Agricultural science park planning method and system based on big data

Abstract

The invention discloses an agricultural science and technology park planning method and system based on big data, and the method comprises the steps: obtaining multi-dimensional agricultural data and related data of a preset agricultural science and technology park, and constructing a parameterized agricultural science and technology park model through the multi-dimensional agricultural data and the related data; performing demand analysis according to the market data to obtain demanded crops, and performing land suitability analysis on the agricultural science park according to the demanded crops and the soil data to obtain candidate areas; importing the candidate regions and the demand crops into the parameterized agricultural science and technology park model, and performing dynamic planning adjustment on the parameterized agricultural science and technology park model based on the insect data and the environment data to obtain optimized region data; and constructing an agricultural science and technology park planning model according to the adjustment and optimization area data and the demand crops, optimizing the agricultural science and technology park planning model, inputting to-be-planned data into the agricultural science and technology park planning model, and outputting a planning result.

Description

Technical Field

[0001] The present invention relates to the field of planning, and particularly to a method and system for planning an agricultural science and technology park based on big data. Background Art

[0002] With the rapid development of big data technology, its applications in various fields have become increasingly widespread, and the agricultural field is no exception. Especially in the planning of agricultural science and technology parks, the application of big data technology provides strong support for the scientific planning, efficient management, and sustainable development of the parks.

[0003] Although the prior art has begun to attempt to use big data for the planning of agricultural science and technology parks, there are still some obvious defects: the data collected by traditional methods are often limited to one or several dimensions, lacking comprehensiveness and systematicness, resulting in possible one-sidedness in planning decisions. Due to diverse data sources, uneven data quality, and limitations in data processing and analysis technologies, the data accuracy in the prior art needs to be improved; the planning of agricultural science and technology parks needs to reflect the actual situation and development trends of the parks in real time and dynamically, but the data update speed in the prior art often lags behind the actual needs; due to the lack of high-precision data and advanced planning models, the planning results in the prior art often fail to meet the refined management requirements of the parks.

[0004] Therefore, a new method for planning an agricultural science and technology park based on big data is needed. By integrating multi-dimensional agricultural data and related data, a parametric agricultural science and technology park model is constructed, achieving comprehensive coverage and in-depth analysis of the park planning; advanced data processing and analysis technologies are adopted to clean, verify, and integrate the collected data, ensuring the accuracy and reliability of the data; the real-time analysis function in big data technology is used to achieve real-time monitoring and dynamic update of the park data, providing timely and accurate information support for planning decisions; based on the optimized regional data and required crops, an agricultural science and technology park planning model is constructed, and through continuous optimization and improvement, refined and intelligent management of the park planning is achieved. Summary of the Invention

[0005] The object of the present invention is to provide a method for planning an agricultural science and technology park based on big data.

[0006] To achieve the above object, the present invention is implemented according to the following technical solution:

[0007] The present invention includes the following steps:

[0008] Obtain multi-dimensional agricultural data and related data of a preset agricultural science and technology park, and construct a parametric agricultural science and technology park model through the multi-dimensional agricultural data and the related data; the related data includes environmental data, market data, insect data, and soil data;

[0009] Demand crops are obtained through demand analysis based on the market data, and candidate areas are obtained through land suitability analysis of the agricultural science and technology park based on the demand crops and the soil data;

[0010] The candidate areas and the demand crops are imported into the parametric agricultural science and technology park model, and the parametric agricultural science and technology park model is dynamically planned and adjusted based on the insect data and the environmental data to obtain optimized area data;

[0011] An agricultural science and technology park planning model is constructed based on the optimized area data and the demand crops, the agricultural science and technology park planning model is optimized, and the data to be planned is input into the agricultural science and technology park planning model to output a planning result.

[0012] Furthermore, the method for constructing the parametric agricultural science and technology park model includes constructing a parametric agricultural science and technology park model based on a multimodal fusion algorithm in combination with environmental data, market data, insect data, and soil data.

