An intelligent soil deep tillage optimization method and system

Abstract

The present invention provides an intelligent soil deep loosening and tillage optimization method and system, including: determining the physical and chemical property indicators of soil samples; calculating the correlation between different physical and chemical property indicators and crop yields, and screening out physical and chemical property indicators with a correlation less than a preset threshold; inputting the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model; using the crop yield prediction model to predict the yield of target crops under different soil deep loosening and tillage methods to obtain a predicted value; selecting the soil deep loosening and tillage method corresponding to the maximum predicted value as the optimal tillage method. The present invention can predict the potential yield of target crops under different tillage methods by establishing a crop yield prediction model. This method allows staff to understand the effects of different strategies before actual farming, so as to select the farming method that is most likely to bring the highest yield, thereby increasing the total yield of crops.

Description

Technical Field

[0001] The present invention belongs to the technical field of soil deep tillage, and more specifically, relates to an intelligent soil deep tillage optimization method and system. Background Art

[0002] Agriculture is the foundation of human society, and soil quality directly affects the yield and quality of crops. Traditional soil cultivation methods include shallow tillage and deep plowing, among which deep tillage is an important agronomic measure that can break the hardened layer at the bottom of the soil, improve soil structure, and increase soil porosity, thereby improving soil air permeability, water permeability and fertility, which is beneficial to the development of crop roots and the absorption of water and nutrients.

[0003] However, traditional deep tillage relies on manual or mechanical experience, and it is difficult to accurately control the tillage depth and frequency, resulting in unstable soil structure improvement effects. Summary of the invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide an intelligent soil deep tillage optimization method and system.

[0005] An intelligent soil deep tillage optimization method comprises the following steps:

[0006] Step 1: Treat the cultivated land using different soil deep tillage methods and take soil samples;

[0007] Step 2: Determine the physical and chemical properties of soil samples;

[0008] Step 3: Calculate the correlation between different physical and chemical property indicators and crop yield, and remove the physical and chemical property indicators with correlation less than a preset threshold to obtain the screened physical and chemical property indicators;

[0009] Step 4: Input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model;

[0010] Step 5: Use the crop yield prediction model to predict the yield of target crops under different soil deep tillage methods to obtain the predicted value;

[0011] Step 6: Select the soil deep tillage method corresponding to the maximum predicted value as the optimal tillage method.

[0012] Preferably, the step 3: calculating the correlation between different physical and chemical property indicators and crop yield, and removing the physical and chemical property indicators with correlation less than a preset threshold, to obtain the screened physical and chemical property indicators, includes:

[0013] Step 3.1: Perform data cleaning on all physical and chemical property indicators to obtain the physical and chemical property indicators after data cleaning;

[0014] Step 3.2: Calculate the correlation between the physical and chemical property indicators after data cleaning and the crop yield, and construct a correlation matrix based on the correlation;

[0015] Step 3.3: Obtain the eigenvalues ​​of the correlation matrix and arrange them in descending order to form an eigenvalue sequence;

[0016] Step 3.4: Calculate the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence;

[0017] Step 3.5: The physical and chemical property indicators whose contribution rates are greater than the threshold are taken as the screened physical and chemical property indicators.

[0018] Preferably, the step 3.1: performing data cleaning on all physical and chemical property indicators to obtain physical and chemical property indicators after data cleaning includes:

[0019] Step 3.1.1: Arrange the corresponding physical and chemical property indicators in the order of increasing crop yield to form an increasing sequence of crop yield and a sequence of physical and chemical property indicators;

[0020] Step 3.1.2: Subtract the normalized crop yield increasing sequence from the physical and chemical property index sequence to form an outlier detection sequence;

[0021] Step 3.1.3: Calculate the abnormal evaluation value of each group of physical and chemical property index sequence using the abnormal value detection sequence; wherein the abnormal evaluation value calculation formula is:

[0022]

