Intelligent decision support system based on big data

Abstract

The invention discloses an intelligent decision support system based on big data, which relates to the technical field of decision support systems, and is characterized in that limiting elements are generated based on historical planting scenes, a plurality of parameter spaces are generated by using a random generation tool according to a preset space number, and a loss degree value of each parameter space is calculated through an information function; the loss degree values of all the parameter spaces are substituted into a dynamic screening algorithm for calculation, after part of the parameter spaces are screened out, content randomization operation is carried out on the reserved parameter spaces, space sets are established for all the parameter spaces completing the randomization operation, multiple sets of space sets are output, loss degree value comparison is carried out on the multiple sets of space sets, and the loss degree values of all the parameter spaces are obtained. And selecting the parameter space with the minimum loss degree value as a management scheme of the current planting scene of the enterprise. According to the support system, through dynamic scheme generation and screening, the optimal management scheme can be globally selected to adapt to the current planting scene of an enterprise, the adaptability is high, and the decision making efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of decision support systems, and in particular to an intelligent decision support system based on big data. Background Art

[0002] Agricultural production is affected by a variety of factors, including climate conditions (temperature, precipitation, light), soil quality, crop varieties and management measures. The assessment system is a technical system used to analyze crops. Its main purpose is to scientifically evaluate the production status, environmental conditions and future yields of crops, thereby providing a basis for agricultural management and decision-making.

[0003] The prior art has the following deficiencies:

[0004] When implementing current agricultural planting management, existing companies usually set up management plans based on manual experience or expert knowledge. However, this kind of manual setting method, firstly, fails to take into account the environment and actual planting conditions of the planting area in recent years, which may easily lead to the management plan being unsuitable for the current planting scenario; secondly, the manual setting is highly subjective, which may result in the management plan not being the optimal solution and reduce decision-making efficiency.

[0005] Based on this, the present invention proposes an intelligent decision support system based on big data. Through dynamic solution generation and screening, it can select the optimal management solution from the global perspective to adapt to the current planting scenario of the enterprise. It has strong adaptability and is conducive to improving decision-making efficiency. Summary of the invention

[0006] The purpose of the present invention is to provide an intelligent decision support system based on big data to solve the shortcomings of the background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: an intelligent decision support system based on big data, the support system comprising the following steps:

[0008] S1: The collection end connects with the enterprise agricultural management platform through the API interface, obtains the enterprise's current planting scene through the enterprise agricultural management platform, and obtains the historical planting scene matching the current planting scene from the big database. Then, the restriction elements are generated based on the historical planting scene, and a number of parameter spaces are generated using the random generation tool according to the preset number of spaces;

[0009] S2: Calculate the loss value of each parameter space through the information function, substitute the loss values ​​of all parameter spaces into the dynamic screening algorithm for calculation, and after screening out some parameter spaces, perform content randomization operation on the remaining parameter spaces, and establish a space set for all parameter spaces that have completed the randomization operation;

[0010] S3: Repeat step S2 for the established spatial set to iterate. After the convergence condition is met, output multiple sets of spatial sets, compare the loss values ​​of the multiple sets of spatial sets, and select the parameter space with the smallest loss value as the management plan for the company's current planting scenario.

[0011] In a preferred embodiment, a random generation tool is used to generate a number of parameter spaces according to a preset number of spaces, and the contents of the parameter spaces include irrigation amount, fertilization amount and planting density.

[0012] In a preferred embodiment, calculating the loss value of each parameter space by an information function comprises the following steps:

[0013] The loss value of each parameter space is calculated by the information function, and the information function expression is:

[0014] In the formula, loss z is the loss value, θ is the prediction error, δ is the predicted yield mean, τ is the land utilization rate, β, α, and γ are the proportional coefficients of the prediction error, the predicted yield mean, and the land utilization rate, respectively, and β, α, and γ are all greater than 0. The larger the loss value, the worse the performance of the parameter space applied to the current planting scenario.

