Intelligent prediction method for corn yield

Abstract

The invention discloses an intelligent prediction method for corn yield, and the method comprises the steps: obtaining a multi-layer feedforward neural network model through training, training the multi-layer feedforward neural network model through a genetic method, and enabling an obtained prediction model to have higher prediction precision when predicting the corn yield; according to the method, by assimilating vegetation net primary productivity data having a great influence on the corn yield, more accurate prediction data of the corn growth state can be obtained, and the network model is updated based on the prediction data, so that the model better adapts to the current corn growth data, and the accuracy of corn yield prediction is improved; according to the method, the prediction operation speed is increased through the multi-layer feedforward neural network model, and the corn yield can be predicted more scientifically and reasonably, so that corn production and management measures can be accurately adjusted in time more conveniently.

Description

Technical Field

[0001] The present invention relates to the technical field of corn yield prediction, and in particular to an intelligent prediction method for corn yield. Background Art

[0002] Corn has attracted much attention due to its rich nutrition, wide range of uses, and great market potential. At present, the corn planting area in my country is gradually expanding. The efficient prediction of corn yield is of great significance for making accurate management decisions during its growth period. The WOFOST prediction model is a dynamic explanatory model driven by daily meteorological data. It simulates the growth process of crops from sowing to maturity, including the simulation of basic physiological processes of crops such as photosynthesis, respiration, transpiration, and dry matter distribution. Although the model has strong adaptability, the data and calculation process used in the model are large in amount. When predicting corn yield, the final yield value cannot be obtained quickly and accurately, and the operation flexibility is poor. The existing BP (backpropagation) multi-layer feedforward neural network model can obtain prediction results quickly, but there are problems of low test accuracy and poor robustness in the prediction.

[0003] Therefore, it is necessary to develop an intelligent prediction method for corn yield to address the above defects. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent prediction method for corn yield, which can solve the problem that the existing prediction model cannot take into account both accuracy and response speed and has poor flexibility.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0006] The present invention provides an intelligent prediction method for corn yield, comprising:

[0007] Step 1: Collect remote sensing observation data and growth-related data of corn planting status in the test area, and use the CASA model to process the remote sensing observation data to obtain vegetation net primary productivity data;

[0008] Step 2: Select corn growth-related data from the database, and assimilate the WOFOST model using the ensemble Kalman filter assimilation method based on the corresponding vegetation net primary productivity data, and use the assimilated WOFOST model to obtain the predicted data of multiple groups of growth-related data after the current state of corn in the tested area using the collected current vegetation net primary productivity data and growth-related data;

[0009] Step 3: Select corn growth-related and corresponding corn yield data from the database, train the multi-layer feedforward neural network model based on the genetic method, input multiple sets of prediction data obtained by running the assimilated WOFOST model into the trained multi-layer feedforward neural network model, and obtain the primary corn yield prediction result;

[0010] Step 4: Combine the growth-related data of the previous stage with the primary corn yield prediction results of the adjacent subsequent stage to obtain an updated data set, and perform update training on the multi-layer feedforward neural network model based on a genetic method;

[0011] Step 5: Use the updated neural network model to obtain the corn yield prediction results corresponding to the corn in the tested area using the collected growth-related data.

[0012] Furthermore, the specific steps of WOFOST model assimilation in step 2 include:

[0013] Step 2.1, using the existing growth data in the database to obtain the relevant parameter values ​​of the WOFOST model, and creating the WOFOST model based on the relevant parameters;

[0014] Step 2.2, using the WOFOST model to obtain further prediction values ​​of corn growth data in the tested area;

[0015] Step 2.3: Use the ensemble Kalman filter assimilation method to assimilate the obtained vegetation net primary productivity data into further prediction values ​​to obtain the assimilated prediction data.

[0016] Furthermore, the training steps of the multi-layer feedforward neural network model in step 3 include:

[0017] Step 3.1, obtaining corn growth-related and corresponding corn yield data in the database, and preprocessing the obtained data, wherein the corn growth-related data includes meteorological data, field environment data, plant characteristics and fertilization information;

[0018] Step 3.2, generate an initialized population based on the preprocessed data and calculate the individual fitness value;

[0019] Step 3.3, performing optimal iteration on the generated population based on a genetic method;

[0020] Step 3.4, obtain the optimal position that meets the conditions after iteration and calculate the optimal weight threshold;

[0021] Step 3.5: Input the obtained optimal solution into the multi-layer feedforward neural network model for training and prediction analysis to obtain a trained multi-layer feedforward neural network model.

