Agricultural production planning method based on genetic algorithm and BP neural network model

Abstract

The invention discloses an agricultural production planning method based on a genetic algorithm and a BP neural network model, and belongs to the technical field of agricultural production planning. The problem that the optimal agricultural production planning scheme is difficult to obtain by a traditional decision-making method based on artificial intelligence is solved. According to the method, the BP neural network model is trained by using the temperature data, rainfall data, agricultural operation planning and agricultural product yield data of each year in history, and the trained BP neural network model can be used for predicting the yield of agricultural products. And establishing a genetic algorithm taking the total income as a fitness function, and optimizing a crop planting scheme in combination with the genetic algorithm and the yield predicted by the trained BP neural network model, so that an optimal crop planting scheme can be obtained, and the optimality of agricultural production planning is realized. The method can be applied to agricultural production planning.

Description

Technical Field

[0001] The present invention belongs to the technical field of agricultural production planning, and particularly relates to an agricultural production planning method based on a genetic algorithm and a BP neural network model. Background Art

[0002] In the process of agricultural production planning, initially, production plans were obtained based on the decisions of decision-makers. The production plans obtained through this method were greatly affected by the subjectivity of decision-makers. With the continuous development of artificial intelligence, many production departments have begun to change the way of obtaining production plans to refer to artificial intelligence methods, and more reliable production plan decisions can be obtained through artificial intelligence methods.

[0003] However, the disadvantage of traditional artificial intelligence-based decision-making methods is that they often consider fewer factors, cannot take into account all aspects of agricultural production, lack systematic ecological considerations, and have insufficient informatization and data support. Considering only a few factors will result in the generated results not conforming to the actual situation, and it is difficult to obtain the best agricultural production planning scheme. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem that it is difficult to obtain the best agricultural production planning scheme by traditional artificial intelligence-based decision-making methods, and an agricultural production planning method based on a genetic algorithm and a BP neural network model is proposed.

[0005] The technical solution adopted by the present invention to solve the above technical problems is: an agricultural production planning method based on a genetic algorithm and a BP neural network model, and the method specifically includes the following steps:

[0006] Step 1: Obtain the types of crops planted in each plot in each historical quarter, the yields of each plot in each historical quarter, the average temperature in each historical quarter, and the daily average precipitation in each historical quarter, respectively;

[0007] Step 2: Use the data obtained in Step 1 to train the BP neural network model to obtain a trained BP neural network model;

[0008] Step 3: Establish a fitness function for the genetic algorithm:

[0009] f = b - c

[0010] where f represents the net profit; b represents the total revenue, and b is calculated based on the yield predicted by the BP neural network model; c represents the total cost;

[0011] Step 4: Use the genetic algorithm to generate a crop planting plan for the next quarter.

[0012] Further, the specific process of Step 2 is:

[0013] Define the inputs of the BP neural network model as follows: the type of crop planted in the i-th plot in the j-th historical quarter, the normalized average temperature of the i-th plot in the j-th historical quarter, the normalized average daily precipitation of the i-th plot in the j-th historical quarter, and the reduction rate of the i-th plot in the j-th historical quarter, where i = 1, 2, …, I (I represents the total number of plots) and j = 1, 2, …, J (J represents the total number of quarters).

[0014] Define the output label of the BP neural network model as: the yield of the i-th plot in the j-th historical quarter.

[0015] Normalize the average temperature of each historical quarter obtained in Step 1, and normalize the average daily precipitation of each historical quarter obtained in Step 1.

[0016] Use the type of crop planted in the i-th plot in the j-th historical quarter, the reduction rate of the i-th plot in the j-th historical quarter, the normalized average temperature of the j-th historical quarter, the normalized average daily precipitation of the j-th historical quarter, and the yield of the i-th plot in the j-th historical quarter to train the BP neural network model.

[0017] Furthermore, the reduction rate of the i-th plot in the j-th historical quarter is as follows:

[0018] If the type of crop planted in the i-th plot in the j-th historical quarter is the same as the type of crop planted in the i-th plot in the (j - 1)-th historical quarter, then the reduction rate of the i-th plot in the j-th historical quarter is R.

[0019] If the type of crop planted in the i-th plot in the j-th historical quarter is different from the type of crop planted in the i-th plot in the (j - 1)-th historical quarter, then the reduction rate of the i-th plot in the j-th historical quarter is 0.