[0015] In a preferred embodiment, the loss degree values ​​of all parameter spaces are substituted into the dynamic screening algorithm calculation, comprising the following steps:

[0016] The fitness value of the parameter space is calculated based on the loss value, and the expression is: In the formula, Atd is the fitness value, loss z is the loss value;

[0017] The fitness values ​​of all parameter spaces are summed up to obtain the total fitness value, the selection probability of each parameter space is obtained by dividing the fitness value by the total fitness value, and the selection probability of each parameter space is mapped to the sector area of ​​the virtual roulette wheel;

[0018] Start the virtual wheel to rotate. When the virtual wheel stops, point the virtual pointer to the parameter space corresponding to the sector area and select it;

[0019] The virtual wheel is repeatedly started to rotate. When the number of selected parameter spaces is equal to a preset number threshold, the remaining parameter spaces on the virtual wheel are screened out and the selected parameter spaces are retained.

[0020] In a preferred embodiment, after filtering out part of the parameter space, performing a content randomization operation on the remaining parameter space, and establishing a space set for all parameter spaces that have completed the randomization operation, including the following steps:

[0021] After screening out some parameter spaces, the retained parameter spaces are subjected to content randomization operations. The randomization operations include content exchange operations and mutation operations. The content exchange operations include exchanging part of the content between two parameter spaces. When all parameter spaces have completed the content exchange operations, all parameter spaces that have completed the content exchange operations are subjected to content random mutation operations. A space set is established for all parameter spaces that have completed the mutation operations.

[0022] In a preferred embodiment, the established space set is iterated by repeating step S2, and after the convergence condition is met, multiple groups of space sets are output, including the following steps:

[0023] Repeat step S2 for the established space set to iterate. When there is a loss value less than or equal to the loss threshold or the number of iterations is equal to the number threshold in any space set, it is judged that the convergence condition is met and multiple groups of space sets {kj1, kj2, kj3, ..., kj m}, where m represents the number of spatial sets, kj i Represents the i-th group of spatial sets.

[0024] In a preferred embodiment, multiple sets of space sets are compared for loss values, and the parameter space with the smallest loss value is selected as the management solution for the current planting scenario of the enterprise, including the following steps:

[0025] Obtain the content information of all parameter spaces in each group of spatial sets and the loss value information corresponding to the parameter space. In the spatial set, sort all parameter spaces from small to large according to the loss value, compare the loss values ​​of the parameter spaces ranked first in all spatial sets, and select the parameter space with the smallest loss value as the management plan for the company's current planting scenario.

[0026] In a preferred embodiment, the calculation expression of the prediction error is: In the formula, θ is the prediction error, n is the number of predictions of the yield prediction model, is the output value predicted by the output prediction model for the i-th time, and y is the corresponding historical actual output value;

[0027] The calculation expression of the predicted output mean is: In the formula, δ is the predicted yield mean, n is the number of predictions of the yield prediction model, is the output value predicted by the output prediction model for the i-th time;

[0028] The calculation logic of the land utilization rate is: obtain the total planting area in the parameter space, obtain the total planting area of ​​the enterprise, and divide the total planting area by the total planting area to obtain the land utilization rate.

[0029] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0030] The present invention generates restriction elements based on historical planting scenarios, generates several parameter spaces using random generation tools according to a preset number of spaces, calculates the loss value of each parameter space through an information function, substitutes the loss values ​​of all parameter spaces into the dynamic screening algorithm for calculation, and after screening out some parameter spaces, performs content randomization operations on the retained parameter spaces, and establishes a space set for all parameter spaces that have completed the randomization operation, outputs multiple sets of space sets, compares the loss values ​​of the multiple sets of space sets, and selects the parameter space with the smallest loss value as the management plan for the current planting scenario of the enterprise. The support system can select the optimal management plan from the global perspective to adapt to the current planting scenario of the enterprise through dynamic plan generation and screening, has strong adaptability, and is conducive to improving decision-making efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0032] Figure 1 It is a system module diagram of the present invention.

[0033] Figure 2 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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.

[0035] Example 1: Please refer to Figure 1 As shown, the intelligent decision support system based on big data described in this embodiment includes a collection module, a space set establishment module, and a solution output module;

[0036] Collection module: connects with the enterprise agricultural management platform through the API interface, obtains the enterprise's current planting scene through the enterprise agricultural management platform, and obtains the historical planting scene matching the current planting scene from the big database, generates restriction elements based on the historical planting scene, and uses the random generation tool to generate several parameter spaces according to the preset number of spaces, and sends the parameter spaces to the space collection establishment module;

[0037] Space set establishment module: Calculate the loss value of each parameter space through the information function, substitute the loss value of all parameter spaces into the dynamic screening algorithm calculation, and after screening out some parameter spaces, perform content randomization operation on the retained parameter spaces, and establish a space set for all parameter spaces that have completed the randomization operation, and send the space set to the solution output module;

[0038] Solution output module: repeatedly iterate the established spatial set, and after meeting the convergence conditions, output multiple sets of spatial sets, compare the loss values ​​of the multiple sets of spatial sets, and select the parameter space with the smallest loss value as the management solution for the company's current planting scenario.

