Farming approval management method based on deep learning
Through the agricultural approval management method based on deep learning, the agricultural approval process is automatically handled, and the problem of inefficient manual approval in traditional agricultural management is solved, and more efficient, accurate and transparent approval management is achieved.
Patent Information
- Application Number
- CN202411476427.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2025-02-18
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In traditional agricultural management, manual approval and supervision are inefficient, resulting in improper management of resources and funds, and lack of transparency and traceability.
Adopt deep learning-based agricultural approval management methods, and by collecting and cleaning historical approval data, establishing prediction models to automate the approval process, and improving the accuracy and transparency of approvals.
Significantly accelerate the approval speed, improve the accuracy and transparency of approval, reduce long-term operation costs, and reduce losses caused by incorrect approvals.
Smart Images
Figure CN119477192A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an agricultural approval management method, in particular to an agricultural approval management method based on deep learning, and belongs to the technical field of smart agriculture. Background Art
[0002] In traditional agricultural management, the problem of "leakage" is usually handled by manual approval and supervision, which not only consumes a lot of human resources, but also may lead to inadequate supervision due to insufficient manpower, resulting in the loss, waste or improper management of resources and funds in the process of allocation and use. In addition, manual processing of a large number of approval requests is usually inefficient, especially in peak seasons, the approval process may become procrastinated. By implementing a strict approval mechanism and strengthening supervision and management, we can ensure the effective use of agricultural resources and funds, reduce losses and improve efficiency. The use of computer technology can automate the processing of a large amount of data and approval requests, significantly improve processing speed and efficiency, and accurately process and analyze data to ensure the consistency of approval standards and reduce human errors. Decision tree algorithms and data analysis tools in some machine learning can also be used to make approval decisions based on a large amount of real-time and historical data to improve the accuracy of decisions. In addition, the use of computer technology also ensures the transparency and traceability of approval, and all approval processes are recorded on the computer for easy tracking and verification. Although the initial investment may be high, in the long run, the automation and computerization process can reduce labor costs and improve resource utilization efficiency.
[0003] Therefore, the introduction of computer technology can not only solve many limitations of traditional methods, but also bring fundamental changes to modern agricultural management through efficient and precise processing methods. This technological innovation helps to achieve more sustainable and transparent agricultural production and management methods, thereby improving the productivity and reliability of the entire agricultural sector and promoting sustainable agricultural development. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide an agricultural approval management method based on deep learning.
[0005] To achieve the above object, the present invention adopts the following technical solution:
[0006] A method for agricultural approval management based on deep learning, comprising the following steps:
[0007] Step 1: Collect data: Collect the approved and unapproved applications in the historical approval process to build a data set; the data set includes the submitter, crop number, total acres of the field, acres of the field that meet the trajectory standards, weather code, cost, and application approval amount;
[0008] Step 2: Data cleaning: It consists of the following specific steps:
[0009] Step 2-1: Identify and handle missing values: Check the missing values in the data set, mark the missing values with preset symbols, and delete the rows containing missing values if the missing value ratio is less than the preset ratio; if the missing value ratio is greater than the preset ratio, fill the missing values with the mean of the corresponding data in the data set, or fill them with the corresponding values at the previous moment;
[0010] Step 2-2: Identify outliers: First, sort the total acres of fields, the number of acres that meet the trajectory standards, the weather number, the cost, and the application approval amount from small to large, find the 25% position in the sorted data as the first quartile Q1, and the 75% position in the sorted data as the third quartile Q3, calculate the interquartile range IQR = Q3-Q1, determine the lower limit of the outlier as Q1-1.5×IQR, and the upper limit of the outlier as Q3+1.5×IQR. Data points that are less than the lower limit or greater than the upper limit are considered outliers;
[0011] Step 2-3: Handle outliers: Use normal values before and after to replace outliers;
[0012] Step 3: Data conversion: Use Z-score to normalize the total acres of fields, acres of track compliance, weather number, cost, and application approval amount in the data set; the Z-score calculation method is:
[0013]
[0014] Among them, mean and std are the mean and standard deviation of the corresponding data respectively; x is the data value of each feature;
[0015] Step 4: Dataset division: Use Bootstrap sampling method to divide the data set into training set and test set; randomly select a sample from the data set, copy it and put it into the training set, repeat the operation m times, and the samples in the data set that do not appear in the training set constitute the test set;
[0016] Step 5: Build a prediction model: The probability of approval is:
[0017]
[0018] A represents the approval event, β0, β1, β2, β3, β4, β5, β6 are model parameters, Crop is the crop code, Total A is the total number of acres of land, Target A is the number of acres of track that meet the standard, W is the weather number, Cost is the cost, and TotalAmount is the amount of application approval;
[0019] Use the data in the training set to train the prediction model parameters β0, β1, β2, β3, β4, β5, β6 until the preset probability of passing the approval is reached or the preset number of training times is reached;
[0020] Step 6: Predict the probability of approval: Use the prediction model to predict the probability of approval. If the probability of approval is greater than the preset value, it will prompt that the approval has been passed, otherwise it will prompt that the approval has not been passed.
[0021] The beneficial effects of the present invention are:
[0022] 1. By automating the approval process, the approval speed is significantly accelerated, and the decision tree algorithm and standardized process are used to improve the accuracy of approval and reduce human judgment bias.
[0023] 2. Record the approval process in detail, provide clear approval logic, and increase the transparency and traceability of the process.
[0024] 3. Automated processes reduce long-term operating costs and reduce losses caused by erroneous approvals. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0026] The following examples serve to illustrate the invention.
