Crop yield prediction management system

Abstract

The invention relates to the technical field of resource planning, in particular to a crop yield prediction management system, which comprises a farming task prediction module for acquiring crop types, planting areas and weather forecast information, calculating crop yield and establishing expected crop yield data. According to the invention, the expected yield is finely predicted by obtaining the crop type, the planting area and the weather forecast information, the agricultural operation is identified in combination with the preset agricultural plan, and the specific task type and task list are generated, so that the relation between yield prediction and task management is established, and the cooperation precision of crop yield expectation and farm task arrangement is improved. The blindness of farming operation is reduced; according to the method, the task list and the labor skill database are connected, dynamic adaptation of the worker skill and the task requirement is achieved, a worker task distribution pool is formed according to the skill adaptation level and the time matching degree, and precision and details are achieved on the personnel management level.

Description

Technical Field

[0001] The present invention relates to the technical field of resource planning, and in particular, to a crop yield prediction management system. Background Art

[0002] The technical field of resource planning is a technical field that comprehensively plans, rationally allocates, and precisely configures various production factors such as human, material, and financial resources by using data analysis, prediction models, optimization algorithms, and information technology.

[0003] In the prior art, in human resource management, resource allocation is mostly based on experience and existing task lists, lacking the analysis of skill matching degree and working hour constraint conditions, often resulting in asymmetry between personnel allocation and task requirements, and then causing the situation of resource idleness or over - utilization; in the process of task arrangement, there is a lack of careful control of real - time task costs and strict inspection of working hour constraints, often causing problems such as budget over - run and low operation efficiency. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a crop yield prediction management system.

[0005] To achieve the above - mentioned purpose, the present invention adopts the following technical solutions: A crop yield prediction management system includes: An agricultural task prediction module, which obtains crop type, planting area, and weather forecast information, calculates the crop yield, and establishes expected crop yield data; associates the expected crop yield data with a preset agronomic plan, identifies necessary agricultural operations, determines the corresponding task types of the operations, and generates a predicted agricultural task list; A labor skill matching module, based on the predicted agricultural task list and the labor database, retrieves the skills, working hours, and time windows required for the tasks, compares the skills required for the tasks with the skills mastered by the labor force, and establishes a task - worker skill adaptation set; based on the task - worker skill adaptation set, matches the task time windows according to the available time of the labor force, calculates the allocation fitness score, and screens to form a qualified worker task allocation pool; A dynamic scheduling module, based on the qualified worker task allocation pool and the predicted agricultural task list, selects workers to pair with tasks, establishes a preliminary scheduling allocation plan, analyzes the task completion timeliness and total labor cost of the preliminary scheduling allocation plan, adjusts the allocations with conflicts or exceeding constraints, and generates a farm dynamic operation schedule; The insurance resource allocation module estimates the amount of financial losses under different yield scenarios based on the expected crop yield data. According to the amount of financial losses, it compares the costs and coverage levels of agricultural insurance products with the farm insurance budget amount, and selects a combination of agricultural insurance products and coverage levels for different plots or crops to obtain an agricultural insurance portfolio.

[0006] Preferably, the steps for obtaining the expected crop yield data are as follows: Based on the planting crop number parameter mapped by the crop type, retrieve the corresponding reference output value per unit area in the crop variety matching table. Combine the planting area, growth stage interval, and stage time series input by the user, and average and summarize the temperature, precipitation, wind speed, sunshine duration, and relative humidity by stage to establish a stage crop environmental factor input matrix. Calculate the stage yield prediction value according to the stage crop environmental factor input matrix. According to the stage yield prediction value, multiply and accumulate it with the allocated area of each stage to generate the total output value of all stages combined, and obtain the expected crop yield data.

[0007] Preferably, the steps for obtaining the predicted farm work task list are as follows: Based on the different crop types and stage yield prediction values in the expected crop yield data, match the agronomic plan task template according to the crop type. Combine the stage time index of the yield data with the task start interval in the template, and compare the task start time point with the yield formation time correspondingly. Screen and match the task records with established start conditions in the agronomic plan to generate a screening result of agronomic plan tasks related to crop yield. According to the screening result of agronomic plan tasks related to crop yield, extract the configured farm work operation content in the triggered task records item by item. Call the job number in the task record to obtain the task type, required skills, estimated working hours, and executable time window parameters of each farm work operation, and establish a structured set of farm work task attributes. Based on the structured set of farm work task attributes, perform similarity judgments on different farm work operations in terms of task type, required skills, and time window dimensions. Call the repeated or overlapping task records, and merge and organize the operation items in the order of crop number and time window to generate a predicted farm work task list.

[0008] Preferably, the steps for obtaining the task worker skill adaptation set are as follows: Based on the task type, required skill codes, required skill levels, estimated execution man-hours, and task time window parameters of each task record in the predicted farm task list, retrieve the worker numbers, mastered skill codes, skill level values, hourly labor costs, calendar available time periods, and worker type identifiers included in the labor database. Pre-screen the workers who have the skill codes and whose skill levels are not lower than the required skill levels, and screen out the workers with non-overlapping time windows to generate a candidate list of task workers with preliminary allocable qualifications; According to the candidate list of task workers with preliminary allocable qualifications, calculate the skill adaptation value for each worker's corresponding task; Based on the skill adaptation values between each worker and the task, screen the workers to generate a skill adaptation set of task workers.

[0009] Preferably, the steps for obtaining the qualified worker task allocation pool are as follows: Based on the combination of each task number and worker number in the skill adaptation set of task workers, retrieve the daily work schedule, historical completed task numbers, skill adaptation values, start and end times of the task time window, and whether there are continuous idle sections for each worker. Screen out the matching records that do not meet the conditions for continuous allocation or have task conflicts in the past three days to obtain an execution mapping set of task workers with allocation candidate qualifications; According to the execution mapping set of task workers with allocation candidate qualifications, calculate the allocation fitness score; Based on the allocation fitness scores between each task and the worker, extract the worker numbers that meet the conditions for each task, classify them by task number to generate a candidate set of workers, and generate a qualified worker task allocation pool.

[0010] Preferably, the steps for obtaining the preliminary scheduling and allocation plan are as follows: Based on the task numbers, worker numbers, start and end times of the task time window, single-task estimated execution man-hours, and daily remaining available man-hour limits of each worker in the qualified worker task allocation pool, call the task priority parameters and task corresponding time window density values in the predicted farm task list to generate a pre-group set of schedulable task workers; According to the pre-group set of schedulable task workers, calculate the scheduling priority index; Based on the scheduling priority index, sort all workers by task number and select the one with the highest scheduling priority index for time period filling. Combine the daily total man-hour limit of the worker and the total cost limit of the farm to perform overrun screening to generate a preliminary scheduling and allocation plan.

[0011] Preferably, the steps for obtaining the farm dynamic operation schedule are as follows: Based on the task number, worker number, start and end times of the task time window, worker idle interval distribution code, and task scheduling interval code in each scheduling record of the preliminary scheduling assignment plan, determine whether there are continuous non-assignable or overlapping conflict situations, summarize the conflicting records, and generate a scheduling exception record set; According to the scheduling exception record set, calculate the assignment constraint strength factor for each record; Based on the assignment constraint strength factor, sort all worker records in ascending order of the assignment constraint strength factor by task number, and perform worker replacement and time period shift for the records with lower priority to generate a farm dynamic operation schedule.