[0039] like Figure 2 As shown: The workflow of the support system is:

[0040] S1: The collection end connects with the enterprise agricultural management platform through the API interface, obtains the enterprise's current planting scene through the enterprise agricultural management platform, and obtains the historical planting scene matching the current planting scene from the big database. Then, the restriction elements are generated based on the historical planting scene, and a number of parameter spaces are generated using the random generation tool according to the preset number of spaces;

[0041] S2: Calculate the loss value of each parameter space through the information function, substitute the loss values ​​of all parameter spaces into the dynamic screening algorithm for calculation, and after screening out some parameter spaces, perform content randomization operation on the remaining parameter spaces, and establish a space set for all parameter spaces that have completed the randomization operation;

[0042] S3: Repeat step S2 for the established spatial set to iterate. After the convergence condition is met, output multiple sets of spatial sets, compare the loss values ​​of the multiple sets of spatial sets, and select the parameter space with the smallest loss value as the management plan for the company's current planting scenario.

[0043] This application generates restriction elements based on historical planting scenarios, uses random generation tools to generate several parameter spaces according to the preset number of spaces, calculates the loss value of each parameter space through information functions, substitutes the loss values ​​of all parameter spaces into the dynamic screening algorithm for calculation, and after screening out some parameter spaces, performs content randomization operations on the retained parameter spaces, and establishes space sets for all parameter spaces that have completed the randomization operation, outputs multiple groups of space sets, compares the loss values ​​of the multiple groups of space sets, and selects the parameter space with the smallest loss value as the management plan for the company's current planting scenario. This support system generates and screens dynamic plans, so that it can select the optimal management plan from the global perspective to adapt to the company's current planting scenarios. It has strong adaptability and is conducive to improving decision-making efficiency.

[0044] The evaluation system includes a parameter space initialization module, a space set establishment module, and a management plan output module;

[0045] Parameter space initialization module: connects with the enterprise agricultural management platform through the API interface, obtains the enterprise's current planting scene through the enterprise agricultural management platform, and obtains the historical planting scene matching the current planting scene from the big database, generates restriction elements based on the historical planting scene, and uses the random generation tool to generate several parameter spaces according to the preset number of spaces, and sends the parameter space to the space collection establishment module;

[0046] Space set establishment module: Calculate the loss value of each parameter space through the information function, substitute the loss value of all parameter spaces into the dynamic screening algorithm calculation, and after screening out some parameter spaces, perform content randomization operation on the retained parameter spaces, and establish a space set for all parameter spaces that have completed the randomization operation, and send the space set to the management plan output module;

[0047] Management plan output module: Repeat step S2 for the established space set to iterate. After the convergence condition is met, multiple groups of space sets are output, and the loss values ​​of the multiple groups of space sets are compared. The parameter space with the smallest loss value is selected as the management plan for the company's current planting scenario.

[0048] Embodiment 2: The collection end is connected to the enterprise agricultural management platform through the API interface, and the enterprise's current planting scene is obtained through the enterprise agricultural management platform. After obtaining the historical planting scene matching the current planting scene from the big database, the restriction elements are generated based on the historical planting scene, and a random generation tool is used to generate several parameter spaces according to the preset number of spaces. The parameter space content includes irrigation amount, fertilization amount and planting density, including the following steps:

[0049] Step 1: Get the current planting scene and historical data

[0050] Connect to the enterprise agricultural management platform through the API interface: The collection end connects to the enterprise agricultural management platform through the API interface to obtain real-time information on the company's current planting scenarios, including basic data such as crop type, growth stage, and regional climate characteristics.

[0051] Obtain historical planting scenarios: Through the large database of the enterprise agricultural management platform, filter historical planting scenarios that match the characteristics of the current planting scenarios, such as:

[0052] Similar climatic conditions (such as rainfall, temperature, etc.).

[0053] Similar soil types (e.g. pH, organic matter content, etc.).

