AI-based crop growth management decision system

Abstract

The application discloses an AI-based crop growth management decision system and belongs to the technical field of crop growth management, comprising a data acquisition module, a state analysis module, a disease and pest prediction module and a management subsystem, wherein the management subsystem comprises an adjustment scheme generation module, a treatment scheme generation module and a prevention strategy generation module. The application constructs the disease and pest prediction module and the prevention strategy generation module based on previous data, the disease and pest prediction module constructs a prediction model based on previous data, the prediction model analyzes data of a future period, judges the probability of occurrence of diseases and pests in the future, and calculates the damage degree of the diseases and pests, which can help relevant agricultural personnel to prevent in advance and formulate disease and pest prevention strategies, facilitate relevant agricultural personnel to take measures in advance, effectively reduce the influence of diseases and pests on crops, and thus reduce economic losses caused by diseases and pests.

Description

Technical Field

[0001] This invention relates to the field of crop growth management technology, specifically to an AI-based crop growth management decision system. Background Technology

[0002] Crops include food crops (such as cereals, tubers, and beans), cash crops (such as oil crops, vegetable crops, specialty crops, industrial raw material crops, and medicinal crops), and forage crops. Currently, in agricultural production, remote sensing monitoring technology can be used to track and monitor the progress of pests and diseases, enabling precise prevention and control, timely detection and treatment, and facilitating early prevention.

[0003] Among existing patents, CN107315381B discloses a method for monitoring crop diseases and pests that utilizes drones and cloud computing. This method includes steps such as establishing a crop disease and pest database, using drones to capture images of crops in the field, calculation, early warning, precise shooting, secondary calculation, and guidance on disease and pest control. Based on the established crop disease and pest database, the method uses drones to capture images of crops in the field, analyzes and compares these images through cloud computing, promptly issues early warnings about crop diseases and pests, and guides disease and pest control, thereby improving agricultural production levels, reducing agricultural production and management costs, and increasing the yield, quality, and efficiency of agricultural products.

[0004] However, the central idea of ​​the solution described in the aforementioned patent is to take pictures, compare the real-time pictures with previously recorded picture data, determine whether pests or diseases have appeared by judging the similarity, and then issue an early warning. Like the existing technology, the above solution is mainly used for the rapid detection and treatment of pests and diseases. However, in practical applications, in the early stages of pests and diseases, the plant's reaction is not obvious, and it is impossible to detect them in time by comparing pictures. When the plant shows the characteristics of pests and diseases, the pests and diseases have already caused certain damage to the crops. At this time, timely treatment can only reduce the subsequent damage to the crops caused by pests and diseases, but cannot eliminate the economic losses caused by pests and diseases. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides an AI-based crop growth management decision-making system. This system constructs a pest and disease prediction module and a prevention strategy generation module based on historical data. The pest and disease prediction module builds a prediction model based on historical data, analyzes data for future periods, determines the probability of future pests and diseases, and calculates the severity of damage. This helps agricultural personnel to take preventative measures and formulate pest and disease prevention strategies, facilitating proactive responses and effectively reducing the impact of pests and diseases on crops, thereby minimizing economic losses. The system demonstrates good performance and promising application prospects.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] An AI-based crop growth management decision-making system includes a data acquisition module, a status analysis module, a pest and disease prediction module, and a management subsystem.

[0010] The data acquisition module obtains real-time crop data, soil data, and current and future environmental data of the area where the crops are located from the crop monitoring platform;

[0011] The status analysis module acquires crop planting data stored in the database, builds a status analysis model based on the crop planting data, and inputs the collected crop data and soil data into the status analysis model. The status analysis model compares these with normal growth data to analyze the growth status and pest and disease status of the crops.

[0012] The pest and disease prediction module acquires pest and disease data stored in the database, builds a prediction model based on the pest and disease data, and inputs collected soil data and current and future environmental data of the crop's location into the prediction model. The prediction model predicts the probability of pests and diseases occurring in the future period and the degree of damage caused by pests and diseases.

[0013] The governance subsystem includes an adjustment plan generation module, a treatment plan generation module, and a prevention strategy generation module;

[0014] The adjustment scheme generation module receives growth status data and generates adjustment schemes based on the growth status data and the scheme data stored in the database.

[0015] The treatment plan generation module receives pest and disease status data and generates a treatment plan based on the pest and disease status data and the plan data stored in the database.

[0016] The prevention strategy generation module receives data on the probability of pests and diseases occurring in the predicted future period and the degree of damage caused by pests and diseases, and formulates pest and disease prevention strategies based on this data.

[0017] Preferably, it also includes a scheme scoring module, which records adjustment schemes, treatment schemes and pest and disease prevention strategies, scores them, and then classifies and stores the scored schemes in the database.