[0023] Among them, Δ j (k)=|x0(k)-x j (k)|, j = 1, 2, ..., n, x0(k) represents the normalized crop yield in the kth year, x j (k) represents the jth physical and chemical property index in the kth year, Δ j (k) represents the absolute value of the difference in the kth year, m represents the minimum value in the outlier detection sequence, M represents the maximum value in the outlier detection sequence, ξ represents the preset coefficient, θ represents the outlier evaluation value, and r 0j (k) represents anomaly detection parameters;

[0024] Step 3.1.4: The physical and chemical property indicators corresponding to the abnormal evaluation values ​​within the preset range are used as the physical and chemical property indicators after data cleaning.

[0025] Preferably, the step 3.2: calculating the correlation between the physical and chemical property index after data cleaning and the crop yield, and constructing a correlation matrix based on the correlation, includes:

[0026] Step 3.2.1: Use the correlation calculation formula to obtain the correlation between the physical and chemical property indicators after data cleaning and the crop yield; wherein the correlation calculation formula is:

[0027]

[0028] Among them, r ab represents the correlation between the ath physical and chemical property index and the bth crop yield, x ka represents the ath physical and chemical property index in the kth year, Represents the mean value of the physical and chemical properties, y kb represents the yield of the bth crop in the kth year, represents the mean value of crop yield;

[0029] Step 3.2.2: Use the correlation degree to construct the correlation matrix; where the correlation matrix is:

[0030]

[0031] Preferably, the step 3.4: calculating the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence includes:

[0032] Using the formula:

[0033]

[0034] Calculate the contribution rate corresponding to each physical and chemical property index; among them, represents the contribution rate corresponding to the i-th physical and chemical property index, λ i represents the i-th value in the eigenvalue sequence, λ k Represents the kth value in the eigenvalue sequence.

[0035] Preferably, the step 4: inputting the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model comprises:

[0036] Step 4.1: inputting the screened physical and chemical property indicators as samples into the BP neural network;

[0037] Step 4.2: Replace the loss function in the training process with the fitness function, and use the fitness function to optimize the BP neural network to obtain a crop yield prediction model; wherein the fitness function is:

[0038]

[0039] Among them, n represents the number of samples, F represents the fitness of individual k, individual is the network parameter, and f ik represents the crop yield value output by the BP neural network of the i-th sample under the k-th individual, f(x i ) represents the actual crop yield value.

[0040] Preferably, in step 4.2, the BP neural network is optimized using the fitness function to obtain a crop yield prediction model, including:

[0041] Step 4.2.1: Calculate the fitness of all individuals;

[0042] Step 4.2.2: Determine the probability of selecting the next generation of individuals based on the fitness of the individuals; the probability of selecting the next generation of individuals is:

[0043]

[0044] Among them, p i represents the probability of individual selection in the next generation, F i represents the fitness of individual i, and N represents the number of individuals;

[0045] Step 4.2.3: Based on the probability of selecting the next generation of individuals, two individuals are selected as parents to perform crossover and mutation operations to obtain new individuals, and return to step 4.2.1 until the fitness of an individual is within the preset range;

[0046] Step 4.2.4: Extract the network parameters corresponding to individuals with fitness within the preset range;

[0047] Step 4.2.5: Utilize the network parameters to optimize the BP neural network and obtain a crop yield prediction model.

[0048] Preferably, in step 4.2.3, the formula is used:

[0049]

[0050] The parent generation is subjected to crossover and mutation operations to obtain new individuals; where a i ' j represents the parent generation after crossover, a kj Indicates the first parent, a lj represents the second parent, b represents a random number between [0,1], represents a new individual, a min Indicates the minimum threshold value, a max represents the maximum threshold value, g represents the number of iterations, G max represents the maximum number of evolutions, G minIndicates the minimum number of evolutions.