[0054] Same or similar crop types (such as wheat, rice, etc.).

[0055] Step 2: Generate restriction features (constraints)

[0056] Extract constraints based on historical scenarios: Analyze the reasonable range of irrigation volume, fertilization volume, and planting density from the matched historical planting scenarios, for example: upper and lower limits of irrigation volume (such as 50-150 cubic meters per mu), upper and lower limits of fertilization volume (such as the range of nitrogen fertilizer, phosphate fertilizer, and potassium fertilizer), and the range of planting density (such as the number of crops per mu).

[0057] Dynamic adjustment based on environmental and policy factors: The above range will be revised based on the current water resources situation, soil fertility, and government policies (such as water restriction policies) in the region to form restriction factors.

[0058] Step 3: Randomly generate parameter space

[0059] Determine the number of parameter spaces: Based on enterprise needs or algorithm design, generate a number of parameter spaces by default, for example, generate 100 sets of parameter combinations.

[0060] Use random generation tools: Randomly generate specific values ​​of irrigation amount, fertilizer amount, and planting density within the range of restricted elements to form a parameter space, for example:

[0061] P1 = {irrigation volume = 120m 3 , fertilizer amount = 20kg, planting density = 3000 plants / mu};

[0062] P2 = {irrigation volume = 100m 3 , fertilizer amount = 25kg, planting density = 2800 plants / mu};

[0063] Form the parameter space set:

[0064] The set of all randomly generated parameter spaces is recorded as: {P1,P2,…,P N}.

[0065] Through the above steps, historical scenarios that match the current planting scenarios can be extracted from the enterprise agricultural management platform and historical data, and the parameter space of irrigation amount, fertilization amount, and planting density can be randomly generated under the conditions of restricted factors to provide support for subsequent scenario optimization and management plan generation.

[0066] The loss value of each parameter space is calculated by the information function, including the following steps:

[0067] The loss value of each parameter space is calculated by the information function, and the information function expression is:

[0068] In the formula, loss z is the loss value, θ is the prediction error, δ is the predicted yield mean, τ is the land utilization rate, β, α, and γ are the proportional coefficients of the prediction error, the predicted yield mean, and the land utilization rate, respectively, and β, α, and γ are all greater than 0. The larger the loss value, the worse the performance of the parameter space applied to the current planting scenario.

[0069] The calculation expression of the prediction error is: In the formula, θ is the prediction error, n is the number of predictions of the yield prediction model, is the yield value predicted by the yield prediction model for the i-th time, and y is the corresponding historical actual yield value. The larger the prediction error, the greater the loss value. When the prediction error is large, it indicates that the combination of parameter space (such as irrigation amount, fertilizer amount and planting density) leads to inaccurate estimation of the actual yield by the model, and the reliability of the current parameter space is reduced.

[0070] The calculation expression for the predicted mean yield is: In the formula, δ is the predicted yield mean, n is the number of predictions of the yield prediction model, is the yield value predicted by the yield prediction model for the i-th time. The lower the predicted yield mean, the greater the loss value. When the predicted yield mean is low, it means that the resource allocation efficiency of the current parameter space is poor, it is difficult to achieve the high yield target, and the superiority of the current solution is low.

[0071] In this application, the prediction model adopts the multivariate linear regression model in the prior art. By inputting the current irrigation amount, fertilizer amount and planting density into the trained multivariate linear regression model, the multivariate linear regression model can output the predicted yield. This technology belongs to the prior art and will not be repeated in this application.

[0072] The calculation logic of land utilization rate is: obtain the total planting area of ​​the parameter space, obtain the total planting area of ​​the enterprise, and divide the total planting area by the total planting area to obtain the land utilization rate. The lower the land utilization rate, the greater the loss value. When the land utilization rate is low, it means that the planting density or land use method of the current parameter space fails to make full use of land resources, and the scheme effect is poor.

[0073] Substituting the loss values ​​of all parameter spaces into the dynamic screening algorithm calculation includes the following steps:

[0074] The fitness value of the parameter space is calculated based on the loss value, and the expression is: In the formula, Atd is the fitness value, loss z is the loss value, and the fitness value is the inverse of the loss value. Therefore, the larger the fitness value of the parameter space is, the smaller the loss value of the parameter space is, and the better the performance is when applied to the current planting scene.