[0018] Preferably, the steps for using a state analysis model to analyze whether crops are in a state of pests or diseases are as follows:

[0019] Preprocess the acquired crop data;

[0020] The preprocessed image data is input into a convolutional neural network for pest and disease classification and prediction, and the maximum value Py among the predicted pest and disease values ​​is obtained.

[0021] Compare Py with the set standard value Ps. When Py≥Ps, it means that the crop is infected with the pests and diseases corresponding to the maximum value Py.

[0022] When Py is less than Ps, it means that the crops are free from pests and diseases.

[0023] Preferably, the steps for analyzing whether crops are in a state of overgrowth are as follows:

[0024] Obtain the periodic average consumption of nitrogen, phosphorus, and potassium elements in crops during different growth stages. and The calculation formula is as follows:

[0025] ;

[0026] In the formula, N represents the average nitrogen consumption over several cycles within a growth period. i Let N be the nitrogen content detected at the beginning of the i-th period. i+1 The nitrogen content detected at the beginning of the (i+1)th cycle;

[0027] Obtain the duration range [T, T+L] of different growth stages of crops and the periodic element consumption range of different growth stages. -δ, +δ]、[ -λ, +λ] and [ -ε, +ε], L is the normal fluctuation value, where δ, λ and ε are the fluctuation values ​​of crop cycle absorption in different growth stages;

[0028] The system acquires real-time data on the content of nitrogen, phosphorus, and potassium elements in the soil and calculates the differences in nitrogen, phosphorus, and potassium elements in the soil in the previous cycle (ΔN, ΔP, ΔK). The differences are then compared with the cyclical consumption intervals of different growth stages to determine the growth status of crops.

[0029] When any two sets of data from the differences in nitrogen, phosphorus, and potassium elements in the soil during the previous cycle match the consumption interval of the next growth stage, the actual growth time T´ of the crop is obtained and compared with the duration interval of the crop in different growth stages.

[0030] When T´ is greater than T+L, plant growth is too slow;

[0031] When T' is less than T, the plant grows too fast;

[0032] When T≤T´≤T+L, the plant grows normally.

[0033] Preferably, the steps for the predictive model to predict the probability of pests and diseases occurring in future periods are as follows:

[0034] Analyze the pest and disease data obtained from the database to make a preliminary judgment on the influencing factors of the pest and disease data;

[0035] Correlation coefficient analysis was performed on pest and disease data to obtain the types of influencing factors in group D of pest and disease occurrence, and the maximum damage data of each type of influencing factor was screened and set.

[0036] Calculate the difference between the influencing factor data and the maximum damage data, and then calculate the correlation coefficient between this difference and the pest and disease data. The calculation formula is as follows:

[0037] ;

[0038] In the formula, Let be the correlation coefficient between the difference between the data of the influencing factor C and the data of the maximum harm, and the data of pests and diseases; and let n be the difference between the data of the influencing factor C and the data of the maximum harm, and the number of data points of pests and diseases. This represents the average of the differences between the data for factor C and the data for the maximum hazard. C represents the average value of the pest and disease data. i S is the i-th data point in the difference between the data of the C-th influencing factor and the data of the maximum hazard. i This is the i-th data point in the pest and disease data;

[0039] The weighting of the difference between the influencing factor data and the maximum hazard data is calculated using the following formula:

[0040] ;

[0041] In the formula, W cThis is the weighted value of the difference between the data of the influencing factor in item C and the data of the maximum hazard.

[0042] By acquiring real-time soil data and current and future environmental data of the crop's location, the classification values ​​of pests and diseases for several future periods are calculated. The calculation formula is as follows:

[0043] ;

[0044] In the formula, Bchflz is the pest and disease classification value, and CZ is the disease and disease classification value. O W represents the difference between the data for the Oth influencing factor and the data for the maximum hazard. O The weight value is the difference between the data of the 0th influencing factor and the data of the maximum hazard.

[0045] By comparing the calculated pest and disease classification values ​​with the set probability range, the likelihood of pest and disease occurrence in the future cycle can be determined.

[0046] Preferably, the set possibility ranges are [0, P], [P, 2P], [2P, 3P] and [3P, +∞), where P, 2P and 3P are all set range values;

[0047] When the calculated pest and disease classification value falls within the range of [0, P], the probability of pest and disease occurrence within this period is above 60%.

[0048] When the calculated pest and disease classification value falls within the range of [P, 2P], the probability of pest and disease occurrence within that period is 40-60%.

[0049] When the calculated pest and disease classification value falls within the range of [2P, 3P], the probability of pest and disease occurrence within that period is less than 40%.

[0050] When the calculated pest classification value falls within the range of [3P, +∞), no pests will occur during that period.

