Agricultural product supply and demand prediction and analysis method based on multi-dimensional data

Abstract

The present application relates to the technical field of intelligent agricultural management, and more particularly to a method for predicting and analyzing supply and demand of agricultural products based on multidimensional data, comprising: determining a floating demand point based on the demand stability of a target agricultural product, and determining whether to adjust the establishment method of a regional scheduling set; in stage density acquisition, obtaining an initial regional scheduling set based on a stage density change value, determining a discrete attribution benchmark based on distribution dispersion, and merging to obtain a regional scheduling set; in point attribute acquisition, covering the range of demand floating points based on a stage demand difference value to obtain a regional scheduling set; determining whether to perform interactive node analysis for the floating proportion in the regional scheduling set; obtaining a supply and demand tension information combination of the floating demand point by predicting the supply and demand of the regional scheduling set; and sequentially calling the predicted tension value of the floating demand point. The present application improves the data acquisition accuracy and flexibility of agricultural product prediction.

Description

Technical Field

[0001] This invention relates to the field of smart agricultural management technology, and in particular to a method for predicting and analyzing the supply and demand of agricultural products based on multidimensional data. Background Technology

[0002] With the rapid development of artificial intelligence technology, various predictive models are being applied more and more widely in the agricultural field. Predictive methods based on deep learning, time series analysis, and multi-source data fusion not only break through the limitations of traditional agricultural management that relies solely on order transaction data or market experience, but also can identify risk areas in advance before supply and demand tensions occur, thereby effectively alleviating local product overstocking or inventory shortages and reducing fresh produce spoilage caused by supply and demand imbalances. However, existing agricultural product dispatching systems mostly use administrative regions or large wholesale markets as basic units for supply and demand coordination. Dispatch decisions typically rely on macro-level statistical data, lacking refined analysis of the demand characteristics and spatiotemporal distribution patterns of various sales locations within a local area. Agricultural products have significant seasonal characteristics, with substantial differences in maturity periods, shelf lives, and market demand fluctuations. Furthermore, the historical purchasing behavior, inventory turnover capacity, and radiating influence on surrounding areas vary from sales point to sales point. During cross-regional transportation, the failure to fully capture the individual differences and dynamic changes in demand points within local areas often leads to sudden short-term supply shortages, affecting the timeliness and effectiveness of dispatch response. Therefore, how to obtain stable supply and demand forecast data by reasonably dividing supply and demand areas is a problem of great concern to those skilled in the art.

[0003] Chinese Patent Publication No. CN122288235A discloses a method for predicting agricultural product supply and demand based on multi-source time-series data. The method includes: constructing a causal state-space model of the agricultural product supply and demand system; collecting multi-source time-series data streams and identifying event instances; mapping these event instances to the causal state-space model to update their corresponding state variable values; using the current state of the causal state-space model as a starting point; performing forward reachability analysis to deduce all possible future state sets; and finally, intersecting the supply and demand prediction request with the future state sets to generate the prediction result, thus achieving causally interpretable prediction of agricultural product supply and demand. However, this technical solution suffers from the following problems: it does not fully consider the historical demand behavior characteristics and spatial distribution patterns of sales points, and cannot flexibly adjust the establishment method of the scheduling set according to the actual distribution, resulting in a lack of adaptive scheduling response capability when regional demand fluctuates. Summary of the Invention

[0004] To address this, the present invention provides a method for predicting and analyzing the supply and demand of agricultural products based on multidimensional data. This method overcomes the problem in existing technologies that do not fully consider the historical demand behavior characteristics and spatial distribution patterns of sales points, and cannot flexibly adjust the establishment method of the scheduling set according to the actual distribution situation, resulting in a lack of adaptive scheduling response capability when regional demand fluctuates differently.

