Screening method, storage medium and equipment for peanut varieties suitable for mechanized harvesting

Abstract

The present invention relates to a method, storage medium, and equipment for screening peanut varieties suitable for mechanized harvesting, and belongs to the field of agricultural technology. To address the problem that differences between experimental fields and planted fields may cause a significant difference in the total loss rate of peanuts in actual planted areas during mechanical harvesting compared to the experimental fields, further leading to inaccurate estimates of the total loss rate in actual planted areas, the present invention first obtains basic indicator data for peanut variety i to be screened and the corresponding consolidation degree of peanut variety i; then uses a neural network to obtain the total retention rate corresponding to variety i; and then, based on the consolidation degree corresponding to peanut variety i and the consolidation degree F of the plot to be sown, calculates the total retention rate corresponding to variety i. ij' Get the consolidation difference ratio R ij' And use the neural network to get the total retention rate difference; finally, according to the total retention rate and total retention rate difference Δ corresponding to variety i ij' The total retention rate of peanut variety i in the plot to be sown is obtained, and then the peanut variety to be planted is determined. The present invention is suitable for screening peanut varieties for mechanized harvesting.

Description

Technical Field

[0001] The invention relates to a method for screening peanut varieties suitable for mechanized harvesting, and belongs to the technical field of agriculture. Background Art

[0002] Peanuts are an important food and oil crop worldwide. my country accounts for over 50% of the global demand for peanut oil. China leads the world in peanut cultivation area, total production, and export volume. With the rapid development of peanut cultivation and processing technology, competition with major peanut exporting countries is intensifying. However, my country's relatively low level of peanut cultivation and production, particularly mechanized harvesting, hinders its export competitiveness and profitability.

[0003] Peanut harvesting is a crucial step in the entire peanut cultivation process. It consumes a significant amount of farming time, is highly seasonal, and employs approximately two-thirds of the labor force. Peanut harvesting, in particular, remains primarily animal-powered and manual, supplemented by mechanical harvesting. Mechanized peanut harvesting encompasses multiple processes, primarily digging and removing soil, laying out the peanuts for drying, and picking and sifting the pods. Compared to manual harvesting, mechanical peanut harvesting offers the advantages of lowering costs and labor requirements. While this reduces peanut harvesting costs, it also increases pod loss during the harvest process.

[0004] Currently, there are not many studies on mechanical harvesting of peanuts. The few studies that exist basically test and analyze the agronomic traits of peanuts to determine whether they meet the requirements of mechanized harvesting. This approach has the following problems:

[0005] This method basically requires planting trials to determine various agronomic traits, and the trials are all based on experimental fields. When actually planting, it is sufficient to select varieties with appropriate mechanical requirements. However, this method ignores the fact that there are differences between experimental fields and actual widely planted experimental fields, and there may even be huge differences. Generally, there will be large differences between experimental fields and widely planted areas if they are not in adjacent planting areas. If the two plots are far apart, or even not in the same jurisdiction, there will be huge differences. For example, the land properties of Huanan and Gannan, which belong to the northeastern planting area, are very different. Even different farms in the same administrative city will have differences. Such differences in land properties will lead to large differences in the degree of consolidation affected by soil texture, such as heavy weight. Therefore, there are large differences in the mechanical properties of the land during fruit picking, which in turn leads to a small total loss rate (fruit drop rate + fruit burial rate) in the experimental field, but a high total loss rate in field planting. Therefore, the accuracy of the screening effect of the above method is difficult to guarantee.

[0006] In order to solve the above problems, it is possible to set up test fields in the planned planting areas and conduct planting trials. Although the above problems can be solved under certain conditions, there are specific setting and testing requirements for test fields, such as analysis of soil composition and planting conditions of previous crops, so it is difficult to set up test fields directly in the planned planting areas. In actual planting production, planting trials will be carried out in the test fields first, and then suitable varieties will be screened. Then, small-scale trial planting will be carried out in the planned planting areas. If the effect is good, large-scale planting will be promoted. Although this method is relatively effective and widely used, it still has problems. For example, there are still differences between test fields and small-scale areas, which may lead to unsatisfactory total loss rates when small-scale trial planting is still present, and may even waste a "planting-maturity" cycle, resulting in a very long time to actually screen the right varieties. Summary of the Invention

[0007] The present invention aims to solve the problem that due to the differences between the experimental fields and the planting fields, the total loss rate of peanuts in the actual planting area may be greatly different from that in the experimental fields during mechanical harvesting, which further leads to inaccurate estimation of the total loss rate in the actual planting area.