[0051] The preferred formula for calculating the degree of damage caused by pests and diseases is as follows:

[0052] ;

[0053] In the formula, Whcd is the numerical value of the damage degree of pests and diseases, T1 is the number of periods in which the pest and disease classification value belongs to the interval [0, P], α is the damage coefficient of the pest and disease classification value belonging to the interval [0, P], T2 is the number of periods in which the pest and disease classification value belongs to the interval [P, 2P], β is the damage coefficient of the pest and disease classification value belonging to the interval [P, 2P], T3 is the number of periods in which the pest and disease classification value belongs to the interval [2P, 3P], γ is the damage coefficient of the pest and disease classification value belonging to the interval [2P, 3P], and ζ is the correction coefficient.

[0054] Preferably, the scheme scoring module records the adjustment scheme, treatment scheme, and pest and disease prevention strategy, and sends the results to experts for scoring after summarizing them. Excess and deficiency marks are added to the specific content of the adjustment scheme, treatment scheme, and pest and disease prevention strategy.

[0055] Preferably, the steps for generating treatment plans and prevention strategies in the treatment plan generation module and prevention strategy generation module are as follows:

[0056] Retrieve the highest-scoring treatment plan or highest-scoring prevention strategy stored in the database for the corresponding category, and read the type marked in the highest-scoring treatment plan or highest-scoring prevention strategy;

[0057] Adjust the proportion of the drug in the highest-scoring treatment plan or prevention strategy according to the type of label. The proportion of adjustment is μ. Summarize the data after adjustment to obtain the treatment plan or pest prevention strategy.

[0058] Preferably, the steps for generating the adjustment scheme are as follows:

[0059] Retrieve the highest score adjustment scheme stored in the database for the corresponding category, and read the type of the marker in the highest score adjustment scheme;

[0060] Adjust the soil element data recorded in the highest-scoring scheme by up or down according to the type of marker, with the adjustment ratio being μ. Summarize the soil element data after the adjustment and down to generate soil element scheme data.

[0061] Calculate the difference between the soil element data in the proposed scheme and the actual soil element data, summarize the difference data, and generate an adjustment scheme.

[0062] (III) Beneficial Effects

[0063] The present invention has the following beneficial effects:

[0064] 1. The system of this invention constructs a pest and disease prediction module and a prevention strategy generation module based on previous data. The pest and disease prediction module builds a prediction model based on previous data, analyzes data for future periods, judges the probability of pests and diseases occurring in the future, and calculates the degree of damage caused by pests and diseases. This can help relevant agricultural personnel to take preventive measures in advance and formulate pest and disease prevention strategies, making it convenient for relevant agricultural personnel to take countermeasures in advance. It can effectively reduce the impact of pests and diseases on crops, thereby reducing the economic losses caused by pests and diseases. It has good results and good application prospects.

[0065] The system of this invention is based on existing data and comprises a state analysis module, a solution generation module, and a treatment solution generation module. The state analysis module analyzes soil and crop data. By analyzing soil data, it can predict the growth status of crops in advance, helping agricultural personnel understand the growth trend of crops and generate adjustment plans based on the growth trend, facilitating better fertilization management. The crop data analysis can identify pest and disease data, helping staff understand the severity of plant pests and diseases and generate treatment plans, enabling agricultural personnel to better treat pests and diseases. The entire process eliminates the need for manual plan development by relevant personnel, reducing the workload of technical staff and enabling more scientific management of crop growth. It has good performance and promising application prospects. Attached Figure Description

[0066] Figure 1 This is a modular structure diagram of the present invention. Detailed Implementation

[0067] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0068] Example 1: As Figure 1 As shown, the AI-based crop growth management decision-making system includes a data acquisition module, a status analysis module, a pest and disease prediction module, a management subsystem, and a scheme scoring module.

[0069] The entire system's data sources are a database and a crop monitoring platform. The database stores data on various crops under different conditions at different times in the past. The data in the crop monitoring platform is collected in real time. The database and crop monitoring platform can be public databases, monitoring platforms, or websites that store relevant data. They can also be built using public technologies, as long as they can achieve the relevant functions. For example:

[0070] The crop monitoring platform collects crop data through equipment installed around the crops, such as displacement sensors that monitor changes in crop diameter and height, infrared instruments that monitor leaf temperature, camera devices that take pictures of plant growth, and drones that take dynamic pictures. Soil data is detected by soil fertilizer nutrient analyzers installed around the soil, and environmental data is measured by meteorological instruments set up around the area. Future data will be meteorological data released by the meteorological bureau.

[0071] Like all analysis systems, this system primarily utilizes the comparison between real-time collected data and historical data. Therefore, it is necessary to process and analyze the historical data first.

[0072] Data collection

[0073] Data acquisition is based on a data acquisition module, which obtains crop data in real time from the crop monitoring platform;

[0074] Data Analysis

[0075] The analysis of past planting data is based on the state analysis model. The state analysis module obtains crop planting data stored in the database and builds a state analysis model based on the crop planting data.