[0013] Furthermore, the method for obtaining demand crops through demand analysis based on the market data includes:

[0014] The market data is input into the market demand analysis model, and future market data is predicted based on historical market data according to the crop market principle;

[0015] Among them, the crop market principle: when the market demand is fixed, the crop demand is inversely proportional to the crop seed sales volume and the crop sales volume in the previous quarter in the market, and the crop demand is proportional to the regional disaster loss;

[0016] Calculate the relative difference of the market data:

[0017]

[0018] Among them, the relative difference of the v-th market data at the s-th moment is The actual value of the v-th market data at the s-th moment is k v (s), and the predicted value of the v-th market data at the s-th moment is The number of market data is N v ;

[0019] Calculate the prediction coefficient of variation:

[0020]

[0021] Among them, the prediction coefficient of variation of the v-th market data is θ v , and the upper limit of the observation time is T s ;

[0022] Calculate the combined predicted value:

[0023]

[0024] where the prediction weight of the v-th market data is The combined prediction value at the s-th moment is D s ;

[0025] Crop demand is predicted through the combined prediction value, and the expression is:

[0026]

[0027] where the predicted crop demand Q of the c-th crop at the s-th moment c,s , the ratio of the regional disaster loss area to the total planting area at the s-th moment is B s , the seed sales volume of the c-th crop at the s-th moment is M c,s , the sales volume of the c-th crop in the previous quarter is L c , the market demand of the c-th crop at the s-th moment is H s,c , the combined prediction value of the c-th crop at the s-th moment is D s,c ;

[0028] Sort the crops in descending order according to the predicted crop demand, compare the plant species that can be planted in the agricultural science and technology park with the sorted crops, and take the first 7 crops as the demanded crops.

[0029] Furthermore, the method for obtaining the candidate area includes:

[0030] Input the demanded crops and soil data into the crop-soil suitability analysis model, and use the historical soil division as the initial division area; the crop-soil suitability analysis model is constructed based on climate, terrain, and soil data;

[0031] Extract the soil characteristics from the soil data within the initial division area to obtain soil features, use the soil features as independent variables, and predict the crop growth values of the demanded crops within the initial division area based on the single variable principle;

[0032] Calculate the contribution degree of the soil features according to the regional growth values:

[0033]

[0034] where the a-th soil feature is h a , the soil feature subset of the b-th initial division area is V b , the contribution degree of the soil feature h a in the b-th initial division area is The number of soil features is N h , the original out-of-bag error of the c-th one is Uc For the c-th out-of-bag error after adding noise to the a-th one, it is U a (c), the number of added noises is n c , the soil feature subset V b The number of soil features of is |V b |, the soil feature h a The eigenvalue of is |h a |, after changing the soil feature h a the crop growth value is R a , without changing the soil feature h a the crop growth value is R o ;

[0035] Calculate the suitability of the initial partition area:

[0036]

[0037] Among them, the comprehensive index of the b-th initial partition area is f b , the weight coefficient of the a-th soil feature is λ a ;

[0038] Sort the initial partition areas in descending order according to the suitability of the candidate crops, and take the first three initial partition areas corresponding to the suitability as the candidate areas.

[0039] Furthermore, the method for obtaining the optimized area data includes:

[0040] Perform yield prediction based on soil data and crop data to obtain the predicted crop yield, and perform time series impact analysis of crop yield based on long short-term memory network on insect data and crop data respectively to obtain the pest impact coefficient and the environmental impact coefficient;

[0041] Perform grid division on the candidate areas to obtain multiple sub-grids, introduce an optimization algorithm, take the crop plans of the sub-grids as particles, for the aggregation area of the sub-grids around the same crop less than 0.5 mu, take the crop with the maximum fitness as the planted crop of the surrounding sub-grids, and the sub-grids are superimposed with a square shape;

[0042] Calculate the fitness of the particles according to the predicted crop yield and the actual crop yield of the candidate areas:

[0043]