[0022] Furthermore, the fitness function F of the individual fitness value in step 3.2 is expressed as:

[0023] Among them, f i is the expected output value, f(x i ) is the actual output value, i is the corresponding code of the input layer, and n is the number of input layers.

[0024] Furthermore, the specific steps of step 3.3 include:

[0025] Step 3.3.1: Select excellent samples from the original population through selection operation. The probability function of individual selection in the selection operation is expressed as Among them, F i is the population fitness value of individual i, and N is the population size;

[0026] Step 3.3.2: Generate new individual samples with strong adaptability from excellent samples through crossover operation;

[0027] Step 3.3.3: Generate mutant individuals from the new individual samples through mutation operation to obtain a new population.

[0028] Furthermore, the specific steps of step 3.4 include obtaining the fitness values ​​of individuals in the newly generated population, repeating the optimal iteration until the fitness value meets the requirements or the number of evolutions reaches the maximum value, and obtaining the weights and thresholds of the multilayer feedforward neural network model optimized by the genetic method.

[0029] Furthermore, the function of the assimilated prediction data in step 2.3 is expressed as: t =X t，t-1 +P t，t-1 H T (HP t，t-1 H T +R T ) -1 (Y t -HX t，t-1 ), where X t represents the predicted data after assimilation at time t, Y t represents the net primary productivity of vegetation at time t, X t，t-1 represents the further predicted value at time t, P t，t-1 represents the covariance of further predictions, R t is the mean of the observation equation error, and H represents the observation operator.

[0030] Furthermore, the function of the further predicted value in step 2.2 is expressed as X t，t-1 =f(X t-1 , U t , θ), where Xt-1 represents the estimated value at time t-1, X t，t-1 Represents X t-1 The further predicted value at time t obtained after the WOFOST model prediction, U t is the growth-related data input for running the WOFOST model; θ is the relevant parameter of the WOFOST model.

[0031] Furthermore, the function of the crossover operation in step 3.3.2 is expressed as a kj =a kj (1-b)+a ij , where a kj are two individuals a in the excellent sample k and a i A new individual is obtained by the crossover operation at j, and b is a random number.

[0032] Furthermore, the function of the mutation operation in step 3.3.3 is expressed as:

[0033] Among them, a max 、a min for a ij The boundary conditions, r2 is a random number, g is the number of iterations, G max is the maximum number of evolutions, a ij is individual a in the new individual sample i exist j A mutation occurs and a new individual is formed.

[0034] Compared with the prior art, the beneficial technical effects of the present invention are:

[0035] The invention discloses an intelligent prediction method for corn yield. The multi-layer feedforward neural network model is obtained by training. The multi-layer feedforward neural network model is trained by a genetic method. The obtained prediction model overcomes the blindness and uncertainty of parameter selection of a traditional neural network model when predicting corn yield, and has higher prediction accuracy. The method can obtain more accurate prediction data of corn growth status by assimilating vegetation net primary productivity data that has a great influence on corn yield. Localized updating of the network model based on the prediction data can make the model better adapt to current corn growth data, and further improve the accuracy of corn yield prediction. The method can predict corn yield more scientifically and reasonably by improving the speed of prediction operation through the multi-layer feedforward neural network model, so as to facilitate timely and accurate adjustment of corn production and management measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention will be further described below in conjunction with the accompanying drawings.

[0037] Figure 1 The present invention is a process flowchart of the intelligent prediction method for corn yield. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] See also Figure 1 Now, a specific implementation of an intelligent prediction method for corn yield provided by the present invention is described. The intelligent prediction method for corn yield includes:

[0040] S1. Collect remote sensing observation data and growth-related data of corn planting status in the test area, and use CASA (Carnegie-Ames-Stanford approach) model to process the remote sensing observation data to obtain vegetation net primary productivity data;

[0041] S2. Select corn growth-related data from the database, and assimilate the WOFOST model using the ensemble Kalman filter assimilation method based on the corresponding vegetation net primary productivity data, and use the assimilated WOFOST model to obtain the predicted data of multiple groups of growth-related data after the current state of corn in the tested area based on the collected current vegetation net primary productivity data and growth-related data;