[0076] The analysis content of the state analysis model includes pest and disease analysis and plant growth status analysis;

[0077] During pest and disease analysis, convolutional neural networks are used for pest and disease classification and prediction.

[0078] The activation function used in convolutional neural networks is the Sigmoid function, and its calculation formula is as follows:

[0079] ;

[0080] The output layer uses a Softmax classifier.

[0081] This system uses the ResNet residual network model for prediction. Since this model is commonly used in existing technologies, the pest and disease analysis in this system can be directly achieved through this model. This system has not improved the accuracy and structure of the ResNet residual network model, so it will not be described in detail.

[0082] The data acquisition module transmits the acquired data to a convolutional neural network for pest and disease classification and prediction, thereby obtaining the types of pests and diseases to be classified and predicted. The steps for using a convolutional neural network for pest and disease classification are as follows:

[0083] Preprocess the acquired crop data;

[0084] Preprocessing includes data normalization and image resizing.

[0085] The preprocessed image data is input into a convolutional neural network for pest and disease classification and prediction, and the maximum value Py among the predicted pest and disease values ​​is obtained.

[0086] Compare Py with the set standard value Ps. When Py≥Ps, it means that the crop is infected with the pests and diseases corresponding to the maximum value Py.

[0087] When Py is less than Ps, it means that the crops are free from pests and diseases.

[0088] The standard value Ps is the predicted value of pests and diseases, which is usually taken as 30%. In the above prediction process, it is assumed that the probability of the yth type of pest and disease in the crop is the highest, which is Py. When the probability of the crop being infected with the pest and disease corresponding to the maximum value Py is less than 30%, that is, when Py < 30%, it is considered that the crop is free of pests and diseases.

[0089] When the probability of a crop being infected with the pest or disease corresponding to the maximum value Py is ≥30%, further analysis can be performed. The second largest predicted value of the pest or disease can be retrieved and compared with 30% to determine whether the crop is infected with both pests or diseases.

[0090] Because crop management is now quite scientific and strict, it is unlikely that crops will be infected with two different pests and diseases under normal circumstances. Therefore, usually only one comparison is needed, that is, to determine the size of Py and Ps.

[0091] When it is determined that a crop is affected by a certain type of pest or disease, a corresponding treatment plan needs to be developed.

[0092] Treatment plan generation

[0093] Treatment plans are developed based on the treatment plan generation module.

[0094] Analysis of past pest and disease control data based on the treatment plan generation module;

[0095] The treatment plan generation module receives pest and disease status data and generates treatment plans based on the pest and disease status data and the treatment plan data stored in the database.

[0096] The steps for generating a treatment plan using the treatment plan generation module are as follows:

[0097] Retrieve the highest-scoring treatment plan stored in the database for the corresponding category, and read the type marked in the highest-scoring treatment plan;

[0098] Adjust the proportion of the highest-scoring medication in the treatment plan by increasing or decreasing the proportion by μ, where μ is 1%. Sum the data after increasing and decreasing the proportion to obtain the treatment plan.

[0099] The markings are generated by the scheme scoring module. The scheme scoring module records the adjustment scheme, treatment scheme and pest and disease prevention strategy, and summarizes the results and sends them to experts for expert scoring. Excess and deficiency marks are added to the specific content of the adjustment scheme, treatment scheme and pest and disease prevention strategy.

[0100] For example, when corn leaves show yellow spots, 25% triadimefon wettable powder, 30% solid lime sulfur mixture, 25% difenoconazole emulsifiable concentrate, and 12.5% ​​sphaperam wettable powder are usually used in combination. During treatment, the treatment plan and feedback on the plant treatment are obtained. The plan and feedback are sent to experts for evaluation. The dosage of the added drugs is analyzed based on the detected plant condition to determine whether the drug is added in excess or insufficient, and excess or insufficient drugs are marked.

[0101] When the treatment plan scores the highest, it will be used for the same condition in the future, with the μ ratio increased or decreased. Over time, the treatment plan will be gradually optimized, and the μ value can be adjusted as needed. In the later stages, when the plan is more precise, the μ ratio can be reduced.

[0102] Because existing crops are monitored in real time, treatment can begin in the early stages of pests and diseases. As a result, the severity of pests and diseases is similar, so the treatment plan can be adjusted in real time based on the highest score to obtain the optimal treatment plan.