[0044] s.t.S nb (z)≥0.5

[0045] Among them, the fitness of the w-th particle is The number of sub-grids included in the particle is N u , the pest impact coefficient at the s-th moment is ρ s, the environmental impact coefficient at the s-th moment is χ s , the actual yield of the z-th crop in the u-th sub-grid is HC u,b , the predicted demand yield of the z-th crop in the u-th sub-grid is HC u , the suitability of the z-th crop in the u-th sub-grid is f u,z , the predicted yield of the z-th crop in the u-th sub-grid is HC u,z , the superimposed area of the neighboring sub-grids of the z-th crop is S nb (z);

[0046] Taking the minimum fitness as the search target, calculate the position of the particle:

[0047]

[0048] Among them, the fitness of the random particle is The original position of the w-th particle is The position of the w-th particle at the t-th iteration is The step size control coefficient is ω, and the random position of the w-th particle is The average value of the particle positions is

[0049] Update the position of the particle according to the perturbation factor to obtain the perturbed position, and the expression is:

[0050]

[0051]

[0052] Among them, the perturbed position of the w-th particle at the (t + 1)-th iteration is The perturbation factor is η, the current iteration number is t, and the maximum iteration number is t max , the random numbers from 0 to 1 are r1 and r2 respectively, and the random position of the w-th particle at the t-th iteration is

[0053] Hunt the particle to update the particle position and search speed, and the expression is:

[0054]

[0055] Among them, the hunting position of the w-th particle at the (t + 1)-th iteration is The perturbed position of the w-th particle at the t-th iteration is , the random numbers from 0 to 1 are r3, r4, and r5 respectively, and the average value of the particle positions at the t-th iteration is The speed of the w-th particle at the (t + 1)-th iteration is V w (t + 1), and the evolution factor is

[0056]

[0057] Iterate continuously until the fitness value is less than 0.0131, otherwise continue to perturb the particles;

[0058] Output the sub-grid positions and planted crops in the output particles as the optimized area data.

[0059] Furthermore, a method for constructing an agricultural science and technology park planning model based on the optimized area data and the required crops includes:

[0060] Weightedly sum the crop profit and loss functions of the optimized area data as the objective function of the agricultural science and technology park planning model;

[0061] The agricultural science and technology park planning model includes an agricultural science and technology park model simulation algorithm, a long short-term memory network, and a deep learning algorithm;

[0062] The agricultural science and technology park model simulation algorithm performs simulation on the agricultural science and technology park model according to the input data to obtain an agricultural science and technology park simulation model;

[0063] The long short-term memory network learns the long sequence dependence relationship of the input data through memory units and gating mechanisms to obtain time series features;

[0064] The deep learning algorithm predicts the planning and development trend of the agricultural science and technology park by extracting time series features and optimizing the objective function, and uses a stacked structure and non-linear transformation ability to obtain the agricultural science and technology park planning result.

[0065] In the second aspect, an agricultural science and technology park planning system based on big data includes:

[0066] Data acquisition and model construction module: used to obtain multi-dimensional agricultural data and related data of a preset agricultural science and technology park, and construct a parameterized agricultural science and technology park model through the multi-dimensional agricultural data and the related data; the related data includes environmental data, market data, insect data, and soil data;

[0067] Demand adaptability module: used to perform demand analysis based on the market data to obtain required crops, and perform land suitability analysis on the agricultural science and technology park according to the required crops and the soil data to obtain candidate areas;

[0068] Dynamic programming adjustment module: used to import the candidate areas and the required crops into the parameterized agricultural science and technology park model, and perform dynamic programming adjustment on the parameterized agricultural science and technology park model based on the insect data and the environmental data to obtain optimized area data;

[0069] Building output module: It is used to construct an agricultural science and technology park planning model according to the optimized area data and the required crops, input the data to be planned into the agricultural science and technology park planning model, and output the planning result. The beneficial effects of the present invention are:

[0070] The present invention relates to a method and system for planning an agricultural science and technology park based on big data. Compared with the prior art, the present invention has the following technical effects:

[0071] By constructing steps such as a parametric agricultural science and technology park model, demand analysis, land suitability analysis, dynamic planning adjustment, and model construction, the present invention can improve the accuracy of the agricultural science and technology park planning, thereby improving the precision of the agricultural science and technology park planning. Optimizing the agricultural science and technology park planning can greatly save resources, improve work efficiency, realize the intelligent planning of the agricultural science and technology park, dynamically plan and adjust the agricultural science and technology park planning in real time, which is of great significance for the agricultural science and technology park planning, and can adapt to different standards of agricultural science and technology park planning and different agricultural science and technology park planning requirements, having a certain universality. Description of the Drawings

[0072] Figure 1 It is a flowchart of the steps of the method for planning an agricultural science and technology park based on big data of the present invention. Detailed Embodiments

[0073] The present invention will be further described below through specific embodiments. The illustrative embodiments and explanations of the present invention are used to explain the present invention, but do not limit the present invention.

[0074] The method and system for planning an agricultural science and technology park based on big data of the present invention include the following steps:

[0075] As Figure 1 shown, in this embodiment, it includes the following steps:

[0076] Obtain the multi-dimensional agricultural data and related data of the preset agricultural science and technology park, and construct a parametric agricultural science and technology park model through the multi-dimensional agricultural data and the related data; the related data includes environmental data, market data, insect data, and soil data;

[0077] Multi-dimensional agricultural data includes the types of crops, growth cycles, yields, sown areas, precipitation, temperature, humidity, wind speed, sunshine hours, physical and chemical properties, such as soil pH value, soil organic matter content, soil texture, climate, soil; environmental data includes precipitation, temperature, humidity, wind speed, sunshine hours, availability of surface water and groundwater, water quality, sulfur dioxide concentration, nitrogen oxide concentration, biodiversity, vegetation coverage, ecosystem health status; market data includes prices, fluctuations, demand for agricultural products within regions and nationwide, sales volume, sales amount, sales channels, relevant policies, sales data of relevant crop seeds, sales data of relevant pesticides; insect data includes types, densities, occurrence periods, control effects, population numbers, activity ranges; soil data includes clay, loam, sand, organic matter content, nitrogen, phosphorus, potassium, particle size, porosity, permeability, pH value;

[0078] In actual evaluation, a certain citrus agricultural science and technology park is taken as the research object, with a soil pH value of 6.2, an organic matter content of 2.1%, a daily sunshine duration of 8.2 h, and a pest density of 0.3 per square meter. The market data is that the wholesale price of citrus fluctuates between 4.8 - 6.2 yuan / kg, the search volume on the e-commerce platform is 12,000 times per month on average, the historical yield per mu is 2,500 kg, and the fertilization amount, with a nitrogen-phosphorus-potassium ratio of 3:1:2;

[0079] Demand crops are obtained through demand analysis based on the market data, and candidate areas are obtained through land suitability analysis of the agricultural science and technology park according to the demand crops and the soil data;

[0080] In actual evaluation, the demand crop models are red tangerines, honey tangerines, emperor mandarins, ponkan oranges, ponkan tangerines, navel oranges, and blood oranges;

[0081] Input: Soil moisture content, the proportion of red soil is 82%, and the accumulated temperature ≥ 10℃ is 5,200℃; Candidate areas: Grid code is A03, suitability score is 9.61, Grid B01, suitability score is 9.17, Grid B13, suitability score is 8.94;

[0082] The candidate areas and the demand crops are imported into the parametric agricultural science and technology park model, and the parametric agricultural science and technology park model is dynamically planned and adjusted based on the insect data and the environmental data to obtain optimized area data;

[0083] In actual evaluation, the optimized area data is navel oranges, planting density is 70 plants per mu, expected sugar content is 14%, risk coefficient is 0.05, grid code is A03, Grid B01;

[0084] An agricultural science and technology park planning model is constructed according to the optimized area data and the demand crops, the agricultural science and technology park planning model is optimized, and the data to be planned is input into the agricultural science and technology park planning model to output the planning results.

[0085] In this embodiment, the method for constructing the parametric agricultural science and technology park model includes constructing a parametric agricultural science and technology park model based on a multimodal fusion algorithm in combination with environmental data, market data, insect data, and soil data.

