Agricultural planting big data analysis task scheduling system

Abstract

The invention relates to the technical field of agricultural information management, in particular to an agricultural planting big data analysis task scheduling system which comprises a time sequence recognition module, an urgency evaluation module, an index weight generation module, a priority evaluation module and a task distribution module. According to the method, the fluctuation characteristics of the environment temperature time sequence data are analyzed, the time axis is divided into stable and violent fluctuation sections, the characteristics are extracted, the physiological urgency degree of crops under the environment pressure is quantified, and contribution weights of key indexes to the yield are mined in combination with historical yield data; a graph reflecting a data flow dependency relationship is constructed to evaluate the priority, and a task queue is rearranged by using a dynamic weighting algorithm containing a real-time load factor to ensure that a core task which is most valuable for crop growth and yield is preferentially processed under an environment sudden change; and intelligent on-demand distribution of computing resources according to crop biological requirements and environmental urgency is realized.

Description

Technical Field

[0001] This invention relates to the field of agricultural information management technology, and in particular to an agricultural planting big data analysis task scheduling system. Background Technology

[0002] The field of agricultural information management technology involves the collection, organization, and management of information such as environmental parameters, crop growth data, and operation records generated during agricultural production, and the planning and scheduling of agricultural planting activities using data statistical analysis, process management methods, and task organization methods.

[0003] In agricultural planting big data analysis task scheduling systems, data such as temperature and humidity, light intensity, and soil moisture content acquired by sensors are typically arranged chronologically. Then, data cleaning steps, such as outlier removal, sampling period alignment, and data format standardization, are performed according to preset task attributes. Subsequently, statistical processing is performed on data from different sources according to a predetermined data analysis workflow. Finally, various analysis tasks are sequentially allocated to computing nodes according to preset scheduling rules to perform steps such as growth stage segmentation, yield correlation calculation, or comparative analysis of data from specified periods, thereby completing the big data analysis task scheduling process in agricultural planting scenarios.

[0004] The existing agricultural planting big data analysis task scheduling model only arranges tasks linearly according to the time sequence of data collection, ignoring the differences in the physiological impact of environmental changes on crops. As a result, high-urgency tasks still need to be queued according to the conventional process when there are drastic temperature changes, and the calculation order cannot be adjusted according to the stress level of crops. This causes a lag in the analysis of key indicators. In addition, the fixed scheduling rules do not consider the adaptation of the real-time load of nodes to the biological value of tasks, which causes computing resources to be occupied by low-priority tasks. This results in a delay in the response of core tasks that have a decisive impact on yield, making it difficult to meet the timeliness requirements of precision agriculture for disaster early warning. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing an agricultural planting big data analysis task scheduling system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an agricultural planting big data analysis task scheduling system includes: The time-series identification module collects time-series data of the ambient temperature of the planting base according to a specified period, and divides the temperature fluctuation into stable temperature fluctuation ranges and temperature fluctuation ranges exceeding the standard based on the rate of change and amplitude of temperature over time, thereby obtaining the agricultural environment fluctuation state sequence. The urgency assessment module extracts the average temperature change rate of the stable temperature fluctuation segment and the peak temperature change rate of the temperature fluctuation segment exceeding the temperature limit in the agricultural environment fluctuation state sequence to construct the time urgency coefficient of the agricultural computing task. The indicator weight generation module obtains a preset list of agricultural crop yield prediction factors, identifies key biological growth indicators from it, and quantifies the basic biological sensitivity weights of the indicators. The priority assessment module constructs a dependency graph of agricultural computing tasks that shows the flow relationship between agricultural temperature time-series data and key biological growth indicators. It determines the correlation between agricultural computing tasks in the graph and calculates the priority score of agricultural computing tasks by combining the basic sensitivity weights of the biological indicators. The task distribution module integrates the time urgency coefficient and priority score of the agricultural computing task, maps them to the task index in the waiting queue, and obtains the task scheduling results for analysis.

[0007] As a further aspect of the present invention, the agricultural environment fluctuation state sequence includes a stable temperature fluctuation range and a temperature exceeding standard fluctuation range. The agricultural computing task time urgency coefficient is specifically a quantitative value constructed based on the average temperature change rate of the stable temperature fluctuation range and the peak temperature change rate of the temperature exceeding standard fluctuation range. The biological indicator basic sensitivity weight is specifically a value of the contribution degree of key biological growth indicators quantified based on the agricultural crop yield prediction factor association list. The agricultural computing task priority score is specifically a ranking score calculated by combining the task relevance in the agricultural computing task dependency graph with the biological indicator basic sensitivity weight. The analysis task scheduling result includes a comprehensive score after fusion calculation and a task index mapped to the waiting queue.

[0008] As a further aspect of the present invention, the timing identification module includes: The rate of change calculation submodule connects to the agricultural temperature time series data collected by the sensors in the planting base, sets a time window on the time axis, calculates the cumulative rate of change of agricultural temperature in the agricultural temperature time series data within each time window, calculates the numerical difference of the cumulative rate of change of agricultural temperature corresponding to adjacent time windows, and obtains the fluctuation range of temperature change rate. The fluctuation state marking submodule filters the time periods in which the fluctuation amplitude of the temperature change rate is within the preset temperature fluctuation threshold range and marks the corresponding time periods as temperature stable indicators. It also filters the time periods in which the fluctuation amplitude of the temperature change rate exceeds the temperature fluctuation threshold range and marks the corresponding time periods as temperature unstable indicators, thus obtaining a set of temperature fluctuation indicators. The segment aggregation mapping submodule calls the temperature fluctuation identifier set, identifies the temperature stability identifiers continuously distributed on the time axis and summarizes them into temperature fluctuation stability segments, identifies the temperature instability identifiers continuously distributed on the time axis and summarizes them into temperature exceedance fluctuation segments, and performs time sequence splicing and state mapping on all temperature fluctuation stability segments and temperature exceedance fluctuation segments according to the chronological order of the time axis to generate an agricultural environment fluctuation state sequence.