[0051] The present invention also provides an intelligent soil deep tillage optimization system, comprising:

[0052] A soil sample acquisition module is used to treat the cultivated land using different soil deep tillage methods and to dig soil samples;

[0053] Physical and chemical property index determination module, used to determine the physical and chemical property indexes of soil samples;

[0054] A correlation screening module is used to calculate the correlation between different physical and chemical property indicators and crop yields, and remove physical and chemical property indicators with correlations less than a preset threshold to obtain screened physical and chemical property indicators;

[0055] A training module, used to input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model;

[0056] The yield prediction module is used to predict the yield of target crops under different soil deep tillage methods using the crop yield prediction model to obtain the predicted value;

[0057] The tillage method determination module is used to select the soil deep loosening tillage method corresponding to the maximum predicted value as the optimal tillage method.

[0058] The present invention also provides an electronic device, comprising a bus, a transceiver, a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory and the processor are connected via the bus, and wherein the computer program, when executed by the processor, implements the steps in the above-mentioned intelligent soil deep loosening and tillage optimization method.

[0059] The beneficial effect of the intelligent soil deep tillage optimization method and system provided by the present invention is that: compared with the prior art, the present invention can predict the potential yield of target crops under different tillage methods by establishing a crop yield prediction model. This method allows workers to understand the effects of different strategies before actual tillage, so as to select the tillage method that is most likely to bring the highest yield, thereby increasing the total yield of crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 A flow chart of an intelligent soil deep tillage optimization method provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of an intelligent soil deep tillage optimization system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0064] To achieve the above object, the technical solution adopted by the present invention is:

[0065] See also Figure 1 , an intelligent soil deep tillage optimization method, comprising the following steps:

[0066] Step 1: Treat the cultivated land using different soil deep tillage methods and take soil samples;

[0067] Step 2: Determine the physical and chemical properties of soil samples; physical and chemical property indicators include: soil bulk density, soil specific gravity, porosity, soil moisture content, soil compaction, pH value, organic matter content, soil heavy metal content and other physical and chemical property indicators.

[0068] Step 3: Calculate the correlation between different physical and chemical property indicators and crop yield, and remove the physical and chemical property indicators with correlation less than a preset threshold to obtain the screened physical and chemical property indicators;

[0069] Furthermore, step 3 includes:

[0070] Step 3.1: Perform data cleaning on all physical and chemical property indicators to obtain the physical and chemical property indicators after data cleaning;

[0071] Furthermore, step 3.1 includes:

[0072] Step 3.1.1: Arrange the corresponding physical and chemical property indicators in the order of increasing crop yield to form an increasing sequence of crop yield and a sequence of physical and chemical property indicators;

[0073] Step 3.1.2: Subtract the normalized crop yield increasing sequence from the physical and chemical property index sequence to form an outlier detection sequence;

[0074] Step 3.1.3: Calculate the abnormal evaluation value of each group of physical and chemical property index sequence using the abnormal value detection sequence; wherein the abnormal evaluation value calculation formula is:

[0075]

[0076] Among them, Δ j (k)=|x0(k)-x j (k)|, j = 1, 2, ..., n, x0(k) represents the normalized crop yield in the kth year, x j (k) represents the jth physical and chemical property index in the kth year, Δ j (k) represents the absolute value of the difference in the kth year, m represents the minimum value in the outlier detection sequence, M represents the maximum value in the outlier detection sequence, ξ represents the preset coefficient, θ represents the outlier evaluation value, and r 0j (k) represents anomaly detection parameters;

[0077] Step 3.1.4: The physical and chemical property indicators corresponding to the abnormal evaluation values ​​within the preset range are used as the physical and chemical property indicators after data cleaning.

[0078] The present invention obtains an outlier detection sequence by subtracting the normalized crop yield increment sequence from the physical and chemical property index sequence, which can quickly identify outliers in the data and help decision makers find problem points more quickly.