[0012] Preferably, the steps for obtaining the agricultural insurance portfolio are as follows: Based on the various crop types, corresponding plot numbers, and estimated yields in the expected crop yield data, judge the market prices of the crops corresponding to the plots at different yield levels, calculate the difference between the crop income and the expected income at different yield levels step by step, and generate the financial loss amounts under different yield scenarios; According to the financial loss amounts under different yield scenarios, respectively retrieve the unit cost, loss coverage ratio, and insured amount range of each product in the agricultural insurance product database, compare and screen the product unit cost with the budget amount item by item in accordance with the farm insurance budget amount, and match the loss amount with the insured amount range of the insurance product one by one to generate a screening and matching result of insurance product cost coverage; According to the screening and matching result of insurance product cost coverage, map the corresponding plot numbers and crop types to the agricultural insurance products that meet the cost constraint and insured amount range requirements item by item, determine the agricultural insurance products matched by each plot or crop, and generate an agricultural insurance portfolio.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by obtaining crop types, planting areas, and weather forecast information, the expected yield is accurately predicted. By combining a preset agronomic plan, farming operations are identified, specific task types and task lists are generated, and thus the connection between yield prediction and task management is established, improving the coordination accuracy between crop yield prediction and farm task arrangement and reducing the blindness of farming operations. By docking the task list with the labor skill database, the dynamic adaptation of worker skills to task requirements is achieved. Based on the skill adaptation level and time matching degree, a worker task allocation pool is formed, achieving precision and detail in personnel management. Based on the worker task allocation, a shift scheduling and allocation plan is established, and real-time analysis and adjustment are carried out according to the timeliness of task completion and labor costs to ensure the rationality of operation arrangements and the feasibility of execution, reducing the risks of labor waste and cost overruns caused by improper shift scheduling. Based on multi-scenario analysis of financial loss estimation, comparison of insurance products, and matching with budget amounts, the optimal insurance combination is configured for different crops or plots, improving the agricultural risk prevention and control ability, reducing the economic loss risk caused by crop yield fluctuations, and enhancing the economic security of agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a system flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] In order to make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0016] Please refer to Figure 1 , the present invention provides a technical solution: A crop yield prediction and management system includes: A farming task prediction module, which obtains crop types, planting areas, and weather forecast information, calculates the crop yield, and establishes expected crop yield data; associates the expected crop yield data with a preset agronomic plan, identifies necessary farming operations, determines the corresponding task types of the operations, and generates a predicted farming task list; A labor skill matching module, based on the predicted farming task list and the labor database, retrieves the skills, working hours, and time windows required for the tasks, compares the skills required for the tasks with the skills mastered by the labor force, and establishes a task-worker skill adaptation set; based on the task-worker skill adaptation set, according to the available time of the labor force to match the task time window, calculates the allocation fitness score, and screens to form a qualified worker task allocation pool; The dynamic shift scheduling module selects workers and tasks for pairing based on the qualified worker task allocation pool and the predicted farm task list, establishes a preliminary shift allocation plan, analyzes the task completion timeliness and total labor cost of the preliminary shift allocation plan, adjusts the allocations with conflicts or those exceeding the constraints, and generates a dynamic farm work shift schedule. The insurance resource allocation module estimates the amount of financial losses under different yield scenarios based on the expected crop yield data, compares the costs, coverage levels of agricultural insurance products with the farm insurance budget amount according to the amount of financial losses, and selects a combination of agricultural insurance products and coverage levels for different plots or crops to obtain an agricultural insurance portfolio.

[0017] The steps for obtaining the expected crop yield data are as follows: Based on the planting crop number parameter mapped by the crop type, retrieve the corresponding reference output value per unit area in the crop variety matching table, and combine the planting area, growth stage interval, and stage time series input by the user. Average and summarize the temperature, precipitation, wind speed, sunshine duration, and relative humidity by stage to establish a stage crop environmental factor input matrix. According to the stage crop environmental factor input matrix, calculate the stage yield prediction value, and the expression is: ; where is the predicted yield value for the stage, is the regression intercept term, , , , , are all regression coefficients respectively, is the daily average temperature, is the total precipitation, is the average wind speed, is the cumulative sunshine duration, is the stage average relative humidity; According to the stage yield prediction value, multiply and accumulate it with the allocated area of each stage to generate the total output value of all stages combined, and obtain the expected crop yield data.

[0018] Specifically, based on the parameter content of the planting crop number mapped by the crop type, first retrieve the unit area reference output value matching the crop number from the corresponding crop variety matching table, and associate the retrieved unit area reference output value with the planting area parameter submitted by the user, and then extract the five types of environmental information, namely temperature, precipitation, wind speed, sunshine duration and relative humidity, one by one according to the time period corresponding to the growth stage interval and the stage time series. All environmental information is obtained through the measurement equipment deployed in the field in the early stage. These devices can record the temperature value, cumulative precipitation, air flow rate, sunshine duration and air humidity of a single plot within a fixed time interval. All records are compared with the established valid range, for example, the temperature data corresponds to the range of -30℃ to 50℃, the precipitation data corresponds to the range of 0 mm to 1000 mm, the wind speed data corresponds to the range of 0 meters per second to 20 meters per second, and the sunshine duration data corresponds to 0 hours to 400 hours. Range, corresponding to the range of 0% to 100% for air humidity, only records that meet the range are marked as valid data, and these valid data are summarized in chronological order and organized by stage. The stage division is based on the growth cycle nodes determined by the user. For example, the interval between adjacent nodes is set to 10 days or 15 days. According to different stages, temperature, precipitation, wind speed, sunshine duration and air humidity are statistically summarized in the same time period and recorded as the stage average or cumulative amount, so as to be placed in the same data structure together with the planting area parameters of the stage. Then, the five types of environmental factors are uniformly arranged in a two-dimensional matrix with the stage number, plot number and crop type, where the row index represents the stage number, and the column index includes five fields of temperature, precipitation, wind speed, sunshine duration and relative humidity, as well as other auxiliary fields. A completeness check is performed on this matrix to eliminate row records with missing values ​​or non-compliant values, and after the check is completed, a stage crop environmental factor input matrix is ​​formed.

[0019] formula: The benefit of the formula is that by incorporating the influence of five environmental factors, namely, average daily temperature, total precipitation, average wind speed, cumulative sunshine duration, and average relative humidity of the stage, and coordinating the overall balancing effect of the regression intercept term, it is possible to make a refined estimate of the stage yield and avoid dependence on a single environmental factor. Combined with the scientific setting of each regression coefficient, a more representative stage yield forecast value can be obtained under the joint action of multi-dimensional factors.

[0020] is the regression intercept term, representing the baseline value of the yield when all environmental factors are zero. This value is obtained through linear regression or multiple regression on historical observation data of the same crop variety. The acquisition steps include sorting out a large dataset with records of yield and environmental factors, matrix processing the environmental fields such as temperature, precipitation, wind speed, sunshine duration, and humidity in each complete record, setting the yield field as the dependent variable, and then performing parameter estimation using the least squares method or the gradient descent method. Taking the least squares method as an example, first construct the matrix to record all environmental factors and the constant column 1, and then construct the vector to record the corresponding yield values. Let be the regression coefficient vector, then is the standard estimation formula, where is the first component of the vector . Subsequently, perform necessary statistical tests and residual analysis on , confirm that there are no significant anomalies, and then fix it as the baseline intercept value in subsequent production scenarios. In an actual example, through the regression analysis of 600 growth data over five consecutive years, has been calculated to be 3.50.

[0021] is the regression coefficient of the impact of temperature on yield, obtained through the multiple regression estimation process, indicating the gain or loss of the yield value when the temperature rises or falls by 1 unit within a reasonable range. To obtain this coefficient, first extract the daily average temperature corresponding to each growth stage and the stage yield measured during the current period from the historical data, and then construct a multi-dimensional sample in combination with the values of other factors, and use the same least squares estimation method as above to obtain 's value, and check its significance level and regression residual distribution. If this coefficient is significantly different from 0 statistically, it can reflect the important impact of temperature on yield. In a specific example, by summarizing 840 records in a certain area over the past seven years, is obtained. When the daily average temperature rises by 1℃, the stage yield increases by an average of 0.08 kg.

[0022] is the regression coefficient of the impact of precipitation on yield. This value needs to be fitted between the cumulative precipitation and yield data. The methods include least squares regression or other optimization regression means. The steps are similar to those for obtaining . First, extract the precipitation field from the multi-dimensional environmental factor dataset, and construct an equation system in combination with the yield field, and use to obtain . In the actual dataset, a certain experimental field obtained after multiple precipitation monitoring.

[0023] is the regression coefficient of the impact of average wind speed on yield, and its calculation is also based on the multiple regression method. After organizing the wind speed data into the corresponding fields, it is input into the regression equation together with other environmental factors. In the result, is usually positive or negative, depending on the beneficial or harmful impact of wind speed during the actual growth stage, and is calculated using the least squares method. When calculating, first form a vector for each complete record. with yield as the corresponding scalar. Then perform matrix inversion and calculation to obtain this coefficient. In a certain example dataset, is 0.03, with the unit of kilograms per meter per second, meaning that for every 1-meter-per-second increase in wind speed, the stage yield increases by approximately 0.03 kilograms.

[0024] is the regression coefficient of the impact of sunshine duration on yield, which is used to reflect the contribution of photosynthesis time to the final yield. The acquisition method also adopts incorporating the sunshine duration field into the least squares regression model. When other environmental factors remain unchanged, for every 1-hour increase in sunshine duration, the yield gain is reflected by The parameter estimation formula is still the same as described above. Treat the sunshine duration as a column of independent variables and incorporate it into the matrix and then solve. Then regard its corresponding component as Through the database, 1200 sunshine measurement records from different plots are integrated for regression to obtain .

[0025] is the regression coefficient of the impact of the stage average relative humidity on yield. Its meaning is the change in the yield value when the humidity value changes by 1% while other factors are fixed. The acquisition process continues to use the multiple regression method, performing matrix calculations on the matching records of historical humidity and yield, and then extracting the value by and combining statistical tests to confirm the credibility and significance level. After analyzing 520 pairs of humidity and yield records, .

[0026] is the average daily temperature, that is, the average value obtained by summing the measured temperatures for each day during the stage and then dividing by the number of days. To avoid including missing data in the statistics, it is necessary to screen the temperature measurement records according to the previously established valid range, such as excluding invalid records, and then taking the arithmetic mean of the remaining temperature data to obtain , the unit of the value is degrees Celsius. The acquisition process usually requires an independent temperature recorder to conduct 24-hour temperature detection in the farmland, store the collected data per hour into a table, calculate the daily average value sequence within each stage, and summarize it to form a staged indicator , taking an example to illustrate: In the k = 5 stage, referring to the records of the temperature field in the database for 10 days, first check one by one whether the temperature range is within -30°C to 50°C, and then sum up the temperatures of the remaining 10 days. If the sum value is 256°C, the average value for 10 days is 25.6°C, and this 25.6 is the .