[0020] Furthermore, the specific method for normalizing the average temperature of each historical quarter obtained in Step 1 is as follows:

[0021] Define that a total of data for each plot in the past N years is obtained in Step 1, then normalize the average temperature of the k-th quarter in the past N years:

[0022]

[0023] where W n,k represents the average temperature of the k-th quarter in the n-th past year, represents the normalized value of W n,k , represents the minimum value of the average temperature of the k-th quarter in the past N years, Denote the maximum value among the average temperatures in the k-th quarter of the past N years, where k = 1, 2, 3, 4.

[0024] Furthermore, the normalization process for the daily average precipitation in each historical quarter obtained in Step 1 is as follows:

[0025]

[0026] where h n,k denotes the daily average precipitation in the k-th quarter of the n-th past year, denotes the normalization value of h n,k , denotes the minimum value among the daily average precipitations in the k-th quarter of the past N years, denotes the maximum value among the daily average precipitations in the k-th quarter of the past N years, where k = 1, 2, 3, 4.

[0027] Furthermore, the genetic algorithm is specifically as follows:

[0028] Step 1: Initialize a population containing M individuals. Each individual in the population represents a set of planting plans. Denote the m-th individual in the population as

[0029] Step 2: Initialize the iteration count l = 1;

[0030] Step 3: Select individuals according to the fitness function values and add the selected individuals to the l-th generation population;

[0031] Step 4: Perform crossover and mutation operations on the individuals selected in Step 3, and add the individuals that meet the constraint conditions obtained from the crossover and mutation operations to the l-th generation population;

[0032] Step 5: Calculate the maximum fitness function value in the l-th generation population;

[0033] If the maximum fitness function value does not converge, set l = l + 1 and return to execute Step 3;

[0034] If the maximum fitness function value converges, take the planting plan represented by the individual with the maximum fitness function value as the crop planting plan for the next quarter.

[0035] Furthermore, in Step 1, the individuals in the initialized population include randomly generated planting plans and planting plans with a net profit higher than a set threshold in previous years' planting plans.

[0036] Furthermore, the individuals in the initialized population meet the following constraint conditions:

[0037] (1) A dry field is planted with legumes at least once in K consecutive quarters;

[0038] (2) Rice is not planted in the dry field.

[0039] Furthermore, the specific process of the crossover operation is as follows:

[0040] Denote the vector composed of the coding results of the planting plan represented by an individual as x, x = [x 1 , x 2 , x 3 , x 4 , …, x q , where x 1 , x 2 , x 3 , x 4 , …, x q respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots;

[0041] Denote the vector composed of the coding results of the planting plan represented by another individual as y, y = [y 1 , y 2 , y 3 , y 4 , …, y q , where y 1 , y 2 , y 3 , y 4 , …, y q respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots;

[0042] Randomly select the crossover points, and perform crossover operations on the elements in x and y according to the selected crossover points to obtain the individual after the crossover operation.

[0043] Even further, the specific process of the mutation operation is as follows:

[0044] Generate the mutation rate:

[0045] r l+1 =(a * r l + c) mod m

[0046] where m = 2 31 - 1, a represents a constant relatively prime to m, c represents a non - zero integer relatively prime to m, r l represents the mutation rate of the l - th iteration, and r l+1 represents the mutation rate of the (l + 1) - th iteration;

[0047] If the mutation rate r l+1 is greater than the mutation threshold, no mutation operation is required;

[0048] If the mutation rate r l+1 is less than or equal to the mutation threshold, a bit flip operation is performed on the coding result of the crop planted in each plot of the individual, and the type of crop planted in each plot of the individual is re-determined according to the result of the bit flip operation, that is, a new individual after the mutation operation is obtained.

[0049] The beneficial effects of the present invention are as follows:

[0050] The present invention uses the temperature data, rainfall data, agricultural operation plans, and agricultural product yield data of each historical year to train a BP neural network model. The trained BP neural network model can be used to predict the agricultural product yield. And a genetic algorithm with the total revenue as the fitness function is established. Combining the genetic algorithm with the yield predicted by the trained BP neural network model to optimize the crop planting plan, the best crop planting plan can be obtained, and the optimality of agricultural production planning can be realized. Description of the Drawings

[0051] Figure 1 is a schematic structural diagram of a BP neural network model;

[0052] Figure 2 is a flow chart of a genetic algorithm;

[0053] Figure 3 is a schematic diagram of the planting plan of the previous quarter;

[0054] Figure 4 is a relationship diagram between the number of iterations of the genetic algorithm and the total revenue;