[0079] Step 3.2: Calculate the correlation between the physical and chemical property indicators after data cleaning and the crop yield, and construct a correlation matrix based on the correlation;

[0080] Furthermore, step 3.2 includes:

[0081] Step 3.2.1: Use the correlation calculation formula to obtain the correlation between the physical and chemical property indicators after data cleaning and the crop yield; wherein the correlation calculation formula is:

[0082]

[0083] Among them, r ab represents the correlation between the ath physical and chemical property index and the bth crop yield, x ka represents the ath physical and chemical property index in the kth year, Represents the mean value of the physical and chemical properties, y kb represents the yield of the bth crop in the kth year, represents the mean value of crop yield;

[0084] Step 3.2.2: Use the correlation degree to construct the correlation matrix; where the correlation matrix is:

[0085]

[0086] Step 3.3: Obtain the eigenvalues ​​of the correlation matrix and arrange them in descending order to form an eigenvalue sequence;

[0087] Step 3.4: Calculate the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence;

[0088] Wherein, in step 3.4, the formula may be used:

[0089]

[0090] Calculate the contribution rate corresponding to each physical and chemical property index; among them, represents the contribution rate corresponding to the i-th physical and chemical property index, λ i represents the i-th value in the eigenvalue sequence, λ k Represents the kth value in the eigenvalue sequence.

[0091] Step 3.5: The physical and chemical property indicators whose contribution rates are greater than the threshold are taken as the screened physical and chemical property indicators.

[0092] The present invention calculates the correlation between the physical and chemical property indicators after data cleaning and the crop yield, which can be quantified by the Pearson correlation coefficient calculation formula, which can help decision makers understand the strength and direction of the relationship between them. At the same time, the correlation matrix constructed using the correlation degree can intuitively display the correlation between different physical and chemical property indicators and crop yield, which helps to quickly identify indicators with strong correlation.

[0093] Step 4: Input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model;

[0094] Furthermore, step 4 includes:

[0095] Step 4.1: inputting the screened physical and chemical property indicators as samples into the BP neural network;

[0096] Step 4.2: Replace the loss function in the training process with the fitness function, and use the fitness function to optimize the BP neural network to obtain a crop yield prediction model; wherein the fitness function is:

[0097]

[0098] Among them, n represents the number of samples, F represents the fitness of individual k, individual is the network parameter, and f ik represents the crop yield value output by the BP neural network of the i-th sample under the k-th individual, f(x i ) represents the actual crop yield value;

[0099] In step 4.2, the BP neural network is optimized using the fitness function to obtain a crop yield prediction model, including:

[0100] Step 4.2.1: Calculate the fitness of all individuals;

[0101] Step 4.2.2: Determine the probability of selecting the next generation of individuals based on the fitness of the individuals; the probability of selecting the next generation of individuals is:

[0102]

[0103] Among them, p i represents the probability of individual selection in the next generation, F i represents the fitness of individual i, and N represents the number of individuals;

[0104] Step 4.2.3: Based on the probability of selecting the next generation of individuals, two individuals are selected as parents to perform crossover and mutation operations to obtain new individuals, and return to step 4.2.1 until the fitness of an individual is within the preset range;

[0105] In step 4.2.3, the formula may be used:

[0106]

[0107] The parent generation is subjected to crossover and mutation operations to obtain new individuals; where a i ' j represents the parent generation after crossover, a kj Indicates the first parent, a lj represents the second parent, b represents a random number between [0,1], represents a new individual, a min Indicates the minimum threshold value, a max represents the maximum threshold value, g represents the number of iterations, G max represents the maximum number of evolutions, G min represents the minimum number of evolutions;

[0108] Step 4.2.4: Extract the network parameters corresponding to individuals with fitness within the preset range;

[0109] Step 4.2.5: Utilize the network parameters to optimize the BP neural network and obtain a crop yield prediction model.

[0110] Since the BP neural network has the problem of local optimal solutions, the present invention can help the BP neural network escape the local optimal solution by utilizing the global search capability of the genetic algorithm, and can make the BP neural network better fit the training data, thereby improving the generalization ability of the network and accelerating the convergence speed of the BP neural network.