[0027] is the total precipitation, referring to the total amount of precipitation accumulated within the stage. In the agricultural production scenario, a rain gauge or an automated precipitation detection device is often used to obtain the precipitation data for each day within the stage, and after summing it up, it is recorded as . The acquisition steps are to first query the precipitation data list in the database between the start date and the end date of the k stage. If the start time of a certain stage is from the 20th day to the 30th day, and the sum of the precipitation records within 10 days is 210 millimeters, then at this time .

[0028] is the average wind speed, referring to the average value of the wind speed monitoring values within the stage. The acquisition methods include setting up an anemometer in the farmland for continuous monitoring, and then statistically averaging the wind speed data after the stage ends. In specific operations, record the wind speed readings per hour or every half hour. Taking an example: For a 7-day stage, if 168 wind speed readings are observed and the sum of all valid readings is 392 meters per second, then meters per second.

[0029] is the cumulative sunshine duration, referring to the total amount of sunshine hours accumulated within the stage. By installing a light sensor and combining with a timestamp to determine whether it meets the light threshold to count the sunshine time period. The specific acquisition steps are to first establish a light detection device in each plot and cross-verify it with the data of the local weather station. After reading, include the time periods that meet the light intensity not lower than a certain set standard in the statistics, and sum up all the time periods that meet the standard in hours within the k stage. If the k stage spans 15 days and the sum of all the durations that meet the light standard is 100 hours, then at this time .

[0030] is the stage average relative humidity. The relative humidity data is directly read from the humidity sensors placed in the plots, and the average value of the humidity values for all days can be obtained after the stage ends When the humidity sensor records the value every 30 minutes, a large number of observation entries can be accumulated in a single stage. Add up all the humidity records within the range of 0% to 100%, and then divide by the total number of valid records. If 720 humidity data are collected in stage k and the sum of the values is 50400, then %.

[0031] Calculation and derivation process: The following gives an actual example. Substitute the aforementioned regression coefficients into the calculation process to obtain the predicted yield of the stage numerical result.

[0032] First step, select the regression coefficients: ; Second step, select the actual monitored values of the stage environmental factors: ; Third step, substitute into the formula ; Fourth step, expand the calculation step by step: 1) Calculate , 2) Calculate , 3) Calculate , 4) Calculate , 5) Calculate .

[0033] Fifth step, add the above partial sum to : ; Sixth step, combine to get ; The result shows that under the current parameter configuration, the The predicted yield value for the stage is 13.14 kg. When this value is in a higher range, it indicates that the stage has good environmental conditions and there is a certain growth space for stage yield prediction. If the stage yield value is below certain critical points, it prompts the need to verify environmental factors or the cultivation management process in order to continue calculating the total yield value of all stages in subsequent links.

[0034] According to the multiplication accumulation logic between the aforementioned stage yield prediction value and the allocated area of each stage, it is necessary to lock the yield result of each stage by comparing the stage numbers one by one. Then, read out the land area value corresponding to this stage. During the execution process, find the corresponding relationship between the stage and the land from the land management record. Through the mapping of the land number, know how much land area is covered by the current stage. Then multiply the yield prediction value by this area to obtain the quantitative result of the stage yield for the corresponding stage, and accumulate the results into the storage variable of the total yield in sequence. During the whole process, it is necessary to check whether the relationship between each land record and the stage number is in the same growth cycle. The definition of the growth cycle comes from the previous division of the planting plan. For example, a certain piece of land has completed two stages in the first growth cycle, while another piece of land may still be in the first stage. All lands need to be managed according to strict numbering to ensure that the area corresponds one-to-one with the stage yield value. Then, when accumulating the product results, check the storage precision and keep the accumulated result within the floating-point operation precision requirements. Perform a difference check for each accumulation, and check the difference range with reference to the range from 0 to tens of thousands or even hundreds of thousands of kilograms. If the accumulated result of a certain time exceeds the reasonable prediction range, mark it at the record level for secondary inspection. The inspection methods include checking whether the stage yield prediction value of this stage is in the normal range, such as between 0 kg and 30,000 kg, and whether the land area of this land is consistent with the previously registered area and falls within the range of 0 to 500 mu or more. After confirming compliance, continue to accumulate until the final completion to obtain the total yield value of all stages. Mark the result obtained by this accumulation as the final stage yield data, and then make a simple comparison with the historical data of the same land. If the comparison value is within the acceptable range, it is officially output as the expected crop yield data.

[0035] The steps to obtain the predicted farming task list are as follows: Based on the different crop types and stage yield prediction values in the expected crop yield data, match the agronomic plan task template according to the crop type. Combine the stage time index of the yield data with the task start interval in the template, and correspondingly compare the task start time point with the yield formation time. Screen and match the task records in the agronomic plan where the start conditions are met to generate the screening result of the crop yield-related agronomic plan tasks; According to the screening results of agronomic plan tasks associated with crop yields, extract item by item the farming operation content configured in the triggered task records, call the job numbers in the task records, obtain the task types, required skills, estimated working hours, and executable time window parameters for each farming operation, and establish a structured set of farming task attributes; Based on the structured set of farming task attributes, perform similarity judgments on different farming operations in terms of task type, required skills, and time window dimensions, call the repeated or overlapping task records, and merge and organize the operation items in the order of crop numbers and time windows to generate a predicted list of farming tasks.

[0036] Specifically, based on the expected crop yield data and stage yield prediction values, combined with the data indexes of different crop types and the matching content between crop types and agronomic plan task templates, use the time series labels of each crop type as a comparison basis, and match the specific time points when the same type of crop forms yields at each growth stage according to the task start intervals listed in the template. Refer to the yield formation dates of each stage and the start conditions recorded in the task template, and compare the condition descriptions with the actual observed values item by item. For example, check whether the yield indicators exceed the stage benchmark values of specific crops or meet the numerical interval requirements. These benchmark values are obtained by statistically analyzing the historical yield records of the same variety of crops under similar climate and soil conditions, and a corresponding benchmark coefficient is set to define the minimum and maximum yield values for tasks that can be started. The benchmark coefficient is calculated from historical statistics. First, average the yield data of each past growth stage of all the same variety of crops, and then set thresholds by comparing the three quantiles of large, medium, and small. For example, when the historical average yield value of a certain crop in the third stage is 450 kg, 420 kg can be set as the lower threshold and 480 kg as the upper threshold. If the actually monitored yield is between 420 and 480 kg, the start condition is met. In this process, it is necessary to compare the corresponding plot numbers and crop numbers of each record to confirm that the crop type matches the records in the template, and then mark the records that meet the conditions and record the corresponding time indexes, so as to output these screening results to the next implementation step, and finally obtain the screening results of agronomic plan tasks associated with crop yields.

[0037] Based on the screening results of the agronomic plan tasks associated with crop yields obtained from the previous paragraph, break down the triggered farming operation contents item by item. First, extract the operation numbers from the determined task records, and then read the task types, required skills, estimated working hours, and executable time windows and other information item by item from the corresponding farming operation descriptions. Here, the skill requirements need to be clearly listed. For example, whether it is necessary to drive agricultural machinery or operate sowing equipment, etc. Each skill has a specific numerical identifier in the skill library. If the skill number is S101, it represents basic planting skills, and if it is S202, it represents agricultural machinery operation skills, corresponding to different personnel matching schemes. At the same time, in this process, compare the established workload intervals in combination with the estimated working hour information. For example, when the estimated working hours are determined to be within the range of 0 to 12 hours, it is classified as a small-batch operation. If it exceeds 12 hours, it is recorded as a large-batch operation and different manpower scheduling strategies are arranged in the subsequent stage. Mark the executable time window in the same structured record. The upper and lower limits of this time window are determined by the statistical evaluation of the local climate and field topography. For example, in areas where excessive waterlogging is not likely to occur, the executable time window can be extended to more days, while in plots with insufficient drainage capacity, the time window needs to be tightened and corresponding time thresholds are set. When further allocating tasks, compare and combine these windows with the skill numbers. Finally, combine the relevant attributes of all farming operations into a structured set of farming task attributes.

[0038] Based on the structured set of farming task attributes established above, compare different farming operations one by one in three dimensions: task type, required skills, and time window. Identify the records with the same task type, the same skill requirements, and similar executable time periods, and then confirm whether they belong to duplicate or overlapping tasks. The identification logic usually first reads the task type field. If both records are of the harvesting type, it is considered that there is a similarity in task type. At the same time, check whether the required skill numbers are exactly the same. If both require S202, it is considered that they also match exactly in terms of skills. Then, compare the start and end dates of the executable time windows of these two records. If the overlapping days of the windows are greater than the set threshold value, it is determined that there are overlapping tasks. This threshold value can be set according to past work statistics, such as 3 days or 5 days, and set reasonably according to the degree of differentiation of field data. If it is greater than this threshold value, it is considered an overlapping task, and a merge operation is performed in the subsequent processing. When merging, it is necessary to ensure that the records with the same crop number appear in the same sequence first, and then arrange the time window distributions in sequence. After these steps are completed, write the merged and sorted operation items into the final result set to obtain the predicted farming task list.