[0055] Generation represents the number of iterations, Total Profit represents the total revenue, and 1e7 represents 10 to the power of 7;

[0056] Figure 5 is a relationship diagram between the number of iterations of the simulated annealing algorithm and the total revenue;

[0057] Figure 6 is a training flow chart of the BP neural network model. Detailed Embodiments

[0058] Detailed Embodiment 1: For an agricultural production planning method based on a genetic algorithm and a BP neural network model described in this embodiment, it is assumed that only one type of crop is planted in a plot, the growth cycle of each planted crop is one quarter, the average temperature of each plot is equal within the same quarter, and the daily average precipitation of each plot is equal within the same quarter; the method specifically includes the following steps:

[0059] Step 1: Obtain the type of crops planted in each plot in each historical quarter, the yield of each plot in each historical quarter, the average temperature in each historical quarter, and the daily average precipitation in each historical quarter respectively;

[0060] Step 2: Use the data obtained in Step 1 to train the BP neural network model to obtain a trained BP neural network model;

[0061] Step 3: Establish the fitness function of the genetic algorithm:

[0062] f = b - c

[0063] where f represents the net profit; b represents the total revenue, and b is calculated based on the yield predicted by the BP neural network model; c represents the total cost;

[0064] Step 4: Use the genetic algorithm to generate a crop planting plan for the next quarter.

[0065] Summarize the method of the present invention into the process shown in Table 1:

[0066] Table 1

[0067]

[0068] Specific Embodiment 2: Combine Figure 1 and Figure 6 to illustrate this embodiment. The difference between this embodiment and Specific Embodiment 1 is that the specific process of Step 2 is as follows:

[0069] Define the input of the BP neural network model as: the type of crops planted in the i-th plot in the j-th historical quarter, the normalized average temperature in the i-th plot in the j-th historical quarter, the normalized daily average precipitation in the i-th plot in the j-th historical quarter, and the reduction rate in the i-th plot in the j-th historical quarter, where i = 1, 2,..., I, I represents the total number of plots, and j = 1, 2,..., J, J represents the total number of quarters;

[0070] Define the output label of the BP neural network model as: the yield of the i-th plot in the j-th historical quarter;

[0071] Perform normalization processing on the average temperature in each historical quarter obtained in Step 1, and perform normalization processing on the daily average precipitation in each historical quarter obtained in Step 1;

[0072] Use the type of crops planted in the i-th plot in the j-th historical quarter, the reduction rate in the i-th plot in the j-th historical quarter, the normalized average temperature in the j-th historical quarter, the normalized daily average precipitation in the j-th historical quarter, and the yield of the i-th plot in the j-th historical quarter to train the BP neural network model.

[0073] Other steps and parameters are the same as those in the first specific implementation manner.

[0074] For the current research area, various crops can be encoded according to the total types of crops planted in all plots within the current research area, and the corresponding encoding of the crops is used as the input of the BP neural network model. For example, corn is encoded as 0001, rice is encoded as 0010, etc.

[0075] This implementation manner takes into account that the planting of a single crop in a plot is affected by the following factors: continuous planting of the same crop weakens soil fertility and leads to reduced yields, the soil fertility of the plot in each quarter is affected by temperature and yields, and the rainfall in each quarter of the plot affects yields. For each plot, a BP neural network model is established, with the type of planted crop, the average temperature of the quarter, the reduction rate of repeated cuttings, and the average precipitation of the quarter as inputs, and the predicted yield as the output of the BP neural network model. The input layer includes 4 neurons, the output layer includes 1 neuron, and the training data for yield prediction is obtained from the yield records of recent plantings in this area. The number of neurons in the hidden layer is calculated by the following formula:

[0076]

[0077] where H h represents the number of neurons in the hidden layer, H s represents the number of neurons in the input layer, H in represents the number of hidden layers, H out represents the number of neurons in the output layer, and α represents the empirical coefficient for adjusting the proportion of the number of neurons in the hidden layer.

[0078] Specific implementation manner three: The difference between this implementation manner and the first or second specific implementation manner is that the reduction rate of the i-th plot in the j-th historical quarter is:

[0079] If the type of crop planted in the i-th plot in the j-th historical quarter is the same as the type of crop planted in the i-th plot in the (j - 1)-th historical quarter, then the reduction rate of the i-th plot in the j-th historical quarter is R;

[0080] If the type of crop planted in the i-th plot in the j-th historical quarter is different from the type of crop planted in the i-th plot in the (j - 1)-th historical quarter, then the reduction rate of the i-th plot in the j-th historical quarter is 0.