[0042] The historical observation data of key growth periods of corn in the database are used to determine the relevant parameter values ​​of the WOFOST model, including the effective accumulated temperature from sowing to seedling stage, the effective accumulated temperature from seedling stage to flowering stage, and the effective accumulated temperature from flowering stage to maturity stage. The ensemble Kalman filter assimilation method can be used for data assimilation of nonlinear systems with less computational effort. The specific steps of WOFOST model assimilation in S2 include:

[0043] S2.1, using the existing growth data in the database to obtain the relevant parameter values ​​of the WOFOST model, and creating the WOFOST model based on the relevant parameters;

[0044] S2.2, using the WOFOST model to obtain further prediction values ​​of corn growth data in the tested area;

[0045] The function of the further predicted value in S2.2 is expressed as X t，t-1 =f(X t-1 , U t , θ), where X t-1 represents the estimated value at time t-1, Xt，t-1 Represents X t-1 The further predicted value at time t obtained after the WOFOST model prediction, U t is the growth-related data input for running the WOFOST model; θ is the relevant parameter of the WOFOST model.

[0046] S2.3. Use the ensemble Kalman filter assimilation method to assimilate the obtained vegetation net primary productivity data into further prediction values ​​to obtain the assimilated prediction data.

[0047] The function of the assimilated prediction data in S2.3 is expressed as:

[0048] X t =X t，t-1 +P t，t-1 H T (HP t，t-1 H T +R T ) -1 (Y t -HX t，t-1 ), P t，t-1 =<(X t，t-1 -X t )(X t，t-1 -X t ) T > Among them, X t represents the predicted data after assimilation at time t, Y t represents the net primary productivity of vegetation at time t, X t，t-1 represents the further predicted value at time t, P t，t-1 represents the covariance of further predictions, R t is the mean of the observation equation error, and H represents the observation operator.

[0049] S3, selecting corn growth-related and corresponding corn yield data from the database, training the multi-layer feedforward neural network model based on the genetic method, inputting multiple sets of prediction data obtained by running the assimilated WOFOST model into the trained multi-layer feedforward neural network model, and obtaining the primary corn yield prediction result;

[0050] The training steps of the multi-layer feedforward neural network model in S3 include:

[0051] S3.1. Obtain corn growth-related and corresponding corn yield data in the database, and pre-process the acquired corn growth-related data. The corn growth-related data includes meteorological data, field environment data, plant characteristics and fertilization information;

[0052] Before constructing the model in S3.1, the topology of the neural network model needs to be determined. The meteorological data in S3.1 include atmospheric humidity, atmospheric temperature, and rainfall. The database is the field Internet of Things or corn growth data collected on the spot. The corn yield is significantly correlated with the monthly minimum soil temperature, monthly average soil temperature, monthly maximum atmospheric temperature, monthly average atmospheric humidity, monthly maximum temperature, monthly average soil moisture content, monthly average atmospheric temperature, monthly rainfall, and monthly average atmospheric humidity. The model is trained with the monthly minimum soil temperature, monthly average soil temperature, monthly maximum atmospheric temperature, monthly average atmospheric humidity, monthly maximum temperature, monthly average soil moisture content, monthly average atmospheric temperature, monthly rainfall, and monthly average atmospheric humidity as the main data. With the selected multiple indicators as input and corn yield as output, a multi-layer feedforward neural network model with multiple node input layers and one node output layer is constructed, and training and prediction are performed.

[0053] S3.2, based on the preprocessed data, set the population size to generate the initial population and calculate the individual fitness value. The population mainly includes the weights and thresholds of the neural network model;

[0054] The fitness function F of the individual fitness value in S3.2 is expressed as:

[0055] Among them, f i is the expected output value, f(x i ) is the actual output value, i is the corresponding code of the input layer, and n is the number of input layers.