[0075] The fitness values ​​of all parameter spaces are summed up to obtain the total fitness value, and the selection probability of each parameter space is obtained by dividing the fitness value by the total fitness value. The selection probability of each parameter space is mapped to the sector area of ​​the virtual roulette wheel. The greater the selection probability of the parameter space, the larger the sector area occupied by the parameter space on the virtual roulette wheel. The virtual roulette wheel is started to rotate. When the virtual roulette wheel stops, the virtual pointer is pointed to the parameter space corresponding to the sector area to be selected, and the virtual roulette wheel is repeatedly started to rotate. When the number of selected parameter spaces is equal to the preset number threshold, the remaining parameter spaces on the virtual roulette wheel are screened out, and the selected parameter spaces are retained.

[0076] After filtering out some parameter spaces, the content of the remaining parameter spaces is randomized, and a space set is established for all parameter spaces that have completed the randomization operation, including the following steps:

[0077] After screening out some parameter spaces, the retained parameter spaces are subjected to content randomization operations, the randomization operations including content exchange operations and mutation operations, the content exchange operations including exchanging parts of the content of the parameter spaces between each other (in order to avoid the situation where an odd number of parameter spaces cannot realize content exchange, the quantity threshold of the present application is preset to an even number so that all the retained parameter spaces can perform content exchange). When all parameter spaces complete the content exchange operation, all parameter spaces that have completed the content exchange operation are subjected to content random mutation operations, thereby increasing the randomness of the parameter spaces, facilitating global optimization, and improving the robustness of the evaluation system, and establishing a spatial set of all parameter spaces that have completed the mutation operation.

[0078] Initial reserved parameter space:

[0079] Assume that 4 parameter spaces are reserved (because the number threshold is an even number), and the content of each parameter space includes irrigation amount (m 3 / mu), fertilizer amount (kg / mu) and planting density (plants / mu), as shown below:

[0080] Parameter space 1: [120, 30, 2800];

[0081] Parameter space 2: [100, 25, 3000];

[0082] Parameter space 3: [110, 28, 2600];

[0083] Parameter space 4: [115, 35, 2900];

[0084] 1. Content exchange operation

[0085] The reserved parameter spaces are randomly grouped in pairs and some of their contents are exchanged.

[0086] Randomization:

[0087] [parameter space 1, parameter space 2];

[0088] [parameter space 3, parameter space 4];

[0089] Exchange rules:

[0090] Each group randomly selects a variable to exchange. For example:

[0091] Group 1 (parameter spaces 1 and 2): swap irrigation amounts

[0092] Parameter space 1: [100, 30, 2800];

[0093] Parameter space 2: [120, 25, 3000];

[0094] Group 2 (parameter spaces 3 and 4): swapping planting density

[0095] Parameter space 3: [110, 28, 2900];

[0096] Parameter space 4: [115, 35, 2600];

[0097] 2. Random content mutation operation

[0098] All parameter spaces after the content exchange operation are randomly mutated to increase randomness and avoid falling into local optimality.

[0099] Mutation rule: Randomly select a variable and increase or decrease its value by a random number within a certain range (generated according to parameter constraints).

[0100] For example: irrigation volume increases or decreases by 5 to 10 m 3 / mu, adjust the fertilizer amount to 2-5kg / mu, and adjust the planting density to 50-100 plants / mu.

[0101] Mutation results:

[0102] Parameter space 1: irrigation amount 100→105 (increase by 5), other parameters remain unchanged. [105, 30, 2800];

[0103] Parameter space 2: Fertilizer amount 25→28 (increase by 3), other parameters remain unchanged. [120, 28, 3000];

[0104] Parameter space 3: Planting density 2900→2850 (reduced by 50), other parameters remain unchanged. [110, 28, 2850];

[0105] Parameter space 4: irrigation amount 115→120 (increase by 5), other parameters remain unchanged. [120, 35, 2600];

[0106] 3. Create a spatial collection

[0107] After the content randomization operation is completed, all parameter spaces are formed into a new space set: ={[105, 30, 2800], [120, 28, 3000], [110, 28, 2850], [120, 35, 2600]};

[0108] Content exchange operation: select variables to exchange in pairs to promote information interaction in parameter space and improve diversity.

[0109] Content mutation operation: Randomly adjust variables to increase exploration capabilities and avoid falling into local optimal solutions.

[0110] Establish a spatial set: provide a new parameter candidate space for subsequent iterations, and finally screen out the global optimal solution.