[0009] As a further aspect of the present invention, the urgency assessment module includes: The feature value extraction submodule analyzes the distribution of stable temperature fluctuation segments and temperature exceeding fluctuation segments in the agricultural environment fluctuation state sequence on the time axis, traverses the cumulative change rate of agricultural temperature in the stable temperature fluctuation segments, calculates the average temperature change rate, traverses the cumulative change rate of agricultural temperature in the temperature exceeding fluctuation segments, and filters the peak temperature change rate. The stationarity parameter calculation submodule calculates the difference ratio between the peak temperature change rate and the average temperature change rate to obtain a stationarity deviation value, which characterizes the crop state change attribute and obtains the crop growth state transition stationarity parameter. The urgency coefficient mapping submodule takes the crop growth state transition stability parameter as an input variable and substitutes it into the forward mapping transformation function to perform the operation. It then quantifies the transformation result in combination with the crop rhythm mutation requirements to generate the time urgency coefficient of agricultural computing tasks.

[0010] As a further aspect of the present invention, the index weight generation module includes: The indicator identification submodule obtains a pre-defined list of agricultural crop yield prediction factors, analyzes the mapping relationship between each agricultural crop yield prediction factor and crop yield, and identifies the indicator items that play a dominant role in determining crop yield from all agricultural crop yield prediction factors based on the mapping relationship, generating key biological growth indicators. The coefficient extraction submodule uses the key biological growth index as an index key value to perform a search and matching in the agricultural crop yield prediction factor association list, locates the entry position corresponding to the key biological growth index, extracts the recorded regression coefficient quantitative data that represents the quantitative dependence between the key biological growth index and crop yield, and obtains the yield regression coefficient. The weight quantification submodule calculates the proportion of the absolute value of the current indicator's output regression coefficient in the sum of the absolute values ​​of the output regression coefficients of all key indicators, thus obtaining the basic sensitivity weight of the biological indicator.

[0011] As a further aspect of the present invention, the priority evaluation module includes: The graph path tracing submodule constructs an agricultural computing task dependency graph of the flow relationship between agricultural temperature time series data and key biological growth indicators. Taking the key biological growth indicator node at the end of the graph as the starting anchor point, it performs reverse path tracing to the agricultural temperature time series data node at the beginning of the graph to determine the spatial distribution of all associated paths and obtain the set of reverse flow paths. The weight generation submodule extracts the input and output parameters of adjacent-level agricultural computing tasks in the reverse flow path set, calculates the Pearson correlation coefficient between the output parameters of the current-level agricultural computing task and the input parameters of the next-level agricultural computing task, introduces the basic sensitivity weight of biological indicators as the initial weight benchmark, performs a transfer correction on the weight inherited by the next-level agricultural computing task based on the Pearson correlation coefficient, and generates task transfer weights. The priority score calculation submodule traverses each agricultural computing task instance distributed in the agricultural computing task dependency graph, retrieves the task transfer weight of the agricultural computing task instance on each flow path, performs full-path numerical accumulation on all the task transfer weights associated with the same instance, and generates an agricultural computing task priority score.

[0012] As a further aspect of the present invention, the task distribution module includes: The execution score calculation submodule performs numerical fusion of the time urgency coefficient of the agricultural computing task and the priority score of the agricultural computing task to quantify the execution priority of each agricultural computing task to be executed at the current moment and obtain the agricultural task scheduling execution score. The queue descending order reordering submodule calls the task index interface of the agricultural big data computing server waiting queue, maps the agricultural task scheduling execution score to the index attribute of the corresponding task, sets it as the reference benchmark for queue sorting, performs descending order reordering processing on all agricultural computing tasks residing in the waiting queue, and generates a reordered task sequence. The task node distribution submodule determines the order of the rearranged task sequence indexes, extracts agricultural computing task instances to be processed, synchronously scans and detects the real-time load status of agricultural task computing nodes, filters computing node resources whose current load value is lower than the preset idle state judgment threshold, and distributes the extracted agricultural computing tasks to the filtered idle agricultural task computing nodes, generating analysis task scheduling results.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by analyzing the fluctuation characteristics of environmental temperature time-series data, the time axis is divided into stable and drastic fluctuation segments and features are extracted. The physiological urgency of crops under environmental stress is quantified. By combining historical yield data to mine the contribution weight of key indicators to yield, a graph reflecting the data flow dependency relationship is constructed to evaluate priority. A dynamic weighting algorithm including real-time load factors is used to rearrange the task queue to ensure that the core tasks most valuable to crop growth and yield are processed first under environmental changes. This realizes the intelligent on-demand allocation of computing resources based on crop biological needs and environmental urgency. Attached Figure Description

[0014] Figure 1This is a system flowchart of the present invention; Figure 2 This is a flowchart of the timing recognition module of the present invention; Figure 3 This is a flowchart of the urgency assessment module of the present invention; Figure 4 This is a flowchart of the indicator weight generation module of the present invention; Figure 5 This is a flowchart of the priority evaluation module of the present invention; Figure 6 This is a flowchart of the task distribution module of the present invention. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the 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 merely illustrative and not intended to limit the invention.