[0086] In this embodiment, the method for obtaining the required crops through demand analysis based on the market data includes:

[0087] Input the market data into the market demand analysis model, and predict the future market data based on the historical market data according to the crop market principle;

[0088] Among them, the crop market principle: when the market demand is fixed, the crop demand is inversely proportional to the sales volume of crop seeds in the market and the sales volume of crops in the previous quarter, and the crop demand is directly proportional to the regional disaster losses;

[0089] Calculate the relative difference of the market data:

[0090]

[0091] Among them, the relative difference of the v-th market data at the s-th moment is The actual value of the v-th market data at the s-th moment is k v (s), and the predicted value of the v-th market data at the s-th moment is The number of market data is N v ;

[0092] Calculate the prediction coefficient of variation:

[0093]

[0094] Among them, the prediction coefficient of variation of the v-th market data is θ v , and the upper limit of the observation time is T s ;

[0095] Calculate the combined predicted value:

[0096]

[0097] Among them, the prediction weight of the v-th market data is The combined predicted value at the s-th moment is D s ;

[0098] Predict the crop demand through the combined predicted value, and the expression is:

[0099]

[0100] Among them, the predicted crop demand Q of the c-th crop at the s-th moment c,s, the ratio of the regional disaster loss area to the total planting area at the s-th moment is B s , the sales volume of the c-th crop seeds at the s-th moment is M c,s , the sales volume of the c-th crop in the previous quarter is L c , the market demand for the c-th crop at the s-th moment is H s,c , the combined prediction value of the c-th crop at the s-th moment is D s,c ;

[0101] Sort the crops in descending order according to the predicted crop demand, compare the plantable species in the agricultural science and technology park with the sorted crops, and take the first 7 crops as the demand crops.

[0102] In this embodiment, the method for obtaining the candidate area includes:

[0103] Input the demand crops and soil data into the crop and soil suitability analysis model, and use the historical soil division as the initial division area; among them, the crop and soil suitability analysis model is constructed based on climate, terrain, and soil data;

[0104] Extract the soil characteristics from the soil data in the initial division area to obtain soil features, use the soil features as independent variables, and predict the crop growth values of the demand crops in the initial division area based on the single variable principle to obtain crop growth values;

[0105] Calculate the contribution degree of the soil features according to the regional growth values:

[0106]

[0107] Where the a-th soil feature is h a , the soil feature subset of the b-th initial division area is V b , the contribution degree of the soil feature h a in the b-th initial division area is The number of soil features is N h , the original out-of-bag error of the c-th is U c , the out-of-bag error of the c-th after adding noise to the a-th is U a (c), the number of added noises is n c , the soil feature subset V b has the number of soil features |V b |, the eigenvalue of the soil feature h a is |h a |, the crop growth value after changing the soil feature h a is R a , the crop growth value without changing the soil feature h a is R o ;

[0108] Calculate the suitability of the initial division area:

[0109]

[0110] Among them, the comprehensive index of the b-th initial division area is f b , and the weight coefficient of the a-th soil characteristic is λ a ;

[0111] Sort the initial division areas in descending order according to the suitability of the candidate crops, and take the initial division areas corresponding to the top three suitability as candidate areas.

[0112] In this embodiment, the method for obtaining the optimized area data includes:

[0113] Perform yield prediction based on soil data and crop data to obtain the predicted crop yield, and respectively perform time series impact analysis of crop yield based on long short-term memory network on insect data and crop data to obtain the pest impact coefficient and the environmental impact coefficient;

[0114] Perform grid division on the candidate areas to obtain multiple sub-grids, introduce an optimization algorithm, take the crop plan of the sub-grids as particles, for the aggregation area of the sub-grids around the same crop less than 0.5 mu, take the crop with the maximum fitness as the planted crop of the surrounding sub-grids, and the superimposed sub-grid shape is square;

[0115] Calculate the fitness of the particles according to the predicted crop yield and the actual crop yield of the candidate areas:

[0116]

