specific formula is as follows: g=max[max(lon)-min(lon),max(lat)-min(lat)]*110 Where lon is the longitude of biofertilizer application, lat is the latitude of biofertilizer application; max(lon) is the maximum longitude of biofertilizer application, min(lon) is the minimum longitude of biofertilizer application; max(lat) is the maximum latitude of biofertilizer application, min(lat) is the minimum latitude of biofertilizer application, (3.1.2) Select regional production data, perform spatial interpolation, and obtain the base value c; (3.1.3) Weight ratio calculation Calculate the weighted longitude and latitude w(lon) and w(lat) of the same type of biological fertilizer as follows: w(lon)=g / c*lon w(lat)=g / c*lat The weighted longitude and latitude of all biological fertilizers are calculated using the same method; (3.2) Cluster analysis Using the clustering model, cluster analysis is performed on all indicators and weight latitude and longitude in the case dimension reduction data set to obtain n cluster data sets; Said n≥3; (3.3) Data partitioning (3.3.1) Based on the clustering data set, search for the production data in the corresponding case information data set, merge the production data with the clustering data set, and obtain n new clustering data sets; (3.3.2) All indicators in the n new clustering data sets are randomly divided into two data sets, namely a training set and a test set; the random division ratio and method are consistent in all the clustering data sets; (4) Data augmentation The same data augmentation model is used to augment the n training set data respectively, and the augmented data and the training set data are merged into a new training set; (5) Construction of production prediction model (5.1) Model construction Combine the data of the new training set as input, select the prediction model, adjust the parameters, and perform repeated training until R 2 ≥0.7; (5.2) Model verification (5.2.1) merging the data of the test set; (5.2.2) Extract all indicators except yield from the test set data and input them into the trained model to obtain predicted yield data; (5.2.3) Linearly fit the test set yield index data with the model predicted yield data to obtain cR 2 ; When cR 2 ≥0.7, the yield prediction model training is completed and the next step is entered. 2 <0.7, return to the data amplification step and repeat the subsequent steps until cR 2 ≥0.7; (6) Prediction of specific biological fertilizer production (6.1) Production forecast The climate data in the historical climate data set of the research area is used to replace the climate data in the dimension reduction data set of the research area, and matrix multiplication operation is performed with the component matrix to obtain the climate dimension reduction data set; the yield prediction model is input to obtain the yield prediction data of all biofertilizers; (6.2) Curve fitting (6.2.1) The climate data in the historical climate dataset of the study area are classified from small to large to obtain the amount of climate classification data r; The number of classifications r≥5; (6.2.2) Establish a fitting curve based on the biofertilizer production data and the corresponding climate classification data. Obtain a fitting curve graph for each type of biofertilizer, and calculate the fitting curve area s of all fitting curve graphs; (6.3) Yield forecast distribution map Based on the fitting curve area and meteorological classification data, the yield m of each category of biological fertilizer is calculated, and the predicted distribution map of the biological fertilizer yield under different climatic conditions of all categories is drawn. The calculation formula is as follows: m=s / r.

2. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The fitting curve adopts the species sensitivity distribution method.

3. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The case information of the case information data set includes: one or more of biological fertilizer characteristics, implementation location, implementation year, soil properties, climate information, and crop seed property characteristics.

4. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The characteristics of the biological fertilizer include: type, effective live bacteria abundance, effective live bacteria count, N content, P content, K content, organic matter content; the implementation site includes longitude, latitude, soil properties, including one or more of N, P, K, pH, Ca, Mg, S, Fe, Cu, B, Zn, Mo, Se, Mn, I, and Cl; the climate information includes temperature, humidity, and rainfall; the crop seed trait characteristics include crop type, growth period, protein content, and yield.

5. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The method for obtaining the implementation location is: when the biological fertilizer is implemented by region, the area of ​​the region is <500 square meters, the longitude and latitude of the four boundary points and the center point of the region are collected; the area of ​​the region is ≥500 square meters, in addition to the longitude and longitude of the four boundary points and the center point, the longitude and latitude information in the region is evenly collected according to demand.

6. The method for predicting crop yields by applying biofertilizers under multiple climate models according to claim 1, characterized in that: Relevant information of the study area was collected, the indicators corresponded to the database cases, and the units of the yield data were unified, including kg / mu, t / mu, kg / ha, and t / ha.

7. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The interpolation methods include: ordinary kriging, simple kriging, co-kriging, and pan-kriging. Preferably, the clustering models include GMM, K-means, and hierarchical clustering.

8. The method for predicting crop yields by applying biofertilizers under multiple climate models according to claim 1, characterized in that: The data volume ratio of the training set and the test set includes: 1-9:9-1; further, the data volume ratio of the training set and the test set is 5 / 5, 6 / 4, 7 / 3, 8 / 2 or 9 / 1.

9. The method for predicting crop yield by applying biofertilizer under multiple climate modes according to claim 1, characterized in that: The data augmentation models include SMOT and GNN, and the prediction models include random forest, support vector machine, Markov chain and decision tree.

10. The method for predicting crop yield by applying biofertilizer under multiple climate models according to claim 1, characterized in that: The data volume of the climate dimension reduction data set of the study area is 5-15 times the data volume of the climate dimension reduction data set of the study area, that is, one climate data is replaced by 5-15 historical climate data.

Images