[0016] Please see Figure 1 An agricultural planting big data analysis task scheduling system includes: The time-series identification module collects time-series data of the ambient temperature of the planting base according to a specified period, and divides the temperature fluctuation into stable temperature fluctuation ranges and temperature fluctuation ranges exceeding the standard based on the rate of change and amplitude of temperature over time, thereby obtaining the agricultural environment fluctuation state sequence. The urgency assessment module extracts the average temperature change rate in the stable temperature fluctuation range and the peak temperature change rate in the temperature exceeding fluctuation range in the agricultural environmental fluctuation state sequence, and constructs the time urgency coefficient of agricultural computing tasks. The indicator weight generation module obtains a preset list of agricultural crop yield prediction factors, identifies key biological growth indicators from it, and quantifies the basic biological sensitivity weights of the indicators. The priority assessment module constructs a dependency graph of agricultural computing tasks that shows the flow relationship between agricultural temperature time series data and key biological growth indicators. It judges the correlation between agricultural computing tasks in the graph and calculates the priority score of agricultural computing tasks by combining the basic sensitivity weight of biological indicators. The task distribution module integrates the time urgency coefficient and priority score of agricultural computing tasks, maps them to the task index in the waiting queue, and obtains the task scheduling results for analysis.

[0017] The agricultural environment fluctuation state sequence includes stable temperature fluctuation ranges and temperature exceeding ranges. The agricultural computing task time urgency coefficient is a quantitative value constructed based on the average temperature change rate in the stable temperature fluctuation range and the peak temperature change rate in the temperature exceeding range. The basic sensitivity weight of biological indicators is a value of the contribution of key biological growth indicators quantified based on the correlation list of agricultural crop yield prediction factors. The priority score of agricultural computing tasks is a ranking score calculated by combining the task relevance in the agricultural computing task dependency graph with the basic sensitivity weight of biological indicators. The task scheduling results analysis includes the comprehensive score after fusion calculation and the task index mapped to the waiting queue.

[0018] Please see Figure 2 The timing recognition module includes: The rate of change calculation submodule connects to the agricultural temperature time series data collected by the sensors in the planting base, sets a time window on the time axis, calculates the cumulative rate of change of agricultural temperature in the agricultural temperature time series data within each time window, calculates the numerical difference of the cumulative rate of change of agricultural temperature corresponding to adjacent time windows, and obtains the fluctuation range of temperature change rate. Agricultural temperature time-series data collected by sensors in the planting base is accessed. This data is uploaded in real time by digital temperature sensors distributed at different spatial locations in the planting area according to a predetermined sampling frequency, set to collect data once every 10 minutes. A time window is set on the time axis. The length of this time window is determined based on statistical data on the physiological lag time of crops in response to drastic changes in environmental temperature. By statistically analyzing the average time from the onset of a temperature change to the appearance of a stress response in a specified crop in historical disaster records, the average time is set to 120 minutes. One-tenth of this time, i.e., 12 minutes, is taken as the time window length, rounded down to 10 minutes. The cumulative rate of change of agricultural temperature time-series data within each time window is calculated. For any time window, the slope is obtained by subtracting the temperature collected at the beginning of the window from the temperature collected at the end of the window, and dividing the difference by the time span of the time window. The end temperature is 25.5 degrees Celsius, the beginning temperature is 24.5 degrees Celsius, and the time span is 10 minutes. The slope is calculated using numerical values. Calculate the numerical difference in the cumulative rate of change of agricultural temperature for adjacent time windows. Extract the rate of change value of the current time window and the rate of change value of the previous adjacent time window, perform subtraction, and take the absolute value. The current window's rate of change is 0.1, and the previous window's rate of change is 0.04. The fluctuation range of the temperature change rate is obtained.

[0019] The fluctuation state marking submodule filters time periods where the temperature change rate fluctuation amplitude is within the preset temperature fluctuation threshold range and marks the corresponding time periods as temperature stable indicators. It also filters time periods where the temperature change rate fluctuation amplitude exceeds the temperature fluctuation threshold range and marks the corresponding time periods as temperature unstable indicators, thus obtaining a set of temperature fluctuation indicators. The process for setting the preset temperature fluctuation threshold is as follows: Retrieve historical temperature time-series data for the past three complete crop growth cycles from the planting base. Calculate the arithmetic mean and standard deviation of the temperature change rate fluctuation amplitude across all adjacent time windows in the historical data. Add twice the standard deviation to the arithmetic mean to obtain the preset temperature fluctuation threshold. The arithmetic mean of the historical data fluctuation amplitude is 0.02, and the standard deviation is 0.015. The threshold is calculated as follows: The real-time calculated temperature change rate fluctuation amplitude is compared with this threshold. The comparison result is categorized into two cases: if the calculated fluctuation amplitude is less than or equal to 0.05, the temperature change within that time period is determined to be within the stable range that the crop can adapt to, and the corresponding time period is marked as temperature stable; if the calculated fluctuation amplitude is greater than 0.05, the temperature change within that time period is determined to be beyond the crop's normal adaptability, and the corresponding time period is marked as temperature unstable. The aforementioned calculation result of 0.06 is greater than 0.05, therefore it is determined to be unstable, and the corresponding time period is marked as temperature unstable. By traversing all time windows and performing the above comparison and marking operations, a set of temperature fluctuation indicators is obtained.

[0020] The segment aggregation mapping submodule calls the temperature fluctuation identifier set, identifies the temperature stability identifiers continuously distributed on the time axis and summarizes them into temperature fluctuation stability segments, identifies the temperature instability identifiers continuously distributed on the time axis and summarizes them into temperature exceedance fluctuation segments, and performs time sequence splicing and state mapping on all temperature fluctuation stability segments and temperature exceedance fluctuation segments according to the chronological order of the time axis to generate an agricultural environment fluctuation state sequence. This process detects adjacent time markers on the timeline. If three or more consecutive time windows are marked as temperature stable, the start and end times of these consecutive time windows are merged into a single continuous time interval. For example, three consecutive windows from 10:00 to 10:30 are all marked as stable, thus forming a stable segment from 10:00 to 10:30. Continuously distributed temperature instability markers on the timeline are identified and categorized as temperature fluctuation segments exceeding limits. Any consecutive temperature instability markers, regardless of their number, are merged into a fluctuation segment. Based on the chronological order of the timeline, all stable and temperature fluctuation segments are spliced ​​and mapped in a temporal sequence. The segments are arranged in ascending order according to their start timestamps, connecting the scattered segments end-to-end to form a state sequence covering the entire monitoring period, generating an agricultural environmental fluctuation state sequence.