[0008] A method for screening peanut varieties suitable for mechanized harvesting, the method comprising the following steps:

[0009] S1. Obtain the basic index data of peanut variety i that needs to be screened and the consolidation degree F corresponding to peanut variety i i ;

[0010] Normalize the basic indicator data of variety i; and perform One-Hot encoding and normalization on variety i;

[0011] Then, the normalized value of the One-Hot encoding of variety i and the normalized value of the basic indicator data are input into the total retention rate network model to obtain the total retention rate corresponding to variety i;

[0012] The total retention rate network model is a multi-layer neural network model;

[0013] S2. Obtain the consolidation degree F for the plot j' to be sown ij' ;

[0014] According to the consolidation degree F corresponding to peanut variety i i and the degree of consolidation F of the plot to be sown ij' Get the consolidation difference ratio R ij' ;

[0015] Then the One-Hot encoding normalized value of variety i and the consolidation difference ratio R ij' Input the total retention rate difference network model to obtain the total retention rate difference Δij' ;

[0016] The total retention rate difference network model is a multi-layer neural network model;

[0017] S3, the total retention rate corresponding to variety i and the total retention rate difference Δ ij' Get the total retention rate of peanut variety i in the plot to be sown;

[0018] S4. Determine the peanut variety to be planted based on the total retention rate of peanut variety i in the plot to be sown.

[0019] Furthermore, the consolidation degree difference ratio

[0020] Furthermore, the consolidation degree corresponding to peanut variety i Among them F ij represents the j-th consolidation degree among the J land consolidation degree samples corresponding to peanut variety i.

[0021] Furthermore, the degree of consolidation F ij' For land resistance.

[0022] Furthermore, the soil resistance is obtained by burying an auxiliary measuring device in the soil during crop planting and measuring the maximum force obtained by pulling out the auxiliary measuring device during harvesting.

[0023] Furthermore, the basic indicator data of the peanut variety include: grain width, grain thickness, grain vertical pressure, fruit stalk strength on the 7th day, vine stalk strength on the 7th day, fruit stalk strength when digging, vine stalk strength when digging, grain positive pressure, grain lateral pressure, grain length, fruiting range, shell positive pressure, shell lateral pressure and shell vertical pressure.

[0024] Furthermore, the process of normalizing the basic indicator data of variety i includes the following steps:

[0025] The basic indicator data are recorded as X1-X 14 , the normalized value of each indicator Xn is calculated as follows:

[0026] Xn=(Xn-Xnmin) / λ(Xnmax-Xnmin), n=1,2,...,14,

[0027] Among them, Xnmax and Xnmin are the maximum and minimum values ​​of the indicator Xn in all peanut varieties; λ is the normalized adjustment ratio.

[0028] Furthermore, the normalized adjustment ratio λ is set to 1.1.

[0029] A storage medium stores at least one instruction, which is loaded and executed by a processor to implement the method for screening peanut varieties suitable for mechanized harvesting.

[0030] A screening device for peanut varieties suitable for mechanized harvesting comprises a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the screening method for peanut varieties suitable for mechanized harvesting.

[0031] Beneficial effects:

[0032] By using the present invention, before actual planting, as long as the soil resistance of the plot to be planted is obtained, or the soil resistance of the plot to be planted is obtained in advance by a soil resistance auxiliary measurement device, then the soil resistance F corresponding to the peanut variety i can be calculated based on the basic indicator data provided by the seed provider. i A prior estimate of the total peanut retention rate of the planned planting plots can be made, so that even if there are large differences between the experimental fields and the planned planting plots, a relatively accurate estimate of the total loss rate / total retention rate of peanuts in the actual planting area during mechanical harvesting can be achieved, thereby effectively controlling a satisfactory loss amount after the actual planting production and harvest, thereby effectively controlling the actual loss.

[0033] The present invention can be used directly for relatively large-scale planting based on the established database, eliminating the need for small-scale trial planting, thereby saving the "planting-maturity" cycle and further shortening the time period for actually screening suitable varieties. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is the result diagram of fruit stalk strength;

[0035] Figure 2 This is the result diagram of the shell strength;

[0036] Figure 3 This is the correlation analysis diagram of each trait;

[0037] Figure 4 Cluster analysis diagram for 14 traits;

[0038] Figure 5 It is an auxiliary measuring device for land resistance. DETAILED DESCRIPTION

[0039] If a peanut variety database exists (agricultural science observation and experimental stations are built throughout the country, and these experimental stations have a large amount of crop variety data, such as the 126 peanut production varieties and farmer materials collected and preserved by the Peanut Research Institute of Shenyang Agricultural University, a scientific observation and experimental station for crop cultivation in the Northeast region of the Ministry of Agriculture and Rural Affairs. These agricultural science observation and experimental stations have built or are building a crop variety database based on the data), the present invention can directly conduct small-scale trial planting using the relevant data of peanut varieties, which can largely ensure that the mechanical properties of the soil in a small area meet the final requirements, greatly reducing the probability of an unsatisfactory total loss rate during small-scale trial planting, thereby ensuring that the total loss rate of the trial planting is close to the total loss rate of the experimental field, and greatly improving the seed selection success rate during trial planting. Therefore, the "planting-maturity" cycle can be saved, and the time period for truly screening for suitable varieties can be shortened. In fact, the present invention can be directly used for relatively large-scale planting (not within the scope of trial planting), which can save the link of small-scale trial planting, further save the "planting-maturity" cycle, and further shorten the time period for truly screening for suitable varieties.

