Banana production deduction method and system

Abstract

The invention relates to the technical field of computer systems based on specific calculation models, in particular to a banana production deduction method and system, and the method comprises the steps: obtaining a biological data set of bananas in a target region; obtaining final soil nutrient data based on the biological data set; based on the final soil nutrient data, obtaining soil fertility comprehensive evaluation parameters; constructing a biomass model based on the biological data set; constructing a yield prediction model based on the soil fertility comprehensive evaluation parameters; based on the final soil nutrient data, obtaining soil fertility obstruction factor data and fertilization recommendation data; and based on the soil fertility obstruction factor data and the fertilization recommendation data, obtaining a banana planting production deduction scheme in the target area. The problem that in the prior art, banana production cannot be reasonably deduced to achieve reasonable management of banana production is solved.

Description

Technical Field

[0001] The present application relates to the technical field of computer systems based on specific computing models, and in particular, to a method and system for deducing banana production. Background Art

[0002] Banana (Musa spp.) belongs to the genus Musa, family Musaceae, order Scitamineae, and is an evergreen perennial tropical and subtropical monocotyledonous large herb. According to statistics from the Food and Agriculture Organization of the United Nations (FAO), more than 130 countries or regions in the world are engaged in banana cultivation, with an annual output of more than 100 million tons, making it the fourth largest crop in the world after rice, wheat and corn in terms of output value. Understanding the soil fertility of banana orchards, clarifying the dominant factors of yield, and formulating a reasonable nutrient integrated management strategy are important ways to achieve green, safe and ecologically sustainable development of China's banana industry. However, in the existing technology, there is no method that can reasonably deduce banana production to achieve rational management of banana production. Summary of the invention

[0003] The purpose of the present application is to provide a method and system for deducing banana production, which solves the problem in the prior art that banana production cannot be reasonably deduced to achieve rational management of banana production.

[0004] The technical solution of this application:

[0005] The present application provides a method and system for deducing banana production, including:

[0006] Obtain a biological dataset of bananas in the target area;

[0007] Based on the biological data set, the final soil nutrient data was obtained;

[0008] Based on the final soil nutrient data, comprehensive evaluation parameters of soil fertility are obtained;

[0009] Based on biological data sets, build biomass models;

[0010] Based on the comprehensive evaluation parameters of soil fertility, a yield prediction model was constructed;

[0011] Based on the final soil nutrient data, obtain soil fertility barrier factor data and fertilization recommendation data;

[0012] Based on the soil fertility obstacle factor data and fertilization recommendation data, a banana planting production deduction plan for the target area is obtained.

[0013] Furthermore, the final soil nutrient data is obtained as follows:

[0014] Extract indicator parameter sets based on biological data sets;

[0015] Based on the indicator parameter set, biomass indicator data, physical indicator data, plant nutrient data, and initial soil nutrient data of banana plants in the target area are generated;

[0016] The final soil nutrient data is calculated based on biomass index data, property index data, plant nutrient data, and initial soil nutrient data.

[0017] Furthermore, the indicator parameter set includes at least: soil pH parameter, bulk density parameter, organic matter parameter, alkaline nitrogen parameter, effective phosphorus parameter, available potassium parameter, effective sulfur parameter, exchangeable calcium parameter, exchangeable magnesium parameter, effective iron parameter, effective manganese parameter, effective copper parameter, and effective zinc parameter.

[0018] Furthermore, the biomass model is constructed as follows:

[0019] Based on the biological data set, pseudostem data, leaf data, bud data, and fruit data are obtained;

[0020] Normalize pseudostem data, leaf data, bud data, and fruit data;

[0021] Based on the normalized pseudostem data, a diameter biomass model was constructed;

[0022] Based on the normalized leaf data, a leaf biomass model was constructed;

[0023] Based on the normalized flower bud data, a flower biomass model was constructed;

[0024] Based on the normalized fruit data, a fruit biomass model was constructed;

[0025] Based on the diameter biomass model, leaf biomass model, flower biomass model and fruit biomass model, the banana organ nutrient parameter set was obtained.

[0026] Furthermore, the construction of the yield prediction model is specifically as follows:

[0027] The aforementioned 13 indicator parameters are divided into training set and test set;

[0028] Build a yield prediction model based on the training set and test set;

[0029] Based on grid search and 5-fold cross validation method, the optimal hyperparameters of the yield prediction model are obtained;

[0030] The optimal hyperparameters and training set are input into the yield prediction model for training.