[0103] Example 2: As Figure 1 As shown, the solution described in this embodiment is to analyze whether crops are in an over-growth state;

[0104] Data collection

[0105] Data acquisition is based on a data acquisition module, which obtains soil data in real time from the crop monitoring platform;

[0106] Data Analysis

[0107] The analysis of past planting data is based on the state analysis module. The steps of the state analysis module to analyze whether crops are in an over-growth state are as follows:

[0108] Obtain the periodic average consumption of nitrogen, phosphorus, and potassium elements in crops during different growth stages. and The calculation formula is as follows:

[0109] ;

[0110] In the formula, N represents the average nitrogen consumption over several cycles within a growth period. i Let N be the nitrogen content detected at the beginning of the i-th period. i+1 The nitrogen content detected at the beginning of the (i+1)th cycle;

[0111] ;

[0112] In the formula, P represents the average consumption of phosphorus over several cycles within a growth period. i P represents the phosphorus content detected at the beginning of the i-th period. i+1 The phosphorus content detected at the beginning of the (i+1)th cycle;

[0113] ;

[0114] In the formula, K represents the average potassium consumption over several cycles within a growth period. i K represents the potassium content detected at the beginning of the i-th period. i+1 The potassium content detected at the beginning of the (i+1)th cycle;

[0115] Plants have different requirements for different elements at different growth stages. Therefore, when the growth period of a plant changes, it is necessary to replenish the content of the corresponding elements in a timely manner.

[0116] Obtain the duration range [T, T+L] of different growth stages of crops and the periodic element consumption range of different growth stages. -δ, +δ]、[ -λ, +λ] and [ -ε, +ε], L is the normal fluctuation value, where δ, λ and ε are the fluctuation values ​​of crop cycle absorption in different growth stages;

[0117] Different plants absorb different amounts of elements at different growth stages, and the fluctuations in these elements also vary. Therefore, each plant needs to be assessed individually.

[0118] The system acquires real-time data on the content of nitrogen, phosphorus, and potassium elements in the soil and calculates the differences in nitrogen, phosphorus, and potassium elements in the soil in the previous cycle (ΔN, ΔP, ΔK). The differences are then compared with the cyclical consumption intervals of different growth stages to determine the growth status of crops.

[0119] When any two sets of data from the differences in nitrogen, phosphorus, and potassium elements in the soil during the previous cycle match the consumption interval of the next growth stage, the actual growth time T´ of the crop is obtained and compared with the duration interval of the crop in different growth stages.

[0120] When T´ is greater than T+L, plant growth is too slow;

[0121] When T' is less than T, the plant grows too fast;

[0122] When T≤T´≤T+L, the plant grows normally.

[0123] Different crops have different cycles; crops with a planting cycle of more than 30 days have a cycle of 1 day.

[0124] For example, in the seedling and developmental stages of crops, the cyclical consumption range during the seedling stage is much shorter than that during the developmental stage. When monitoring... and When any two of the cycle consumption intervals switch from the seedling stage to the development stage, the seedling stage of the crop is considered to have ended and the crop has entered the development stage.

[0125] At this point, calculate the actual seedling stage duration T´ of the crop and compare it with the standard seedling stage duration (which is affected by the environment and is generally within a range, with L usually being 3 days). For example, the normal seedling stage duration for corn is T=15 and L=3, meaning 15-18 days is normal. If the calculation shows that the corn seedling stage exceeds 18 days, it indicates slow development. If the corn seedling stage is less than 18 days, it indicates rapid development. In most cases, this is caused by excessive or insufficient fertilizer, so the amount of fertilizer needs to be adjusted, and an adjustment plan needs to be developed.

[0126] The steps to generate the adjustment plan are as follows:

[0127] The adjustment plan is based on the adjustment plan generation module, which receives growth status data and generates an adjustment plan based on the growth status data and the plan data stored in the database.

[0128] Retrieve the highest score adjustment scheme stored in the database for the corresponding category, and read the type of the marker in the highest score adjustment scheme;

[0129] Adjust the soil element data recorded in the highest-scoring scheme by up or down according to the type of marker, with the adjustment ratio being μ. Summarize the soil element data after the adjustment and down to generate soil element scheme data.

[0130] Calculate the difference between the soil element data in the proposed scheme and the actual soil element data, summarize the difference data, and generate an adjustment scheme.

[0131] For example, if the amount of nitrogen fertilizer in the soil after the original highest score adjustment plan is Nz, and the nitrogen fertilizer amount needs to be increased after evaluation, the increased nitrogen fertilizer amount is Nz*(1+μ), and the current amount of nitrogen fertilizer in the soil is Nj, the amount to be adjusted is Nz*(1+μ)-Nj. The adjustment plan can be obtained by summing up the adjustment amounts of all elements.

[0132] Example 3: As Figure 1 As shown, the solution described in this embodiment is for predicting the likelihood of pests and diseases and the degree of harm caused by them.

[0133] Data collection

[0134] Data acquisition is based on a data acquisition module, which obtains real-time environmental data of the area where the crops are located from the crop monitoring platform, including current and future environmental data.

[0135] Data Analysis

[0136] The analysis of past pest and disease data is based on the pest and disease prediction module. The pest and disease prediction module obtains pest and disease data stored in the database and builds a prediction model based on the pest and disease data. The collected soil data and the current and future environmental data of the crop area are input into the prediction model. The prediction model predicts the probability of pests and diseases occurring in the future period and the degree of damage caused by pests and diseases.