[0117] s.t.S nb (z)≥0.5

[0118] Among them, the fitness of the w-th particle is The number of sub-grids included in the particle is N u , the pest impact coefficient at the s-th moment is ρ s , the environmental impact coefficient at the s-th moment is χ s , the actual yield of the z-th crop in the u-th sub-grid is HC u,b , the predicted demand yield of the z-th crop in the u-th sub-grid is HC u , the suitability of the z-th crop in the u-th sub-grid is f u,z , the predicted yield of the z-th crop in the u-th sub-grid is HC u,z , the superimposed area of the adjacent sub-grids of the z-th crop is S nb (z);

[0119] Take the minimum fitness as the search target and calculate the position of the particle:

[0120]

[0121] where the fitness of the random particles is The original position of the w-th particle is The position of the w-th particle at the t-th iteration is The step size control coefficient is ω, and the random position of the w-th particle is The average value of the particle positions is

[0122] Update the position of the particle according to the perturbation factor to obtain the perturbed position, and the expression is:

[0123]

[0124] where the perturbed position of the w-th particle at the (t + 1)-th iteration is The perturbation factor is η, the current iteration number is t, and the maximum iteration number is t max , the random numbers from 0 to 1 are r1 and r2 respectively, and the random position of the w-th particle at the t-th iteration is

[0125] Hunt the particles to update the particle positions and search speeds, and the expression is:

[0126]

[0127] where the hunting position of the w-th particle at the (t + 1)-th iteration is The perturbed position of the w-th particle at the t-th iteration is The random numbers from 0 to 1 are r3, r4, and r5 respectively, and the average value of the particle positions at the t-th iteration is The speed of the w-th particle at the (t + 1)-th iteration is V w (t + 1), and the evolution factor is

[0128]

[0129] Iterate continuously until the fitness value is less than 0.0131, otherwise continue to perturb the particles;

[0130] Output the sub-grid positions and planted crops in the output particles as the optimized area data.

[0131] In this embodiment, the method for constructing an agricultural science and technology park planning model according to the optimized area data and the required crops includes:

[0132] Weighted sum the crop profit and loss functions of the optimized area data as the objective function of the agricultural science and technology park planning model;

[0133] The agricultural science and technology park planning model includes the agricultural science and technology park model simulation algorithm, long short-term memory network, and deep learning algorithm;

[0134] The agricultural science and technology park model simulation algorithm performs the simulation of the agricultural science and technology park model based on the input data to obtain the agricultural science and technology park simulation model;

[0135] The long short-term memory network learns the long sequence dependence relationship of the input data through memory units and gating mechanisms to obtain time series features;

[0136] The deep learning algorithm predicts the planning and development trend of the agricultural science and technology park by extracting time series features and optimizing the objective function, and utilizes the stacked structure and non-linear transformation ability to obtain the agricultural science and technology park planning result.

[0137] In the second aspect, an agricultural science and technology park planning system based on big data includes:

[0138] The acquisition and construction model module: used to obtain the multi-dimensional agricultural data and related data of a preset agricultural science and technology park, and construct a parametric agricultural science and technology park model through the multi-dimensional agricultural data and the related data; the related data includes environmental data, market data, insect data, and soil data;

[0139] The demand adaptability module: used to perform demand analysis based on the market data to obtain the demand crops, and perform land suitability analysis on the agricultural science and technology park according to the demand crops and the soil data to obtain the candidate areas;

[0140] The dynamic planning adjustment module: used to import the candidate areas and the demand crops into the parametric agricultural science and technology park model, and perform dynamic planning adjustment on the parametric agricultural science and technology park model based on the insect data and the environmental data to obtain the optimized area data;