[0039] The steps to obtain the task worker skill adaptation set are as follows: Based on the task type, required skill code, required skill level, estimated execution man-hours, and task time window parameters recorded in the predicted farming task list, retrieve the worker number, mastered skill code, skill level value, hourly labor cost, calendar available time period, and worker type identifier included in the labor database. Prescreen the workers who possess the skill code and whose skill level is not lower than the required skill level, and eliminate the workers with non-overlapping time windows to generate a candidate list of task workers with preliminary assignable qualifications; According to the candidate list of task workers with preliminary assignable qualifications, calculate the skill adaptation value for each worker's corresponding task. The calculation formula is: ; where, is the skill adaptation value of the th worker for the th task, is the skill level of the skill required for the th worker corresponding to the th task, is the required skill level of the th task, is the estimated execution man-hours of the th task, is the maximum available continuous man-hours of the th worker within the task time window, is the current number of tasks assigned to the th worker, is the hourly labor cost of the th worker, is the current consecutive working days of this worker; Based on the skill adaptation value between each worker and the task, screen the workers to generate a task worker skill adaptation set.

[0040] Specifically, based on the information such as the task types, corresponding skill codes, skill levels, estimated execution man-hours, and task time windows listed in each record of the predicted farming task list, combined with the fields such as worker numbers, mastered skill codes, skill level values, hourly labor costs, available time periods, and worker types pre-recorded in the labor database, compare and determine the skill satisfaction level by matching the skill codes and levels, and compare the estimated man-hours with the available time periods of the workers one by one. For example, first take 0 to 12 hours as one working duration interval, and then take 12 hours to 24 hours as another interval. If the estimated execution man-hours of the target task fall within the current interval and the worker has an available time period within the same interval, record a successful match. Further, conduct an overlap check on the time spans of the task time window and the worker's calendar. During the execution process, first confirm whether the start and end dates of the task are within the available dates of the worker, and strictly compare the time coincidence degree. If the time coincidence degree is greater than 3 days, mark it as continuous workable; if the time coincidence degree is only 2 days or less, mark it as low continuity. At the same time, set a threshold to filter out entries with low continuity. This threshold is based on the average continuous working days obtained from the statistics of the same type of operations in the past three years, and an inductive record forms a minimum continuous days benchmark coefficient. This coefficient can be obtained by summing up all the recorded continuous days and then dividing by the total number of records to get the average value, and then refer to adjacent quantiles to exclude extreme values and set a final value to define whether the continuity requirement is met. List all the matching results that meet the skill and time allocation requirements into the candidate records one by one, and record the corresponding worker numbers, available time periods, and the situation of meeting the skill levels. Finally, during the data integration stage, deduplicate multiple matching entries with the same worker number. If multiple duplicate matches are found within the same time period, only retain the first generated entry and mark the subsequent entries as duplicates, so as to summarize into a candidate list of task workers with preliminary allocable qualifications.

[0041] Formula: ; The benefit of the formula is that it combines multiple factors such as the deviation of the worker's skill level, the difference between the task man-hours and the available man-hours, and the number of assigned tasks into the numerator, and amplifies the cost and fatigue effects through the combination of the hourly labor cost and the logarithmic function of the continuous working days, so as to more accurately measure the matching degree of workers and tasks in terms of skills and time allocation.

[0042] This parameter represents the th worker's actual level value of the skills required for the th task, which is between 1 and 10 and is obtained from the hierarchical monitoring of the worker's operation ability. The hierarchical monitoring process is quantified through phased skill tests or work performance sampling, and the score values of each evaluation session are integrated into Calculate the comprehensive score later. For example, first sum up the four assessment scores, and then divide by 4 to get the average value, denoted as , and then according to the pre-established grade division table, falls into the corresponding section to form the specific skill level value . Taking worker number 005 as an example, the four assessment scores are 80, 85, 82, and 88 respectively. The sum is 335, and dividing by 4 gives 83.75. Looking at the grade table, it may correspond to level 7. Therefore, the of this worker = 7.

[0043] This parameter is the required value of the skill level for the th task, and it is also between 1 and 10. It is uniformly configured in the previous agricultural operation list. The acquisition process is to classify and count the operation difficulty and equipment operation complexity corresponding to this task. The benchmark coefficient method is adopted in the statistics. The corresponding score is determined by observing the average level required for historical similar tasks, and then fine-tuned according to special situations. For example, if the operation of the harvesting equipment is defined as a task with a level above 5, then the current task is marked as level 5.

[0044] This parameter represents the estimated execution man-hours of the th task. It is comprehensively estimated based on information such as equipment, plot area, and crop characteristics in the operation preparation stage. Usually, it is in the range of 0 to 48 hours. It is obtained by checking the operation area and operation method. If it takes 2 hours to harvest per mu of land and this plot has 10 mu, then the estimated execution man-hours of this task can be obtained as 20 hours.

[0045] This parameter refers to the maximum available continuous man-hours of the th worker within the required task time window. The range is around 0 to 12 hours. It is determined by the worker's personal schedule and other previously scheduled work tasks. The acquisition method is to query the worker's shift management records, analyze these records by time period, extract the length of the available interval, and take the intersection within the start and end ranges of this demand. For example, if a worker can work 8 hours per day and has been assigned a 4-hour task previously, then the maximum available continuous man-hours within the current window is 4 hours.

[0046] This parameter is the number of tasks currently assigned to the th worker. The calculation method is to count the total number of tasks that have been successfully assigned and are in the in-progress or to-be-executed state from the worker task assignment table, which is represented by an integer. The higher this value, the heavier the current burden of the worker. To avoid over-scheduling of workers, this value will have an impact on the algorithm. An example of obtaining it is that worker number 002 has received 3 subsequent tasks, so it is equal to 3.

[0047] This parameter is the hourly labor cost of the th worker, which is obtained through a pre-signed labor contract or salary standard. The common range is between 20 yuan and 60 yuan. If the hourly wage standard corresponding to worker number 006 is found to be 35 yuan in the actual query result, then it is equal to 35.

[0048] This parameter represents the number of consecutive working days of the worker, which is used to measure the continuous fatigue degree of the worker. The acquisition method is usually to query the worker's previous clock-in or sign-in records, judge whether there is work every day in the past several days, and then make a cumulative total. Taking worker number 003 as an example, if the worker has been in a working state for 5 consecutive days, then it is equal to 5. If it enters a rest day, this value will be reset to 0 when reallocated.

[0049] Calculation process: Select regression parameters and substitute initial values for calculation examples: First step, let the task information in the , hours; Second step, let the worker 's , hours, , yuan / hour, days; Third step, substitute into the formula: ; Fourth step, calculate item by item: 1) , 2) , 3) , 4) The numerator part , 5) The denominator part , so the denominator = around, 6) Finally .

[0050] This result indicates that the worker 's skill adaptation value for the task is approximately 0.076. When this value is much less than 1, it means that the skill level difference, available working hours deviation, and the number of assigned tasks of this worker are relatively excellent in the overall range, compared with other workers selected later The result comparison can show who is more suitable for the task.

[0051] Based on the skill adaptation values between each worker and the task, screen the workers item by item against the previously compiled list of adaptation scores. Sort the adaptation values of all workers under the same task, select the worker numbers with relatively higher sorting results, and record their corresponding tasks and skill matching entries. During the execution process, the adaptation values of the workers can be numerically segmented first. For example, values from 0 to 0.5 are regarded as high matching degrees, values from 0.5 to 1.0 are regarded as medium matching degrees, and values greater than 1.0 are regarded as low matching degrees. When explaining the source of this segmentation threshold, the adaptation values can be concentrated in a matrix from a large number of past task assignment records, and then the mean, median, and standard deviation are calculated respectively. Finally, a upper and lower segmentation value is selected with reference to the distribution characteristics. In the example, the mean value of 0.6 is used as the demarcation point to form a preliminary interval, and the standard deviation is statistically analyzed in combination with multiple operation scenarios to set another threshold. After comparing the aforementioned adaptation value results with these intervals, they are marked with corresponding levels. If a certain adaptation value is lower than 0.5, it is marked as a high matching degree; if it is between 0.5 and 1.0, it is marked as a medium matching degree; if it is greater than 1.0, it is marked as a low matching degree. After listing these marks, the candidates with high matching degrees are preferentially entered into the candidate list, those with medium matching degrees are the second, and those with low matching degrees are considered later. Subsequently, summarize the candidate numbers marked as high or medium matching degrees, and finally form a task-worker skill adaptation set.