[0005] To achieve the above objectives, this invention provides a method for predicting and analyzing the supply and demand of agricultural products based on multidimensional data, comprising: Based on the demand stability of the target agricultural products within the supply and demand forecast coverage area, determine whether the demand point is a floating demand point, and based on the local clustering density of the floating demand point, determine whether to adjust the establishment method of the regional scheduling set from point attribute acquisition to stage density acquisition. In the process of obtaining the phase density, it is determined whether the sub-regions within the supply and demand forecast coverage area are clustered sub-regions based on the floating density, and the initial regional scheduling set is determined based on the phase density change value of the clustered sub-regions. The discrete attribution benchmark is determined based on the distribution dispersion of the discrete sub-regions for merging, and the regional scheduling set is obtained. In the point attribute acquisition process, the range of demand floating points is covered based on the stage demand difference value to obtain the regional scheduling set; To determine whether to perform interaction node analysis based on the floating proportion in the regional scheduling set, the selection criteria for the region and the number of regions to be selected are determined based on the product maturity and stability of the floating demand points. Supply and demand forecasting is performed on the regional scheduling set to obtain the supply and demand tension information combination of floating demand points, and the predicted tension value of floating demand points is used to retrieve the information from the exchange nodes in sequence.

[0006] Furthermore, the point where the demand stability of the target agricultural product is less than the preset demand stability during the stage forecast period is denoted as the floating demand point. The phased prediction period consists of consecutive historical target time periods.

[0007] Furthermore, for the condition that the local clustering density of floating demand points is greater than or equal to the preset local clustering density, the method for establishing the regional scheduling set is changed from point attribute acquisition to stage density acquisition. The local cluster density is the ratio of the dense quantity range to the supply and demand forecast coverage area.

[0008] Furthermore, the supply and demand forecast coverage area is uniformly divided to obtain several sub-regions with the same area. Sub-regions with a floating density greater than the preset floating density are recorded as clustered sub-regions. The phase coverage analysis is performed sequentially according to the demand point density of the clustered sub-regions from large to small, and the sub-regions whose phase density change value is less than the preset phase density change value are constructed as the initial region scheduling set. Subregions that did not participate in the construction of the initial region scheduling set are denoted as discrete subregions.

[0009] Furthermore, for the condition that the distribution dispersion of the discrete sub-region is greater than or equal to the preset distribution dispersion, the discrete attribution benchmark is determined to be the center distance ratio; For the condition that the dispersion of the discrete sub-regions is less than the preset dispersion, the discrete attribution benchmark is determined to be the initial floating quantity ratio; Each discrete sub-region is merged into the initial region scheduling set with the largest discrete attribution benchmark to obtain the region set.

[0010] Furthermore, considering the condition that the local clustering density of floating demand points is less than the preset local clustering density, the method for establishing the regional scheduling set is determined to be point attribute acquisition, including: Arrange the demand fluctuation points in descending order of the stage demand difference value to obtain the order list; Following the order of establishment, the range of demand floating points is covered sequentially, and the demand points that meet the range coverage conditions are constructed into a regional scheduling set.

[0011] Furthermore, for the regional scheduling set whose floating percentage is greater than or equal to the preset floating percentage, it is denoted as the scheduling analysis set, and interaction node analysis is determined. The selected number of exchange nodes are obtained sequentially within the selected area in descending order of the differential adjustment value.

[0012] Furthermore, in the interaction node analysis, for the condition that the product maturity stability of floating demand points in the scheduling analysis set is less than the preset product maturity stability, the selection area of ​​the exchange node is determined as the coverage area of ​​the scheduling analysis set, and the confirmation criterion for the selection number of exchange nodes is the product maturity stability.

[0013] Furthermore, based on the condition that the product maturity stability of floating demand points in the scheduling analysis set is greater than or equal to the preset product maturity stability, the selection area for exchange nodes is determined to be the expanded overlapping area, and the confirmation criterion for the number of exchange nodes is the neighborhood connectivity density.

[0014] Furthermore, the forecasting baseline parameters of each regional scheduling set are obtained to forecast the supply and demand of each demand point to obtain a combination of supply and demand tension information for each floating demand point. Target agricultural products are then retrieved from floating demand points and exchange nodes whose forecast tension values ​​are greater than the preset forecast tension values ​​in descending order.

[0015] Compared with the prior art, the beneficial effects of the present invention are that it reflects the stability of the continuous demand for the target agricultural product by the demand stability of the demand point within the stage forecast period, thereby determining the possibility of supply and demand deviation of the demand point in the future supply and demand forecast period, releasing higher attention in advance and responding to the demand of the demand point for the target agricultural product in a timely manner, further improving the accuracy of problem point identification in supply and demand forecasting, and providing a construction benchmark for the subsequent establishment of regional scheduling sets.