[0081] Other steps and parameters are the same as those in the first or second specific implementation manner.

[0082] The reduction rate R can be determined based on historical data. That is, in the historical planting process, if the types of crops planted in a plot in two consecutive quarters are the same, then the reduction rate R is:

[0083] Yn = Y 0 -RY 0

[0084] Among them, Y n represents the yield after continuous cropping, Y 0 represents the yield of the initial planting, and R represents the reduction rate.

[0085] Specific Embodiment 4: What is different from one of Embodiments 1 to 3 is that the average temperature of each historical quarter obtained in Step 1 is normalized as follows:

[0086] It is defined that the data of each plot in the past N years are obtained in Step 1, then the average temperature of the kth quarter in the past N years is normalized:

[0087]

[0088] Among them, W n,k represents the average temperature of the kth quarter in the nth past year, represents W n,k 's normalized value, represents the minimum value of the average temperature of the kth quarter in the past N years, represents the maximum value of the average temperature of the kth quarter in the past N years, k = 1, 2, 3, 4.

[0089] Other steps and parameters are the same as one of Embodiments 1 to 3.

[0090] Specific Embodiment 5: What is different from one of Embodiments 1 to 4 is that the daily average precipitation of each historical quarter obtained in Step 1 is normalized, and the specific process is as follows:

[0091]

[0092] Among them, h n,k represents the daily average precipitation of the kth quarter in the nth past year, represents h n,k 's normalized value, represents the minimum value of the daily average precipitation of the kth quarter in the past N years, represents the maximum value of the daily average precipitation of the kth quarter in the past N years, k = 1, 2, 3, 4.

[0093] Other steps and parameters are the same as one of Embodiments 1 to 4.

[0094] Specific Embodiment 6: Combining Figure 2Describe this embodiment. The difference between this embodiment and any one of the first to fifth specific embodiments is that the genetic algorithm is specifically as follows:

[0095] Step 1: Initialize a population containing M individuals. Each individual in the population represents a set of planting plans (the planting plan includes the types of crops to be planted in each plot in the next quarter). Denote the m-th individual in the population as

[0096] Step 2: Initialize the iteration number l = 1;

[0097] Step 3: Select individuals according to the fitness function values (that is, select individuals with high fitness function values), and add the selected individuals to the l-th generation population; Step 4: Perform crossover and mutation operations on the individuals selected in Step 3, and add the individuals that meet the restrictive conditions obtained from the crossover and mutation operations to the l-th generation population;

[0098] Step 5: Calculate the maximum fitness function value in the l-th generation population;

[0099] If the maximum fitness function value has not converged, let l = l + 1, and return to execute Step 3;

[0100] If the maximum fitness function value has converged, use the planting plan represented by the individual with the maximum fitness function value as the crop planting plan for the next quarter.

[0101] Other steps and parameters are the same as any one of the first to fifth specific embodiments.

[0102] Other steps and parameters are the same as any one of the first to sixth specific embodiments.

[0103] The present invention uses a genetic algorithm to optimize the crop planting plan and can obtain the best crop planting plan.

[0104] Specific Embodiment Seven: The difference between this embodiment and any one of the first to sixth specific embodiments is that in Step 1, the individuals in the initialized population include randomly generated planting plans and planting plans with a net profit higher than a set threshold in previous years' planting plans.

[0105] Other steps and parameters are the same as any one of the first to sixth specific embodiments.

[0106] And the individuals in the initialized population need to meet the restrictive conditions.

[0107] Specific Embodiment Eight: The difference between this embodiment and any one of the first to seventh specific embodiments is that the individuals in the initialized population meet the following restrictive conditions:

[0108] (1) A dry field should be planted with legumes at least once in K consecutive quarters;

[0109] (2) Rice is not planted in dry fields.

[0110] Other steps and parameters are the same as those in any one of the first to seventh specific embodiments.

[0111] Specific Embodiment Nine: The difference between this embodiment and any one of the first to eighth specific embodiments is that the specific process of the crossover operation is as follows:

[0112] Denote the vector composed of the coding results of the planting scheme represented by an individual as x, x = [x 1 , x 2 , x 3 , x 4 , …, x q , where x 1 , x 2 , x 3 , x 4 , …, x q respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots;

[0113] Denote the vector composed of the coding results of the planting scheme represented by another individual as y, y = [y 1 , y 2 , y 3 , y 4 , …, y q , where y 1 , y 2 , y 3 , y 4 , …, y q respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots;

[0114] Randomly select the crossover points, and perform the crossover operation on the elements in x and y according to the selected crossover points to obtain the individual after the crossover operation.