[0052] The steps to obtain the qualified worker task assignment pool are as follows: Based on the combinations of task numbers and worker numbers in the task-worker skill adaptation set, retrieve the daily shift schedule, historical completed task numbers, skill adaptation values, start and end times of the task time window, and whether there are continuous idle sections for each worker. Screen out the matching records that do not meet the condition of continuous assignment or have task conflicts in the past three days to obtain a task-worker execution mapping set with candidate assignment qualifications; According to the task-worker execution mapping set with candidate assignment qualifications, calculate the assignment fitness score. The calculation formula is: ; Among them, is the assignment fitness score of the th worker for the th task, is the skill adaptation value between the th worker and the th task, is the on-time completion rate of this worker in the recent seven days, is the daily maximum task load of this worker, is the concurrent conflict level of this task, is the current consecutive working days of this worker; Based on the assignment fitness scores of each task and worker, extract the worker numbers whose scores meet the conditions for each task, classify them by task number to generate a worker candidate set, and generate a qualified worker task assignment pool.

[0053] Specifically, based on the combination relationship between each task number and worker number in the task-worker skill adaptation set, retrieve information such as the daily shift schedule and historical completed task numbers corresponding to each worker, and combine the current skill adaptation value of the same worker and the task start and end time ranges to check whether the worker has a continuous idle period during the corresponding time period. If it is identified that a worker's daily shift schedule has been occupied by other projects from the current date to the next two days, it means there is a task conflict. It is necessary to compare whether the conflict overlaps with the time range of the latest task requirements, measure the overlapping part in hours and compare it with the set threshold. If the overlapping hours are greater than the threshold, directly exclude the matching relationship between the worker and this task. If the overlapping hours are less than or equal to the threshold, it is considered that a complete available period can be arranged in the remaining time period. The setting of the threshold is obtained by summarizing historical scheduling data. For example, in the historical scheduling data of one year, record the proportion of conflict scheduling hours of each worker and take the average value, then compare the upper and lower quartiles of this average value, and take the side closer to the lower quartile as the threshold value range and fix it at 4 hours. For special cases such as certain task types, fine-tuning can also be carried out on this basis. Eliminate all matching records with conflict hours greater than 4 within the current three days, and the remaining ones are considered to still have potential assignable qualifications. Then, according to the detection method of continuous idle periods, read the length of the blank periods of each worker on the shift schedule one by one. If the consecutive period is not enough to cover the shortest execution time of the task, the worker is also excluded. If the consecutive period is sufficient to support the shortest execution time, it is marked as meeting the conditions. Integrate these matching records that meet continuous idleness and have no major conflicts into a candidate data and associate the corresponding worker numbers, task numbers, and available period information. After the above screening, summarize the results to generate a task-worker execution mapping set with assignment candidate qualifications.

[0054] Formula: ; The benefit of the formula is that it simultaneously introduces the larger value of the skill adaptation value and the on-time completion rate in the last seven days in the numerator part, which can take into account both the actual matching degree of the worker and the punctuality performance of the recently completed tasks, and in the denominator part, through the combination of the daily maximum task load, concurrent conflict level, and consecutive working days, regulate the assignment priority of the worker.

[0055] This parameter is for the worker ; The on-time completion rate in the last seven days. The acquisition method needs to view the comparison between the actual progress and the planned progress of the worker's task completion in the last seven days. By setting a timely completion mark for each completed task, and then summing up the timely completion marks of all tasks in the last seven days and dividing by the total number of tasks to obtain a value. The specific calculation formula can be set as , where =1 indicates that the k-th task is completed as planned, 0 indicates delay or cancellation. Taking worker number 002 as an example, a total of 5 tasks were completed in the last seven days, among which 4 were completed on time and 1 was delayed. Then .

[0056] This parameter represents the worker ; The maximum daily task load. It can be set by statistically analyzing the current work intensity of the worker and combining with daily labor laws and regulations. The acquisition steps generally involve summarizing the daily order-receiving quantities of the worker in the past month and taking the maximum value, and then adjusting according to the physical condition or contract agreement. For example, worker number 005 completed at most 7 independent small-scale tasks in one day in the past month, so it is equal to 7.

[0057] This parameter is the ; Concurrent conflict level of the k-th task, used to measure the mutual influence of multiple tasks carried out simultaneously. The acquisition method includes two steps. The first step is to statistically analyze the concurrent difficulty from the historical conflict situations of similar tasks, and the second step is to revise it by comprehensively evaluating the occupation of the current plot or equipment resources. For example, a mark of 2 means there is a general degree of conflict demand, and a mark of 4 means that this task needs to preempt key equipment, thus making it easier to generate conflicts.

[0058] This parameter is for the worker ; Current consecutive working days. The acquisition method can be through statistical analysis of the clock-in data in the past 30 days, accumulating all consecutive clock-in days, and restarting the count as 0 if there is a one-day break. If worker number 010 has clocked in for work in the past 9 days, it is equal to 9.

[0059] Calculation process: Select parameters and substitute into actual examples: First step, determine that the task number = 15, worker number = 001, and the previous record , ; Second step, obtain the maximum daily task load of this worker , the task concurrent conflict level , the consecutive working days of this worker ; Third step, the numerator part ; In the fourth step, calculate the denominator part first , and then calculate , so the brackets in the denominator , and then add 1, which is ; In the fifth step, .

[0060] This result indicates that when the assignment fitness score is less than 1, it means that the comprehensive cooperation degree of the worker is in a relatively stable range. If the value is larger compared with the scores of the same task of other workers, the worker is more suitable to be assigned this task. If the value is too small, the worker can be considered to be transferred to other tasks.

[0061] Based on the assignment fitness scores of each task and the worker, compare them one by one according to the task number and the worker number from the mapping set with the qualification for assignment. Screen and mark the corresponding worker numbers for all the records that have been calculated and the assignment fitness scores are greater than the preset lower limit. The preset lower limit value can be determined by calculating the average value and variance of a large amount of past task scheduling data. For example, first summarize all the values into an array, calculate the mean value m, then calculate the standard deviation of the array to get d, and regard m minus 0.5 times d as the lower limit threshold or reference value. If m is 0.6 and d is 0.2, the lower limit can be set to 0.5, and the scores greater than 0.5 are regarded as meeting the scheduling requirements. Count the worker numbers marked as meeting the scheduling requirements and organize them into the corresponding categories according to the task numbers. Then check whether there are duplicate or conflicting records in the list. If it is found that the same worker is listed multiple times for the same task, only keep the first generated entry and delete the remaining duplicate entries. Finally, summarize and form a worker candidate set, and then output this candidate set as a qualified worker task assignment pool.

[0062] The steps to obtain the preliminary scheduling assignment plan are as follows: Based on the task numbers, worker numbers, start and end times of the task time window, the estimated execution man-hours of a single task and the upper limit of the remaining available man-hours per day of each worker in the qualified worker task assignment pool, call the task priority parameters and the task corresponding time window density values in the predicted agricultural task list to generate a pre-group set of paired task worker schedules; According to the pre-group set of paired task worker schedules, calculate the scheduling priority index. The calculation formula is: ; Among them, is the scheduling priority index of the th worker for the th task, is the time window density value of the th task, is the remaining available working hours of the th worker, is the starting time when the worker plans to execute this task, is the latest working time of the worker on that day, is the number of unassigned tasks of the worker on that day, is the execution priority level of this task; Based on the scheduling priority index, all workers are sorted by task number, and the one with the highest scheduling priority index is selected for time slot filling. An overlimit screening is performed in combination with the daily total working hour limit of the worker and the total cost limit of the farm to generate a preliminary scheduling allocation plan.

[0063] Specifically, based on information such as task numbers and worker numbers in the qualified worker task assignment pool, one-to-one comparison is made between the estimated execution man-hours for each single task and the upper limit of the remaining available man-hours per day for each worker. First, read the available time periods for the day from the worker's existing work schedule data, and check the overlap with the start and end times of the task time window item by item. The overlap check method is to make a minute-by-minute comparison between the task start time to end time and the worker's schedulable time period. When making the comparison, a minimum continuous working hour threshold needs to be set, which is obtained by statistically analyzing the work schedule data in the past year. For example, first extract the shortest working sections of the worker in each historical work schedule record to form an array of shortest sections, then sort the array and select the average value of the median position and several positions before and after it, and then remove extreme outliers and take an integer closest to the vast majority of cases. For example, the final threshold can be set at 3 hours. If the worker's schedulable time period is less than 3 hours, it is considered that the task cannot be scheduled. If it is equal to or higher than 3 hours, it is marked as passing the preliminary time matching. In the subsequent steps, re-screening needs to be carried out in combination with the task priority parameter and the time window density value. Here, first retrieve the priority of each task from the predicted agricultural task list. The larger the priority number, the more it needs to be scheduled first during allocation. The acquisition of the priority record comes from the planting progress management table of the plot management department. A fixed priority value is marked for each stage of operation in the management table. For example, fertilization is 3 and harvesting is 7. The task corresponding time window density value is determined by statistically analyzing the progress distribution of similar operations. If multiple operations are concentrated in a very compact time period, the density value is higher; otherwise, it is lower. After aggregating these elements, a combined mark is made for each task-worker matching entry, and then a mapping structure is established, which includes fields such as worker number, task number, start and end times of the time window, and priority and density values. To ensure data consistency, the upper limit of the remaining available man-hours also needs to be corrected. This available man-hour limit is obtained from the statistical results of the worker's current physiological load and previously accepted tasks. If the number of hours of previously accepted tasks is high, the available man-hours will be further reduced. Combining the average actual working hours of the worker in the recent statistical period, the record of the available man-hour limit is correspondingly reduced by several hours. After completing all comparisons and corrections, one or more matching options can be generated for each worker. Then, filter out the options where the available man-hours are already lower than the estimated execution man-hours for a single task. The remaining combinations after screening are considered to have potential scheduling feasibility. Aggregate all combinations that meet the conditions and arrange them in ascending order according to the task number to form a pre-set of schedulable task-worker schedules.