[0137] The steps involved in predicting the likelihood of pests and diseases occurring within a future timeframe using a predictive model are as follows:

[0138] Analyze the pest and disease data obtained from the database to make a preliminary judgment on the influencing factors of the pest and disease data;

[0139] This plan selects the occurrence conditions of the most severe pests and diseases (the average value of data within a period), that is, the maximum damage data of the influencing factors of this type;

[0140] The difference between the data on influencing factors and the data on the maximum harm was used as the independent variable, that is, the difference with the most severe pests and diseases.

[0141] For example, temperature. The average temperature at which a certain pest is most harmful is 20℃. The influencing factor data is the average data for each period. The difference between the average data for each period and the maximum harm data is used as the basis. The larger the calculated value, the lower the probability of pests and diseases occurring. They show an inverse correlation.

[0142] Correlation coefficient analysis was performed on pest and disease data to obtain the types of influencing factors in group D of pest and disease occurrence, and the maximum damage data of each type of influencing factor was screened and set.

[0143] Calculate the difference between the influencing factor data and the maximum damage data, and then calculate the correlation coefficient between this difference and the pest and disease data. The calculation formula is as follows:

[0144]

[0145] In the formula, Let be the correlation coefficient between the difference between the data of the influencing factor C and the data of the maximum harm, and the data of pests and diseases; and let n be the difference between the data of the influencing factor C and the data of the maximum harm, and the number of data points of pests and diseases. This represents the average of the differences between the data for factor C and the data for the maximum hazard. C represents the average value of the pest and disease data. i S is the i-th data point in the difference between the data of the C-th influencing factor and the data of the maximum hazard. i This is the i-th data point in the pest and disease data.

[0146] The weighting of the difference between the influencing factor data and the maximum hazard data is calculated using the following formula:

[0147] ;

[0148] In the formula, W c This is the weighted value of the difference between the data of the influencing factor in item C and the data of the maximum hazard.

[0149] For example, the difference between the data of the influencing factor in item C and the data of the maximum hazard. It is -0.8, while the total relevant absorbance is... Then the weight value W c =|-0.8| / 2.5=0.32.

[0150] By acquiring real-time soil data and current and future environmental data of the crop's location, the classification values ​​of pests and diseases for several future periods are calculated. The calculation formula is as follows:

[0151] ;

[0152] In the formula, Bchflz is the pest and disease classification value, and CZ is the disease and disease classification value. O W represents the difference between the data for the Oth influencing factor and the data for the maximum hazard. O The weight value is the difference between the data of the 0th influencing factor and the data of the maximum hazard.

[0153] By comparing the calculated pest and disease classification values ​​with the set probability range, the likelihood of pest and disease occurrence in the future cycle can be determined.

[0154] The possible ranges are set as [0, P], [P, 2P], [2P, 3P] and [3P, +∞), where P, 2P and 3P are all set range values;

[0155] When the calculated pest and disease classification value falls within the range of [0, P], the probability of pest and disease occurrence within this period is above 60%.

[0156] When the calculated pest and disease classification value falls within the range of [P, 2P], the probability of pest and disease occurrence within that period is 40-60%.

[0157] When the calculated pest and disease classification value falls within the range of [2P, 3P], the probability of pest and disease occurrence within that period is less than 40%.

[0158] When the calculated pest classification value falls within the range of [3P, +∞), no pests will occur during that period.

[0159] This calculation is performed separately, based on the assumption that no pests or diseases occurred before the start of the cycle, or that appropriate measures were taken to ensure that the impact of the previous day would not affect the following day.

[0160] This tool can help staff understand the trend of pests and diseases. For example, if the Bchflz values ​​for the next five consecutive cycles are located at [0, P], [0, P], [P, 2P], [P, 2P], and [2P, 3P], it can be determined that the probability of pests and diseases occurring is decreasing. Therefore, appropriate preventive measures can be taken a few days in advance.

[0161] The formula for calculating the degree of damage caused by pests and diseases is as follows:

[0162] ;

[0163] In the formula, Whcd is the numerical value of the damage degree of pests and diseases, T1 is the number of periods in which the pest and disease classification value belongs to the interval [0, P], α is the damage coefficient of the pest and disease classification value belonging to the interval [0, P], T2 is the number of periods in which the pest and disease classification value belongs to the interval [P, 2P], β is the damage coefficient of the pest and disease classification value belonging to the interval [P, 2P], T3 is the number of periods in which the pest and disease classification value belongs to the interval [2P, 3P], γ is the damage coefficient of the pest and disease classification value belonging to the interval [2P, 3P], and ζ is the correction coefficient.