[0016] Furthermore, this invention uses local clustering density to reflect the density of floating demand points in a local area within the supply and demand coverage range, thereby identifying local clustering areas where floating demand points exist. This may lead to fluctuations in demand clustering of floating demand points, affecting the overall supply and demand forecasting accuracy of the surrounding area. In this case, stage density is used to obtain a regional scheduling set that can cover locally clustered floating demand points, and demand points in discrete sub-regions are used for demand buffering to further stabilize the basic demand in the region, thereby alleviating supply and demand fluctuations within the overall regional set and further improving the flexibility of constructing the regional scheduling set.

[0017] Furthermore, this invention uses the stage demand difference value to reflect the degree of difference in demand changes of each floating demand point within each stage forecast period, thereby determining the order of establishing regional scheduling sets. Higher priority is given to the construction of regional scheduling sets for more unstable floating demand points, thus obtaining regional scheduling sets that may contain more demand points, further increasing the adaptability of regional scheduling sets.

[0018] Furthermore, this invention reflects the proportion of floating demand points within a regional scheduling set by the floating percentage, thereby determining a preliminary prediction of the stability of supply and demand within the regional scheduling set during the future supply and demand forecasting period. Subsequently, different exchange node setting methods are adopted to optimize the scheduling of target agricultural products, further improving the flexibility of forecast data acquisition and providing a data foundation for product scheduling within the region.

[0019] Furthermore, this invention reflects the stability of the target agricultural products in the regional scheduling set in terms of maturity through product maturity stability, determines the setting location and quantity benchmark of the exchange nodes, and thus prioritizes scheduling within the regional scheduling set when the maturity stability of the target agricultural products is poor to reduce transportation losses. When the maturity stability of the target agricultural products is high, it can exchange with the outside through the edge portal to obtain a more sufficient source of target agricultural products, further improving the flexibility of supply and demand forecasting. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the agricultural product supply and demand forecasting and analysis method based on multidimensional data, as described in an embodiment of the present invention. Figure 2 This is a flowchart illustrating how to determine whether a demand point is a floating demand point based on demand stability, according to an embodiment of the present invention. Figure 3 This is a flowchart illustrating whether to adjust the method of establishing the regional scheduling set from point attribute acquisition to stage density acquisition based on the local cluster density of floating demand points in an embodiment of the present invention. Figure 4 This is a flowchart illustrating how to determine whether to perform interactive node analysis based on the floating percentage in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0022] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] It should be noted that in the description of this invention, the terms "upper," "lower," "left," "right," "inner," and "outer," etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is merely for ease of description and does not indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0024] Please see Figure 1 As shown, it is a schematic diagram of the agricultural product supply and demand forecasting and analysis method based on multidimensional data according to an embodiment of the present invention. The method includes: Based on the demand stability of the target agricultural products within the supply and demand forecast coverage area, determine whether the demand point is a floating demand point, and based on the local clustering density of the floating demand point, determine whether to adjust the establishment method of the regional scheduling set from point attribute acquisition to stage density acquisition. In the process of obtaining the phase density, it is determined whether the sub-regions within the supply and demand forecast coverage area are clustered sub-regions based on the floating density, and the initial regional scheduling set is determined based on the phase density change value of the clustered sub-regions. The discrete attribution benchmark is determined based on the distribution dispersion of the discrete sub-regions for merging, and the regional scheduling set is obtained. In the point attribute acquisition process, the range of demand floating points is covered based on the stage demand difference value to obtain the regional scheduling set; To determine whether to perform interaction node analysis based on the floating proportion in the regional scheduling set, the selection criteria for the region and the number of regions to be selected are determined based on the product maturity and stability of the floating demand points. Supply and demand forecasting is performed on the regional scheduling set to obtain the supply and demand tension information combination of floating demand points, and scheduling analysis is performed sequentially using the predicted tension values ​​of floating demand points.

[0025] In this embodiment of the invention, the information acquisition process of demand points includes: at the end of the stage prediction period, detecting the circulation information of the target agricultural products of each demand point to obtain a regional scheduling set, wherein the demand point is a sales point of the target agricultural products with storage capacity, such as a wholesale market in the place of origin, a distributor, and an agricultural cooperative, and the circulation information includes, but is not limited to, purchase information and location information.