[0064] Formula: ; The advantage of the formula is that when scheduling shifts, it simultaneously compares the time window density of tasks with the remaining available working hours of workers. It reflects the urgency of the task time window and the worker's available working period in the numerator, and adds the difference between the start time and the latest available working time in the denominator. Additionally, it incorporates the combination of the number of tasks not assigned to the worker on the day and the task execution priority level, enabling the final shift scheduling priority index to comprehensively reflect the work urgency, the amount of free time of workers, and the overall level requirements of tasks.

[0065] This parameter is the time window density value of the th task, which is used to measure whether there are a large number of tasks of the same type or cross - type concentrated in the same or similar time periods. This value is obtained by statistically analyzing the superposition degree of past tasks of the same type in the same season, on the same plot or adjacent plots in the agricultural scheduling system. The statistical process first extracts the start and end time points and the number of tasks of all relevant tasks, calculates the average number of relevant operations occurring within a specific week, and then compares it with the number of regular operations to obtain a density coefficient. Finally, it is mapped to the range of 1 to 10 or 1 to 20 according to the comparison result. If the operations are highly concentrated, a higher value is taken; if the operations are scattered, a lower value is taken. For example, when seven tasks of the same type need to be completed within three days in a certain month, might be set to 8; if only a small number of tasks are distributed over a longer period,

[0066] This parameter is the remaining available working hours of the th worker, and the value usually ranges from 0 to 12 hours, indicating the continuous working hours that the worker can still invest on the day or during the current time period. The acquisition process requires checking the worker's personal shift record and the total workload for the day. First, add up the total working hours already scheduled for the worker on the day, and then subtract it from the daily total working hour limit to obtain the remaining hours. If factors such as physiological load or the number of consecutive working days are further considered, this part of the remaining available working hours will be adjusted again. In practical applications, if the daily limit of worker No. 010 is 10 hours and 4 hours have been scheduled currently, then

[0067] This parameter represents the worker's The start time of a task, recorded as a specific moment in a day. For example, 8 o'clock is recorded as 8, and 13 o'clock is recorded as 13. It is necessary to read a starting work time point through the preliminary scheduling decision of workers and tasks, and ensure that this time falls within the calendar time when the worker can work. If there is a conflict between this time and the actual situation, scheduling cannot be executed. Therefore, when calculating the formula, the difference between this value and the latest workable time of the worker is used to measure the interval between the two.

[0068] This parameter is for the worker The latest workable time on the current day, which is generally set according to the worker's work and rest schedule or labor compliance. For example, if worker number 007 can work until 20 o'clock every day in the contract, then .

[0069] This parameter is the number of unassigned tasks of the worker on the current day. The value usually ranges from 0 to 10, which is a description of the worker's idle degree on the current day. The acquisition method is to first count the number of tasks already assigned to the worker on the current day, and then subtract this number from the maximum allowable daily task number stipulated in the worker's contract, that is, the number of tasks that the worker can be reassigned is obtained. When used in this formula, it will be placed in the additive part of the numerator. If is large, it means that the worker still has more gaps to take tasks on the current day, which will make the scheduling priority index increase accordingly. For example, by querying the scheduling arrangement of worker number 003, it is found that 2 tasks have been assigned to him, and the maximum number of tasks that can be received per day is 5, then .

[0070] This parameter is the execution priority level of the task. It is used to distinguish the urgency of tasks on the same day, and the range is from 1 to 10. The larger the value, the more the task needs to be processed first. The acquisition process can directly retrieve the priority field carried by the task in the agricultural task list. This field comes from the planting plan or the importance assessment of the task by the management department. For example, the priority levels of sowing and fertilizing are often lower than those of harvesting or pest control. If it is shown in the record that the task belongs to high-risk pest control, then will be set relatively high.

[0071] Calculation process: The following gives a numerical example to show how to substitute each parameter and calculate the scheduling priority index: First step, given a certain task number = 12, its , , a certain worker number = 009, its hours, o'clock, o'clock, ; In the second step, first calculate ; In the third step, calculate in the denominator, so ; In the fourth step, calculate . At this time , , so ; In the fifth step, add the two parts: . After adding 0.894 ; In the sixth step, perform a floor operation on the result: , so .

[0072] This result indicates that the scheduling priority index of worker number 009 for task number 12 is 1. Compared with the calculation results of the same task for other workers, if a worker obtains a scheduling priority index greater than 1, it means that the worker is more suitable to be arranged for this task first. If the value is much less than 1, it means that the worker will be ranked lower in the scheduling.

[0073] Based on the scheduling priority index, sort all candidate workers one by one in descending order according to the task number, read the information of the worker ranked first and fill in his time period in the scheduling table. During this process, dynamically verify the daily total working hour limit of each worker. For example, if the daily upper limit of a worker is 8 hours, after allocating the current task to the worker, calculate the used working hours plus the working hours of this task execution. If the sum exceeds 8, it is determined that an overlimit has occurred, and the scheduling of this worker is immediately interrupted and the next worker ranked lower is selected to continue the detection. At the same time, in terms of the farm total cost limit check, first multiply the hourly cost of the worker by the working hours of this task to obtain a one-time cost, and then accumulate it into the daily overall cost account for comparison. If it exceeds the pre-set cost benchmark, it is excluded and the next worker is replaced. This cost benchmark is obtained by summarizing and statistically analyzing the daily cost data in multiple production cycles in the previous year. For example, first collect all daily expenditure items to obtain a total sum S and the total number of days N, calculate S÷N to get the average daily cost M, and then according to the upper and lower quartile method, discard the extreme abnormal days, and set the value near the high end in the quartile range as the highest daily expenditure allowed. If the current farm records this high end as T, then after adding the worker to the scheduling, judge whether the cost accumulation is greater than T. If it is greater, mark that the worker cannot be allocated continuously. Finally, write the information of the successfully allocated workers into the scheduling result and detect all task and worker combinations in order, and summarize to obtain a preliminary scheduling allocation plan.

[0074] The steps to obtain the farm dynamic operation schedule are as follows: Based on the task number, worker number, start and end times of the task time window, worker idle interval distribution code, and task scheduling interval code in each scheduling record of the preliminary scheduling assignment plan, determine whether there are continuous non-assignable or overlapping conflict situations, summarize the conflicting records, and generate a scheduling exception record set; According to the scheduling exception record set, calculate the assignment constraint strength factor for each record. The calculation formula is: ; where, is the assignment constraint strength factor of the -th worker for the -th task, is the scheduling priority index of this combination, is the scheduling interval code of the -th task, is the fatigue index of the -th worker, is an indicator variable indicating whether the worker has a time window gap for the task. If there is a gap, it is 1; otherwise, it is 0, is the urgency level of the -th task; Based on the assignment constraint strength factor, sort all worker records in ascending order of the assignment constraint strength factor by task number, and perform worker replacement and time period shift on the records with lower priority to generate a farm dynamic operation schedule.

[0075] Specifically, based on the task number, worker number, start and end times of the task time window, distribution code of the worker's idle intervals, and task scheduling interval code in each scheduling record of the preliminary scheduling and allocation plan, first extract the correspondence between the task number and the worker number from the scheduling records one by one. Conduct a matching analysis on the distribution code of the idle intervals of each worker, and record whether there is an overlap between the available time period of the worker and the required time period of the task. If it is found that the overlapping interval is less than the set effective working hours threshold, mark that it is difficult for this worker to form a stable connection with this task. If the overlapping interval is higher than or equal to the effective working hours threshold, mark it as having allocation feasibility. To further determine whether there are consecutive non-allocation or overlapping conflicts, it is necessary to compare the scheduling intervals of the workers with the task scheduling interval codes one by one, and mark each scheduling time in the hourly sequence of the day, so as to find out whether there are repeated allocations in multiple identical hourly periods. If the repeated allocation at the hourly granularity is greater than a certain empirical threshold, it is regarded as a conflict. This empirical threshold is obtained by summarizing the scheduling data of the past year. For example, count all the occurrences of scheduling conflicts, extract the average overlapping hours hc at the time of conflict, calculate the median and the offsets before and after for hc in the full set of conflict records, and take the smallest integer in these offset ranges as the threshold. If the overlapping hours exceed this threshold, it is determined that the conflict is established. Mark this record and the worker number with a conflict mark and organize them into a conflict list. Then, compare the entries in this conflict list one by one from the overall scheduling records. If the worker number and the task number are the same and the start and end intervals of the time window are also the same, it means that this conflict has occurred repeatedly. It is necessary to further judge whether the remaining capacity of this worker is completely insufficient according to the distribution code of the worker's idle intervals. If the remaining capacity is insufficient, confirm that the worker cannot continue to undertake this task, and add this record to the exception list. When generating the exception list, record the worker number, task number, conflict type, and conflict time period information in the corresponding data structure, and also record the daily scheduling interval synchronously to verify whether there are associated conflicts in other hourly periods. Finally, unify all the records with conflicts, accurately register the conflicting task and worker permutations in this exception list, and re-check these conflict contents in a cumulative statistical manner. Merge multiple conflicts of the same worker on the same day. If the conflict time periods overlap with each other, merge them into one conflict section and update the start and end times of the conflict. After all the merging operations are completed, if there are continuous non-allocation situations that cannot be resolved or multiple overlapping conflicts, further classify and summarize these records according to different conflict types (such as pure time period conflicts or equipment dependency conflicts) so that the abnormal repetition rate of the same worker or the same task can be tracked more accurately in the future. Finally, export and summarize all the conflict information into a scheduling exception record set.

