Agricultural informatization management platform based on big data

Abstract

The invention relates to the field of agricultural planting, and discloses an agricultural informatization management platform based on big data, and the platform comprises a data collection module which is used for collecting the over-the-year trading volume data of each crop seedling of each e-commerce platform and the over-the-year average purchase price of each crop; the data analysis module is used for obtaining a growth month and a planting area of each kind of crops, determining that ground disasters, weather disasters and planting subsidies exist in the planting area of the crops over the years in the growth month of the crops, and obtaining alternative years; the data clustering module is used for clustering the alternative years and performing deviation analysis on the alternative years according to the average purchase price of the clustered years to obtain a correction function; and the data processing module is used for correcting the average purchasing price of each crop over the years according to the correction function to obtain standard data.

Description

Technical Field

[0001] The present invention relates to the field of agricultural planting, and more specifically, to an agricultural information management platform based on big data. Background Art

[0002] In modern agricultural production, farmers often face uncertainty caused by market price fluctuations when making planting decisions. However, since market prices and sales volumes are affected by many factors, farmers often lack scientific data support, which makes it easy for them to blindly follow the trend or miss market opportunities during the planting process.

[0003] Existing agricultural information management platforms mainly focus on providing real-time market data and decision-making tools, but they are insufficient in dealing with the lag in prices during agricultural planting. When crops mature, decisions based on real-time market data can only adjust their profit margins within a limited range. Coupled with the limitation of agricultural product storage costs, this decision-making model imposes a greater constraint on the upper limit of farmers' profits. Summary of the invention

[0004] The present invention provides an agricultural information management platform based on big data to solve the technical problems raised in the background technology.

[0005] The present invention provides an agricultural information management platform based on big data, including a data acquisition module, a data analysis module, a data clustering module, a data processing module, a prediction module and a planning module: The data collection module is used to collect the transaction volume data of each crop seedling on each e-commerce platform over the years and the average purchase price of each crop over the years; The data analysis module is used to obtain the growing months and planting areas of each crop, and determine the presence of ground disasters, weather disasters and planting subsidies in the planting areas of crops in the growing months of the crops in previous years, and obtain alternative years; A data clustering module is used to cluster the candidate years and perform deviation analysis on the candidate years according to the average purchase price of the clustered years to obtain a correction function; A data processing module is used to correct the average purchase price of each crop over the years according to the correction function to obtain standard data; The forecasting module is used to input the standard data into the preset forecasting model to obtain the average selling price of each crop in the current year; The planning module is used to obtain the planting plan for the current year based on the average selling price of each crop in the current year combined with the optimization algorithm.

[0006] Furthermore, it is determined that the planting areas of previous years have ground disasters, weather disasters and planting subsidies in the growing months of crops, and alternative years are obtained, including: Get the Multiple growing regions for the crop and the growing months in each growing region; Determine If there are ground disasters, weather disasters and planting subsidies in the growing months of the k-th planting area of ​​the crop in previous years, the years in which the ground disasters, weather disasters and planting subsidies occur will be repeatedly divided into the first alternative year, the second alternative year and the third alternative year.

[0007] Furthermore, the candidate years are clustered, including: Determine the e-commerce platform transaction volume and average purchase price in the year before each candidate year; For the previous year, the transaction volume and average purchase price of the e-commerce platform are matched to the year in which the transaction volume and average purchase price of the e-commerce platform meet the preset rules to obtain the matching year; The year following the matching year that is not a candidate year is used as the first screening year; The transaction volume of the e-commerce platform in the first screening year and the transaction volume of the e-commerce platform in the alternative year are matched according to preset rules to obtain the second screening year, and the average acquisition price in the second screening year is recorded.

[0008] Furthermore, preset rules include: Compare the financial report data of e-commerce platforms over the years to obtain the transaction volume correction factors of crop seedlings over the years; among which, the financial report data represents the profit parameters of the e-commerce platform in the corresponding year; The pth year and the qth year are matched according to the transaction volume correction factor of crop seedlings in previous years, including: Determine the year that comes before the p-th year and the q-th year, and calculate the year interval; Correct the previous year to the next year year by year to get the corrected trading volume; If the difference between the corrected trading volume in the previous year and the trading volume in the subsequent year is determined to be less than a preset difference threshold, then the matching is completed.