[0111] Step 5: Use the crop yield prediction model to predict the yield of target crops under different soil deep tillage methods to obtain the predicted value;

[0112] Step 6: Select the soil deep tillage method corresponding to the maximum predicted value as the optimal tillage method.

[0113] The present invention can predict the potential yield of target crops under different farming methods by establishing a crop yield prediction model. This method allows workers to understand the effects of different strategies before actual farming, so as to choose the farming method that is most likely to bring the highest yield, thereby increasing the total yield of crops.

[0114] See also Figure 2 The present invention also provides an intelligent soil deep tillage optimization system, comprising:

[0115] A soil sample acquisition module is used to treat the cultivated land using different soil deep tillage methods and to dig soil samples;

[0116] Physical and chemical property index determination module, used to determine the physical and chemical property indexes of soil samples;

[0117] A correlation screening module is used to calculate the correlation between different physical and chemical property indicators and crop yields, and remove physical and chemical property indicators with correlations less than a preset threshold to obtain screened physical and chemical property indicators;

[0118] A training module, used to input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model;

[0119] The yield prediction module is used to predict the yield of target crops under different soil deep tillage methods using the crop yield prediction model to obtain the predicted value;

[0120] The tillage method determination module is used to select the soil deep loosening tillage method corresponding to the maximum predicted value as the optimal tillage method.

[0121] Compared with the prior art, the beneficial effects of the intelligent soil deep loosening and tillage optimization system provided by the present invention are the same as the beneficial effects of the intelligent soil deep loosening and tillage optimization method described in the above technical scheme, which will not be repeated here.

[0122] The present invention also provides an electronic device, including a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, and when the computer program is executed by the processor, the steps in the above-mentioned intelligent soil deep loosening and tillage optimization method are implemented. Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as the beneficial effects of the intelligent soil deep loosening and tillage optimization method described in the above-mentioned technical solution, and will not be described in detail here.

[0123] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent soil deep tillage optimization method, characterized in that: The following steps are involved: Step 1: Treat the cultivated land using different soil deep tillage methods and take soil samples; Step 2: Determine the physical and chemical properties of soil samples; Step 3: Calculate the correlation between different physical and chemical property indicators and crop yield, and remove the physical and chemical property indicators with correlation less than a preset threshold to obtain the screened physical and chemical property indicators; The step 3: calculating the correlation between different physical and chemical property indicators and crop yields, and removing the physical and chemical property indicators with correlations less than a preset threshold, to obtain the screened physical and chemical property indicators, including: Step 3.1: Perform data cleaning on all physical and chemical property indicators to obtain the physical and chemical property indicators after data cleaning; The step 3.1: performing data cleaning on all physical and chemical property indicators to obtain physical and chemical property indicators after data cleaning, including: Step 3.1.1: Arrange the corresponding physical and chemical property indicators in the order of increasing crop yield to form an increasing sequence of crop yield and a sequence of physical and chemical property indicators; Step 3.1.2: Subtract the normalized crop yield increasing sequence from the physical and chemical property index sequence to form an outlier detection sequence; Step 3.1.3: Calculate the abnormal evaluation value of each group of physical and chemical property index sequence using the abnormal value detection sequence; wherein the abnormal evaluation value calculation formula is: Among them, Δ j (k)=|x0(k)-x j (k)|, j = 1, 2, ..., n, x0(k) represents the normalized crop yield in the kth year, x j (k) represents the jth physical and chemical property index in the kth year, Δ j (k) represents the absolute value of the difference in the kth year, m represents the minimum value in the outlier detection sequence, M represents the maximum value in the outlier detection sequence, ξ represents the preset coefficient, and θ represents the outlier assessment value; Step 3.1.4: The physical and chemical property index corresponding to the abnormal evaluation value within the preset range is used as the physical and chemical property index after data cleaning; Step 3.2: Calculate the correlation between the physical and chemical property indicators after data cleaning and the crop yield, and construct a correlation matrix based on the correlation; The step 3.2: calculating the correlation between the physical and chemical property index after data cleaning and the crop yield, and constructing a correlation matrix based on the correlation, including: Step 3.2.1: Use the correlation calculation formula to obtain the correlation between the physical and chemical property indicators after data cleaning and the crop yield; wherein the correlation calculation formula is: Among them, r ab represents the correlation between the ath physical and chemical property index and the bth crop yield, x ka represents the ath physical and chemical property index in the kth year, Represents the mean value of the physical and chemical properties, y kb represents the yield of the bth crop in the kth year, represents the mean value of crop yield; Step 3.2.2: Use the correlation degree to construct the correlation matrix; where the correlation matrix is: Step 3.3: Obtain the eigenvalues ​​of the correlation matrix and arrange them in descending order to form an eigenvalue sequence; Step 3.4: Calculate the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence; The step 3.4: calculating the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence, includes: Using the formula: Calculate the contribution rate corresponding to each physical and chemical property index; among them, represents the contribution rate corresponding to the i-th physical and chemical property index, λ i represents the i-th value in the eigenvalue sequence, λ k represents the kth value in the eigenvalue sequence; Step 3.5: The physical and chemical property indicators whose contribution rates are all greater than the threshold are taken as the screened physical and chemical property indicators; Step 4: Input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model; Step 5: Use the crop yield prediction model to predict the yield of target crops under different soil deep tillage methods to obtain the predicted value; Step 6: Select the soil deep tillage method corresponding to the maximum predicted value as the optimal tillage method.