[0164] As the cycle lengthens, when the cycle falls within the range of pest and disease occurrence, the degree of harm will gradually increase without treatment. 1≤Whcd≤10, level 1 is the smallest, indicating a low probability of pest and disease and a short duration, while level 10 is the largest, indicating a high probability of pest and disease and a long duration.

[0165] The steps for generating a prevention strategy are as follows:

[0166] Retrieve the highest-scoring prevention strategy stored in the database for the corresponding category, and read the type of the tag in the highest-scoring prevention strategy;

[0167] Adjust the proportion of the highest-scoring pesticide in the prevention strategy according to the type of label. The proportion of adjustment is μ, and μ is 1%. Summarize the data after adjustment to obtain the pest and disease prevention strategy.

[0168] The markings are generated by the scheme scoring module. The scheme scoring module records the adjustment scheme, treatment scheme and pest and disease prevention strategy, and summarizes the results and sends them to experts for expert scoring. Excess and deficiency marks are added to the specific content of the adjustment scheme, treatment scheme and pest and disease prevention strategy.

[0169] For example, when a certain pest occurs, it is necessary to spray a mixture of multiple pesticides in advance. When using it, a prevention plan and feedback will be obtained. The plan and feedback will be sent to experts for evaluation. Based on the detected pest occurrence, the amount of pesticide added will be analyzed to determine whether the pesticide is added too much or too little, and excessive or insufficient pesticides will be marked.

[0170] When the plan scores the highest, it will be used to prevent the same disease in the future. The μ ratio will be increased or decreased based on the plan, and the plan will be used for prevention. Over time, the prevention plan will be gradually optimized, and the μ value can be adjusted as needed. In the later stages, when the plan is more accurate, the μ ratio can be reduced.

[0171] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0172] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

Claims

1. An AI-based crop growth management decision-making system, characterized in that, include: The data acquisition module obtains real-time crop data, soil data, and current and future environmental data of the area where the crops are located from the crop monitoring platform; The status analysis module acquires crop planting data stored in the database, builds a status analysis model based on the crop planting data, and inputs the collected crop data and soil data into the status analysis model. The status analysis model compares these data with normal growth data to analyze the growth status and pest and disease status of the crops. The pest and disease prediction module acquires pest and disease data stored in the database, builds a prediction model based on the pest and disease data, and inputs collected soil data and current and future environmental data of the crop's location into the prediction model. The prediction model predicts the probability of pests and diseases occurring in the future period and the degree of damage caused by pests and diseases. The governance subsystem includes an adjustment plan generation module, a treatment plan generation module, and a prevention strategy generation module; The adjustment scheme generation module receives growth status data and generates adjustment schemes based on the growth status data and the scheme data stored in the database. The treatment plan generation module receives pest and disease status data and generates a treatment plan based on the pest and disease status data and the plan data stored in the database. The prevention strategy generation module receives data on the probability of pests and diseases occurring in the predicted future period and the degree of damage caused by pests and diseases, and formulates pest and disease prevention strategies based on the data. The steps to analyze whether crops are in a state of overgrowth are as follows: Obtain the periodic average consumption of nitrogen, phosphorus, and potassium elements in crops during different growth stages. and The calculation formula is as follows: ； In the formula, N represents the average nitrogen consumption over several cycles within a growth period. i Let N be the nitrogen content detected at the beginning of the i-th period. i+1 The nitrogen content detected at the beginning of the (i+1)th cycle; Obtain the duration range [T, T+L] of different growth stages of crops and the periodic element consumption range of different growth stages. -δ, +δ]、[ -λ, +λ] and [ -ε, +ε], L is the normal fluctuation value, where δ, λ and ε are the fluctuation values ​​of crop cycle absorption in different growth stages; The system acquires real-time data on the content of nitrogen, phosphorus, and potassium elements in the soil and calculates the differences in nitrogen, phosphorus, and potassium elements in the soil in the previous cycle (ΔN, ΔP, ΔK). The differences are then compared with the cyclical consumption intervals of different growth stages to determine the growth status of crops. When any two sets of data from the differences in nitrogen, phosphorus, and potassium elements in the soil during the previous cycle match the consumption interval of the next growth stage, the actual growth time T´ of the crop is obtained and compared with the duration interval of the crop in different growth stages. When T´ is greater than T+L, plant growth is too slow; When T' is less than T, the plant grows too fast; When T≤T´≤T+L, the plant grows normally; The steps of a predictive model to predict the likelihood of pests and diseases occurring within a future timeframe are as follows: Analyze the pest and disease data obtained from the database to make a preliminary judgment on the influencing factors of the pest and disease data; Correlation coefficient analysis was performed on pest and disease data to obtain the types of influencing factors in group D of pest and disease occurrence, and the maximum harm data of each type of influencing factor was screened and set. Calculate the difference between the influencing factor data and the maximum damage data, and then calculate the correlation coefficient between this difference and the pest and disease data. The calculation formula is as follows: ； In the formula, Let be the correlation coefficient between the difference between the data of the influencing factor C and the data of the maximum harm, and the data of pests and diseases; and let n be the difference between the data of the influencing factor C and the data of the maximum harm, and the number of data points of pests and diseases. This represents the average of the differences between the data for factor C and the data for the maximum hazard. C represents the average value of the pest and disease data. i S is the i-th data point in the difference between the data of the C-th influencing factor and the data of the maximum hazard. i This is the i-th data point in the pest and disease data; The weighting of the difference between the influencing factor data and the maximum hazard data is calculated using the following formula: ； In the formula, W c This is the weighted value of the difference between the data of the influencing factor in item C and the data of the maximum hazard. By acquiring real-time soil data and current and future environmental data of the crop's location, the classification values ​​of pests and diseases for several future periods are calculated. The calculation formula is as follows: ； In the formula, Bchflz is the pest and disease classification value, and CZ is the disease and disease classification value. O W represents the difference between the data for the Oth influencing factor and the data for the maximum hazard. O The weight value is the difference between the data of the 0th influencing factor and the data of the maximum hazard. By comparing the calculated pest and disease classification values ​​with the set probability range, the probability of pest and disease occurrence in the future cycle can be determined. The possible ranges are set as [0, P], [P, 2P], [2P, 3P] and [3P, +∞), where P, 2P and 3P are all set range values; When the calculated pest and disease classification value falls within the range of [0, P], the probability of pest and disease occurrence within this period is above 60%. When the calculated pest and disease classification value falls within the range of [P, 2P], the probability of pest and disease occurrence within that period is 40-60%. When the calculated pest and disease classification value falls within the range of [2P, 3P], the probability of pest and disease occurrence within that period is less than 40%. When the calculated pest classification value falls within the interval [3P, +∞), no pests will occur during that period. The formula for calculating the degree of damage caused by pests and diseases is as follows: ； In the formula, Whcd is the numerical value of the damage degree of pests and diseases, T1 is the number of periods in which the pest and disease classification value belongs to the interval [0, P], α is the damage coefficient of the pest and disease classification value belonging to the interval [0, P], T2 is the number of periods in which the pest and disease classification value belongs to the interval [P, 2P], β is the damage coefficient of the pest and disease classification value belonging to the interval [P, 2P], T3 is the number of periods in which the pest and disease classification value belongs to the interval [2P, 3P], γ is the damage coefficient of the pest and disease classification value belonging to the interval [2P, 3P], and ζ is the correction coefficient.