[0040] Of course, if there is no peanut variety database, a database can be established through experimental field planting. Although the experimental field planting link is still required, it should be noted that this link is only required during the first seed selection. In the subsequent seed selection process, as long as the variety is in the database, the above-mentioned effect will still be achieved. It should also be noted that even if experimental field planting data is required when the database is first established, the entire process, including establishing the database and directly using the database for trial planting or large-scale planting, as long as the database is established, it can be directly used for relatively large-scale planting based on the established database, and the small-scale trial planting link can be omitted, thereby saving the "planting-maturity" cycle and further shortening the time period for actually screening suitable varieties.

[0041] In order to fully illustrate the present invention, the present invention will now be described in detail in conjunction with the process of establishing a database. Specific implementation method one:

[0043] This embodiment is a method for screening peanut varieties suitable for mechanized harvesting, which specifically includes the following steps:

[0044] 1. Establishing a peanut parameter database through experiments:

[0045] (1) Resource garden planting experiment:

[0046] The test was conducted at the Shenyang Agricultural University experimental field (41°N, 123°E), part of the Northeast China Crop Cultivation Science Observation and Experimental Station of the Ministry of Agriculture and Rural Affairs. The soil type was brown soil with 101.5 mg / kg of alkaline-hydrolyzable nitrogen, 25.4 mg / kg of available phosphorus, 112.7 mg / kg of available potassium, a pH of 6.7, and 9.1 g / kg of organic carbon. The terrain was flat and well-drained, and the previous crop was corn.

[0047] Experimental materials: The materials used were 126 materials collected and preserved by the Peanut Research Institute of Shenyang Agricultural University (Table 1), including cultivated varieties and farmer materials, and the indicators related to mechanized planting production were measured.

[0048] Table 1 Peanut varieties tested

[0049]

[0050]

[0051]

[0052] A resource garden planting experiment: 126 peanut germplasm accessions were selected from the resource garden of the Peanut Research Institute of Shenyang Agricultural University. A single-ridge, small double-row sowing method was used, with a density of 200,000 plants per hectare, and each variety was planted in single rows at 4-meter intervals. Field management followed conventional field practices, and harvesting was performed manually at harvest time. Each variety was measured for pedicel strength (seedling-pedicel, fruit-pedicel), shell strength (positive pressure, lateral pressure, and vertical pressure), kernel strength (positive pressure, lateral pressure, and vertical pressure), fruiting range, kernel length, width, and thickness, kernel shape (including pod mouth visibility, pod constriction, pod surface texture, and pod type), and other indicators related to seed testing and yield assessment.

[0053] The plant status parameters of each peanut variety were measured, including: fruit stalk strength, fruit shell strength, seed strength, fruiting range, fruiting depth, fruit needle length, seed length, width and thickness, yield measurement-related and mechanization-related indicators (soil carrying rate, fruit drop rate, fruit buried rate, and total loss rate).

[0054] Measurement indicators:

[0055] Fruit stalk strength: On the first and seventh days after the harvest, the peanut plants were naturally air-dried. The peanut vines and pods were fixed on a force gauge. The fruit stalk strength (fruit-stalk strength, vine-stalk strength) of the peanuts was accurately measured using a digital force gauge. 15 pods of each variety were measured and the average value was taken.

[0056] Shell and kernel strength: Determine the pod-cracking and kernel-cracking forces on sun-dried peanut pods and kernels. Place the pod or kernel under the force gauge of a digital dynamometer and slowly press down until the pod or kernel cracks. The force applied at this point is the shell and kernel strength. Each parameter is measured repeatedly on 10 pods or kernels, and the average value is taken.

[0057] Fruiting range: the distance from the main stem as the central axis to the outermost fruit.

[0058] Fruiting depth: the distance from the ground surface to the deepest pod location.

[0059] Fruit needle length: Measure the length of all fruit needles of a single peanut plant and calculate the average value as the fruit needle length.

[0060] Grain length, width and thickness: Use a digital vernier caliper to measure the length, width and thickness of the grains. Measure 20 kernels for each variety and take the average value.