[0026] The acquisition of purchase information includes: demand points purchasing the target agricultural product from the supplier of the target agricultural product through online channels or direct offline transactions. The supplier can be, but is not limited to, the planting base of the target agricultural product. A big data model is used to obtain the time when each demand point receives the target agricultural product within the stage prediction period, which is recorded as the detection time of each demand point. The weight of the target agricultural product already present at the demand point at that detection time is recorded as the remaining inventory, and the weight of the received target agricultural product is recorded as the newly added inventory. The target agricultural product can be, but is not limited to, fruits and vegetables.

[0027] The process of obtaining location information includes: using the geocoding function of the GIS platform to convert the text addresses of each demand point into spatial coordinates on the road network, and using the path planning engine to calculate the passable path between any two demand points.

[0028] In this embodiment of the invention, the process of acquiring the supply and demand forecasting model includes: the supply and demand forecasting model is acquired using a time-series-based LSTM model. Specifically, the prediction benchmark parameters for each forecasting period within the most recent historical observation period are used as training samples. The labeled training samples are input into the LSTM model for iterative training until the supply and demand forecasting accuracy reaches a level that meets user needs, at which point training stops, and the trained supply and demand forecasting model is used for forecasting. At the beginning of each supply and demand forecasting period, the prediction benchmark parameters for each demand point within each regional scheduling set are input into the supply and demand forecasting model to obtain the supply and demand tension information combination within the forecasting window. After each forecasting period ends, the actual supply and demand tension information combination is compared with the supply and demand tension information combination predicted by the model for update training.

[0029] Please see Figure 2 As shown, it is a flowchart of an embodiment of the present invention for determining whether a demand point is a floating demand point based on demand stability. For demand points where the demand stability of the target agricultural product is less than the preset demand stability within the stage forecast period, they are recorded as floating demand points. The stage prediction period consists of continuous historical target time periods. The condition for continuity is: for a single historical target time period, if the time interval between its start time and the start time of an adjacent historical target time period is less than or equal to the duration of that single time period, then the historical target time period and the adjacent historical target time period can be considered continuous historical target time periods. The historical target time period is a historical time period where the newly added parameter is greater than a preset newly added parameter. The newly added parameter is the ratio of the sum of the newly added inventory at the detection time of each demand point within that historical time period to the duration of the historical time period. It is worth noting that the historical time period refers to the period prior to the current time. It can be understood that the newly added parameter is used to quantify the overall activity level of the target agricultural product entering the supply and demand prediction coverage within the historical time period. The larger the newly added parameter, the more intensive the replenishment behavior of the demand points within that historical time period, the more active the market circulation, and the more representative that period is of the normal supply and sales rhythm of the target agricultural product.

[0030] In this embodiment of the invention, a time period is used to cyclically acquire target agricultural product data from various demand points. The acquisition process of the time period includes: simulating the acquisition of historical data from demand points under different time period conditions by collecting multiple sets of such data; setting different time period thresholds for data acquisition testing; and selecting the average value of the time period thresholds that meet user requirements for supply and demand forecast accuracy as the duration of the time period. In this embodiment, the longer the time period, the more demand point data is acquired in a single acquisition, but the lower the acquisition frequency. The higher the user's requirement for data acquisition accuracy, the shorter the time period value. The data includes, but is not limited to, the newly added inventory and remaining inventory of target agricultural products at each detection time within the time period. It is understood that the supply and demand forecast accuracy rate is the proportion of the number of demand points actually experiencing supply and demand tensions in the forecast results after supply and demand forecasting of the regional scheduling set, to the total number of demand points predicted to experience supply and demand tensions. The higher the supply and demand forecast accuracy rate, the higher the accuracy of the system's construction of the stage forecast cycle, and the greater the user's sensitivity to this indicator. Therefore, when selecting a preset value using the supply and demand forecast accuracy rate as an evaluation indicator, the requirement for accuracy is higher. The setting of the evaluation indicator is content already understood by those skilled in the art and will not be elaborated further.