2. An intelligent soil deep tillage optimization method as claimed in claim 1, characterized in that: The step 4: inputting the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model, including: Step 4.1: inputting the screened physical and chemical property indicators as samples into the BP neural network; Step 4.2: Replace the loss function in the training process with the fitness function, and use the fitness function to optimize the BP neural network to obtain a crop yield prediction model; wherein the fitness function is: Among them, n represents the number of samples, F represents the fitness of individual k, individual is the network parameter, and f ik represents the crop yield value output by the BP neural network of the i-th sample under the k-th individual, f(x i ) represents the actual crop yield value.

3. An intelligent soil deep tillage optimization method as claimed in claim 2, characterized in that: In step 4.2, the BP neural network is optimized using the fitness function to obtain a crop yield prediction model, including: Step 4.2.1: Calculate the fitness of all individuals; Step 4.2.2: Determine the probability of selecting the next generation of individuals based on the fitness of the individuals; the probability of selecting the next generation of individuals is: Among them, p i represents the probability of individual selection in the next generation, F i represents the fitness of individual i, and N represents the number of individuals; Step 4.2.3: Based on the probability of selecting the next generation of individuals, two individuals are selected as parents to perform crossover and mutation operations to obtain new individuals, and return to step 4.2.1 until the fitness of an individual is within the preset range; Step 4.2.4: Extract the network parameters corresponding to individuals with fitness within the preset range; Step 4.2.5: Utilize the network parameters to optimize the BP neural network and obtain a crop yield prediction model.

4. An intelligent soil deep tillage optimization method as claimed in claim 3, characterized in that: In step 4.2.3, the formula is used: The parent generation is subjected to crossover and mutation operations to obtain new individuals; where a i ' j represents the parent generation after crossover, a kj Indicates the first parent, a lj represents the second parent, b represents a random number between [0,1], represents a new individual, a min Indicates the minimum threshold value, a max represents the maximum threshold value, g represents the number of iterations, G max represents the maximum number of evolutions, G min Indicates the minimum number of evolutions.