[0141] The construction and output module: used to construct an agricultural science and technology park planning model according to the optimized area data and the demand crops, input the data to be planned into the agricultural science and technology park planning model, and output the planning result. The above is only the preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The agricultural science and technology park planning method based on big data is characterized by: The following steps are involved: Acquire multidimensional agricultural data and related data of a preset agricultural science and technology park, and construct a parameterized agricultural science and technology park model through the multidimensional agricultural data and the related data; the related data include environmental data, market data, insect data and soil data; Performing demand analysis based on the market data to obtain demand crops, and performing land suitability analysis on the agricultural science and technology park based on the demand crops and the soil data to obtain candidate areas; The candidate area and the required crop are imported into the parameterized agricultural science and technology park model, and the parameterized agricultural science and technology park model is dynamically adjusted based on the insect data and the environmental data to obtain the optimized area data; An agricultural science and technology park planning model is constructed according to the tuning area data and the demand crops, the agricultural science and technology park planning model is optimized, the data to be planned is input into the agricultural science and technology park planning model, and the planning result is output.

2. The agricultural science and technology park planning method based on big data according to claim 1 is characterized in that: The method for constructing the parameterized agricultural science and technology park model includes constructing the parameterized agricultural science and technology park model based on a multimodal fusion algorithm in combination with environmental data, market data, insect data, and soil data.

3. The agricultural science and technology park planning method based on big data according to claim 1 is characterized in that: The method for obtaining the demand crops by performing demand analysis according to the market data comprises: Input market data into the market demand analysis model and predict future market data based on historical market data based on crop market principles; The crop market principle: when the market demand is fixed, the crop demand is inversely proportional to the crop seed sales volume and the crop sales volume in the previous quarter according to the market, and the crop demand is directly proportional to the regional disaster losses; Calculate the relative difference in market data: The relative difference of the vth market data at the sth time is The actual value of the vth market data at the sth time is k v (s), the predicted value of the vth market data at the sth time is The number of market data is N v ; Calculate the predicted coefficient of variation: The predicted coefficient of variation of the vth market data is θ v , the upper limit of the observation time is T s ; Calculate the combined forecast value: The prediction weight of the vth market data is The combined prediction value at time s is D s ; The crop demand is predicted by combining the predicted values, and the expression is: The predicted crop demand for the cth crop at time s is Q c,s , the ratio of the regional disaster loss area to the total planting area at the sth moment is B s , the sales volume of the cth crop seed at the sth time is M c,s , the sales volume of the cth crop in the previous quarter was L c , the market demand for the cth crop at the sth time is H s,c , the combined prediction value of the cth crop at the sth time is D s,c ; Crops were sorted in descending order according to the predicted crop demand, the cultivable species in the agricultural science and technology park were compared with the sorted crops, and the top seven crops were selected as demand crops.

4. The agricultural science and technology park planning method based on big data according to claim 1 is characterized in that: The method for obtaining the candidate region includes: The crop and soil data are required to be input into the crop and soil suitability analysis model, and the historical soil division is used as the initial division area; the crop and soil suitability analysis model is constructed based on climate, topography and soil data; Extract soil data in the initial divided area to obtain soil characteristics, use soil characteristics as independent variables, and predict the growth of required crops in the initial divided area based on the single variable principle to obtain crop growth values; Calculate the contribution of soil characteristics based on regional growth values: The ath soil characteristic is h a , the soil characteristic subset of the bth initialization sub-region is V b , the soil characteristics h in the bth initialization sub-region a The contribution of a ), the number of soil characteristics is N h , the original c-th out-of-bag error is U c , the out-of-bag error of the cth sample after adding noise is U a (c), the number of noise added is n c , soil characteristic subset V b The number of soil characteristics is |V b |, soil characteristics h a The characteristic value of h a |, change soil characteristics h a The crop growth value after is R a , without changing soil characteristics a The crop growth value after is R o ; Calculate the suitability of the initial partitioned area: The comprehensive index of the bth initial division area is f b , the weight coefficient of the ath soil characteristic is λ a ; The candidate crops are sorted in descending order in the initial divided areas according to their suitability, and the initial divided areas corresponding to the first three suitabilities are taken as candidate areas.