[0031] The process of obtaining demand stability includes: demand stability is determined based on the average value of the detection difference in each complete stage prediction period within the historical observation period, and the demand stability and the average value of the detection difference are positively correlated. In this embodiment, each complete stage prediction period within the historical observation period is obtained, and the demand stability is the average value of the detection difference of a single demand point in each stage prediction period. For a single stage detection period, the detection difference is the ratio of the number of detection moments in the stage prediction period where the new inventory of the target agricultural product is greater than the preset new inventory to the average number of detection moments of each demand point in the stage prediction period. It can be understood that demand stability is used to quantify the continuous stability of the demand for the target agricultural product by a demand point within the stage prediction period. The greater the demand stability, the more consistently the new inventory of the demand point is higher than the preset threshold in the historical target period, the more stable the demand for the target agricultural product, and the less likely the demand point is to experience supply and demand deviations within the prediction period; the smaller the demand stability, the more concentrated the replenishment behavior of the demand point is in a few moments during the historically high-activity period, the more unstable the demand for the target agricultural product, and the higher the probability of supply and demand tensions for the demand point within the prediction period.

[0032] In this embodiment of the invention, the process of obtaining the preset new parameters includes: simulating multiple sets of historical time period data under different new parameter conditions; setting different new parameter thresholds for stage prediction cycle construction tests; and selecting the average value of the new parameter thresholds whose supply and demand prediction accuracy meets user requirements as the preset new parameters. The higher the user's requirement for supply and demand prediction accuracy, the smaller the value of the preset new parameters.

[0033] The process of obtaining the preset new inventory quantity includes: simulating demand data under multiple sets of different new inventory quantity conditions; calculating and testing demand stability by setting different new inventory quantity thresholds; and selecting the average value of the new inventory quantity thresholds that meet user needs in terms of supply and demand forecast accuracy as the preset new inventory quantity. In this embodiment, the higher the user's requirement for the accuracy of identifying floating demand points, the smaller the value of the preset new inventory quantity.

[0034] The process of obtaining the preset demand stability includes: simulating multiple sets of demand point data with different demand stability characteristics; setting different demand stability thresholds for floating demand point determination tests; and selecting the average value of the demand stability thresholds that meet user needs in terms of supply and demand forecast accuracy as the preset demand stability. In this embodiment, the higher the user's requirement for the accuracy of floating demand point identification, the smaller the value of the preset demand stability.

[0035] The process of obtaining the preset historical observation period includes: simulating demand data collected under multiple sets of different historical observation period conditions; calculating and testing demand stability by setting different historical observation period thresholds; and selecting the average value of the historical observation period thresholds that meet user requirements for supply and demand forecast accuracy as the preset historical observation period. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the larger the value of the preset historical observation period.

[0036] Please see Figure 3 As shown, it is a flowchart of an embodiment of the present invention for determining whether to change the method of establishing the regional scheduling set from point attribute acquisition to stage density acquisition based on the local cluster density of floating demand points. For the condition that the local cluster density of floating demand points is greater than or equal to the preset local cluster density, it is determined that the method of establishing the regional scheduling set is changed from point attribute acquisition to stage density acquisition. The local cluster density is the ratio of the dense quantity range to the supply and demand forecast coverage area.

[0037] In this embodiment of the invention, the process of obtaining the dense quantity range includes: the dense quantity range is determined based on the smallest circular area that can contain the dense quantity of floating demand points, and the dense quantity range and the dense quantity are positively correlated. In this embodiment, the dense quantity range is the smallest circular area that can contain the dense quantity of floating demand points, and the process of obtaining the dense quantity includes: if the number of floating demand points within the supply and demand forecast coverage area is greater than a preset number of floating demand points, then the dense quantity is the product of a floating adjustment coefficient and a base quantity, where the floating adjustment coefficient is the ratio of the number of floating demand points to the preset number of floating demand points; if the number of floating demand points within the supply and demand forecast coverage area is less than or equal to the preset number of floating demand points, then the dense quantity is the base quantity. It is worth noting that the dense quantity is an integer rounded up. Understandably, the dense quantity range is used to quantify the size of the spatial range of local clustering of floating demand points within the supply and demand forecast coverage area. The larger the dense quantity range, the more clustered the spatial distribution of floating demand points in the area, and the higher the local clustering density. In the scenario of dense clustering of floating demand points, stage density analysis is performed at the sub-region level to avoid the fragmented boundary division of the regional scheduling set caused by the point attribute acquisition method when dealing with dense distribution due to the range coverage of each point. This would not effectively cover the overall demand characteristics of the clustered area and ensure that the establishment method of the regional scheduling set matches the actual distribution pattern of floating demand points.