[0031] Furthermore, the construction of the yield prediction model further includes:

[0032] Perform data normalization on data in biological datasets;

[0033] Extraction of key variable sets from biological datasets based on data normalization;

[0034] Based on the key variable set, the indicator parameter set is obtained; among them, the standardized formula is:

[0035]

[0036] Where: Z represents the standardized score, x represents the sample variable, μ represents the sample mean, and σ represents the sample standard deviation.

[0037] Furthermore, the soil fertility comprehensive evaluation parameters are obtained, specifically:

[0038] The weight of soil fertility index distribution in the target area is calculated based on the correlation coefficient method.

[0039] Establish membership function models for different fertility indicators, and obtain membership parameter sets based on the membership function models;

[0040] Based on the membership parameter set, soil property index parameters are obtained;

[0041] The soil property index parameters are converted into dimensionless values ​​between 0.1 and 1.0.

[0042] The present application also provides a banana production deduction system, the system comprising a processor, including:

[0043] A data acquisition module, used to acquire a biological data set of bananas in a target area;

[0044] A data screening module, signal-connected to the data acquisition module, for screening out final soil nutrient data from the biological data set;

[0045] A calculation module is signal-connected to the data screening module and the data acquisition module, and is used to calculate soil fertility comprehensive evaluation parameters according to the biological data set and the final soil nutrient data.

[0046] A biomass model construction module, connected to the calculation module by signal, for constructing a biomass model according to soil fertility comprehensive evaluation parameters;

[0047] A yield prediction model building module, connected to the calculation module signal, is used to build a yield prediction model according to soil fertility comprehensive evaluation parameters;

[0048] An analysis module, connected to the data screening module signal, for obtaining soil fertility obstacle factor data and fertilization recommendation data according to the final soil nutrient data analysis;

[0049] The deduction plan generation module is connected to the analysis module signal and is used to generate a banana planting production deduction plan for the target area based on the soil fertility obstacle factor data and the fertilization recommendation data.

[0050] The technical solution of the present application has at least the following advantages and beneficial effects: the present application provides a deduction method for banana production, which calculates the comprehensive soil fertility index of the banana garden in the target area through the fuzzy comprehensive evaluation method, and clarifies the data of soil fertility obstacle factors; secondly, a banana yield prediction model based on soil nutrient indicators is established by using a machine learning algorithm, which provides a reference for banana yield prediction and clarifies the yield influencing factors; then, the biomass model is used to predict the accumulation of dry matter in each organ of the banana, and the banana nutrient absorption law is obtained in combination with the accumulation of nutrient elements in each organ of the banana; then, based on the final soil nutrient data of the banana garden, fertilization recommendations are made to maximize the efficiency of fertilizer utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of the deduction method provided for this application;

[0052] Figure 2 A schematic diagram of the structure of the deduction system provided for this application;

[0053] Figure 3 Schematic diagram of the correlation coefficient matrix of the yield prediction model variables in this application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0055] Example 1

[0056] It is worth noting that understanding the soil fertility status of banana gardens, clarifying the dominant factors of yield, and formulating reasonable nutrient comprehensive management strategies are important ways to achieve green, safe and ecological sustainable development of the banana industry. This embodiment calculates the soil comprehensive fertility index of banana gardens in the target area by fuzzy comprehensive evaluation method, and clarifies the data of soil fertility barrier factors; secondly, a banana yield prediction model based on soil nutrient indicators is established by using machine learning algorithm to provide reference for banana yield prediction and clarify the yield influencing factors; then, the biomass model is used to predict the accumulation of dry matter in various organs of bananas, and the banana nutrient absorption law is obtained by combining the accumulation of nutrient elements in various organs of bananas; then, according to the final soil nutrient data of banana gardens, fertilization recommendations are proposed to maximize the efficiency of fertilizer utilization; finally, a banana planting production deduction plan is obtained according to the signed content, which solves the problem that banana production cannot be reasonably deduced in the prior art to achieve rational management of banana production.

[0057] Please refer to Figure 1-Figure 3 For the deduction of banana production, this embodiment provides a method for deducing banana production. Taking the emperor banana production area of ​​a certain county in a certain province as an example, the deduction process of this method is explained, which specifically includes:

[0058] First, obtain the biological data set of bananas in the target area: soil samples were collected in the banana producing area of ​​a certain county in a certain province from October 2022 to May 2023, and a total of 163 samples were collected. For each banana garden, 3 to 5 surface soils were randomly collected at a depth of 0 to 20 cm according to the specific fertilization management measures, area and terrain; after the collected soil samples were mixed, 1 kg of soil samples were retained by the quartering method; the collected soil was placed indoors for natural air drying, and impurities such as plant residues, stones, and plastics were removed; the air-dried soil was ground and sieved through 2 mm and 0.149 mm aperture sieves for later use; the longitude and latitude of each sampling point were measured and recorded using a portable GPS. Among them, the pH value was determined by water extraction potential method, with a soil-water ratio of 2.5:1; the organic matter was determined by potassium dichromate oxidation-external heating method; the alkaline nitrogen was determined by alkaline diffusion method; the available phosphorus was determined by ammonium fluoride extraction-molybdenum antimony colorimetric method; the available potassium, exchangeable calcium and exchangeable magnesium were determined by ammonium acetate extraction-inductively coupled plasma emission spectrometry; the available sulfur was determined by phosphate extraction-inductively coupled plasma emission spectrometry; the available iron, available manganese, available copper and available zinc were determined by 0.1 mol / L HCl extraction-inductively coupled plasma emission spectrometry. The bulk density was determined by the ring knife method. Plant samples were collected from 69 healthy and pest-free banana trees in a banana producing area in a county of a certain province from October 2022 to May 2023. 1.5 kg fresh samples of pseudostems (divided into 3 sections, 20 cm was taken from each section and mixed), leaves, flower buds, and fruits were collected. The pseudostems were divided into 3 sections, 20 cm was taken from each section and mixed. The collected samples were wiped clean with a dry towel and placed in a constant temperature drying oven with the temperature set to 105°C for 30 minutes, and then dried at 80°C. The dried samples were crushed and sieved through a 1 mm aperture sieve, and stored dry. The dried, crushed and sieved plant samples were pre-treated by microwave digestion. The semi-micro Kelvin method was used for the determination of nitrogen, the flame spectrophotometry was used for the determination of potassium, and the inductively coupled plasma emission spectrometer was used for the determination of phosphorus, sulfur, calcium, magnesium, iron, manganese, copper and zinc.

[0059] Next, based on the biological data set, the final soil nutrient data is obtained: based on the biological data set, an index parameter set is extracted, wherein the index parameter set includes at least: soil pH parameter, bulk density parameter, organic matter parameter, alkaline nitrogen parameter, effective phosphorus parameter, available potassium parameter, effective sulfur parameter, exchangeable calcium parameter, exchangeable magnesium parameter, effective iron parameter, effective manganese parameter, effective copper parameter, and effective zinc parameter; based on the index parameter set, biomass index data, physical index data, plant nutrient data, and initial soil nutrient data of banana plants in the target area are formed; based on the biomass index data, physical index data, plant nutrient data, and initial soil nutrient data, the final soil nutrient data is calculated. The biological data set obtained from the emperor banana production area in a county in a certain province, the results of the banana garden soil fertility abundance and deficiency index are shown in Table 1 below:

[0060] project Very low Low medium high Very high pH <4.5 4.5～5.8 5.8～6.5 6.5～8.5 >8.5 <![CDATA[Bulk density / (g / cm 3 )]]> <1.00 1.00～1.25 1.25～1.35 1.35～1.45 >1.45 Organic matter / (g / kg) <10 10～20 20～30 30～40 >40 Alkaline nitrogen / (mg / kg) <50 50～100 100～150 150～200 >200 Available phosphorus / (mg / kg) <5 5～10 10～20 20～40 >40 Available potassium / (mg / kg) <50 50～100 100～150 150～250 >250 Available sulfur / (mg / kg) <10 10～16 16～30 30～50 >50 Exchangeable calcium / (mg / kg) <100 100～250 250～1000 1000～2000 >2000 Exchangeable magnesium / (mg / kg) <25 25～50 50～100 100～200 >200 Effective iron / (mg / kg) <2.5 2.5～4.5 4.5～10 10～20 >20 Effective manganese / (mg / kg) <5 5～10 10～20 20～30 >30 Effective copper / (mg / kg) <0.1 0.1～0.2 0.2～1.0 1～2 >2 Effective zinc / (mg / kg) <0.5 0.5～1.0 1.0～2.0 2.0～4.0 >4.0

[0061] Table 1

[0062] Next, based on the final soil nutrient data, the soil fertility comprehensive evaluation parameters are obtained: the weight of the soil fertility index allocation in the target area is calculated based on the correlation coefficient method, and the correlation coefficients of the above 13 indicators are calculated, and the ratio of the average value of the correlation coefficient of a certain indicator to the sum of the average values ​​of the correlation coefficients of all fertility indicators is used as the weight coefficient of the fertility indicator; a membership function model of different fertility indicators is established, and a membership parameter set is obtained based on the membership function model; based on the membership parameter set, soil attribute index parameters are obtained; the soil attribute index parameters are converted into dimensionless values ​​between 0.1 and 1.0; the biological data set obtained from the emperor banana production area of ​​a county in a certain province in this embodiment is combined with the membership function model to obtain the turning point value results of the membership function curve, as shown in Table 2 below:

[0063] index <![CDATA[X1]]> <![CDATA[X2]]> <![CDATA[X3]]> <![CDATA[X4]]> pH 3.5 5.8 6.5 8.5 <![CDATA[Unit weight / (g / cm 3 )]]> 1 1.25 1.35 1.45 Organic matter / (g / kg) 20 30 Alkaline nitrogen / (mg / kg) 100 150 Available phosphorus / (mg / kg) 10 20 Available potassium / (mg / kg) 100 150 Available sulfur / (mg / kg) 16 30 Exchangeable calcium / (mg / kg) 250 1000 Exchangeable magnesium / (mg / kg) 50 100 Effective iron / (mg / kg) 4.5 10 Effective manganese / (mg / kg) 10 20 Effective copper / (mg / kg) 0.2 1 Effective zinc / (mg / kg) 1 2

[0064] Table 2

[0065] Then, based on the biological data set, a biomass model was constructed: based on the biological data set, pseudostem data, leaf data, flower bud data, and fruit data were obtained, and data were extracted from plant samples to obtain plant biological data sets. Specifically, an electronic scale was used to record the total fresh weight of each organ of the emperor banana, including pseudostem, leaf, flower bud, and fruit. According to the fresh weight and dry weight data of the 1.5 kg sample obtained, the total dry weight of each organ of the emperor banana was calculated. Then, a tape measure and a vernier caliper were used to measure the stem diameter (SD), stem height (H), leaf length (LL), leaf width (LW), leaf base length (LBL), flower bud length (FL), flower bud width (FW), and flower bud base diameter (FB). D), bud aspect ratio (FAR), fruit comb number (FrN), fruit stalk diameter (FrBD), fruit stalk length (FrBL), fruit finger inner arc length (FrI), fruit finger outer arc length (FrO), fruit finger diameter (FrC), based on which the semi-micro Kelvin method and other methods were used to determine the element content of each organ of Emperor Banana. In detail, this embodiment selected the morphological indicators of Emperor Banana that are easy to measure and the morphological indicators that have an important influence on each organ as measurement indicators, based on which the diameter biomass model, leaf biomass model, flower biomass model, and fruit biomass model were constructed; the pseudostem data, leaf data, flower bud data, and fruit data Normalization was performed, specifically using the Min-Max normalization method to process the data, and the original data was mapped to between 0 and 1 through linear transformation; based on the normalized pseudostem data, a diameter biomass model was constructed, and based on the normalized leaf data, a leaf biomass model was constructed. Specifically, the stem biomass model and the leaf biomass model used stem diameter (SD), plant height (H), leaf length (LL), leaf width (LW) and leaf base length (LBL) as input variables, and the unit of each variable was meter. The leaf biomass model was a biomass model based on each individual leaf; based on the normalized bud data, a flower biomass model was constructed. Specifically, the flower biomass Model: Bud length (FL), bud width (FW), bud base diameter (FBD) and bud length-to-width ratio (FAR) are used as input variables, and the unit of each variable is centimeter; based on the normalized fruit data, a fruit biomass model is constructed. Specifically, the fruit biomass model: the number of fruit combs (FrN), fruit stalk diameter (FrBD), fruit stalk length (FrBL), fruit finger inner arc length (FrI), fruit finger outer arc length (FrO) and fruit finger diameter (FrC) are used as input variables, and the unit of each variable is centimeter; based on the diameter biomass model, leaf biomass model, flower biomass model, and fruit biomass model, a banana organ nutrient parameter set is obtained. In detail, the normalized pseudostem data, leaf data, bud data, and fruit data are divided into 70% as a training set and 30% as a test set, and the stepwise linear regression method is used to construct and verify the model. The whole process is iterated at least 5 times. The conversion formula (Formula 1) and the normalization conversion formula (Formula 1) are respectively:

[0066]

[0067] In Formula 2, X represents the normalized score, x represents the sample variable, and x min represents the minimum value of the sample, x max Indicates the maximum value of the sample.