5. An intelligent soil deep tillage optimization system, characterized in that: include: A soil sample acquisition module is used to treat the cultivated land using different soil deep tillage methods and to dig soil samples; Physical and chemical property index determination module, used to determine the physical and chemical property indexes of soil samples; A correlation screening module is used to calculate the correlation between different physical and chemical property indicators and crop yields, and remove physical and chemical property indicators with correlations less than a preset threshold to obtain screened physical and chemical property indicators; Among them, the correlation degree of different physical and chemical property indicators to crop yield is calculated, and the physical and chemical property indicators with correlation degree less than the preset threshold are removed to obtain the screened physical and chemical property indicators, including: Step 3.1: Perform data cleaning on all physical and chemical property indicators to obtain the physical and chemical property indicators after data cleaning; The step 3.1: performing data cleaning on all physical and chemical property indicators to obtain physical and chemical property indicators after data cleaning, including: Step 3.1.1: Arrange the corresponding physical and chemical property indicators in the order of increasing crop yield to form an increasing sequence of crop yield and a sequence of physical and chemical property indicators; Step 3.1.2: Subtract the normalized crop yield increasing sequence from the physical and chemical property index sequence to form an outlier detection sequence; Step 3.1.3: Calculate the abnormal evaluation value of each group of physical and chemical property index sequence using the abnormal value detection sequence; wherein the abnormal evaluation value calculation formula is: Among them, Δ j (k)=|x0(k)-x j (k)|, j = 1, 2, ..., n, x0(k) represents the normalized crop yield in the kth year, x j (k) represents the jth physical and chemical property index in the kth year, Δ j (k) represents the absolute value of the difference in the kth year, m represents the minimum value in the outlier detection sequence, M represents the maximum value in the outlier detection sequence, ξ represents the preset coefficient, and θ represents the outlier assessment value; Step 3.1.4: The physical and chemical property index corresponding to the abnormal evaluation value within the preset range is used as the physical and chemical property index after data cleaning; Step 3.2: Calculate the correlation between the physical and chemical property indicators after data cleaning and the crop yield, and construct a correlation matrix based on the correlation; The step 3.2: calculating the correlation between the physical and chemical property index after data cleaning and the crop yield, and constructing a correlation matrix based on the correlation, including: Step 3.2.1: Use the correlation calculation formula to obtain the correlation between the physical and chemical property indicators after data cleaning and the crop yield; wherein the correlation calculation formula is: Among them, r ab represents the correlation between the ath physical and chemical property index and the bth crop yield, x ka represents the ath physical and chemical property index in the kth year, Represents the mean value of the physical and chemical properties, y kb represents the yield of the bth crop in the kth year, represents the mean value of crop yield; Step 3.2.2: Use the correlation degree to construct the correlation matrix; where the correlation matrix is: Step 3.3: Obtain the eigenvalues ​​of the correlation matrix and arrange them in descending order to form an eigenvalue sequence; Step 3.4: Calculate the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence; The step 3.4: calculating the contribution rate corresponding to each physical and chemical property index based on the characteristic value sequence, includes: Using the formula: Calculate the contribution rate corresponding to each physical and chemical property index; among them, represents the contribution rate corresponding to the i-th physical and chemical property index, λ i represents the i-th value in the eigenvalue sequence, λ k represents the kth value in the eigenvalue sequence; Step 3.5: The physical and chemical property indicators whose contribution rates are all greater than the threshold are taken as the screened physical and chemical property indicators; A training module, used to input the screened physical and chemical property indicators as samples into a neural network for training to obtain a crop yield prediction model; The yield prediction module is used to predict the yield of target crops under different soil deep tillage methods using the crop yield prediction model to obtain the predicted value; The tillage method determination module is used to select the soil deep loosening tillage method corresponding to the maximum predicted value as the optimal tillage method.

6. An electronic device, comprising a bus, a transceiver, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the transceiver, the memory, and the processor are connected via the bus, wherein: When the computer program is executed by the processor, the steps of an intelligent soil deep tillage optimization method as described in any one of claims 1-4 are implemented.

Images