[0027] Example 1
[0028] A method for agricultural approval management based on deep learning, comprising the following steps:
[0029] Step 1: Collect data: Collect the submitter, crop, total acres of field, acres of track-compliant acres, weather, cost, and application approval amount from the approved and unapproved approval applications in the historical approval data to build a data set;
[0030] Step 2: Data cleaning: It consists of the following specific steps:
[0031] Step 2-1: Identify and handle missing values: Check the missing values in the data set, mark the missing values with preset symbols, and delete the rows containing missing values if the missing value ratio is less than the preset ratio; if the missing value ratio is greater than the preset ratio, fill the missing values with the mean of the corresponding data in the data set, or fill them with the corresponding values at the previous moment;
[0032] Step 2-2: Identify outliers: First, sort the total acres of fields, acres of track-compliant acres, weather, cost, and application approval amount from small to large, find the 25% position in the sorted data as the first quartile Q1, and the 75% position in the sorted data as the third quartile Q3, calculate the interquartile range IQR = Q3-Q1, determine the lower limit of the outlier as Q1-1.5×IQR, and the upper limit of the outlier as Q3+1.5×IQR. Data points that are less than the lower limit or greater than the upper limit are considered outliers;
[0033] Step 2-3: Handle outliers: Use normal values before and after to replace outliers;
[0034] Step 3: Data conversion: Use Z-score normalization to convert all sensor data into a consistent format and scale; first, calculate the mean and standard deviation of the total acres of fields, acres of track compliance, weather, cost, and application approval amount data. ； Then, use the formula Calculate the Z score of each feature data point, where y represents the Z score after the data x is transformed; x is the original data;
[0035] Step 4: Dataset division: Use Bootstrap sampling method to divide the data set into training set and test set; randomly select a sample from the data set, copy it and put it into the training set, repeat the operation m times, and the samples in the data set that do not appear in the training set constitute the test set;
[0036] Step 5: Build a prediction model: The probability of passing the approval is:
[0037]
[0038] A represents approval, β0, β1, β2, β3, β4, β5, β6 are model parameters, Crop is the crop, TotalA is the total acres of the field, Target A is the number of acres of track that meet the standard, W is the weather, Cost is the cost, and TotalAmount is the amount of application approval;
[0039] Use the data in the training set to train the prediction model parameters β0, β1, β2, β3, β4, β5, β6 until the preset probability of passing the approval is reached or the preset number of training times is reached;
[0040] Step 6: Predict the probability of approval: Use the prediction model to predict the probability of approval. If the probability of approval is greater than the preset value, it will prompt that the approval has been passed, otherwise it will prompt that the approval has not been passed. If the approval application also contains pictures of contracts, transaction vouchers, invoices, etc., the person in charge of approval can further determine whether it has been approved.
[0041] At present, the technical solution of the present invention has been piloted, and a user survey has been conducted in a small range. The survey results show that user satisfaction is high, and preparations for the transformation and promotion of the results have begun. At the same time, a survey on intellectual property risk warning has been carried out.
Claims
1. A method for agricultural approval management based on deep learning, characterized in that: The following steps are involved: Step 1: Collect data: Collect the approved and unapproved applications in the historical approval process to build a data set; the data set includes the submitter, crop number, total acres of the field, acres of the field that meet the trajectory standards, weather code, cost, and application approval amount; Step 2: Data cleaning: It consists of the following specific steps: Step 2-1: Identify and handle missing values: Check the missing values in the data set, mark the missing values with preset symbols, and delete the rows containing missing values if the missing value ratio is less than the preset ratio; if the missing value ratio is greater than the preset ratio, fill the missing values with the mean of the corresponding data in the data set, or fill them with the corresponding values at the previous moment; Step 2-2: Identify outliers: First, sort the total acres of fields, the number of acres that meet the trajectory standards, the weather number, the cost, and the application approval amount from small to large, find the 25% position in the sorted data as the first quartile Q1, and the 75% position in the sorted data as the third quartile Q3, calculate the interquartile range IQR = Q3-Q1, determine the lower limit of the outlier as Q1-1.5×IQR, and the upper limit of the outlier as Q3+1.5×IQR. Data points that are less than the lower limit or greater than the upper limit are considered outliers; Step 2-3: Handle outliers: Use normal values before and after to replace outliers; Step 3: Data conversion: Use Z-score to normalize the total acres of fields, acres of track compliance, weather number, cost, and application approval amount in the data set; the Z-score calculation method is: Among them, mean and std are the mean and standard deviation of the corresponding data respectively; x is the data value of each feature; Step 4: Dataset division: Use Bootstrap sampling method to divide the data set into training set and test set; randomly select a sample from the data set, copy it and put it into the training set, repeat the operation m times, and the samples in the data set that do not appear in the training set constitute the test set; Step 5: Build a prediction model: The probability of approval is: A represents the approval event, β0, β1, β2, β3, β4, β5, β6 are model parameters, Crop is the crop code, TotalA is the total acres of the field, TargetA is the acres of the trajectory that meet the target, W is the weather number, Cost is the cost, and TotalAmount is the amount of application approval; Use the data in the training set to train the prediction model parameters β0, β1, β2, β3, β4, β5, β6 until the preset probability of passing the approval is reached or the preset number of training times is reached; Step 6: Predict the probability of approval: Use the prediction model to predict the probability of approval. If the probability of approval is greater than the preset value, it will prompt that the approval has been passed, otherwise it will prompt that the approval has not been passed.