[0038] In this embodiment of the invention, the process of obtaining the basic quantity includes: simulating the distribution of demand points under multiple sets of different basic quantity conditions; conducting intensive quantity calculation tests by setting different basic quantity thresholds; and selecting the average value of the basic quantity thresholds whose supply and demand forecast accuracy meets user requirements as the basic quantity. In this embodiment, the basic quantity is used to determine whether each floating demand point is clustered. The larger the basic quantity, the more floating demand points are obtained. If the coverage area of ​​the obtained floating demand points is smaller, the degree of clustering of floating demand points is higher. The higher the user's requirement for supply and demand forecast accuracy, the larger the value of the basic quantity.

[0039] The process of obtaining the preset number of floating demand points includes: simulating the distribution of demand points under multiple sets of conditions with different numbers of floating demand points; conducting intensive quantity calculation tests by setting different thresholds for the number of floating demand points; and selecting the average value of the floating demand point thresholds that meet user requirements for supply and demand forecast accuracy as the preset number of floating demand points. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset number of floating demand points.

[0040] The process of obtaining the preset local cluster density includes: simulating the distribution of demand points under multiple sets of conditions with different local cluster densities; setting different local cluster density thresholds to test the regional scheduling set establishment method; and selecting the average value of the local cluster density thresholds that meet user requirements for supply and demand forecast accuracy as the preset local cluster density. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset local cluster density.

[0041] Specifically, the supply and demand forecast coverage area is uniformly divided to obtain several sub-regions with the same area, and the sub-regions with a floating density greater than the preset floating density are recorded as clustered sub-regions. According to the order of demand point density from large to small, the phase coverage analysis is performed on each cluster sub-region in turn, and the sub-regions whose phase density change value is less than the preset phase density change value are constructed as the initial region scheduling set. Subregions that did not participate in the construction of the initial region scheduling set are denoted as discrete subregions.

[0042] The enclosing uniform segmentation includes: obtaining the smallest rectangle that can encompass the supply and demand forecast coverage area, denoted as the enclosing region; uniformly segmenting the enclosing region to obtain several initial sub-regions with the same area; removing the initial sub-regions without demand points; and denoting the remaining initial sub-regions as sub-regions. Simulations are performed by collecting supply and demand forecast coverage segmentation data under multiple sets of conditions with different numbers of sub-regions. Different sub-region number thresholds are set for region division and supply and demand forecast testing. The average value of the sub-region number thresholds that meet user requirements for supply and demand forecast accuracy is selected as the number of sub-regions. In this embodiment, the higher the user's requirement for the fineness of region division, the larger the value of the number of sub-regions.

[0043] The process of performing phase coverage analysis on a single sub-region includes: designating the sub-region as the splicing region, obtaining each adjacent sub-region as a candidate region, and calculating the phase density change value sequentially according to the demand point density of the candidate regions in descending order for merging analysis. In a single merging analysis, if the phase density change value corresponding to the merged splicing region is less than the preset phase density change value, then the merging is determined to meet the merging analysis conditions; otherwise, the merging is determined to not meet the merging analysis conditions, the merging is canceled, and the candidate region is skipped, continuing to perform merging analysis with the remaining candidate regions. This process continues until all candidate regions have undergone merging analysis, then the number of sub-regions in the splicing region is checked. If the number of sub-regions in the splicing region has not changed, the acquisition of candidate regions stops; if the number of sub-regions in the splicing region has changed, the process continues to obtain each adjacent sub-region as a candidate region and perform merging analysis on each candidate region. This cycle continues until none of the candidate regions obtained in a single instance meet the merging analysis conditions, at which point the acquisition of candidate regions stops. The demand points in the last obtained splicing region are recorded as the initial region scheduling set.

[0044] In this embodiment of the invention, the floating density is obtained by determining the floating density based on the ratio of the number of floating demand points within a single sub-region to the area of ​​the sub-region. The floating density is positively correlated with the number of floating demand points and negatively correlated with the area of ​​the sub-region. In this embodiment, the floating density is the ratio of the number of floating demand points within a single sub-region to the area of ​​the sub-region. It can be understood that the floating density is used to quantify the spatial concentration of floating demand points in each sub-region within the supply and demand forecast coverage area. The higher the floating density, the denser the distribution of unstable demand points within that sub-region, and the more concentrated the risk of demand fluctuations.