[0021] Please see Figure 3 The urgency assessment module includes: The feature value extraction submodule analyzes the distribution of stable temperature fluctuation segments and temperature exceeding fluctuation segments in the agricultural environment fluctuation state sequence on the time axis, traverses the cumulative change rate of agricultural temperature in stable temperature fluctuation segments, calculates the average temperature change rate, traverses the cumulative change rate of agricultural temperature in temperature exceeding fluctuation segments, and filters the peak temperature change rate. Extract the rate of change values ​​of all time windows within a stable temperature fluctuation range, perform an arithmetic mean calculation, and calculate the average rate of change of temperature. A certain stable range contains three time windows with rates of change of 0.02, 0.03, and 0.01 respectively. The calculation process is as follows: The cumulative rate of change of agricultural temperature is iterated through the temperature fluctuation range, and the rate of change values ​​of all time windows within the range are extracted. The maximum value is selected by comparing the values ​​and taken as the peak value of the temperature change rate. The rate of change in a certain drastic fluctuation range are 0.08, 0.12 and 0.09 respectively. After comparison and selection, the peak value of the temperature change rate is found to be 0.12.

[0022] The stationarity parameter calculation submodule calculates the ratio of the difference between the peak temperature change rate and the average temperature change rate to obtain the degree of stationarity deviation, which characterizes the crop state change attribute and obtains the stationarity parameter of crop growth state transition. First, calculate the difference between the peak temperature change rate and the average temperature change rate. Then, divide this difference by the average temperature change rate to obtain the degree of deviation from the mean. Substituting these values, the peak temperature change rate is 0.12, and the average temperature change rate is 0.02. The calculation process is as follows: This process requires a preset maximum tolerance deviation threshold, which is obtained through crop stress experiments. Specifically, it involves measuring the deviation value at which irreversible damage begins to appear in the crop under controlled conditions. This threshold is set to 10. The calculated steady-state deviation value is then divided by this threshold. The result is categorized into three cases: if the result is greater than 1, it is taken as 1; if the result is less than 0, it is taken as 0; if the result is between 0 and 1, the original value is retained. The calculation process is as follows: The result is between 0 and 1, so we retain 0.5 to obtain the parameter of crop growth state transition stability.

[0023] The urgency coefficient mapping submodule takes the crop growth state transition stability parameter as the input variable and substitutes it into the forward mapping transformation function to perform the operation. It quantifies the transformation result in combination with the crop rhythm mutation requirements and generates the time urgency coefficient of agricultural computing tasks. The forward mapping transformation function is defined as directly using the numerical values ​​of the input variables, substituting the previously calculated result of 0.5 as the base urgency value. The transformation result is quantified based on the crop's rhythmic mutation requirements, and a rhythm weighting coefficient is set according to the physiological activity level of the crop at different times of the day. This coefficient has two levels: if it is during the peak photosynthetic period (e.g., 10:00 to 14:00), the weight is set to a high value of 0.9; if it is during the nighttime dormancy period or non-peak period, the weight is set to a low value of 0.4. The current time is during the peak photosynthetic period, so a weight of 0.9 is extracted. Since the calculated stability parameter is actually the degree of fluctuation deviation, a higher parameter value indicates more severe fluctuations. Therefore, the time urgency of the calculation task should be higher. The correction logic is to directly use the crop growth state transition stability parameter as the base urgency, and then multiply it by the rhythm weight. The calculation process is 0.5 × 0.9 = 0.45, generating the agricultural calculation task time urgency coefficient.

[0024] Please see Figure 4 The indicator weight generation module includes: The indicator identification submodule obtains a pre-defined list of agricultural crop yield prediction factors, analyzes the mapping relationship between each agricultural crop yield prediction factor and crop yield, and identifies the indicator items that play a dominant role in determining crop yield from all agricultural crop yield prediction factors based on the mapping relationship, generating key biological growth indicators. The list of factors associated with agricultural crop yield forecasts includes: Collect agricultural planting records within a specified period, analyze the agricultural planting records to extract crop yields for the specified period and a set of multidimensional environmental monitoring variables for the same period. The set of multidimensional environmental monitoring variables includes environmental temperature data, environmental humidity data, light intensity data, and soil moisture content data. For the environmental temperature data, environmental humidity data, light intensity data, and soil moisture content data in the multidimensional environmental monitoring variable set, outlier removal and numerical standardization normalization processing with values ​​exceeding 3 times the standard deviation are performed to generate a standardized environmental dataset. Construct a multiple regression analysis model with crop yield for a specified period as the dependent variable and standardized environmental temperature, standardized environmental humidity, standardized light intensity, and standardized soil moisture content from a standardized environmental dataset as independent variables. The least squares method was used to perform parameter fitting operations on the multiple regression analysis model, and the independent regression coefficients of each environmental monitoring variable with respect to crop yield in a specified period were obtained. Each environmental monitoring variable is defined as an agricultural crop yield predictor, and the corresponding independent regression coefficients are encapsulated as multiple regression statistics of agricultural crop yield. Establish a key-value pair index structure with the correspondence between the names of agricultural crop yield predictors and the statistical data of multiple regression of agricultural crop yield, and generate a list of agricultural crop yield predictor associations. After collecting agricultural planting records for a specified historical period (such as the past 5 years), the records are first structured and parsed to obtain two types of core data: one is historical crop yield data summarized according to the historical period (as the yield target value for subsequent modeling), and the other is a set of multidimensional environmental monitoring variables that are concurrent with the yield data in time (as a set of candidate predictive variables, which may include monitoring sequences or statistical values ​​of environmental temperature, environmental humidity, light intensity, soil moisture content, etc. aligned to the same period). Subsequently, data cleaning and standardization were performed on each variable in the multidimensional environmental monitoring variable set: first, the mean and standard deviation of the variable were calculated for the samples in the historical period, and the deviation of each sample from the mean was calculated. Any sample that deviated from the mean by more than 3 times the standard deviation was identified as an outlier and removed. After the outlier removal was completed, numerical standardization and normalization were performed on the variable. Using the range transformation method, the minimum and maximum values ​​of the variable after cleaning were used as the benchmark to linearly map all samples of the variable to the interval between 0 and 1. This allows environmental variables with different dimensions and different value ranges to participate in regression calculations on the same scale, thereby forming a standardized environmental dataset. After obtaining a standardized environmental dataset, a multiple linear regression analysis model was constructed with historical crop yield as the dependent variable and various environmental variables in the standardized environmental dataset as independent variables. The least squares method was used to fit the parameters of this model: aiming to minimize the sum of squared residuals, the regression parameters corresponding to each independent variable were iteratively adjusted until the sum of squared residuals converged or a preset convergence condition was met. This yielded the independent regression coefficient of each environmental variable on historical crop yield (representing the direction and strength of the independent linear influence of that variable on yield change under the condition that other variables remain constant). After obtaining the regression results, each environmental variable was considered a yield predictor, and its mapping relationship with yield was tested for significance. A significance level probability threshold of 0.05 was set, and predictors with a probability value less than 0.05 were selected and determined to have a significant impact on yield. Then, within the set of significant predictors, combined with their independent regression coefficients and other statistical results, the indicators that play a dominant role in determining crop yield were identified from all factors, ultimately generating key biological growth indicators.