[0115] Other steps and parameters are the same as those in any one of the first to eighth specific embodiments.

[0116] For example, encode the field with the minimum planting area M'. If the total area is M, then it is divided into M / M' plots in total. Randomly select crossover points from [pa, pb, .., pn]. For example, the randomly selected crossover points are p2 and p3. Then the new scheme generated by crossing x and y is:

[0117] x' = [x 1 , y 2 , y 3 , x 4 , …, x M’ / M ​

[0118] y’ = [y 1 , x 2 , x 3 , y 4 , …, y M’ / M

[0119] Specific Embodiment Ten: Different from any one of Specific Embodiments One to Nine, the specific process of the mutation operation is as follows:

[0120] Generate the mutation rate:

[0121] r l+1 = (a * r l + c) mod m

[0122] where m = 2 31 - 1 (i.e., m takes the Mersenne prime), a represents a constant relatively prime to m, c represents a non - zero integer relatively prime to m, r l represents the mutation rate of the l - th iteration, r l+1 represents the mutation rate of the (l + 1) - th iteration; r 0 is a constant obtained through initialization;

[0123] If the mutation rate r l+1 is greater than the mutation threshold (the mutation threshold in the present invention is 0.01), then no mutation operation is required;

[0124] If the mutation rate r l+1 is less than or equal to the mutation threshold, then perform a bit - flip operation on the coding result of the crop planted in each plot of the individual, and re - determine the crop type planted in each plot of the individual according to the result of the bit - flip operation, that is, obtain a new individual after the mutation operation.

[0125] Other steps and parameters are the same as any one of Specific Embodiments One to Nine.

[0126] Perform the mutation operation of this embodiment for each selected individual respectively.

[0127] Experimental Part

[0128] Select a certain area for algorithm simulation. The planting plan of the previous quarter in this area is as Figure 3 shown. Perform algorithm simulation on the plots, and use Matplotlib to visualize the algorithm iteration results as Figure 4 , and it can be found that there will be unstable phenomena in the initial stage of the algorithm. However, as the number of iterations increases, the obtained optimal value tends to be stable. Approximately starting from the 40th generation, the growth of profit tends to be stable and reaches a relatively stable level at the 50th generation. The final profit curve region is stable, indicating that the algorithm has converged to a stable solution. The result of using the simulated annealing algorithm to simulate the same area is as Figure 5 ​As shown, it can be seen that compared with the simulated annealing algorithm, the genetic algorithm requires fewer iterations to obtain an approximate optimal solution, and the obtained approximate optimal value is larger.

[0129] The above examples of the present invention are only to illustrate in detail the calculation model and calculation process of the present invention, rather than limiting the implementation manner of the present invention. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is impossible to enumerate all the implementation manners here. Any obvious changes or variations derived from the technical solution of the present invention still fall within the protection scope of the present invention.

Claims

1. An agricultural production planning method based on genetic algorithm and BP neural network model, characterized in that: The method specifically comprises the following steps: Step 1: Obtain the type of crops planted in each historical quarter for each plot, the yield of each plot in each historical quarter, the average temperature in each historical quarter, and the average daily precipitation in each historical quarter; Step 2: Use the data obtained in step 1 to train the BP neural network model to obtain a trained BP neural network model; Step 3: Establish the fitness function of the genetic algorithm: f=bc Among them, f represents net profit; b represents total revenue, which is calculated based on the output predicted by the BP neural network model; c represents total cost; Step 4: Generate a crop planting plan for the next quarter using a genetic algorithm.

2. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 1 is characterized in that: The specific process of step 2 is as follows: The input of the BP neural network model is defined as: the type of crops planted in the i-th plot in the j-th quarter of history, the normalized average temperature of the i-th plot in the j-th quarter of history, the normalized average daily precipitation of the i-th plot in the j-th quarter of history, and the yield reduction rate of the i-th plot in the j-th quarter of history, i = 1, 2, ..., I, I represents the total number of plots, j = 1, 2, ..., J, J represents the total number of quarters; The output label of the BP neural network model is defined as: the output of the i-th plot in the j-th quarter of history; Normalize the average temperature of each historical quarter obtained in step 1, and normalize the average daily precipitation of each historical quarter obtained in step 1; The BP neural network model is trained using the type of crops planted on the ith plot in the jth historical quarter, the yield reduction rate of the ith plot in the jth historical quarter, the normalized average temperature of the jth historical quarter, the normalized average daily precipitation of the jth historical quarter, and the yield of the ith plot in the jth historical quarter.

3. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 2 is characterized in that: The reduction rate of the i-th plot in the j-th quarter of history is: If the crop type planted on the i-th plot in the j-th quarter of history is the same as the crop type planted on the i-th plot in the j-1-th quarter of history, then the yield reduction rate of the i-th plot in the j-th quarter of history is R; If the crop type planted on the i-th plot in the j-th quarter of history is different from the crop type planted on the i-th plot in the j-1-th quarter of history, the yield reduction rate of the i-th plot in the j-th quarter of history is 0.

4. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 3 is characterized in that: The normalization process of the average temperature of each historical quarter obtained in step 1 is specifically as follows: In the definition step 1, the data of each plot in the past N years are obtained in total, and the average temperature of the kth quarter in the past N years is normalized: Among them, W n,k represents the average temperature of the kth quarter in the past nth year, W n,k The normalized value of represents the minimum value of the average temperature in the kth quarter of the past N years, Represents the maximum value of the average temperature in the kth quarter of the past N years, k = 1, 2, 3, 4.

5. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 4 is characterized in that: The normalization process of the historical daily average precipitation in each quarter obtained in step 1 is as follows: Among them, h n,k represents the average daily precipitation in the kth quarter of the past nth year, Indicates h n,k The normalized value of represents the minimum value of the average daily precipitation in the kth quarter of the past N years, It represents the maximum value of the average daily precipitation in the kth quarter of the past N years, where k = 1, 2, 3, 4.

6. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 5 is characterized in that: The genetic algorithm is specifically: Step 1: Initialize a population of M individuals. Each individual in the population represents a set of planting plans. The mth individual in the population is denoted as Step 2, initialize the number of iterations l = 1; Step 3: According to the fitness function value, Make a selection and add the selected individuals to the l-generation population; Step 4: Perform crossover and mutation operations on the individuals selected in step 3, and add the individuals that meet the restriction conditions obtained by the crossover and mutation operations to the first generation population; Step 5, calculate the maximum fitness function value in the lth generation population; If the maximum fitness function value has not converged, set l = l + 1 and return to step 3; If the maximum fitness function value converges, the planting plan represented by the individual with the maximum fitness function value will be used as the crop planting plan for the next quarter.

7. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 6 is characterized in that: In step 1, the individuals in the initialized population include randomly generated planting plans and planting plans whose net profits are higher than a set threshold in the planting plans of previous years.

8. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 7 is characterized in that: The individuals in the initialized population meet the following constraints: (1) A dryland field is planted with beans at least once in K consecutive seasons; (2) Rice is not grown in dry fields.

9. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 8 is characterized in that: The specific process of the crossover operation is as follows: The vector composed of the encoding results of the planting plan represented by an individual is denoted as x, x = [x1, x2, x3, x4, ..., x q ],x1,x2,x3,x4,…,x q Respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots; The vector composed of the encoding results of the planting plan represented by another individual is recorded as y, y = [y1, y2, y3, y4, ..., y q ],y1,y2,y3,y4,…,y q Respectively represent the coding results of the crops planted in the 1st, 2nd, 3rd, 4th, …, qth plots; Randomly select a crossover point, perform a crossover operation on the elements in x and y according to the selected crossover point, and obtain the individuals after the crossover operation.

10. The agricultural production planning method based on genetic algorithm and BP neural network model according to claim 9, characterized in that: The specific process of the mutation operation is as follows: Generate mutation rate: r l+1 =(a*r l +c)modm Where m = 2 31 -1, a is a constant that is relatively prime to m, c is a non-zero integer that is relatively prime to m, r l represents the mutation rate of the lth iteration, r l+1 represents the mutation rate of the l+1th iteration; If the mutation rate r l+1 If it is greater than the mutation threshold, no mutation operation is required; If the mutation rate r l+1 If it is less than or equal to the mutation threshold, a bit flipping operation is performed on the coding result of the crops planted in each plot in the individual, and the type of crops planted in each plot in the individual is re-determined according to the result of the bit flipping operation, that is, a new individual after the mutation operation is obtained.

Images