5. The agricultural science and technology park planning method based on big data according to claim 1 is characterized in that: Methods for obtaining tuning area data include: Yield prediction is performed based on soil data and crop data to obtain predicted crop yields. Time series impact analysis of crop yields based on long short-term memory networks is performed on insect data and crop data to obtain pest impact coefficients and environmental impact coefficients. The candidate area is gridded to obtain multiple sub-grids, and an optimization algorithm is introduced to use the crop scheme of the sub-grid as a particle. For the sub-grid aggregation area around the same crop that is less than 0.5 mu, the crop with the greatest fitness is used as the planting crop of the surrounding sub-grids, and the superimposed sub-grids are square in shape; The fitness of the particle is calculated based on the predicted crop yield and actual crop yield of the candidate area: s.t.S nb (z)≥0.5 The fitness of the wth particle is The number of particles including subgrids is N u , the pest impact coefficient at the sth moment is ρ s , the environmental impact coefficient at the sth moment is χ s , the actual yield of the zth crop in the uth subgrid is HC u,b , the predicted required yield of the zth crop in the uth subgrid is HC u , the suitability of the zth crop in the uth subgrid is f u,z , the predicted yield of the zth crop in the uth subgrid is HC u,z , the overlapping area of ​​the zth crop adjacent subgrid is S nb (z); Taking the minimum fitness as the search target, calculate the position of the particle: The fitness of random particles is The original position of the wth particle is The position of the wth particle at the tth iteration is The step size control coefficient is ω, and the random position of the wth particle is The average particle position is Update the particle position according to the disturbance factor to obtain the disturbance position. The expression is: The perturbed position of the wth particle in the t+1th iteration is The perturbation factor is η, the current number of iterations is t, and the maximum number of iterations is t max , the random numbers from 0 to 1 are r1 and r2 respectively, and the random position of the wth particle in the tth iteration is Hunting particles to update particle positions and search speeds, the expression is: The hunting position of the wth particle in the t+1th iteration is The perturbed position of the wth particle at the tth iteration is The random numbers from 0 to 1 are r3, r4, and r5 respectively. The average value of the particle position at the tth iteration is The velocity of the wth particle at the t+1th iteration is V w (t+1), the evolution factor is Continue to iterate until the fitness value is less than 0.0131, otherwise continue to perturb the particles; Output the subgrid positions and planted crops in the output particles as tuning area data.

6. The agricultural science and technology park planning method based on big data according to claim 1 is characterized in that: The method for constructing an agricultural science and technology park planning model according to the optimized regional data and the demanded crops comprises: The weighted sum of crop profit and loss functions of the optimized regional data is used as the objective function of the agricultural science and technology park planning model; The agricultural science and technology park planning model includes the agricultural science and technology park model simulation algorithm, long short-term memory network, and deep learning algorithm; The agricultural science and technology park model simulation algorithm simulates the agricultural science and technology park model according to the input data to obtain the agricultural science and technology park simulation model; The long short-term memory network learns the long-term dependencies of input data through memory units and gating mechanisms to obtain time series features; The deep learning algorithm extracts time series features and optimizes the objective function, uses the cascading structure and nonlinear transformation capabilities to predict the planning and development trends of agricultural science and technology parks and obtain the planning results of agricultural science and technology parks.

7. An agricultural science and technology park planning system based on big data, used to execute the method described in any one of claims 1 to 6, characterized in that: include: Collection and construction model module: used to obtain multi-dimensional agricultural data and related data of a preset agricultural science and technology park, and to construct a parameterized agricultural science and technology park model through the multi-dimensional agricultural data and the related data; the related data includes environmental data, market data, insect data and soil data; Demand adaptability module: used to perform demand analysis based on the market data to obtain demand crops, and perform land suitability analysis on the agricultural science and technology park based on the demand crops and the soil data to obtain candidate areas; Dynamic programming adjustment module: used for importing the candidate area and the required crop into the parameterized agricultural science and technology park model, and performing dynamic programming adjustment on the parameterized agricultural science and technology park model based on the insect data and the environmental data to obtain the optimized area data; Constructing an output module: It is used to construct an agricultural science and technology park planning model according to the optimization area data and the demand crops, input the data to be planned into the agricultural science and technology park planning model, and output the planning results.

Images