[0025] The coefficient extraction submodule uses key biological growth indicators as index keys to perform a search and match in the agricultural crop yield prediction factor association list, locates the entry position corresponding to the key biological growth indicators, extracts the recorded regression coefficient quantitative data that represents the quantitative dependence between key biological growth indicators and crop yield, and obtains the yield regression coefficient. Iterate through the keys in the list, searching for data items whose names exactly match the current key yield predictor name. Once located, extract the quantified regression coefficient data stored in that item; this data represents the independent regression coefficients calculated in the aforementioned regression analysis. In the example, the key biological growth indicator is "leaf area index," and its corresponding regression coefficient is 0.65 found in the association list; the key environmental impact factor is "soil moisture," with a corresponding regression coefficient of -0.3. Directly read these values ​​to obtain the yield regression coefficients.

[0026] The weight quantification submodule calculates the proportion of the absolute value of the current indicator's output regression coefficient in the sum of the absolute values ​​of the output regression coefficients of all key indicators, thus obtaining the basic sensitivity weight of the biological indicator. Extract the absolute values ​​of the output regression coefficients: 0.65 equals 0.65, and -0.3 equals 0.3. Establish a mapping rule between the absolute values ​​of the output regression coefficients and their weights. This rule uses a normalized proportion calculation method, which calculates the proportion of the current indicator's regression coefficient absolute value in the sum of all key indicator regression coefficient absolute values. Assuming the total sum is 0.95, the weight calculation process for the first indicator is as follows: The basic sensitivity weights of the biometrics were obtained.

[0027] Please see Figure 5 The priority assessment module includes: The graph path tracing submodule constructs an agricultural computing task dependency graph of the flow relationship between agricultural temperature time series data and key biological growth indicators. Taking the key biological growth indicator node at the end of the graph as the starting anchor point, it performs reverse path tracing to the agricultural temperature time series data node at the beginning of the graph to determine the spatial distribution of all associated paths and obtain the set of reverse flow paths. The agricultural computational task dependency mapping for constructing the flow relationship between agricultural temperature time-series data and key biological growth indicators includes: Agricultural temperature time series data is transformed into preset calculation rules for key biological growth indicators, and the preset calculation rules are discretized and decomposed into multiple agricultural computing task instances. Iterate through each agricultural computing task instance and extract the input variable definitions required for the operation and the output variable definitions after the operation is completed. Agricultural temperature time series data is instantiated as the root data node of the graph, key biological growth indicators are instantiated as leaf data nodes of the graph, and all agricultural computing task instances are mapped as intermediate computing nodes of the graph. Perform parameter dependency matching scans on all nodes in the graph, compare the output variable definitions of the preceding node with the input variable definitions of the following node, and when the variable definitions are consistent, establish a directed dependency edge representing the data flow between the two corresponding nodes. Based on directed dependency edges, the root data node, intermediate computing nodes, and leaf data nodes are topologically connected according to the temporal logic of computing flow to generate an agricultural computing task dependency graph. First, we analyze all the computational steps required to transform raw temperature data into the final biological indicator (such as leaf area index), instantiating each step as an independent agricultural computation task node. We traverse these nodes, clarifying the input and output parameter names for each node. We designate the agricultural temperature time-series data as the root node, the key biological growth indicator as the leaf node, and the intermediate computational steps as intermediate nodes. We perform parameter dependency matching; if the output parameter name of a preceding node matches the input parameter name of a subsequent node, we establish a directed edge. Starting from the key biological growth indicator node at the end of the graph as the anchor point, we perform reverse path tracing towards the root node along the opposite direction of the directed edge, recording all traversed node sequences and connecting edges to obtain the set of reverse flow paths.