[0009] Furthermore, deviation analysis of alternative years was conducted based on the average acquisition price of clustered years, including: Let the first correction function be , the second correction function is , the third correction function is ; Determine The number of times the crop is repeated in the first, second and third candidate years in the sth candidate year is obtained. The combined deviation of the crop in the sth candidate year is ;in, , and Represent the number of repetitions of the first alternative year, the second alternative year, and the third alternative year respectively; Calculate the The average purchase price of the crop in the second screening year is The average difference in the average purchase price of the crop in the sth alternative year; Determine the average differences of multiple crops in multiple candidate years respectively, and use the multiple average differences as reference sample values ​​of comprehensive deviations of multiple crops in multiple candidate years; Obtaining multiple reference sample values ​​of the first correction function, the second correction function and the third correction function by linear least squares fitting; Establishing fitting polynomials for the first correction function, the second correction function and the third correction function respectively; The weight parameters and constant parameters in the first correction function, the second correction function and the third correction function are fitted according to a plurality of reference sample values ​​to obtain the fitted first correction function, the second correction function and the third correction function.

[0010] Furthermore, the average purchase price of each crop over the years is corrected according to the correction function to obtain standard data, including: Based on the fitted first correction function, second correction function and third correction function, we get The comprehensive deviation value of the transaction volume data of the crop in the sth alternative year; The first The transaction volume data of the crop in the sth candidate year minus the comprehensive deviation value is obtained Standard data of the crop in the sth candidate year.

[0011] Furthermore, the preset prediction model includes input layer, hidden layer and output layer: Configure the input layer of the prediction model, and input the feature vector constructed by the standard data of previous years; Configure the hidden layer of the prediction model. The hidden layer performs multi-scale sampling on the feature vector, including: Perform global sampling on the feature vectors of previous years to obtain the global features of the standard data of previous years; Perform sliding sampling on the feature vectors of previous years to obtain the local features of the year within the sliding window; Configure the output layer of the prediction model. The output layer is used to weight and merge the global features and local features to obtain the hidden state, and the hidden state is input into the classifier. The classification space of the classifier represents the average purchase price of the crops in the current year. The prediction model updates the weight parameters and bias parameters of the hidden layer through the mean error variance.

[0012] Furthermore, based on the average selling price of each crop in the current year, combined with the optimization algorithm, the planting plan for the current year is obtained, including: Determine the crops to be planted by target farmers; Obtain the transaction volume of seedlings of the crops to be selected for planting, the average purchase price of the crops to be selected for planting, the growing months and planting areas of the crops to be selected for planting, and the ground disasters, weather disasters and planting subsidies in the planting areas of the crops to be selected for planting; The average purchase price of the crop to be selected for planting in previous years is corrected by a correction function to obtain the standard data of the crop to be selected for planting in previous years; Input the standard data of the crops to be selected for planting in previous years into the prediction model to obtain the average purchase price of the crops to be selected for planting in the current year; Randomly generate an initial population with R individuals that meet the constraints, and initialize the current number of iterations to 1; The individual code is represented as: ;in, Expressing the The planting area allocated to the crops to be planted, Y represents the number of types of crops to be planted; The constraints are as follows: ; represents the total planting area of ​​the target farmers; For any two crops L1 and L2 to be planted, if the growing months of L1 and L2 overlap, then: and No intersection; The area of ​​each crop to be planted ; The objective function is as follows: ;in, Indicates The return per unit area of ​​the crop to be planted; Obtain the objective function value of each individual through the objective function; Select the corresponding individual as the parent to update; When the current number of iterations reaches the preset number of iterations, the update stops and the planting plan of the target farmers in the current year is output.

[0013] The beneficial effect of the present invention is that by screening and excluding external environmental factors in historical data, a forecast compliance scheme based only on market factors is constructed, so that the forecast results more purely reflect market changes. This method avoids the interference of external factors, provides farmers with more accurate market-oriented forecasts, helps them obtain higher returns on the optimal planting strategy, and thus maximizes the planting benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is an architecture diagram of an agricultural information management platform based on big data of the present invention; Figure 2 It is a flow chart of an agricultural information management method based on big data of the present invention. DETAILED DESCRIPTION

[0015] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0016] like Figure 1-2 As shown, an agricultural information management platform based on big data includes a data collection module, a data analysis module, a data clustering module, a data processing module, a prediction module and a planning module: The data collection module is used to collect the transaction volume data of each crop seedling on each e-commerce platform over the years and the average purchase price of each crop over the years; The data analysis module is used to obtain the growing months and planting areas of each crop, and determine the presence of ground disasters, weather disasters and planting subsidies in the planting areas of crops in the growing months of the crops in previous years, and obtain alternative years; A data clustering module is used to cluster the candidate years and perform deviation analysis on the candidate years according to the average purchase price of the clustered years to obtain a correction function; A data processing module is used to correct the average purchase price of each crop over the years according to the correction function to obtain standard data; The forecasting module is used to input the standard data into the preset forecasting model to obtain the average selling price of each crop in the current year; The planning module is used to obtain the planting plan for the current year based on the average selling price of each crop in the current year combined with the optimization algorithm.