2. The AI-based crop growth management decision-making system according to claim 1, characterized in that: It also includes a scheme scoring module, which records adjustment schemes, treatment schemes and pest and disease prevention strategies, scores them, and then classifies and stores the scored schemes in the database.

3. The AI-based crop growth management decision-making system according to claim 2, characterized in that: The steps of the state analysis model to analyze whether crops are in a state of pests and diseases are as follows: Preprocess the acquired crop data; The preprocessed image data is input into a convolutional neural network for pest and disease classification and prediction, and the maximum value Py among the predicted pest and disease values ​​is obtained. Compare Py with the set standard value Ps. When Py≥Ps, it means that the crop is infected with the pests and diseases corresponding to the maximum value Py. When Py is less than Ps, it means that the crops are free from pests and diseases.

4. The AI-based crop growth management decision-making system according to claim 2, characterized in that: The scheme scoring module records the adjustment scheme, treatment scheme, and pest and disease prevention strategy, and summarizes the results before sending them to experts for scoring. Excess and deficiency markers are added to the specific content of the adjustment scheme, treatment scheme, and pest and disease prevention strategy.

5. The AI-based crop growth management decision-making system according to claim 4, characterized in that: The steps for generating treatment plans and prevention strategies in the treatment plan generation module are as follows: Retrieve the highest-scoring treatment plan or highest-scoring prevention strategy stored in the database for the corresponding category, and read the type marked in the highest-scoring treatment plan or highest-scoring prevention strategy; Adjust the proportion of the drug in the highest-scoring treatment plan or prevention strategy according to the type of label. The proportion of adjustment is μ. Summarize the data after adjustment to obtain the treatment plan or pest prevention strategy.

6. The AI-based crop growth management decision-making system according to claim 5, characterized in that: The steps to generate the adjustment plan are as follows: Retrieve the highest score adjustment scheme stored in the database for the corresponding category, and read the type of the marker in the highest score adjustment scheme; Adjust the soil element data recorded in the highest-scoring scheme by up or down according to the type of marker, with the adjustment ratio being μ. Summarize the soil element data after the adjustment and down to generate soil element scheme data. Calculate the difference between the soil element data in the proposed scheme and the actual soil element data, summarize the difference data, and generate an adjustment scheme.

Images