[0076] Formula: ; The benefit of the formula is that based on the original scheduling priority index and the scheduling interval code Introduce modulo operation and combine worker fatigue index The index item is adjusted by using the worker time window gap indicator variable and task urgency level The product of is used to explicitly emphasize high-urgency tasks, thereby flexibly evaluating the constraint strength between workers and tasks in the schedule correction stage under the coupling of multiple factors.

[0077] This parameter is the scheduling priority index of the combination. It is necessary to retain the final value from each calculation of the scheduling priority index formula as the input of this formula. If the value calculated in the previous step is 12, then 12 can be directly called at the corresponding position in this formula.

[0078] This parameter is The scheduling interval code of a task is used to characterize the interval requirement when two or more scheduled operations are connected between the tasks. The value is usually between 1 and 10. The average or median interval required for the connection of tasks of the same type in historical scheduling is counted, and then combined with the shortest time for equipment or plot transfer, the interval value is set through integrated analysis and then distributed to specific tasks. For example, a 2-hour gap may be required between the harvesting operation and the subsequent transportation operation. The scheduling interval code can be set to 2. If no gap is required, it can be set to 1. The larger the value, the longer the interval period is required.

[0079] This parameter is The fatigue index of a worker is used to indicate the degree of fatigue accumulated from the worker's recent work. The value is usually between 0 and 10. It is quantified by monitoring the number of consecutive working days, daily working hours and energy consumption. Specifically, it can be based on statistics on the actual working hours of the worker in a day, and accumulated in weekly or monthly dimensions. The part that exceeds the preset healthy working hour standard is counted as fatigue contribution, and then summarized into a fatigue index. If a worker is in high-intensity work for many consecutive days, the fatigue index will rise to 7 or above. If there are rest days or shorter working hours during this period, the fatigue index will fall. For example, worker No. 005 has worked more than 9 hours a day in the past 7 days. The system accumulates the part that exceeds 8 hours a day, and obtains a fatigue index of 5.8 in the weekly calculation.

[0080] This parameter is an indicator variable showing whether a worker has a time window gap for a task, and it only takes two values, 0 or 1. 0 means the worker has no gap to meet the scheduling correction requirement of this task, and 1 means there is a gap for fine-tuning the time period during conflict adjustment. The acquisition process first has the scheduling system query the worker's schedule for the current day or week. If the worker still has available time slots during the current task conflict adjustment period, it is recorded as 1; otherwise, it is recorded as 0. For example, if worker number 002 is not occupied by any task during the period from 16:00 to 18:00 on the current day, and the task itself can be fine-tuned to this period, then 。

[0081] This parameter is the urgency level of the th task, with a value ranging from 1 to 10. The larger the value, the more urgent the task, and it requires a higher priority for correction opportunities during conflict resolution. The acquisition path can be directly read from the task list. The corresponding urgency level is classified by the agricultural management personnel in combination with the plot conditions and crop requirements, so as to set a larger for tasks with higher urgency such as pest control, plowing, or harvesting. For example, a pest control task can be set to 8, and a regular fertilization task can be set to 3.

[0082] Calculation process: The following selects a set of specific numerical calculation examples: First step, for a certain worker-task combination, the , the scheduling interval code of this task , the worker fatigue index , , the task urgency level ; Second step, for the numerator part, first perform , and then this 0 is operated in the exponential term . Since , ; Third step, the second term ; Fourth step, add the two terms together: ; This result shows that the worker-task combination has a distribution constraint strength factor of 2.236. When this value is relatively high, it means that this combination is in a relatively prominent constraint state during the scheduling conflict correction process and needs to be replaced or the time period shifted forward preferentially. Then, compare the values with other combinations to make the final adjustment.

[0083] Based on the calculation results of the distribution constraint strength factor, sort each record of workers and tasks in ascending order. Consider the records at the front as cases with lower constraint strength or being easily retained in conflicts. Records at the back indicate that the combination has a relatively large value in conflict situations and needs to be further replaced or the task execution time period postponed. When replacing a record, first check if there are other conflict-free available time periods for the corresponding worker number. If so, shift the time period within the available slots, advancing in hourly or half-hourly granularity, and determine the shift interval by comparing the remaining time intervals. If the shift cannot meet the minimum execution duration of the task, then search for other candidate replacement workers to execute the task. If the replacement worker cannot undertake it within the same time window, then postpone the entire task to the subsequent available adjacent time period. For entries with a relatively high urgency level for each task, reconfirm during this process to check if there are workers meeting the urgency conditions who can quickly take over the task. If no suitable candidates are found, mark it as a non-conforming record in the original record and continue the next round of investigation. After completing all such adjustment operations item by item, rearrange the records with the adjustments in effect and mark the final start and end times. Then, summarize all task records and worker arrangements within a day and perform a conflict-free verification. If no serious overlaps or irreconcilable section conflicts are found again during the verification process, consider the current result as the latest version of the dynamic work schedule arrangement, summarize the records and mark them as the farm dynamic operation shift schedule.

[0084] The steps for obtaining the agricultural insurance portfolio are as follows: Based on the types of each crop, the corresponding plot numbers, and the expected yields in the expected crop yield data, judge the market prices of the crops corresponding to the plots at different yield levels, calculate the differences between the crop revenues and the expected revenues at different yield levels step by step, and generate the financial loss amounts under different yield scenarios. According to the financial loss amounts under different yield scenarios, respectively retrieve the unit costs, loss coverage ratios, and insured amount ranges of each product in the agricultural insurance product database, compare and screen the product unit costs with the budget amount item by item against the farm insurance budget, and match the loss amounts with the insured amount ranges of the insurance products one by one to generate the screening and matching results of the insurance product cost coverage. According to the screening and matching results of the insurance product cost coverage, map the plot numbers to the types of crops for each agricultural insurance product that meets the cost constraint and insured amount range requirements, determine the agricultural insurance products matched by each plot or crop, and generate the agricultural insurance portfolio.

[0085] Specifically, based on the crop types, corresponding plot numbers, and expected yields in the expected crop yield data, first read the expected yields of the growth cycles from the crop type records of each plot and retrieve the market prices at different yield levels by referring to the agricultural market price list. These market prices are also split into segmented ranges. For example, three levels can be set: within 500 kg, 500 to 1000 kg, and above 1000 kg, and a unit price is matched for each level. Then, the plot numbers and crop types are corresponded one by one, and the actual transaction statistics over the years are retrieved from the farm management records to evaluate the average selling price of each crop within each yield range. These selling prices are obtained by calculating the transaction data of the corresponding crops over the past three years or a longer period. When calculating, first collect the weekly or monthly transaction prices of the same type of crops and perform extreme value removal processing, and then stratify the remaining data according to the yield paragraphs to obtain the unit selling prices corresponding to each yield segment. To ensure sufficient differentiation of the differences in different ranges, a sales volume correction coefficient for the key ranges can also be set. This coefficient is aggregated from the comparison of multiple plots. For example, group and analyze the transaction data of plots with the same type of crops, similar climatic conditions, and similar areas, select the area with concentrated yield distribution and calculate the weighted average price, use this weighted average price as the benchmark unit price for this segment range, and then match the unit price with the situation of each plot to form the segmented market selling price. For each yield level, multiply step by step by the selling price corresponding to this segment to obtain the segmented income, and record the difference between the segmented income and the expected income as the current period difference. If the income of a certain segment is significantly lower than the critical value set at the budget level, for example, the critical value is set at 80% of the expected income, it can be judged that the corresponding scenario represents a significant loss. During the process, all calculation results are recorded item by item and compared, and the scenarios exceeding the preset loss critical value are marked. Finally, the differences between the incomes corresponding to each yield level and the expected income are refined into financial loss data for the next step of insurance coverage analysis.