[0017] In one embodiment of the present invention, determining whether ground disasters, weather disasters, and planting subsidies exist in the planting areas of crops in previous years during the growing months of the crops, and obtaining alternative years, includes: Get the Multiple growing regions for the crop and the growing months in each growing region; Determine If there are ground disasters, weather disasters and planting subsidies in the growing months of the k-th planting area of ​​the crop in previous years, the years in which the ground disasters, weather disasters and planting subsidies occur will be repeatedly divided into the first alternative year, the second alternative year and the third alternative year.

[0018] Specifically, the crop mung bean has planting zones 1, 2, 3 and 4.

[0019] Among them, planting area 1 had government subsidies in 2013 (April-August), planting area 2 had geological disasters in 2013 (March-September), planting area 3 had weather disasters in 2015 (January-October), and planting area 4 had weather disasters in 2013 (June-August); then 2013 and 2015 were alternative years for the crop mung bean, and the crop mung bean was repeatedly divided into the first alternative year, the second alternative year and the third alternative year in 2013, and the crop mung bean was only divided into the second alternative year in 2015.

[0020] In one embodiment of the present invention, clustering the candidate years includes: Determine the e-commerce platform transaction volume and average purchase price in the year before each candidate year; For the previous year, the transaction volume and average purchase price of the e-commerce platform are matched to the year in which the transaction volume and average purchase price of the e-commerce platform meet the preset rules to obtain the matching year; The year following the matching year that is not a candidate year is used as the first screening year; The transaction volume of the e-commerce platform in the first screening year and the transaction volume of the e-commerce platform in the alternative year are matched according to preset rules to obtain the second screening year, and the average acquisition price in the second screening year is recorded.

[0021] Specifically, if the crop mung bean is an alternative year in 2013, 2012 will be used as the clustering index for 2013; the transaction volume and average purchase price in 2012 will be matched one by one with the transaction volume and average purchase price in previous years, and a non-alternative year that can be matched with 2012, such as 2017, will be found, and it will be judged that 2018 may be the same as 2013; then, the transaction volume in 2018 is matched again with the transaction volume in 2013; if the difference between the transaction volume in 2018 and the transaction volume in 2013 is found to be less than the preset threshold, the matching of 2018 and 2013 is completed; this indicates that the average purchase price in 2018 has a reference value for the average purchase price in 2013.

[0022] Through multi-step clustering and screening, we ensure that highly relevant years are obtained from historical data to predict the market trend of the current year. This significantly reduces the volatility and deviation of data, improves the accuracy of prediction, and provides farmers with more reliable market data support when making agricultural production decisions.

[0023] In one embodiment of the present invention, the preset rules include: Compare the financial report data of e-commerce platforms over the years to obtain the transaction volume correction factors of crop seedlings over the years; among which, the financial report data represents the profit parameters of the e-commerce platform in the corresponding year; The pth year and the qth year are matched according to the transaction volume correction factor of crop seedlings in previous years, including: Determine the year that comes before the p-th year and the q-th year, and calculate the year interval; Correct the previous year to the next year year by year to get the corrected trading volume; If the difference between the corrected trading volume in the previous year and the trading volume in the subsequent year is determined to be less than a preset difference threshold, then the matching is completed.

[0024] Specifically, taking into account the development of e-commerce platforms, the transaction volume data of the crop mung beans on e-commerce platforms has gradually shifted from offline transaction volume to e-commerce platforms as the e-commerce platforms develop. Therefore, the transaction volume data needs to be corrected for matching.

[0025] For example, to match the transaction volume of mung beans on the e-commerce platform in 2013 with the transaction volume of mung beans on the e-commerce platform in 2018, it is necessary to correct the transaction volume of mung beans on the e-commerce platform in 2013. For example, to match the transaction volume data of the first year with the transaction volume data of the third year, specifically include: Obtain the total financial report data of the e-commerce platform in the first, second and third years; Determine the ratio of the total financial report data of the first year to the second year as the first correction factor, assuming it is 0.8; Determine the ratio of the total financial report data of the second year to the third year as the second correction factor, assuming it is 0.7; The transaction volume data for the first year is 10,000. is the first revised volume; Then, the first modified volume is passed through the second modification factor, is the second revised volume; Match the second revised trading volume with the volume data of the third year. If the difference is less than 1000, it is considered that the first year and the third year are matched.