[0056] S3.3, the generated population is selected and iterated based on the genetic method, and the genetic method includes population selection, population crossover and population mutation;

[0057] The specific steps of S3.3 include:

[0058] S3.3.1. Select excellent samples from the original population through selection operation. Selection operation refers to selecting excellent samples from the original population with a certain probability and generating the next generation of sample data through reproduction. The probability function of individual selection in the selection operation is expressed as Among them, F i is the population fitness value of individual i, and N is the population size;

[0059] S3.3.2, Generate new individual samples with strong adaptability from excellent samples through crossover operation;

[0060] Crossover operation refers to randomly selecting two samples from the population and generating new individuals with strong adaptability through exchange and combination. For example, two individuals a k 、a i The crossover operation is performed at j. The function of the crossover operation in S3.3.2 is expressed as a kj =akj (1-b)+a ij , a ij =a ij (1-b)+a kj , where a kj are two individuals a in the excellent sample k and a i A new individual is obtained by the crossover operation at j, and b is a random number between [0, 1].

[0061] S3.3.3. Generate mutant individuals from new individual samples through mutation operation to obtain a new population.

[0062] The diversity of the population can be maintained through population mutation. The mutation operation is to randomly select an individual from the population and perform mutation operations on a part of the individual to produce better individuals. The function of the mutation operation in S3.3.3 is expressed as: Among them, a max 、a min for a ij The boundary conditions, r2 is a random number, g is the number of iterations, G max is the maximum number of evolutions, a ij is individual a in the new individual sample i exist j A new individual is formed by mutation.

[0063] S3.4, obtaining the optimal position that meets the conditions after iteration and calculating the optimal weight threshold;

[0064] The specific steps of S3.4 include obtaining the fitness values ​​of individuals in the newly generated population, repeating the optimal iteration until the fitness value meets the requirements or the number of evolutions reaches the maximum value, and obtaining the weights and thresholds of the multi-layer feedforward neural network model optimized by genetic method.

[0065] S3.5. Input the obtained optimal solution into the multi-layer feedforward neural network model for training and prediction analysis to obtain a trained multi-layer feedforward neural network model.

[0066] The multi-layer feedforward neural network model, which optimizes the number of hidden nodes and learning rate of the multi-layer feedforward neural network model using genetic algorithm, overcomes the blindness and uncertainty of parameter selection of the traditional neural network model and effectively improves the prediction accuracy of the model.

[0067] S4, combining the growth-related data of the previous stage with the primary corn yield prediction results of the adjacent subsequent stage to obtain an updated data set, and updating and training the multi-layer feedforward neural network model based on a genetic method;

[0068] Genetic methods are currently the most widely used adaptive heuristic search algorithms for solving optimization problems.

[0069] S5. Use the updated neural network model to obtain the corn yield prediction result corresponding to the corn in the tested area using the collected growth-related data.

[0070] Compared with the prior art, the intelligent prediction method for corn yield obtains a multi-layer feedforward neural network model through training, and the multi-layer feedforward neural network model is trained through a genetic method. The obtained prediction model overcomes the blindness and uncertainty of parameter selection of the traditional neural network model when predicting corn yield, and has higher prediction accuracy; the method can obtain more accurate prediction data of corn growth status by assimilating vegetation net primary productivity data that has a great impact on corn yield, and localized updating of the network model based on the prediction data can make the model better adapt to the current corn growth data, and further improve the accuracy of corn yield prediction; the method improves the speed of prediction operation through the multi-layer feedforward neural network model, has better flexibility, can predict corn yield more scientifically and reasonably, and is more convenient to timely and accurately adjust corn production and management measures, and solves the problem that the existing prediction model cannot take into account both accuracy and response speed and has poor flexibility.

[0071] The embodiments described above are only descriptions of the preferred modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent prediction method for corn yield, characterized in that: include: Step 1: Collect remote sensing observation data and growth-related data of corn planting status in the test area, and use the CASA model to process the remote sensing observation data to obtain vegetation net primary productivity data; Step 2: Select corn growth-related data from the database, and assimilate the WOFOST model using the ensemble Kalman filter assimilation method based on the corresponding vegetation net primary productivity data, and use the assimilated WOFOST model to obtain the predicted data of multiple groups of growth-related data after the current state of corn in the tested area using the collected current vegetation net primary productivity data and growth-related data; Step 3: Select corn growth-related and corresponding corn yield data from the database, train the multi-layer feedforward neural network model based on the genetic method, input multiple sets of prediction data obtained by running the assimilated WOFOST model into the trained multi-layer feedforward neural network model, and obtain the primary corn yield prediction result; Step 4: Combine the growth-related data of the previous stage with the primary corn yield prediction results of the adjacent subsequent stage to obtain an updated data set, and perform update training on the multi-layer feedforward neural network model based on a genetic method; Step 5: Use the updated neural network model to obtain the corn yield prediction results corresponding to the corn in the tested area using the collected growth-related data.