[0061] Soil rate: the mass of soil in the sample / the mass of the sample

[0062] Fruit drop rate: all pods on the ground / (all pods on the ground + pods buried in the soil + pods on peanut plants in the field)

[0063] Buried fruit rate: pods buried in the soil / (all pods on the ground + pods buried in the soil + pods on peanut plants in the field)

[0064] Total loss rate: Total loss rate = fruit drop rate + fruit burial rate;

[0065] Total retention rate: Total retention rate = 1 - total loss rate.

[0066] After the resource garden planting experiment, the total retention rate Y corresponding to peanut variety i is obtained i , and the soil resistance of the corresponding planting area of ​​peanut variety i;

[0067] The total retention rate Y corresponding to peanut variety i i The past process of is as follows,

[0068] First, obtain the total retention rate of all plants corresponding to peanut variety i, and then calculate the average value Y of the total retention rate of all plants i , as the total retention rate Y corresponding to peanut variety i i .

[0069] The process of obtaining the soil resistance of the planting area corresponding to peanut variety i includes the following steps:

[0070] When planting each peanut variety, J soil resistance auxiliary measurement devices are buried in the corresponding planting area. Since the soil quality of the resource garden is relatively uniform, several auxiliary measurement devices are evenly set for each peanut variety area, and the average value can be taken later. The soil resistance auxiliary measurement device includes a buried body and a connecting rod connected to the buried body, such as Figure 5 As shown, during installation, the embedded body is located in the soil, with one end of the connecting rod extending out of the soil. The embedded body of the ground resistance auxiliary measurement device used in this embodiment is spherical with a diameter of 2 cm. In practice, the size of the sphere can be arbitrarily set, and the embedded body is not limited to a sphere. It can also be a regular polyhedron (including a cube), an ellipsoid, or a peanut-shaped shape. Once the embedded body is buried in the soil, it can be later extracted via the connecting rod to measure the maximum force during extraction.

[0071] During the growth of peanuts, the soil resistance auxiliary measuring device is kept stationary in the soil. When harvesting peanuts, the soil resistance auxiliary measuring device is pulled out through one end of the connecting rod, and the maximum pulling force F when the soil resistance auxiliary measuring device is pulled out is obtained. ij , where i represents the peanut variety, j represents the soil resistance data sample corresponding to the peanut variety;

[0072] Then calculate the soil resistance corresponding to peanut variety i

[0073] A benchmark peanut parameter database was established based on the plant status parameters of each peanut variety and the corresponding soil resistance.

[0074] (2) In order to study the influence of soil consolidation state on fruit burial rate and fruit drop rate, the present invention also needs to conduct field mechanization test:

[0075] Different field mechanization experiments were conducted, using a staggered sowing method of small double rows per single ridge. Four ridges were used, with ridge lengths of 20m and widths of 0.5m, and a density of 200,000 plants per hectare. Field management followed the same principles as conventional field practices, employing a two-stage harvesting method. Peanut digging and laying were performed by a peanut harvester, and the buried fruit rate was investigated. After seven days of drying, a gleaning combine harvester completed the picking and plucking process, and the fruit drop rate was investigated.

[0076] It should be noted that since the soil quality of the field mechanization test is not necessarily uniform, and the subsequent test is to determine the influence of soil resistance on the fruit drop rate and fruit burial rate, it is necessary to divide the field corresponding to each peanut variety into several blocks (determined according to actual needs, which can be divided evenly or according to the actual soil quality). Each block of the field is recorded as field block j', and several auxiliary measuring devices are evenly set in j'. Later, the soil resistance F is calculated according to the field block j' of each variety. ij'It should also be noted that in this embodiment, the determination is based on field experiments to plant peanuts. When the present invention is actually used, the soil resistance F is obtained for the area to be planted. ij' The crops planted in the planned planting area may not necessarily be peanuts, but may be other crops. It is only necessary to bury the land resistance auxiliary measuring device in the soil when the crops are planted, and pull out the buried body through the connecting rod during harvesting to measure the maximum force when pulling out.

[0077] At the same time, the average value of the fruit burial rate and fruit drop rate of all peanut plants in the field j' is used as the fruit burial rate and fruit drop rate corresponding to the field j', and the total loss rate and total retention rate of the field j' are obtained based on the fruit burial rate and fruit drop rate;

[0078] A field peanut parameter database is established based on the plant status parameters of each peanut variety and the corresponding soil resistance; at the same time, the field peanut parameter database is divided into a training set and a test set.