[0045] In this embodiment of the invention, the process of obtaining demand point density includes: demand point density is determined based on the ratio of the number of demand points within a single sub-region to the area of ​​the sub-region. The demand point density is positively correlated with the number of demand points and negatively correlated with the area of ​​the sub-region. In this embodiment, demand point density is the ratio of the number of demand points within a single sub-region to the area of ​​the sub-region. It can be understood that demand point density is used to quantify the overall spatial distribution density of demand points in each sub-region within the supply and demand forecast coverage area. The higher the demand point density, the more sales points there are in that sub-region, and the larger the potential scale of supply and demand interaction.

[0046] In this embodiment of the invention, the process of obtaining the stage density change value includes: the stage density change value is determined based on the difference between the demand point density of the spliced ​​region after a single merger and the demand point density of the spliced ​​region before the merger, wherein the stage density change value and the difference in demand point density before and after the merger are positively correlated. In this embodiment, the stage density change value is the difference between the demand point density of the spliced ​​region after a single merger and the demand point density of the spliced ​​region before the merger. It can be understood that the stage density change value is used to quantify the magnitude of the change in demand point density of the spliced ​​region during the sub-region merger process. The smaller the stage density change value, the smoother the change in demand point density of the spliced ​​region before and after the merger, and the less significantly it changes the overall demand distribution characteristics of the spliced ​​region.

[0047] The process of obtaining the preset floating density includes: simulating the supply and demand forecast coverage sub-region data under multiple sets of different floating density conditions; setting different floating density thresholds for clustered sub-region determination tests; and selecting the average value of the floating density thresholds that meet user requirements for supply and demand forecast accuracy as the preset floating density. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset floating density.

[0048] The process of obtaining the preset stage density change value includes: simulating the process by collecting multiple sets of sub-region merged data under different stage density change value conditions; conducting stage coverage analysis and judgment tests by setting different stage density change value thresholds; and selecting the average value of the stage density change value thresholds that meet the user's supply and demand forecast accuracy requirements as the preset stage density change value. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the preset stage density change value.

[0049] Specifically, for the condition that the dispersion of the discrete sub-region is greater than or equal to the preset dispersion, the discrete attribution benchmark is determined as the center distance ratio; For the condition that the dispersion of the discrete sub-regions is less than the preset dispersion, the discrete attribution benchmark is determined to be the initial floating quantity ratio; Each discrete sub-region is merged into the initial region scheduling set with the largest discrete attribution benchmark to obtain the region set, and the sub-regions in the region scheduling set are recorded as the coverage area corresponding to the region scheduling set.

[0050] In this embodiment of the invention, the process of obtaining the distribution dispersion includes: the distribution dispersion is determined based on the ratio of the area of ​​the smallest circular region that can contain all discrete sub-regions to the sum of the areas of the discrete sub-regions. The distribution dispersion is positively correlated with the area of ​​the smallest circular region and negatively correlated with the sum of the areas of the discrete sub-regions. In this embodiment, the distribution dispersion is the ratio of the area of ​​the smallest circular region that can contain all discrete sub-regions to the sum of the areas of the discrete sub-regions. It can be understood that the distribution dispersion is used to quantify the spatial dispersion of discrete sub-regions within the supply and demand forecast coverage area. The larger the distribution dispersion, the wider the spatial distribution range of the discrete sub-regions, the greater the distance between them, and the weaker the spatial correlation between the discrete sub-regions.

[0051] When merging a single discrete sub-region with the initial region scheduling set, the analysis is performed sequentially in ascending order of the minimum distance between the area center point of the discrete sub-region and the boundary of the corresponding region of each initial region scheduling set.

[0052] In this embodiment of the invention, the process of obtaining the center distance ratio includes: the center distance ratio is determined based on the ratio of the distance between the center point of the area of ​​the discrete sub-region and the center of the circle of the ideal region of each initial region scheduling set to the radius of the ideal region. The center distance ratio is positively correlated with the center connection distance and negatively correlated with the radius of the ideal region. In this embodiment, the center distance ratio is the ratio of the radius of the ideal region of the initial region scheduling set to the center connection distance corresponding to the discrete sub-region. The center connection distance is the distance between the center point of the area of ​​the discrete sub-region and the center of the circle of the ideal region of each initial region scheduling set. The ideal region is the smallest circle that can enclose the initial region scheduling set. It can be understood that the center distance ratio is used to quantify the spatial proximity between the discrete sub-region and each initial region scheduling set. The smaller the center distance ratio, the closer the discrete sub-region is to the core region of the initial region scheduling set, and the better the spatial connectivity of the merged region scheduling set.