2. The intelligent prediction method for corn yield according to claim 1, characterized in that: The specific steps of WOFOST model assimilation in step 2 include: Step 2.1, using the existing growth data in the database to obtain the relevant parameter values ​​of the WOFOST model, and creating the WOFOST model based on the relevant parameters; Step 2.2, using the WOFOST model to obtain further prediction values ​​of corn growth data in the tested area; Step 2.3: Use the ensemble Kalman filter assimilation method to assimilate the obtained vegetation net primary productivity data into further prediction values ​​to obtain the assimilated prediction data.

3. The intelligent prediction method for corn yield according to claim 2, characterized in that: The training steps of the multi-layer feedforward neural network model in step 3 include: Step 3.1, obtaining corn growth-related and corresponding corn yield data in the database, and preprocessing the obtained data, wherein the corn growth-related data includes meteorological data, field environment data, plant characteristics and fertilization information; Step 3.2, generate an initialized population based on the preprocessed data and calculate the individual fitness value; Step 3.3, performing optimal iteration on the generated population based on a genetic method; Step 3.4, obtain the optimal position that meets the conditions after iteration and calculate the optimal weight threshold; Step 3.5: Input the obtained optimal solution into the multi-layer feedforward neural network model for training and prediction analysis to obtain a trained multi-layer feedforward neural network model.

4. The intelligent prediction method for corn yield according to claim 3, characterized in that: The fitness function F of the individual fitness value in step 3.2 is expressed as: Among them, f i is the expected output value, f(x i ) is the actual output value, i is the corresponding code of the input layer, and n is the number of input layers.

5. The intelligent prediction method for corn yield according to claim 4, characterized in that: The specific steps of step 3.3 include: Step 3.3.1: Select excellent samples from the original population through selection operation. The probability function of individual selection in the selection operation is expressed as Among them, F i is the population fitness value of individual i, and N is the population size; Step 3.3.2: Generate new individual samples with strong adaptability from excellent samples through crossover operation; Step 3.3.3: Generate mutant individuals from the new individual samples through mutation operation to obtain a new population.

6. The intelligent prediction method for corn yield according to claim 5, characterized in that: The specific steps of step 3.4 include obtaining the fitness values ​​of individuals in the newly generated population, repeating the optimal iteration until the fitness value meets the requirements or the number of evolutions reaches the maximum value, and obtaining the weights and thresholds of the multi-layer feedforward neural network model optimized by the genetic method.

7. The intelligent prediction method for corn yield according to claim 2, characterized in that: The function of the assimilated predicted data in step 2.3 is expressed as: X t =X t，t-1 +P t，t-1 H T (HP t，t-1 H T +R T ) -1 (Y t -HX t，t-1 ), where X t represents the predicted data after assimilation at time t, Y t represents the net primary productivity of vegetation at time t, X t，t-1 represents the further predicted value at time t, P t，t-1 represents the covariance of further predictions, R t is the mean of the observation equation error, and H represents the observation operator.

8. The intelligent prediction method for corn yield according to claim 2, characterized in that: The function of the further predicted value in step 2.2 is expressed as X t，t-1 =f(X t-1 , U t , θ), where X t-1 represents the estimated value at time t-1, X t，t-1 Represents X t-1 The further predicted value at time t obtained after the WOFOST model prediction, U t is the growth-related data input for running the WOFOST model; θ is the relevant parameter of the WOFOST model.

9. The intelligent prediction method for corn yield according to claim 5, characterized in that: The function of the crossover operation in step 3.3.2 is expressed as a kj =a kj (1-b)+a ij , where a kj are two individuals a in the excellent sample k and a i A new individual is obtained by the crossover operation at j, and b is a random number.

10. The intelligent prediction method for corn yield according to claim 5, characterized in that: The function of the mutation operation in step 3.3.3 is expressed as: Among them, a max 、a min for a ij The boundary conditions, r2 is a random number, g is the number of iterations, G max is the maximum number of evolutions, a ij is individual a in the new individual sample i exist j A mutation occurs somewhere and a new individual is formed.

Images