[0079] (3) Research and analyze the experimental results of resource garden planting experiments and field mechanization experiments:

[0080] 1. Study on the mechanical properties of peanut pod falling and cracking

[0081] Mechanical research on the strength of peanut stalk and shell found that ( Figure 1 ), after harvest, the fruit-stalk strength was significantly lower than the stalk-vine strength, and the fruit-stalk and stalk-vine strengths weakened with the harvest time, with no significant difference on the 7th day. Therefore, the fruit-stalk strength during the digging process determines the fruit burial rate and fruit drop rate during digging. Our other related studies have found that the mechanized picking process is related to the fruit-stalk and vine-stalk tension. The study of the pressure of peanut pods in three directions, positive pressure, lateral pressure and vertical pressure, found that ( Figure 2 ), positive pressure and vertical pressure are significantly lower than lateral pressure, so the mixing and breakage caused by peanuts during the picking and threshing process mainly depends on positive pressure and vertical pressure.

[0082] 2. Correlation analysis of different traits of peanut varieties

[0083] In order to explore the relationship between different mechanical properties, we further carried out correlation analysis on each property ( Figure 3 Correlation analysis results showed that the fruiting range, fruiting depth, and fruit needle length of peanuts were negatively correlated with fruit stalk strength and stalk strength. The fruit burial rate during mechanical excavation was negatively correlated with fruit stalk strength and stalk strength during excavation, negatively correlated with the number of fruits per plant and the number of filled fruits, and positively correlated with fruiting depth, fruiting range, and fruit needle length.

[0084] This indicates that the greater the strength of the peanut stalk and vine stem during the digging process, the lower the field buried fruit rate. The mechanized digging process is mainly positively correlated with the strength of the stalk during digging. The larger the digging range and depth, the greater the field buried fruit rate. The fruit drop rate is negatively correlated with plant height and lateral branch length. Too tall peanut plants and long lateral branches are not conducive to mechanical peanut picking. Yield is mainly positively correlated with the strength of the stalk during digging, negatively correlated with the vine stem strength during picking, and negatively correlated with the fruit range in the soil. Therefore, screening for high-yield, high-quality varieties with tough stalks during digging, concentrated fruit sets, and strong vertical pressure resistance is the main goal of mechanized variety screening.

[0085] 3. Principal component analysis of different peanut varieties

[0086] A principal component analysis (PCA) was performed on 14 traits, including pedicle strength, shell strength, and fruiting range. As shown in Table 2, six principal components had characteristic roots greater than 1, with a cumulative contribution rate of 71.43%. These characteristics can effectively replace the 14 mechanical properties for evaluating and judging peanut varieties. The first principal component had the highest contribution rate, at 15.44%, with kernel length and kernel positive pressure having the highest loadings, at 0.39 and 0.36, respectively. The second principal component had a contribution rate of 12.96%, with kernel thickness having the highest loading, at 0.51, followed by kernel width (0.5). The third principal component had a contribution rate of 12.88%, with shell positive pressure and lateral pressure having the highest loadings, at 0.46 and 0.45, respectively. The fourth and fifth principal components had contribution rates of 11.17% and 9.71%, respectively, with the highest loadings for pedicle strength at fruit picking, stalk strength at digging, and stalk strength at digging, respectively. The sixth principal component had a contribution rate of 9.27%, encompassing fruiting range and pedicle strength at digging.

[0087] Table 2 Principal component analysis of mechanical properties of different peanut varieties (lines)

[0088]

[0089] 4. Cluster analysis of different traits of peanut varieties

[0090] Cluster analysis was performed on 14 traits including stalk strength, shell strength, and fruiting range. Figure 4 As shown in the figure, the 14 traits were divided into 5 categories. The first category was divided into kernel width and thickness, kernel vertical pressure, pedicel strength and stalk strength at 7 days, and kernel pedicel strength at the time of digging. The second category was divided into stalk strength at the time of digging. The third category was divided into kernel positive pressure and lateral pressure, and kernel length. The fourth category was divided into fruiting range. The fifth category was divided into shell positive pressure, lateral pressure, and vertical pressure.

[0091] 5. Multiple linear regression analysis of different peanut varieties

[0092] To further explore the correlation between 25 traits such as pedicel strength, shell strength, and fruiting range, a multiple regression analysis was conducted with pedicel strength as the dependent variable and the remaining traits as independent variables. The stepwise multiple regression method was used to exclude independent variables with insignificant regression coefficients (P<0.05) and screen out important traits that were closely related to the mean value of pedicel strength at the time of digging. The regression equation was obtained: 10 =4.861+0.447X 10 +0.192X 20 +0.212X 30 , we can see that the fruit stalk strength (X 10 ), the strength of the stalk when digging (X 20 ), result range (X 30 ) is the main trait affecting the strength of the fruit stalk.