[0111] Through these operations, the parameter space is more random and diverse, allowing the evaluation system to explore better management solutions and improve the overall optimization capability and robustness.

[0112] Repeat step S2 for the established space set to iterate, and after the convergence condition is met, output multiple groups of space sets, including the following steps:

[0113] Repeat step S2 for the established space set to iterate. When there is a loss value less than or equal to the loss threshold or the number of iterations is equal to the number threshold in any space set, it is judged that the convergence condition is met and multiple groups of space sets {kj1, kj2, kj3, ..., kj m}, where m represents the number of spatial sets, kji Represents the i-th group of spatial sets.

[0114] Compare the loss values ​​of multiple sets of space sets and select the parameter space with the smallest loss value as the management plan for the enterprise's current planting scenario, including the following steps:

[0115] Obtain the content information of all parameter spaces in each group of spatial sets and the loss value information corresponding to the parameter space. In the spatial set, sort all parameter spaces from small to large according to the loss value, compare the loss values ​​of the parameter spaces ranked first in all spatial sets, and select the parameter space with the smallest loss value as the management plan for the company's current planting scenario.

[0116] The company evaluated the combination of irrigation amount, fertilization amount and planting density. After multiple rounds of randomization and iteration, it generated multiple space sets. Each parameter space corresponds to a loss value (the smaller the better).

[0117] Assume that there are the following three space sets, each of which contains four parameter spaces and their corresponding loss values:

[0118] Space Collection 1:

[0119] [105,30,2800]Loss value = 0.12;

[0120] [120, 28, 3000 loss value = 0.15;

[0121] [110, 28, 2850] loss value = 0.18;

[0122] [120,35,2600] loss value = 0.22;

[0123] Space Collection 2:

[0124] [100, 25, 2900] loss value = 0.10;

[0125] [110, 30, 2800] loss value = 0.13;

[0126] [115, 28, 3000] loss value = 0.20;

[0127] [105,35,2700]Loss value = 0.25;

[0128] Space Collection 3:

[0129] [115,30,2750] Loss value 3 0.14;

[0130] [105,32,2800]Loss value = 0.16;

[0131] [110, 29, 2900] loss value = 0.19;

[0132] [100, 30, 2850] loss value = 0.23;

[0133] Step 1: Sort the parameter space within each space set Space set 1:

[0134] Sort by loss value:

[0135] 1.[105,30,2800](0.12);

[0136] 2.[120, 28, 3000](0.15);

[0137] 3.[110, 28, 2850](0.18);

[0138] 4.[120,35,2600](0.22);

[0139] Space Collection 2:

[0140] Sort by loss value:

[0141] 1.[100, 25, 2900](0.10);

[0142] 2.[110,30,2800](0.13);

[0143] 3.[115,28,3000](0.20);

[0144] 4.[105,35,2700](0.25);

[0145] Space Collection 3:

[0146] Sort by loss value:

[0147] 1.[115,30,2750](0.14);

[0148] 2.[105,32,2800](0.16);

[0149] 3.[110, 29, 2900](0.19);

[0150] 4.[100,30,2850](0.23);

[0151] Step 2: Extract the first ranked parameter space in each set and select the parameter space with the smallest loss value from each space set: Space set 1: [105, 30, 2800] (0.12);

[0152] Space set 2: [100, 25, 2900] (0.10);

[0153] Space set 3: [115, 30, 2750] (0.14);

[0154] Step 3: Compare the loss values ​​of the selected parameter space

[0155] Compare the loss values ​​of the first place in the three sets:

[0156] Space set 1: 0.12;

[0157] Space collection 2: 0.10;

[0158] Space collection 3: 0.14;

[0159] Select result:

[0160] The smallest loss value is [100, 25, 2900] (0.10).

[0161] Finally, [100, 25, 2900] was selected as the management solution for the company's current planting scenario.

[0162] By sorting the parameter spaces in each space set by the loss value and selecting the first-ranked parameter space for comparison, the global optimal management solution with the smallest loss value can be quickly determined. This method ensures the effective exploration and selection of the global space and improves the rationality and scientificity of the management solution.