[0028] The weight generation submodule extracts the input and output parameters of adjacent-level agricultural computing tasks from the reverse flow path set, calculates the Pearson correlation coefficient between the output parameters of the current-level agricultural computing task and the input parameters of the next-level agricultural computing task, introduces the basic sensitivity weight of biological indicators as the initial weight benchmark, performs a transfer correction on the weights inherited by the next-level agricultural computing task based on the Pearson correlation coefficient, and generates task transfer weights. The correction process is implemented using a weighted attenuation algorithm, and the specific calculation formula is as follows: In the formula, This represents the weights passed to the generation task. These values ​​are calculated using a formula and characterize the effective contribution of the current calculation step to the final biological indicator. This represents the basic sensitivity weight of the biometric, with the value derived from 0.684 calculated in the preceding steps, serving as the initial energy for weight transfer. This represents the sequence of output parameters for the current agricultural computing task, with values ​​derived from the task node execution log. This represents the sequence of input parameters for the next-level agricultural computing task, with values ​​derived from data packets received by the next-level node. It represents the covariance of two sets of parameter sequences, and is used to calculate the degree of joint variation of the two sets of data using statistical formulas; and These represent the standard deviations of the two parameter sequences, respectively. The dispersion of the data is obtained through statistical calculation. The product of the covariance and the standard deviation is the Pearson correlation coefficient. To prevent the smoothing factor from being set to a tiny positive number when the denominator is zero, it is set to 0.0001; This is the correlation attenuation index. The parameter is set based on the noise level of the data transmission channel, with two scenarios: if the channel signal-to-noise ratio (SNR) is below a preset threshold (high-noise environment), it is set to 1.2 to accelerate the elimination of weakly correlated paths; if the SNR is above the preset threshold (low-noise environment), it is set to 1.0 to maintain linear transmission. Here, it is set to 1.2.

[0029] Substitute the aforementioned example values ​​into the formula to calculate: 1. Parameter settings: Set to 0.684; Set to 0.0001; Set to 1.2.

[0030] 2. Setting of raw statistical data: Assume that the covariance of the two sets of sequences is obtained by statistically analyzing the parameter sequences of the two most recent calculation periods. The standard deviation of the current layer output sequence is 0.425. The standard deviation of the input sequence of the next layer is 0.5. It is 1.0.

[0031] 3. Substitute numerical values ​​into the formula: .

[0032] The priority score calculation submodule traverses each agricultural computing task instance distributed in the agricultural computing task dependency graph, retrieves the task transfer weight of the agricultural computing task instance on each flow path, performs full-path numerical accumulation on all task transfer weights associated with the same instance, and generates the priority score of the agricultural computing task. Traverse each agricultural computing task instance distributed in the agricultural computing task dependency graph. For each specific computing task, retrieve the task transfer weights obtained by that task instance on all flow paths in the graph. The allocation of these weights is divided into two cases: if the node participates in only one path, the weight of that path is directly extracted; if the node participates in multiple paths, its transfer weights on each path need to be calculated separately and then summed. For example, a node has a transfer weight of 0.5626 on path 1 and a transfer weight of 0.2 on path 2. Perform full path weight summation; the calculation process is as follows: This value reflects the overall contribution of the computational task to multiple key biological indicators, generating an agricultural computational task priority score.

[0033] Please see Figure 6 The task distribution module includes: The execution score calculation submodule performs numerical fusion of the time urgency coefficient and priority score of agricultural computing tasks, quantifies the execution priority of each agricultural computing task to be executed at the current moment, and obtains the agricultural task scheduling execution score. This fusion process introduces a nonlinear dynamic weighting algorithm based on real-time load conditions. A variant of the sigmoid function is used to adjust weight allocation, ensuring that high-urgency tasks are prioritized under high load, while other high-priority tasks are also addressed under low load. The specific calculation formula is as follows: In the formula, This represents the calculated score for agricultural task scheduling and execution. This represents the time urgency coefficient of agricultural computing tasks. The value is derived from the 0.45 calculated in the previous steps. The larger the value, the more time-sensitive the task is. This represents the priority score for agricultural computing tasks. The value is derived from the 0.7626 calculated in the previous steps. The higher the value, the higher the biological value of the task. This represents the real-time load value of the current computing node, a quantified value that combines the CPU and memory usage. The value is derived from the hardware monitoring interface and ranges from 0 to 100. This represents the rated full load baseline value of the computing node. Based on the hardware monitoring interface, it comprehensively quantifies resource utilization indicators such as CPU utilization and memory usage, and uses a percentage to uniformly represent them, so that the real-time load value varies within a closed range of 0 to 100, where 0 indicates that the computing node is in a completely idle state, and 100 indicates that the computing node has reached the rated full load state of its available computing resources. It is usually fixed at 100. This is the load sensitivity index. The parameter is set based on the system's policy preference regarding overload risk, and there are two scenarios: if the system is configured in "high stability mode" (overload averse), it is set to 2 to amplify the load impact; if configured in "high throughput mode" (overload tolerant), it is set to 1 to mitigate the load impact. Here, it is set to 2.

[0034] Substitute the aforementioned example values ​​into the formula to calculate: 1. Parameter settings: Set to 0.45; Set to 0.7626; the current system load is high. Set to 85; Set to 100; Set to 2.

[0035] 2. Substitute numerical values ​​into the formula: .

[0036] The queue descending order reordering submodule calls the task index interface of the agricultural big data computing server waiting queue, maps the agricultural task scheduling execution score to the index attribute of the corresponding task, sets it as the reference benchmark for queue sorting, performs descending order reordering on all agricultural computing tasks residing in the waiting queue, and generates the reordered task sequence. Retrieve object references for all tasks awaiting execution in the queue. Map the calculated agricultural task scheduling execution scores to the corresponding task's index attribute as a sorting criterion. The algorithm compares the execution scores of tasks one by one. Task A has a score of 0.5367, and Task B has a score of 0.6. Since 0.6 is greater than 0.5367, Task B is placed before Task A. Through multiple rounds of comparison and position swapping, the algorithm ensures that the task with the higher score is at the head of the queue, generating a rearranged task sequence.