[0093] From the above analysis, it can be seen that the 14 mechanical properties both promote and restrict each other. When selecting and utilizing varieties, the relationship between them should be coordinated to screen out peanut varieties suitable for mechanized harvesting and processing or with excellent comprehensive traits. Mechanical properties such as peanut pod stalk strength and shell strength have an important impact on mechanized harvesting and subsequent processing. Therefore, although the above process can achieve peanut variety selection suitable for mechanized harvesting, this method has certain inherent limitations. How to select suitable peanut varieties remains to be further studied. Therefore, the present invention does not reduce the data dimension according to the results of clustering or correlation analysis, and screen varieties based on the reduced dimension data, but instead uses a neural network-based method for screening.

[0094] 2. Designing a Total Retention Rate Network Model

[0095] Research has found that indicators such as kernel width and thickness, kernel vertical pressure, stalk strength on the seventh day, stalk strength on the seventh day, stalk strength at the time of digging, stalk strength at the time of digging, kernel positive and lateral pressure, kernel length, fruiting range, shell positive, lateral, and vertical pressure are affected not only by the peanut variety, but also by the soil in which they are grown. For example, the composition of the soil affects peanut growth, and the looseness and air permeability of the soil affect peanut root growth. However, studying the growth state from the perspective of the peanut variety's genes is extremely complex, and genes are not a single factor, making the problem extremely complex and the research cycle very long. Studying the soil state requires not only soil analysis, but also makes the problem extremely complex and even unsolvable. After research and analysis, the present invention ignores the intrinsic influence of variety genes, and instead considers the peanut variety as a macro-characterization of the gene. The peanut variety is encoded by One-Hot and added to the identification process, which is equivalent to considering it as the influence of the gene; at the same time, the present invention ignores the influence of soil status on growth status, but considers the indicators of peanut growth status as the final result of the influence of soil factors, and uses the indicators of peanut growth status as the characterization of the influence of soil factors for the identification process.

[0096] Build a multi-layer neural network model as the total retention rate network model. The multi-layer neural network model has 15 neurons in the input layer and 1 neuron in the output layer;

[0097] The input of the 15 neurons corresponding to the input layer is recorded as X1-X 15 , where X1-X 14 It is divided into the normalized values ​​corresponding to the indicators of grain width, grain thickness, grain vertical pressure, fruit stalk strength on the 7th day, stalk strength on the 7th day, fruit stalk strength when digging, stalk strength when digging, grain positive pressure, grain lateral pressure, grain length, fruit range, shell positive pressure, shell lateral pressure and shell vertical pressure;

[0098] The normalized value for each metric is calculated as follows:

[0099] Xn=(Xn-Xnmin) / λ(Xnmax-Xnmin),n=1,2,……,14

[0100] Where Xnmax and Xnmin are the maximum and minimum values ​​of indicator Xn across all varieties; λ is the normalization adjustment ratio, 1≤λ≤1.5, and is set to 1.1 in this invention. The normalization adjustment ratio λ is to ensure that the normalized upper limit of the value has a reserved normalization margin, thereby improving the accuracy of the overall value in actual identification.

[0101] The normalized value of each indicator can not only unify each indicator to express it under the same impact scale, but also facilitate the extraction and expression of the impact features of different indicators by the neural network.

[0102] The input of the neuron X 15 The normalized value of the One-Hot encoding of the peanut variety;

[0103] At the same time, since the features after encoding can actually be regarded as continuous features in each dimension, each dimension of the features can be normalized. This allows the impact of the 14 indicators and the impact of the variety to be adjusted by the same amount, which is equivalent to expressing them on the same impact scale. However, the essence of the variety is not the actual value, but the One-Hot encoded value. Therefore, normalization is essentially an equivalent adjustment.

[0104] The output of one neuron corresponding to the output layer is the total retention rate.

[0105] The total retention rate network model was trained using the benchmark peanut parameter database. During the training process, the total retention rate, i.e. (1-total loss rate), was obtained based on the actual measured fruit drop rate and fruit burial rate in the benchmark peanut parameter database and used as the label.

[0106] 3. Designing a Total Retention Rate Difference Network Model

[0107] Studies have found that the degree of consolidation of soil is negatively correlated with the total retention rate. Using the degree of consolidation to determine the impact on the total retention rate does not seem to be very rigorous because it ignores the impact of the soil's component content on the strength of the stalk and the fruit stalk. However, studies have found that the degree of consolidation of soil is somewhat correlated with the content of the soil's component (of course, the degree of consolidation is also affected by other factors such as water). The poorer the soil, the more consolidated it is, and even compacted it is. Therefore, in order to simplify the complexity of the research problem, the present invention decides to treat the impact of soil components on the state of the plant as a black box model, without exploring the specific impact relationship, and directly using the degree of consolidation of the soil as a substitute, that is, directly using the degree of consolidation to determine the impact on the total retention rate. Of course, this approach will still sacrifice accuracy, and it also greatly reduces the interpretability of the research.