[0068] In detail, the method of this embodiment is to establish a stem biomass model by using stepwise linear regression. The stem diameter and plant height have the highest fitting degree for the stem biomass. The model training set R 2 is 0.65, RMSE and MAE are 0.35 and 0.27 respectively, and the test set R 2 The value of the leaf biomass model was 0.64, the RMSE and MAE were 0.33 and 0.26 respectively, the F test result of the model was extremely significant, and the t test of each coefficient was significant. The fitting equation was: SDW = -1.9604 + 4.3563 × SD + 0.3384 × H. The leaf biomass model was established by stepwise linear regression. The stem diameter, leaf width and leaf base length had the highest fitting degree for leaf biomass. The model training set R 2 The test set R 2 The value of RMSE and MAE was 0.03 and 0.03 respectively. The F test result of the model was extremely significant, and the t test of each coefficient was significant. The fitting equation was: LDW = -0.2160 + 0.1691 × SD + 0.3291 × LW + 0.1786 × LBL. The flower biomass model was established by stepwise linear regression. The bud width, bud base diameter and bud length-width ratio had the highest fitting degree for flower biomass. The model training set R 2 is 0.78, RMSE and MAE are 0.01 and 0.01 respectively, and the test set R 2 The value of the model was 0.67, the RMSE and MAE were 0.01 and 0.01 respectively, the F test result of the model was extremely significant, and the t test of each coefficient was significant. The fitting equation was: FDW = -0.1073 + 0.0095 × FW + 0.0068 × FBD + 0.0174 × FAR. The fruit biomass model was established by stepwise linear regression. The fruit comb number, fruit stalk diameter and fruit finger diameter ratio had the highest fitting degree for fruit biomass. The model training set R 2 is 0.90, RMSE and MAE are 0.10 and 0.08 respectively, and the test set R 2 It is 0.71, RMSE and MAE are 0.15 and 0.10 respectively, the model F test result is extremely significant, and the t test of each coefficient is significant. The fitting equation is: FrDW=-1.1945+0.0317×FrN-0.1791×FrBD+0.2041×FrC.

[0069] Then, based on the comprehensive evaluation parameters of soil fertility, a yield prediction model is constructed: the yield prediction model is constructed, specifically: data in the biological data set is performed data standardization, through which the numerical ranges of different features can be unified to the same scale, the influence of the dimension can be eliminated and the model can be more stable and treat each feature fairly, especially in models optimized by gradient descent algorithm such as neural networks, which also helps to improve the convergence speed of the model, and each feature can be fairly evaluated for impact; based on the data-standardized biological data set, a key variable set is extracted. This step can not only reduce the complexity of the model and make it easier to interpret, but also help to improve the performance of the model, reduce the occurrence of overfitting problems, and maintain good prediction ability. In this embodiment, correlation analysis is used to eliminate factors that are not significantly correlated with yield, and then a stepwise multiple regression method is used to screen out variables closely related to yield; based on the key variable set, an indicator parameter set is obtained; the aforementioned 13 indicator parameters are divided into a training set and a test set; based on the training set and the test set, a yield prediction model is constructed; based on the grid search and 5-fold cross-validation method, the optimal hyperparameters of the yield prediction model are obtained; the optimal hyperparameters and the training set are input into the yield prediction model for training. The normalization formula is:

[0070]

[0071] Where: Z represents the standardized score, x represents the sample variable, μ represents the sample mean, and σ represents the sample standard deviation.

[0072] In addition, this embodiment uses and tests multiple models such as RF, SVM, KNN, ANN, etc. Specifically, in the process of obtaining and using the optimal hyperparameters, the optimal hyperparameters of each model are obtained on the training set by using the grid search and 5-fold cross validation method, as shown in Table 3. Then the model is trained on the training set using the optimal hyperparameters, and the model accuracy and generalization performance are evaluated on the test set. The entire model training and verification process is iterated at least 5 times. This embodiment uses the model determination coefficient (R2), root mean square error (RMSE), and mean absolute error (MAE) to evaluate the overall accuracy of the model.

[0073]

[0074]