[0163] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0164] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0165] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0166] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0167] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent decision support system based on big data, characterized by: It includes acquisition module, space set building module and solution output module; Collection module: connects with the enterprise agricultural management platform through the API interface, obtains the enterprise's current planting scene through the enterprise agricultural management platform, obtains the historical planting scene that matches the current planting scene from the big database, generates restriction elements based on the historical planting scene, and uses the random generation tool to generate several parameter spaces according to the preset number of spaces; Space set establishment module: Calculate the loss value of each parameter space through the information function, substitute the loss values ​​of all parameter spaces into the dynamic screening algorithm calculation, and after screening out some parameter spaces, perform content randomization operation on the retained parameter spaces, and establish a space set for all parameter spaces that have completed the randomization operation; Solution output module: repeatedly iterate the established spatial set, and after meeting the convergence conditions, output multiple sets of spatial sets, compare the loss values ​​of the multiple sets of spatial sets, and select the parameter space with the smallest loss value as the management solution for the company's current planting scenario.

2. The intelligent decision support system based on big data according to claim 1, characterized in that: The acquisition module generates a number of parameter spaces using a random generation tool according to a preset number of spaces, and the contents of the parameter spaces include irrigation amount, fertilization amount and planting density.

3. The intelligent decision support system based on big data according to claim 2 is characterized in that: The space set establishment module calculates the loss value of each parameter space through an information function, including the following steps: The loss value of each parameter space is calculated by the information function, and the information function expression is: In the formula, loss z is the loss value, θ is the prediction error, δ is the predicted yield mean, τ is the land utilization rate, β, α, and γ are the proportional coefficients of the prediction error, the predicted yield mean, and the land utilization rate, respectively, and β, α, and γ are all greater than 0. The larger the loss value, the worse the performance of the parameter space applied to the current planting scenario.

4. The intelligent decision support system based on big data according to claim 3 is characterized by: The space set establishment module calculates the fitness value of the parameter space according to the loss value, and the expression is: In the formula, Atd is the fitness value, loss z is the loss value; the fitness values ​​of all parameter spaces are summed up to obtain the total fitness value, the selection probability of each parameter space is obtained by dividing the fitness value by the total fitness value, and the selection probability of each parameter space is mapped to the sector area of ​​the virtual roulette; Start the virtual wheel to rotate. When the virtual wheel stops, point the virtual pointer to the parameter space corresponding to the sector area and select it; The virtual wheel is repeatedly started to rotate. When the number of selected parameter spaces is equal to a preset number threshold, the remaining parameter spaces on the virtual wheel are screened out and the selected parameter spaces are retained.

5. The intelligent decision support system based on big data according to claim 4, characterized in that: After screening out some parameter spaces, the space set establishment module performs a content randomization operation on the retained parameter spaces. The randomization operation includes a content exchange operation and a mutation operation. The content exchange operation includes exchanging part of the content of each parameter space. When all parameter spaces complete the content exchange operation, a content random mutation operation is performed on all parameter spaces that have completed the content exchange operation, and a space set is established for all parameter spaces that have completed the mutation operation.

6. The intelligent decision support system based on big data according to claim 5, characterized in that: The scheme output module repeatedly iterates the established space sets. When there is a loss value in any space set that is less than or equal to the loss threshold or the number of iterations is equal to the number threshold, it is judged that the convergence condition is met and multiple groups of space sets {kj1, kj2, kj3, ..., kj m }, where m represents the number of spatial sets, kj i Represents the i-th group of spatial sets.

7. The intelligent decision support system based on big data according to claim 6, characterized in that: The solution output module obtains the content information of all parameter spaces in each group of space sets and the loss value information corresponding to the parameter space. In the space set, all parameter spaces are sorted from small to large according to the loss value, and the loss values ​​of the parameter spaces ranked first in all space sets are compared, and the parameter space with the smallest loss value is selected as the management solution for the enterprise's current planting scenario.

8. The intelligent decision support system based on big data according to claim 7, characterized in that: The calculation expression of the prediction error is: In the formula, θ is the prediction error, n is the number of predictions of the yield prediction model, is the output value predicted by the output prediction model for the i-th time, and y is the corresponding historical actual output value; The calculation expression of the predicted output mean is: In the formula, δ is the predicted yield mean, n is the number of predictions of the yield prediction model, is the output value predicted by the output prediction model for the i-th time; The calculation logic of the land utilization rate is: obtain the total planting area in the parameter space, obtain the total planting area of ​​the enterprise, and divide the total planting area by the total planting area to obtain the land utilization rate.

Images