[0026] In one embodiment of the present invention, a deviation analysis is performed on the candidate years according to the average purchase price of the clustered years, including: Let the first correction function be , the second correction function is , the third correction function is ; Determine The number of times the crop is repeated in the first, second and third candidate years in the sth candidate year is obtained. The combined deviation of the crop in the sth candidate year is ;in, , and Represent the number of repetitions of the first alternative year, the second alternative year, and the third alternative year respectively; Calculate the The average purchase price of the crop in the second screening year is The average difference in the average purchase price of the crop in the sth alternative year; Determine the average differences of multiple crops in multiple candidate years respectively, and use the multiple average differences as reference sample values ​​of comprehensive deviations of multiple crops in multiple candidate years; Obtaining multiple reference sample values ​​of the first correction function, the second correction function and the third correction function by linear least squares fitting; Establishing fitting polynomials for the first correction function, the second correction function and the third correction function respectively; The weight parameters and constant parameters in the first correction function, the second correction function and the third correction function are fitted according to a plurality of reference sample values ​​to obtain the fitted first correction function, the second correction function and the third correction function.

[0027] For example, the number of geological disasters in the crop planting area in the sth candidate year is , the number of weather disasters is The amount of planting subsidy is , then the total error is ; The average purchase price of all matching completed years of the crop's sth candidate year is taken as the average purchase price of the crop's sth candidate year ; Then there is, ; By solving the above sets in pairs, we can get , and A plurality of different solution values, each solution value is used as a reference sample value of each correction function; Set the correction function to Nonlinear format; Will Fit each corresponding reference sample value respectively to obtain the specific value of the correction function; For example, the reference values ​​are 3.2, 3.1, 3, 3.8, 5, etc., and the correction function is the first correction function corresponding to the geological disaster; but middle, Take the planting area involved in geological disasters, for , and , modify its parameters by gradient descent method to approximate all reference values, and get , and The final first correction function is obtained by taking the value of .

[0028] For example, the reference values ​​are 3.2, 3.1, 3, 3.8, 5, etc., and the correction function is the second correction function corresponding to the weather disaster; but middle, Take the planting area affected by weather disasters, , and , modify its parameters by gradient descent method to approximate all reference values, and get , and The final second correction function is obtained by taking the value of .

[0029] For example, the reference values ​​are 3.2, 3.1, 3, 3.8, 5, etc., and the correction function is the third correction function corresponding to the government subsidy; but middle, Take the planting area covered by government subsidies, for , and , modify its parameters by gradient descent method to approximate all reference values, and get , and The final third correction function is obtained by taking the value of .

[0030] In one embodiment of the present invention, the average purchase price of each crop over the years is corrected according to the correction function to obtain standard data, including: Based on the fitted first correction function, second correction function and third correction function, we get The comprehensive deviation value of the transaction volume data of the crop in the sth alternative year; The first The transaction volume data of the crop in the sth candidate year minus the comprehensive deviation value is obtained Standard data of the crop in the sth candidate year.

[0031] Specifically, after obtaining the correction function through a small amount of data, there is no need to match each crop subsequently, and the historical data of the crop can be directly corrected.

[0032] For example, the transaction volume of Sichuan pepper in the vth alternative year is 100. Through the first correction function, the second correction function and the third correction function, the comprehensive deviation value is 12, and the standard transaction volume of Sichuan pepper in the vth alternative year is 88.

[0033] In one embodiment of the present invention, the preset prediction model includes an input layer, a hidden layer and an output layer: Configure the input layer of the prediction model, and input the feature vector constructed by the standard data of previous years; Configure the hidden layer of the prediction model. The hidden layer performs multi-scale sampling on the feature vector, including: Perform global sampling on the feature vectors of previous years to obtain the global features of the standard data of previous years; Perform sliding sampling on the feature vectors of previous years to obtain the local features of the year within the sliding window; Configure the output layer of the prediction model. The output layer is used to weight and merge the global features and local features to obtain the hidden state, and the hidden state is input into the classifier. The classification space of the classifier represents the average purchase price of the crops in the current year. The prediction model updates the weight parameters and bias parameters of the hidden layer through the mean error variance.