[0075] Table 3

[0076] It should be noted that in order to ensure that the yield forecasting model has sufficient forecasting ability and is not affected by too many variables, the Pearson correlation analysis method is used to eliminate redundant and irrelevant input variables. Figure 3It can be seen that the yield of Emperor Banana is significantly or extremely significantly correlated with 12 input variables such as soil bulk density, pH value, organic matter, alkaline nitrogen, available phosphorus, available potassium, exchangeable calcium, exchangeable magnesium, available iron, available manganese, available copper, and available zinc, but is not significantly correlated with available sulfur. Therefore, this embodiment first eliminates the variable of available sulfur. Then, taking the yield of Emperor Banana as the dependent variable, soil alkaline nitrogen, available phosphorus, available potassium and other soil factors as independent variables, the stepwise regression method is used to calculate the correlation between each variable factor and the yield of Emperor Banana to determine the key variables that have an important impact on the yield of Emperor Banana. As can be seen from Table 4, the VIF (Variance Inflation Factor) values ​​of each variable are all <5, indicating that there is no multicollinearity problem between the variables. Further, the significance test results of each variable in the model show that the P values ​​of 7 variables of available potassium, alkaline nitrogen, exchangeable calcium, exchangeable magnesium, available iron, available manganese, and available zinc are <0.05, indicating that these variables have a significant impact on the target variable. Therefore, this embodiment selects 7 variables, namely, available potassium, alkaline nitrogen, exchangeable calcium, exchangeable magnesium, effective iron, effective manganese, and effective zinc, as input variables of the machine learning prediction model. After testing and comparison, the ANN model is more effective in predicting the yield of Emperor Banana than the SVM, RF, and KNN models. 2 The RMSE and MAE are 0.11 and 0.07 respectively, and the test set R 2 The results of the yield prediction model showed that soil available potassium, alkaline nitrogen, exchangeable calcium, and exchangeable magnesium were important factors affecting the yield of Emperor banana. When the available potassium content in the banana garden exceeded 100 mg / kg, the alkaline nitrogen content exceeded 100 mg / kg, the exchangeable calcium content exceeded 600 mg / kg, and the exchangeable magnesium content exceeded 75 mg / kg, the effect on the yield of Emperor banana changed from inhibition to promotion. In addition, there was a synergistic effect between soil exchangeable calcium and exchangeable magnesium. As the content of both increased, the yield of Emperor banana could be effectively promoted. When the soil exchangeable calcium and magnesium were deficient, increasing the content of soil available manganese and available zinc could not directly promote the yield of Emperor banana, but could alleviate the deficiency of banana garden. The ANN model had no overfitting problem, and the model error had converged under the sample size of 100. The ANN model was more effective in predicting the yield of Emperor banana. The screening results of model variables by stepwise regression analysis are shown in Table 4:

[0077]

[0078]

[0079] Table 4

[0080] Then, based on the final soil nutrient data, the soil fertility barrier factor data and fertilization recommendation data were obtained: the load between each fertility index and IFI in the banana garden soil was analyzed by principal component analysis. There were 4 principal components with characteristic values ​​greater than 1, and the variance contribution rates were 28.51%, 17.21%, 15.86%, and 8.29%, respectively. The cumulative variance contribution rate was 69.87%, which basically reflected the basic information of the soil fertility in the banana garden. The determining factors of the first principal component (PC1) were IFI, soil exchangeable calcium, exchangeable magnesium, effective iron, effective manganese, and effective zinc, and the loading values ​​were all greater than 0.60, indicating that the exchangeable calcium, exchangeable magnesium, effective iron, effective manganese, and effective zinc content in the emperor banana soil had the greatest impact on the comprehensive fertility of the soil. Soil bulk density, organic matter, and alkaline nitrogen were the determining factors of the second principal component (PC2), and the loading values ​​were -0.65, 0.85, and 0.70, respectively. pH, effective phosphorus, effective sulfur, and effective copper were in the third principal component (PC3). Available potassium is the determining factor of the fourth principal component (PC4), with a loading value of 0.54. In the emperor banana production area of ​​a certain county in a certain province in this embodiment, the soil pH value is acidic as a whole, the texture is loose, the alkaline nitrogen content is at a low level, the organic matter, available potassium and exchangeable calcium content are at a medium level, the exchangeable magnesium content is at a high level, and the available phosphorus, available sulfur, available iron, available manganese, available copper, and available zinc content are at extremely high levels.

[0081] Based on the above analysis, the soil in the banana production area of ​​a certain county in a certain province in this embodiment is generally acidic, with a loose texture, low alkaline nitrogen content, medium organic matter, quick-acting potassium and exchangeable calcium content, rich exchangeable magnesium content, and extremely high effective phosphorus, effective sulfur, effective iron, effective manganese, effective copper, and effective zinc content. Its fertilization data and fertilization recommendations are specifically that the pH value and organic matter content of the banana garden have both decreased, which is actually due to excessive nitrogen fertilizer application and insufficient use of organic fertilizer. In addition, attention should still be paid to potassium supplementation during the banana planting process to ensure the normal growth and development of the banana. The application of different varieties of silicon calcium potassium magnesium fertilizers can increase the banana yield by 3.82% to 5.64%. During the banana planting process, attention should be paid to the content of soil calcium and magnesium, and reasonable fertilization measures should be taken to improve the quality and yield of bananas. In terms of fertilization management, potassium fertilizer and organic fertilizer should be added, nitrogen fertilizer should be applied reasonably, and soil bulk density should be improved to maintain the stability of soil fertility while ensuring that the content of exchangeable calcium, exchangeable magnesium, effective iron, effective manganese, effective zinc and other elements in the soil is sufficient. In this embodiment, the average element accumulation per emperor banana plant is: 127.50g nitrogen, 13.07g phosphorus, 304.31g potassium, 51.15g calcium, 13.37g magnesium, 11.38g sulfur, 0.99g iron, 7.76g manganese, 0.07g copper, 0.09g zinc, and the nitrogen, phosphorus and potassium accumulation ratio is 1.00:0.10:2.39. Based on the soil conditions of the emperor banana garden in the emperor banana production area of ​​a certain county in a certain province, it is recommended to add 29.79kg / mu of nitrogen fertilizer and 76.88kg / mu of potassium fertilizer for each harvest of emperor banana. The recommended fertilization formula for each nutrient element of the emperor banana harvest is:

[0082] N(kg)=-209.10+108.09×SD+2.23×H+154.46×LW+83.81×LBL+0.25×FW+0.18×FBD+0.46×FAR+0.50×FrN-2.82×FrBD+3.21×FrC;

[0083] P(kg)=-180.14+33.55×SD+0.98×H+40.64×LW+22.05×LBL+0.12×FW+0.09×FBD+0.23×FAR+0.20×FrN-1.13×FrBD+1.29×FrC;

[0084] K(kg)=-238.50+203.55×SD+10.51×H+132.90×LW+72.11×LBL+0.36×FW+0.26×FBD+0.67×FAR+0.68×FrN-3.83×FrBD+4.37×FrC.

[0085] Ca(kg)=-2854.61+253.51×SD+8.18×H+288.34×LW+156.46×LBL+0.12×FW+0.08×FBD+0.21×FAR+0.27×FrN-1.52×FrBD+1.73×FrC;

[0086] Mg(kg)=-577.78+55.69×SD+1.69×H+66.00×LW+35.82×LBL+0.11×FW+0.08×FBD+0.20×FAR+0.24×FrN-1.38×FrBD+1.58×FrC;

[0087] S(kg)=-391.40+45.71×SD+0.89×H+66.69×LW+36.19×LBL+0.10×FW+0.07×FBD+0.17×FAR+0.14×FrN-0.78×FrBD+0.89×FrC.

[0088] FFe(g)=-374116.61+4527.03×SD+118.31×H+5846.44×LW+3172.37×LBL+3.23×FW+2.31×FBD+5.89×FAR+6.63×FrN-37.44×FrBD+42.66×FrC;

[0089] Mn(g)=-274480.89+33484.64×SD+652.78×H+48814.36×LW+26487.40×LBL+15.44×FW+11.05×FBD+28.20×FAR+50.64×FrN-285.93×FrBD+325.78×FrC;

[0090] Cu(g)=-38608.32+175.32×SD+3.99×H+241.31×LW+130.94×LBL+0.74×FW+0.53×FBD+1.35×FAR+2.87×FrN-16.22×FrBD+18.48×FrC;

[0091] Zn(g)=-17127.35+347.18×SD+11.61×H+384.70×LW+208.74×LBL+1.34×FW+0.96×FBD+2.45×FAR+2.27×FrN-12.81×FrBD+14.60×FrC.

[0092] The results of principal component analysis of soil fertility indicators in banana gardens are shown in Table 5:

[0093]

[0094]

[0095] Table 5

[0096] Finally, based on the soil fertility barrier factor data and fertilization recommendation data, a banana planting production deduction plan for the target area was obtained.

[0097] Example 2

[0098] Please refer to Figure 2 The present application also provides a banana production deduction system, the system comprising a processor, including:

[0099] A data acquisition module, used to acquire a biological data set of bananas in a target area;

[0100] A data screening module, signal-connected to the data acquisition module, for screening out final soil nutrient data from the biological data set;

[0101] A calculation module is signal-connected to the data screening module and the data acquisition module, and is used to calculate soil fertility comprehensive evaluation parameters according to the biological data set and the final soil nutrient data.

[0102] A biomass model construction module, connected to the calculation module by signal, for constructing a biomass model according to soil fertility comprehensive evaluation parameters;

[0103] A yield prediction model building module, connected to the calculation module signal, is used to build a yield prediction model according to soil fertility comprehensive evaluation parameters;

[0104] An analysis module, connected to the data screening module signal, for obtaining soil fertility obstacle factor data and fertilization recommendation data according to the final soil nutrient data analysis;

[0105] The deduction scheme generation module is signal-connected to the biomass model construction module, the yield prediction model construction module, and the analysis module, and is used to generate a banana planting production deduction scheme for the target area based on the soil fertility obstacle factor data and the fertilization recommendation data.