[0086] According to the amount of financial losses generated under different production scenarios, retrieve the unit cost, loss coverage ratio, and available insured amount range of each product from the agricultural insurance product database item by item. Compare the unit cost recorded in each product with the farm insurance budget amount. If the unit cost is higher than the budget amount, mark it as exceeding the limit and exclude it. Then, select the product entries that meet the budget constraint for the next matching. During this matching process, it is necessary to first list the relationship between the financial loss amount and the insured amount range of the product. All loss amounts can be arranged in ascending order and compared with the lower and upper limits of the insured amount of the insurance product. If the loss amount falls within the range covered by a certain product, mark it as meeting the fit requirement. For products that meet the fit requirement, detailed calculations need to be combined with the loss coverage ratio. If the coverage ratio is not sufficient to cover the corresponding loss amount, or the calculation result of the coverage ratio is below the risk control benchmark value default set for this farm, this product will be regarded as having insufficient coverage and this option will be excluded. The method for formulating the benchmark value can be through statistics of actual insurance purchase and loss settlement cases in the past several years. Denote the average loss coverage rate as m, and select 90% of m as a lower threshold value. Any coverage ratio lower than this lower threshold value will not be considered. Next, sort the remaining insurance products that meet the coverage requirements by unit cost to check whether the total will exceed the overall budget of the farm. Then, mark all the entries that pass the screening according to the loss range to form the insurance product cost coverage screening and matching results at this stage.

[0087] According to the above-mentioned insurance product cost coverage screening and matching results, the plot number mapping is performed for the insurance products that meet the cost limit and whose insured range can be connected to the corresponding crop yield loss. The specific method is to classify and view the types of crops planted according to the plot number in the farm management system, and connect them with the crop yield level recorded previously in sections to find a matching insurance product number for each crop. If there are multiple crop types in the same plot, they are matched one by one. If it is found that the insured range of a certain product is not enough to cover the loss level of one of the crops, the combination is skipped and other products that meet the conditions are searched. If it is determined that it can be covered, the plot number is bound to the product number, and then the corresponding unit premium cost and estimated compensation amount of the product are recorded. If the same When there are multiple optional products that need to compete for the same plot of land, the optimality can be judged by comparing the product of premium expenditure and coverage ratio. If the product value is too high, the priority will be relatively reduced. A threshold can be set here to represent the balance between premium and protection level. The threshold is formed by calculating the premium and compensation ratio with reference to past insurance cases. For example, first multiply the premiums of all cases by their corresponding compensation ratios to obtain the value P, and then take the average value p0 of these Ps as the benchmark, and then divide the floating range above and below p0 to obtain segments. If the p value exceeds the upper segment threshold, it is determined that the cost-effectiveness of the product is not ideal, and then the selection is postponed. Finally, after completing the matching of all plots of land and crops, the feasible plans are sorted out to form an agricultural insurance portfolio.

Claims

1. A crop yield forecasting and management system, characterized in that: The system comprises: The agricultural task prediction module obtains crop type, planting area and weather forecast information, calculates crop yield, and establishes expected crop yield data; associates the expected crop yield data with the preset agronomic plan, identifies the necessary agricultural operations, determines the task type corresponding to the operation, and generates a predicted agricultural task list; A labor skill matching module, based on the predicted agricultural task list and the labor database, retrieves the skills, working hours and time windows required by the task, compares the skills required for the task with the skills mastered by the labor, and establishes a skill adaptation set for the task workers; based on the skill adaptation set for the task workers, matches the task time window according to the available time of the labor, calculates the allocation fitness score, and screens to form a task allocation pool of qualified workers; A dynamic shift scheduling module selects workers and tasks based on the qualified worker task allocation pool and the predicted farm task list, establishes a preliminary shift allocation plan, analyzes the task completion timeliness and total labor cost of the preliminary shift allocation plan, adjusts the allocations that conflict or exceed the constraints, and generates a dynamic farm work schedule; The insurance resource allocation module estimates the amount of financial losses under different yield scenarios based on the expected crop yield data, compares the cost of agricultural insurance products, coverage levels and farm insurance budget amounts according to the amount of financial losses, selects a combination of agricultural insurance products and coverage levels for different plots or crops, and obtains an agricultural insurance combination.

2. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the expected crop yield data are: Based on the crop number parameter mapped by the crop type, the corresponding reference output value per unit area in the crop variety matching table is retrieved. Combined with the planting area, growth stage interval and stage time series input by the user, the temperature, precipitation, wind speed, sunshine duration and relative humidity are averaged by stage to establish the stage crop environmental factor input matrix; Calculate the predicted value of the stage yield according to the crop environmental factor input matrix of the stage; According to the yield forecast value of the stage, it is multiplied and added with the allocated area of ​​each stage to generate the total yield value of all stages and obtain the expected crop yield data.

3. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the predicted farming task list are as follows: Based on the different crop types and stage yield prediction values ​​in the expected crop yield data, the agronomic plan task template is matched by crop type, and the task start time point and yield formation time are compared in accordance with the stage time index of the yield data and the task start interval in the template, and the task records matching the start conditions in the agronomic plan are screened to generate the screening results of the crop yield-related agronomic plan tasks; According to the screening results of the crop yield-related agronomic plan tasks, the agricultural operation contents configured in the triggered task records are extracted item by item, the job number in the task record is called, the task type, required skills, estimated working hours and executable time window parameters of each agricultural operation are obtained, and a structured agricultural task attribute set is established; Based on the structured agricultural task attribute set, similarity judgment is performed between different agricultural operations in terms of task type, required skills and time window dimensions, repeated or overlapping task records are called, operation items are merged and sorted according to crop number and time window order, and a predicted agricultural task list is generated.

4. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the task worker skill adaptation set are: Based on the task type, required skill code, required skill level, estimated execution time and task time window parameters of each task record in the predicted agricultural task list, the worker number, skill code, skill level value, hourly labor cost, calendar available time period and job category identification included in the labor database are retrieved, workers with skill codes and skill levels not lower than the required skill levels are pre-screened, and workers with non-overlapping time windows are screened out to generate a candidate list of task workers with preliminary assignable qualifications; Calculate the skill adaptation value of each worker for the task according to the candidate list of task workers with preliminary assignable qualifications; Based on the skill fit value between each worker and the task, the workers are screened to generate a task-worker skill fit set.

5. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the qualified worker task allocation pool are as follows: Based on the combination of each task number and worker number in the task worker skill adaptation set, retrieve the daily shift schedule, the number of historical completed tasks, the skill adaptation value, the start and end time of the task time window, and whether there is a continuous idle segment for each worker, filter out the matching records that do not meet the continuous allocation conditions or have task conflicts in the past three days, and obtain the task worker execution mapping set that is eligible for allocation candidate; Calculate an allocation fitness score based on the task worker execution mapping set that is eligible for allocation candidate; Based on the allocation fitness score of each task and worker, the worker numbers whose scores meet the requirements in each task are extracted, and the worker candidate sets are generated by classifying them according to the task numbers to generate a task allocation pool of qualified workers.

6. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the preliminary shift allocation plan are as follows: Based on the task number, worker number, task time window start and end time, single task estimated execution time and each worker's daily remaining available working time limit in the qualified worker task allocation pool, the task priority parameter and the task corresponding time window density value in the predicted agricultural task list are called to generate a pre-group set of worker shifts that can be matched with tasks; Calculating a scheduling priority index according to the pre-set of scheduling of workers with paired tasks; Based on the scheduling priority index, all workers are sorted by task number and the workers with the highest scheduling priority index are selected for time period filling. The worker's daily total working hour limit and the farm's total cost limit are combined to perform over-limit screening and generate a preliminary scheduling allocation plan.

7. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the farm dynamic work schedule are as follows: Based on the task number, worker number, task time window start and end time, worker idle interval distribution code and task scheduling interval code of each scheduling record in the preliminary scheduling allocation plan, determine whether there is a continuous unassignable or overlapping conflict, summarize the conflicting records, and generate a scheduling exception record set; Calculate the allocation constraint strength factor of each record according to the abnormal scheduling record set; Based on the allocation constraint strength factor, all worker records are sorted in ascending order according to the allocation constraint strength factor by task number, and the workers of the records with lower priority are replaced and the time periods are shifted to generate a dynamic farm work schedule.

8. The crop yield forecasting and management system according to claim 1, characterized in that: The steps for obtaining the agricultural insurance combination are: Based on the crop types, corresponding plot numbers and expected yields in the expected crop yield data, determine the market prices of the crops corresponding to the plots at different yield levels, calculate the difference between the crop income and the expected income at different yield levels step by step, and generate the amount of financial losses under different yield scenarios; According to the amount of financial loss under the different production scenarios, the unit cost, loss coverage ratio and insured amount range of each product in the agricultural insurance product database are retrieved respectively, and the farm insurance budget is compared and screened item by item for the unit cost of the product and the budget amount, and the degree of fit between the loss amount and the insured amount range of the insurance product is matched one by one to generate the insurance product cost coverage screening matching results; According to the matching results of the insurance product cost coverage screening, the agricultural insurance products that meet the cost constraints and insurance amount range requirements are mapped between the plot numbers and the crop types one by one, the agricultural insurance products that match each plot or crop are determined, and the agricultural insurance portfolio is generated.

Images