[0034] Specifically, global sampling is based on the long-scale 1D-CNN Convolution sampling is performed on the standard data of various crops over the years to obtain global features; Specifically, sliding sampling extracts the first Convolution sampling is performed on the standard data of crop time series to obtain local features; Output the current year based on the global feature pair The predicted global average purchase price of the crops; Output the current year based on local features The predicted local average purchase price of the crops; By weighting the global average purchase price and the local average purchase price respectively, the final average purchase price of the i-th crop in the current year is obtained.

[0035] In one embodiment of the present invention, the planting plan for the current year is obtained based on the average selling price of each crop in the current year in combination with an optimization algorithm, including: Determine the crops to be planted by target farmers; Obtain the transaction volume of seedlings of the crops to be selected for planting, the average purchase price of the crops to be selected for planting, the growing months and planting areas of the crops to be selected for planting, and the ground disasters, weather disasters and planting subsidies in the planting areas of the crops to be selected for planting; The average purchase price of the crop to be selected for planting in previous years is corrected by a correction function to obtain the standard data of the crop to be selected for planting in previous years; Input the standard data of the crops to be selected for planting in previous years into the prediction model to obtain the average purchase price of the crops to be selected for planting in the current year; Randomly generate an initial population with R individuals that meet the constraints, and initialize the current number of iterations to 1; The individual code is represented as: ;in, Expressing the The planting area allocated to the crops to be planted, Y represents the number of types of crops to be planted; The constraints are as follows: ; represents the total planting area of ​​the target farmers; For any two crops L1 and L2 to be planted, if the growing months of L1 and L2 overlap, then: and No intersection; The area of ​​each crop to be planted ; The objective function is as follows: ;in, Indicates The return per unit area of ​​the crop to be planted; Obtain the objective function value of each individual through the objective function; Select the corresponding individual as the parent to update; When the current number of iterations reaches the preset number of iterations, the update stops and the planting plan of the target farmers in the current year is output.

[0036] Specifically, the introduction of optimization algorithms can make the agricultural information management platform more intelligent and precise in planting planning, helping farmers maximize their profits by maximizing the benefits per unit planting area, reducing environmental risks, and optimizing resource allocation.

[0037] It should be noted that the above plan is only a universal plan for when no geological disasters, weather disasters or government subsidies occur. Geological disasters, weather disasters or government subsidies are unpredictable factors and are therefore not within the scope of the forecast planning of this plan.

[0038] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. An agricultural information management platform based on big data, characterized in that: Including data acquisition module, data analysis module, data clustering module, data processing module, prediction module and planning module: The data collection module is used to collect the transaction volume data of each crop seedling on each e-commerce platform over the years and the average purchase price of each crop over the years; The data analysis module is used to obtain the growing months and planting areas of each crop, and determine the presence of ground disasters, weather disasters and planting subsidies in the planting areas of crops in the growing months of the crops in previous years, and obtain alternative years; A data clustering module is used to cluster the candidate years and perform deviation analysis on the candidate years according to the average purchase price of the clustered years to obtain a correction function; A data processing module is used to correct the average purchase price of each crop over the years according to the correction function to obtain standard data; The forecasting module is used to input the standard data into the preset forecasting model to obtain the average selling price of each crop in the current year; The planning module is used to obtain the planting plan for the current year based on the average selling price of each crop in the current year combined with the optimization algorithm.

2. According to the big data-based agricultural information management platform of claim 1, it is characterized in that: Determine the presence of ground disasters, weather disasters and planting subsidies in the growing months of crops in the planting areas of previous years, and obtain alternative years, including: Get the Multiple growing regions for the crop and the growing months in each growing region; Determine If there are ground disasters, weather disasters and planting subsidies in the growing months of the k-th planting area of ​​the crop in previous years, the years in which the ground disasters, weather disasters and planting subsidies occur will be repeatedly divided into the first alternative year, the second alternative year and the third alternative year.

3. The agricultural information management platform based on big data according to claim 2 is characterized in that: Clustering of candidate years, including: Determine the e-commerce platform transaction volume and average purchase price in the year before each candidate year; For the previous year, the transaction volume and average purchase price of the e-commerce platform are matched to the year in which the transaction volume and average purchase price of the e-commerce platform meet the preset rules to obtain the matching year; The year following the matching year that is not a candidate year is used as the first screening year; The transaction volume of the e-commerce platform in the first screening year and the transaction volume of the e-commerce platform in the alternative year are matched according to preset rules to obtain the second screening year, and the average acquisition price in the second screening year is recorded.