[0108] The degree of soil consolidation during extraction directly affects the total retention rate. Therefore, the present invention uses factors such as soil consolidation to determine its impact on the total retention rate. Research has also found that soil consolidation has a certain positive correlation with the force required to pull a plant out of the soil. Therefore, the present invention uses ground resistance to determine its impact on the total retention rate. However, it should be noted that the degree of consolidation in the present invention includes, but is not limited to, ground resistance, and other indicators that can represent or characterize soil consolidation can be used.

[0109] Build a multi-layer neural network model as the total retention rate difference network model. The multi-layer neural network model has 2 neurons in the input layer and 1 neuron in the output layer;

[0110] The data of each field plot j' of each variety i is used as an input; the inputs of the two neurons corresponding to the input layer correspond to the normalized value of the one-hot encoding of the peanut variety and the soil resistance difference ratio of field plot j' of variety i, respectively; the output of the one neuron corresponding to the output layer is the total retention rate difference corresponding to field plot j' of variety i;

[0111] For variety i, land resistance difference ratio of field j'

[0112] The total retention rate network model is trained using the field peanut parameter database training set and the benchmark peanut parameter database; during the training process, the total retention rate difference Δ ij' =Y ij' -Y i As labels, where Y ij' is the total retention rate of field block j' corresponding to variety i in the field peanut parameter database.

[0113] 4. Use the field peanut parameter database test set to test and obtain the total retention rate of peanut varieties in real planting plots:

[0114] Directly obtain the basic indicator data of peanut variety i and the soil resistance F corresponding to peanut variety i i Basic index data include: kernel width, kernel thickness, kernel vertical pressure, stalk strength on the 7th day, stalk strength on the 7th day, stalk strength when digging, stalk strength when digging, kernel positive pressure, kernel lateral pressure, kernel length, fruiting range, shell positive pressure, shell lateral pressure and shell vertical pressure;

[0115] Normalize the basic indicator data of variety i; and perform One-Hot encoding and normalization on variety i;

[0116] Input the normalized value of the One-Hot encoding of variety i and the normalized value of the basic indicator data into the total retention rate network model to obtain the total retention rate corresponding to variety i;

[0117] For the plots to be sown (i.e. the plots that need to be selected for mechanized harvesting of peanut varieties), the soil resistance F is obtained in advance. ij' ; The data used in this implementation is from the field peanut parameter database test set;

[0118] According to the soil resistance F corresponding to peanut variety i i and the soil resistance F of the plot where sowing is to be carried out ij'Get the land resistance difference ratio R ij' ;

[0119] Then the One-Hot encoding normalized value of variety i and the land resistance difference ratio R ij' Input the total retention rate difference network model to obtain the total retention rate difference Δ ij' ;

[0120] Then, according to the total retention rate corresponding to variety i and the total retention rate difference Δ ij' The total retention rate of peanut variety i in the plot to be sown is obtained, that is, the difference between the total retention rate corresponding to variety i and the total retention rate Δ ij' of and as the total retention rate of the plots to be sown;

[0121] The peanut variety to be planted is determined based on the total retention rate of peanut variety i in the plot to be sown.

[0122] If more than one peanut variety is determined to be planted, further screening can be carried out based on indicators such as peanut oil yield.

[0123] By using the present invention, before actual planting, as long as the soil resistance of the plot to be planted is obtained, or the soil resistance of the plot to be planted is obtained in advance by a soil resistance auxiliary measurement device, the basic indicator data provided by the seed provider (seed company or agricultural science academy, etc.) and the soil resistance F corresponding to the peanut variety i can be used to calculate the soil resistance of the plot to be planted. i By estimating the total peanut retention rate for the proposed planting plots in advance, a relatively accurate estimate of the total peanut loss rate / total peanut retention rate during mechanical harvesting can be achieved in the actual planting area, even if there are significant differences between the experimental fields and the proposed planting plots. This effectively controls a satisfactory loss amount after the actual planting and harvesting, thereby effectively controlling actual losses. Furthermore, the present invention can be used directly based on the established database for relatively large-scale planting, eliminating the need for small-scale trial plantings, thus shortening the "planting-to-maturity" cycle and further reducing the time period for truly selecting suitable varieties. Specific implementation method 2:

[0125] This embodiment is a storage medium, in which at least one instruction is stored. The at least one instruction is loaded and executed by a processor to implement the method for screening peanut varieties suitable for mechanized harvesting.