[0037] The task node distribution submodule determines the order of the rearranged task sequence indexes and extracts the agricultural computing task instances to be processed in sequence. It simultaneously scans and detects the real-time load status of agricultural task computing nodes, filters computing node resources whose current load value is lower than the preset idle state judgment threshold, and distributes the extracted agricultural computing tasks to the filtered idle agricultural task computing nodes to generate and analyze the task scheduling results. Read the CPU utilization and memory usage of each physical computing node. Set an idle state judgment threshold, which is set according to the optimal system operating conditions. The judgment is divided into two cases: if both CPU utilization and memory usage are below 70%, the node is judged as idle; if either indicator is above or equal to 70%, the node is judged as busy. Filter computing node resources whose current load value is below the preset idle state judgment threshold. Node 1 has a utilization of 50%, which is below 70%, so it is judged as an idle node. Transmit task data packets and execution instructions via network sockets to generate analysis task scheduling results.

[0038] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An agricultural planting big data analysis task scheduling system, characterized in that, The system includes: The time-series identification module collects time-series data of the ambient temperature of the planting base according to a specified period, and divides the temperature fluctuation into stable temperature fluctuation ranges and temperature fluctuation ranges exceeding the standard based on the rate of change and amplitude of temperature over time, thereby obtaining the agricultural environment fluctuation state sequence. The urgency assessment module extracts the average temperature change rate of the stable temperature fluctuation segment and the peak temperature change rate of the temperature fluctuation segment exceeding the temperature limit in the agricultural environment fluctuation state sequence to construct the time urgency coefficient of the agricultural computing task. The indicator weight generation module obtains a preset list of agricultural crop yield prediction factors, identifies key biological growth indicators from it, and quantifies the basic biological sensitivity weights of the indicators. The priority assessment module constructs a dependency graph of agricultural computing tasks that shows the flow relationship between agricultural temperature time-series data and key biological growth indicators. It determines the correlation between agricultural computing tasks in the graph and calculates the priority score of agricultural computing tasks by combining the basic sensitivity weights of the biological indicators. The task distribution module integrates the time urgency coefficient and priority score of the agricultural computing task, maps them to the task index in the waiting queue, and obtains the task scheduling results for analysis.

2. The agricultural planting big data analysis task scheduling system according to claim 1, characterized in that, The agricultural environment fluctuation state sequence includes stable temperature fluctuation ranges and temperature exceeding ranges. The agricultural computing task time urgency coefficient is specifically a quantitative value constructed based on the average temperature change rate of the stable temperature fluctuation range and the peak temperature change rate of the temperature exceeding range. The biological indicator basic sensitivity weight is specifically a value of the contribution of key biological growth indicators quantified based on the agricultural crop yield prediction factor association list. The agricultural computing task priority score is specifically a ranking score calculated by combining the task relevance in the agricultural computing task dependency graph with the biological indicator basic sensitivity weight. The analysis task scheduling result includes a comprehensive score after fusion calculation and a task index mapped to the waiting queue.

3. The agricultural planting big data analysis task scheduling system according to claim 1, characterized in that, The timing recognition module includes: The rate of change calculation submodule connects to the agricultural temperature time series data collected by the sensors in the planting base, sets a time window on the time axis, calculates the cumulative rate of change of agricultural temperature in the agricultural temperature time series data within each time window, calculates the numerical difference of the cumulative rate of change of agricultural temperature corresponding to adjacent time windows, and obtains the fluctuation range of temperature change rate. The fluctuation state marking submodule filters the time periods in which the fluctuation amplitude of the temperature change rate is within the preset temperature fluctuation threshold range and marks the corresponding time periods as temperature stable indicators. It also filters the time periods in which the fluctuation amplitude of the temperature change rate exceeds the temperature fluctuation threshold range and marks the corresponding time periods as temperature unstable indicators, thus obtaining a set of temperature fluctuation indicators. The segment aggregation mapping submodule calls the temperature fluctuation identifier set, identifies the temperature stability identifiers continuously distributed on the time axis and summarizes them into temperature fluctuation stability segments, identifies the temperature instability identifiers continuously distributed on the time axis and summarizes them into temperature exceedance fluctuation segments, and performs time sequence splicing and state mapping on all temperature fluctuation stability segments and temperature exceedance fluctuation segments according to the chronological order of the time axis to generate an agricultural environment fluctuation state sequence.

4. The agricultural planting big data analysis task scheduling system according to claim 3, characterized in that, The urgency assessment module includes: The feature value extraction submodule analyzes the distribution of stable temperature fluctuation segments and temperature exceeding fluctuation segments in the agricultural environment fluctuation state sequence on the time axis, traverses the cumulative change rate of agricultural temperature in the stable temperature fluctuation segments, calculates the average temperature change rate, traverses the cumulative change rate of agricultural temperature in the temperature exceeding fluctuation segments, and filters the peak temperature change rate. The stationarity parameter calculation submodule calculates the difference ratio between the peak temperature change rate and the average temperature change rate to obtain a stationarity deviation value, which characterizes the crop state change attribute and obtains the crop growth state transition stationarity parameter. The urgency coefficient mapping submodule takes the crop growth state transition stability parameter as an input variable and substitutes it into the forward mapping transformation function for calculation. The forward mapping transformation function is defined as the input variable. The transformation result is quantified in combination with the crop rhythm mutation requirements to generate the time urgency coefficient of agricultural computing tasks.