[0106] So far, various embodiments of the present disclosure have been described in detail. In order to avoid obscuring the concept of the present disclosure, some details known in the art are not described. Based on the above description, those skilled in the art can fully understand how to implement the technical solution disclosed here.

[0107] Although some specific embodiments of the present disclosure have been described in detail by way of examples, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present disclosure. It should be understood by those skilled in the art that the above embodiments may be modified or some technical features may be replaced by equivalents without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A method for deducing banana production, characterized in that: include: Obtain a biological dataset of bananas in the target area; Based on the biological data set, the final soil nutrient data was obtained; Based on the final soil nutrient data, comprehensive evaluation parameters of soil fertility are obtained; Based on biological data sets, build biomass models; Based on the comprehensive evaluation parameters of soil fertility, a yield prediction model was constructed; Based on the final soil nutrient data, obtain soil fertility barrier factor data and fertilization recommendation data; Based on the soil fertility obstacle factor data and fertilization recommendation data, a banana planting production deduction plan for the target area is obtained.

2. The deduction method according to claim 1, characterized in that: The final soil nutrient data is obtained as follows: Extract indicator parameter sets based on biological data sets; Based on the indicator parameter set, biomass indicator data, physical indicator data, plant nutrient data, and initial soil nutrient data of banana plants in the target area are generated; The final soil nutrient data is calculated based on biomass index data, property index data, plant nutrient data, and initial soil nutrient data.

3. The deduction method according to claim 2, characterized in that: The indicator parameter set includes at least: soil pH parameter, bulk density parameter, organic matter parameter, alkaline nitrogen parameter, effective phosphorus parameter, available potassium parameter, effective sulfur parameter, exchangeable calcium parameter, exchangeable magnesium parameter, effective iron parameter, effective manganese parameter, effective copper parameter, and effective zinc parameter.

4. The deduction method according to claim 2, characterized in that: The biomass model is constructed as follows: Based on the biological data set, pseudostem data, leaf data, bud data, and fruit data are obtained; Normalize pseudostem data, leaf data, bud data, and fruit data; Based on the normalized pseudostem data, a diameter biomass model was constructed; Based on the normalized leaf data, a leaf biomass model was constructed; Based on the normalized flower bud data, a flower biomass model was constructed; Based on the normalized fruit data, a fruit biomass model was constructed; Based on the diameter biomass model, leaf biomass model, flower biomass model and fruit biomass model, the banana organ nutrient parameter set was obtained.

5. The deduction method according to claim 3, characterized in that: The construction of the yield prediction model is specifically as follows: The aforementioned 13 indicator parameters are divided into training set and test set; Build a yield prediction model based on the training set and test set; Based on grid search and 5-fold cross validation method, the optimal hyperparameters of the yield prediction model are obtained; The optimal hyperparameters and training set are input into the yield prediction model for training.

6. The deduction method according to claim 5, characterized in that: The method of constructing the yield prediction model further includes: Perform data normalization on data in biological datasets; Extraction of key variable sets from biological datasets based on data normalization; Based on the key variable set, the indicator parameter set is obtained; among them, the standardized formula is: Where: Z represents the standardized score, x represents the sample variable, μ represents the sample mean, and σ represents the sample standard deviation.

7. The deduction method according to claim 3, characterized in that: The soil fertility comprehensive evaluation parameters are as follows: The weight of soil fertility index distribution in the target area is calculated based on the correlation coefficient method. Establish membership function models for different fertility indicators, and obtain membership parameter sets based on the membership function models; Based on the membership parameter set, soil property index parameters are obtained; The soil property index parameters are converted into dimensionless values ​​between 0.1 and 1.

0.

8. A banana production deduction system, characterized in that: The system includes a processor, including: A data acquisition module, used to acquire a biological data set of bananas in a target area; A data screening module, signal-connected to the data acquisition module, for screening out final soil nutrient data from the biological data set; A calculation module is signal-connected to the data screening module and the data acquisition module, and is used to calculate soil fertility comprehensive evaluation parameters according to the biological data set and the final soil nutrient data. A biomass model construction module, connected to the calculation module by signal, for constructing a biomass model according to soil fertility comprehensive evaluation parameters; A yield prediction model building module, connected to the calculation module signal, is used to build a yield prediction model according to soil fertility comprehensive evaluation parameters; An analysis module, connected to the data screening module signal, for obtaining soil fertility obstacle factor data and fertilization recommendation data according to the final soil nutrient data analysis; The deduction scheme generation module is signal-connected to the biomass model construction module, the yield prediction model construction module, and the analysis module, and is used to generate a banana planting production deduction scheme for the target area based on the soil fertility obstacle factor data and the fertilization recommendation data.

Images