[0126] It should be understood that any method described herein may be provided as a computer program product, software, or computerized method, which may include a non-transitory machine-readable medium having instructions stored thereon, the instructions being used to program a computer system or other electronic device. The storage medium may include, but is not limited to, magnetic storage media, optical storage media; magneto-optical storage media including: read-only memory (ROM), random access memory (RAM), erasable programmable memory (e.g., EPROM and EEPROM), and flash memory layers; or other types of media suitable for storing electronic instructions. Specific implementation method three:

[0128] This embodiment is a screening device for peanut varieties suitable for mechanized harvesting, the device including a processor and a memory. It should be understood that, including any device including a processor and a memory described in the present invention, the device may also include other units or modules that perform display, interaction, processing, control, and other functions through signals or instructions;

[0129] At least one instruction is stored in the memory, and the at least one instruction is loaded and executed by the processor to implement the method for screening peanut varieties suitable for mechanized harvesting.

[0130] The above examples are merely illustrative of the calculation model and process of the present invention and are not intended to limit the embodiments of the present invention. Persons skilled in the art will readily appreciate that other variations or modifications based on the above description are possible. This list of embodiments is not exhaustive; however, any obvious variations or modifications derived from the technical solution of the present invention remain within the scope of protection of the present invention.

Claims

1. A method for screening peanut varieties suitable for mechanized harvesting, characterized in that: The method comprises the following steps: S1. Obtain the basic index data of peanut variety i that needs to be screened and the consolidation degree F corresponding to peanut variety i i ; Normalizing the basic indicator data of variety i, including: kernel width, kernel thickness, kernel vertical pressure, stalk strength on the 7th day, stalk strength on the 7th day, stalk strength when digging, stalk strength when digging, kernel positive pressure, kernel lateral pressure, kernel length, fruiting range, shell positive pressure, shell lateral pressure, and shell vertical pressure; performing one-hot encoding and normalization on the variety i; Then, the normalized value of the One-Hot encoding of variety i and the normalized value of the basic indicator data are input into the total retention rate network model to obtain the total retention rate corresponding to variety i; The total retention rate network model is a multi-layer neural network model; the multi-layer neural network model corresponding to the total retention rate network model has 15 neurons in the input layer and 1 neuron in the output layer; the input of the 15 neurons corresponding to the input layer is recorded as X1-X 15 , where X1-X 14 is the normalized value corresponding to the basic indicator data; the neuron input X 15 is the normalized value of the One-Hot encoding of the peanut variety; the output of one neuron corresponding to the output layer is the total retention rate; S2. Obtain the consolidation degree F for the plot j' to be sown ij' ; According to the consolidation degree F corresponding to peanut variety i i and the degree of consolidation F of the plot to be sown ij' Get the consolidation difference ratio R ij' ; Then the One-Hot encoding normalized value of variety i and the consolidation difference ratio R ij' Input the total retention rate difference network model to obtain the total retention rate difference Δ ij' ; The total retention rate difference network model is a multi-layer neural network model; S3, the total retention rate corresponding to variety i and the total retention rate difference Δ ij' Get the total retention rate of peanut variety i in the plot to be sown; S4. Determine the peanut variety to be planted based on the total retention rate of peanut variety i in the plot to be sown.

2. The method for screening peanut varieties suitable for mechanized harvesting according to claim 1, characterized in that: The consolidation degree difference ratio 3. The method for screening peanut varieties suitable for mechanized harvesting according to claim 2, characterized in that: Consolidation degree corresponding to peanut variety i Among them F ij represents the j-th consolidation degree among the J land consolidation degree samples corresponding to peanut variety i.

4. The method for screening peanut varieties suitable for mechanized harvesting according to claim 3, characterized in that: The degree of consolidation F ij' For land resistance.

5. The method for screening peanut varieties suitable for mechanized harvesting according to claim 4, characterized in that: The soil resistance is obtained by burying an auxiliary measuring device in the soil during planting and measuring the maximum force obtained by pulling out the auxiliary measuring device during harvesting.

6. The method for screening peanut varieties suitable for mechanized harvesting according to claim 1, characterized in that: The process of normalizing the basic indicator data of variety i includes the following steps: The basic indicator data are recorded as X1-X 14 , the normalized value of each indicator Xn is calculated as follows: Xn=(Xn-Xnmin) / λ(Xnmax-Xnmin), n=1,2,...,14, Among them, Xnmax and Xnmin are the maximum and minimum values ​​of the indicator Xn in all peanut varieties; λ is the normalized adjustment ratio.

7. The method for screening peanut varieties suitable for mechanized harvesting according to claim 6, characterized in that: The normalized adjustment ratio λ is set to 1.

1.

8. A storage medium, characterized in that: The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the method for screening peanut varieties suitable for mechanized harvesting as described in any one of claims 1 to 7.

9. A screening device suitable for mechanized harvesting of peanut varieties, characterized in that: The device includes a processor and a memory, wherein the memory stores at least one instruction, and the at least one instruction is loaded and executed by the processor to implement the method for screening peanut varieties suitable for mechanized harvesting as described in any one of claims 1 to 7.

Images