5. The agricultural planting big data analysis task scheduling system according to claim 4, characterized in that, The indicator weight generation module includes: The indicator identification submodule obtains a pre-defined list of agricultural crop yield prediction factors, analyzes the mapping relationship between each agricultural crop yield prediction factor and crop yield, and identifies the indicator items that play a dominant role in determining crop yield from all agricultural crop yield prediction factors based on the mapping relationship, generating key biological growth indicators. The coefficient extraction submodule uses the key biological growth index as an index key value to perform a search and matching in the agricultural crop yield prediction factor association list, locates the entry position corresponding to the key biological growth index, extracts the recorded regression coefficient quantitative data that represents the quantitative dependence between the key biological growth index and crop yield, and obtains the yield regression coefficient. The weight quantification submodule calculates the proportion of the absolute value of the current indicator's output regression coefficient in the sum of the absolute values ​​of the output regression coefficients of all key indicators, thus obtaining the basic sensitivity weight of the biological indicator.

6. The agricultural planting big data analysis task scheduling system according to claim 5, characterized in that, The list of factors associated with agricultural crop yield forecasts includes: Collect agricultural planting records within a specified period, analyze the agricultural planting records to extract crop yields for the specified period and a set of multidimensional environmental monitoring variables for the same period. The set of multidimensional environmental monitoring variables includes environmental temperature data, environmental humidity data, light intensity data, and soil moisture content data. For the environmental temperature data, environmental humidity data, light intensity data, and soil moisture content data in the multidimensional environmental monitoring variable set, outlier removal and numerical standardization normalization processing with values ​​exceeding 3 times the standard deviation are performed to generate a standardized environmental dataset. Construct a multiple regression analysis model with crop yield for a specified period as the dependent variable and standardized environmental temperature, standardized environmental humidity, standardized light intensity, and standardized soil moisture content from a standardized environmental dataset as independent variables. The least squares method was used to perform parameter fitting operations on the multiple regression analysis model, and the independent regression coefficients of each environmental monitoring variable with respect to crop yield in a specified period were obtained. Each environmental monitoring variable is defined as an agricultural crop yield predictor, and the corresponding independent regression coefficients are encapsulated as multiple regression statistics of agricultural crop yield. Establish a key-value pair index structure with the correspondence between the names of agricultural crop yield predictors and the statistical data of multiple regression of agricultural crop yield, and generate a list of associations for agricultural crop yield predictors.

7. The agricultural planting big data analysis task scheduling system according to claim 5, characterized in that, The priority assessment module includes: The graph path tracing submodule constructs an agricultural computing task dependency graph of the flow relationship between agricultural temperature time series data and key biological growth indicators. Taking the key biological growth indicator node at the end of the graph as the starting anchor point, it performs reverse path tracing to the agricultural temperature time series data node at the beginning of the graph to determine the spatial distribution of all associated paths and obtain the set of reverse flow paths. The weight generation submodule extracts the input and output parameters of adjacent-level agricultural computing tasks in the reverse flow path set, calculates the Pearson correlation coefficient between the output parameters of the current-level agricultural computing task and the input parameters of the next-level agricultural computing task, introduces the basic sensitivity weight of biological indicators as the initial weight benchmark, performs a transfer correction on the weight inherited by the next-level agricultural computing task based on the Pearson correlation coefficient, and generates task transfer weights. The priority score calculation submodule traverses each agricultural computing task instance distributed in the agricultural computing task dependency graph, retrieves the task transfer weight of the agricultural computing task instance on each flow path, performs full-path numerical accumulation on all the task transfer weights associated with the same instance, and generates an agricultural computing task priority score.

8. The agricultural planting big data analysis task scheduling system according to claim 7, characterized in that, The agricultural computational task dependency mapping for constructing the flow relationship between agricultural temperature time-series data and key biological growth indicators includes: Agricultural temperature time series data is transformed into preset calculation rules for key biological growth indicators, and the preset calculation rules are discretized and decomposed into multiple agricultural computing task instances. Iterate through each agricultural computing task instance and extract the input variable definitions required for the operation and the output variable definitions after the operation is completed. Agricultural temperature time series data is instantiated as the root data node of the graph, key biological growth indicators are instantiated as leaf data nodes of the graph, and all agricultural computing task instances are mapped as intermediate computing nodes of the graph. Perform parameter dependency matching scans on all nodes in the graph, compare the output variable definitions of the preceding node with the input variable definitions of the following node, and when the variable definitions are consistent, establish a directed dependency edge representing the data flow between the two corresponding nodes. Based on directed dependency edges, the root data node, intermediate computing nodes, and leaf data nodes are topologically connected according to the temporal logic of computing flow to generate an agricultural computing task dependency graph.

9. The agricultural planting big data analysis task scheduling system according to claim 7, characterized in that, The task distribution module includes: The execution score calculation submodule performs numerical fusion of the time urgency coefficient of the agricultural computing task and the priority score of the agricultural computing task to quantify the execution priority of each agricultural computing task to be executed at the current moment and obtain the agricultural task scheduling execution score. The queue descending order reordering submodule calls the task index interface of the agricultural big data computing server waiting queue, maps the agricultural task scheduling execution score to the index attribute of the corresponding task, sets it as the reference benchmark for queue sorting, performs descending order reordering processing on all agricultural computing tasks residing in the waiting queue, and generates a reordered task sequence. The task node distribution submodule determines the order of the rearranged task sequence indexes, extracts agricultural computing task instances to be processed, synchronously scans and detects the real-time load status of agricultural task computing nodes, filters computing node resources whose current load value is lower than the preset idle state judgment threshold, and distributes the extracted agricultural computing tasks to the filtered idle agricultural task computing nodes, generating analysis task scheduling results.

10. The agricultural planting big data analysis task scheduling system according to claim 9, characterized in that, For agricultural task scheduling and execution scores The formula used is: ； In the formula, This represents the time urgency coefficient for agricultural computing tasks. This indicates the priority score for agricultural computing tasks. This represents the real-time load value of the current computing node, sourced from the hardware monitoring interface, and ranges from 0 to 100. This represents the rated full-load baseline value for the compute node, derived from the hardware monitoring interface, and is set to 100. This is the load sensitivity index, set according to the system's policy preference for overload risk, with a value range of 1 or 2.

Images