4. The agricultural information management platform based on big data according to claim 3 is characterized in that: Preset rules include: Compare the financial report data of e-commerce platforms over the years to obtain the transaction volume correction factors of crop seedlings over the years; among which, the financial report data represents the profit parameters of the e-commerce platform in the corresponding year; The pth year and the qth year are matched according to the transaction volume correction factor of crop seedlings in previous years, including: Determine the year that comes before the p-th year and the q-th year, and calculate the year interval; Correct the previous year to the next year year by year to get the corrected trading volume; If the difference between the corrected trading volume in the previous year and the trading volume in the subsequent year is determined to be less than a preset difference threshold, then the matching is completed.

5. The agricultural information management platform based on big data according to claim 3 is characterized in that: Deviation analysis of alternative years based on the average acquisition price of clustered years includes: Let the first correction function be , the second correction function is , the third correction function is ; Determine The number of times the crop is repeated in the first, second and third candidate years in the sth candidate year is obtained. The combined deviation of the crop in the sth candidate year is ;in, , and Represent the number of repetitions of the first alternative year, the second alternative year, and the third alternative year respectively; Calculate the The average purchase price of the crop in the second screening year is The average difference in the average purchase price of the crop in the sth alternative year; Determine the average differences of multiple crops in multiple candidate years respectively, and use the multiple average differences as reference sample values ​​of comprehensive deviations of multiple crops in multiple candidate years; Obtaining multiple reference sample values ​​of the first correction function, the second correction function and the third correction function by linear least squares fitting; Establishing fitting polynomials for the first correction function, the second correction function and the third correction function respectively; The weight parameters and constant parameters in the first correction function, the second correction function and the third correction function are fitted according to a plurality of reference sample values ​​to obtain the fitted first correction function, the second correction function and the third correction function.

6. The agricultural information management platform based on big data according to claim 5 is characterized in that: The average purchase price of each crop over the years is corrected according to the correction function to obtain standard data, including: Based on the fitted first correction function, second correction function and third correction function, we get The comprehensive deviation value of the transaction volume data of the crop in the sth alternative year; The first The transaction volume data of the crop in the sth candidate year minus the comprehensive deviation value is obtained Standard data of the crop in the sth candidate year.

7. The agricultural information management platform based on big data according to claim 6 is characterized in that: The preset prediction model includes input layer, hidden layer and output layer: Configure the input layer of the prediction model, and input the feature vector constructed by the standard data of previous years; Configure the hidden layer of the prediction model. The hidden layer performs multi-scale sampling on the feature vector, including: Perform global sampling on the feature vectors of previous years to obtain the global features of the standard data of previous years; Perform sliding sampling on the feature vectors of previous years to obtain the local features of the year within the sliding window; Configure the output layer of the prediction model. The output layer is used to weight and merge the global features and local features to obtain the hidden state, and the hidden state is input into the classifier. The classification space of the classifier represents the average purchase price of the crops in the current year. The prediction model updates the weight parameters and bias parameters of the hidden layer through the mean error variance.

8. The agricultural information management platform based on big data according to claim 7 is characterized in that: Based on the average selling price of each crop in the current year and combined with the optimization algorithm, the planting plan for the current year is obtained, including: Determine the crops to be planted by target farmers; Obtain the transaction volume of seedlings of the crops to be selected for planting, the average purchase price of the crops to be selected for planting, the growing months and planting areas of the crops to be selected for planting, and the ground disasters, weather disasters and planting subsidies in the planting areas of the crops to be selected for planting; The average purchase price of the crop to be selected for planting in previous years is corrected by a correction function to obtain the standard data of the crop to be selected for planting in previous years; Input the standard data of the crops to be selected for planting in previous years into the prediction model to obtain the average purchase price of the crops to be selected for planting in the current year; Randomly generate an initial population with R individuals that meet the constraints, and initialize the current number of iterations to 1; The individual code is represented as: ;in, Indicates The planting area allocated to the crops to be planted, Y represents the number of types of crops to be planted; The constraints are as follows: ; represents the total planting area of ​​the target farmers; For any two crops L1 and L2 to be planted, if the growing months of L1 and L2 overlap, then: and No intersection; The area of ​​each crop to be planted ; The objective function is as follows: ;in, Indicates The return per unit area of ​​the crop to be planted; Obtain the objective function value of each individual through the objective function; Select the corresponding individual as the parent to update; When the current number of iterations reaches the preset number of iterations, the update stops and the planting plan of the target farmers in the current year is output.

Images