[0053] The process of obtaining the initial floating quantity ratio includes: the initial floating quantity ratio is determined based on the ratio of the number of floating demand points within the initial regional scheduling set to the center connection distance. The initial floating quantity ratio is positively correlated with the number of floating demand points and negatively correlated with the center connection distance. In this embodiment, for a single initial regional scheduling set, the initial floating quantity ratio is the ratio of the number of floating demand points within that initial regional scheduling set to the center connection distance. It can be understood that the initial floating quantity ratio is used to quantify the overall matching degree between the density of floating demand points already gathered within each initial regional scheduling set and the spatial distance when discrete sub-regions merge into each initial regional scheduling set. A larger initial floating quantity ratio indicates that there are already more floating demand points within the initial regional scheduling set, and the spatial distance between the discrete sub-region and the set is relatively short.

[0054] The process of obtaining the preset distribution dispersion includes: simulating the distribution of discrete sub-regions under multiple sets of data with different distribution dispersion conditions; setting different distribution dispersion thresholds for discrete attribution benchmark determination tests; and selecting the average value of the distribution dispersion thresholds that meet user requirements for supply and demand forecast accuracy as the preset distribution dispersion. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset distribution dispersion.

[0055] Specifically, for the condition that the local clustering density of floating demand points is less than the preset local clustering density, the method for establishing the regional scheduling set is determined to be point attribute acquisition, including: Arrange the demand fluctuation points in descending order of the stage demand difference value to obtain the order list; Following the order of establishment, the range of demand floating points is covered sequentially, and the demand points that meet the range coverage conditions are constructed into a regional scheduling set.

[0056] In this embodiment of the invention, the method for obtaining the stage demand difference value includes: the stage demand difference value is determined by comparing the average value of the detection reference value in each stage prediction period within the historical observation period with the preset detection reference average value; If the average value of the detection reference value of each floating demand point in each stage prediction period within the historical observation period is greater than or equal to the preset average value of the detection reference value, then the stage difference value is determined based on the number of detection times of the floating demand point. It is positively correlated with the number of detection times of a single floating demand point and negatively correlated with the average value of the number of detection times of each floating demand point. Specifically, the stage difference value is the difference obtained by subtracting the average value of the number of detection times of each floating demand point from the number of detection times of a single floating demand point. If the average value of the detection reference value in each stage prediction period within the historical observation period is less than the preset average value of the detection reference value, then the stage difference value is determined based on the new inventory quantity at each detection time and is positively correlated with the standard deviation of the new inventory quantity at each detection time. Specifically, the stage demand difference value is the standard deviation of the new inventory quantity at each detection time within the most recent complete stage period for a single floating demand point. The detection reference value is the standard deviation of the sum of the number of detection moments of each floating demand point in a single stage forecast period. It can be understood that the stage demand difference value is used to quantify the degree of difference in the rhythm of demand change of each floating demand point within the stage forecast period. The larger the stage demand difference value, the more significant the deviation of the number of detection moments of the floating demand point from the overall average level, and the greater the fluctuation of its new inventory. This indicates that the demand behavior of the demand point is more unstable and is more likely to produce supply and demand deviation in subsequent supply and demand forecasts.

[0057] In this embodiment of the invention, the process of obtaining the coverage range includes: recording a single floating demand point as a base point, recording demand points within the construction distance of the base point as new base points, and repeating this process until no new base points appear, at which point the loop stops. Each base point is recorded as a regional scheduling set, and the smallest range that can surround the regional scheduling set is recorded as the coverage range corresponding to the regional scheduling set. It is worth noting that if a floating demand point is already within a regional scheduling set, the construction of the regional scheduling set is no longer performed; for any missed demand points, they are merged into the nearest regional scheduling set.

[0058] The process of obtaining the preset reference average value includes: simulating the historical data of floating demand points under multiple sets of different reference value conditions; setting different threshold values ​​for the reference average value to determine the calculation method of stage demand difference; and selecting the average value of the threshold values ​​that meets user needs for supply and demand forecast accuracy as the preset reference average value. The more sensitive the user is to the frequency of the target agricultural products entering the warehouse at each floating demand point, the smaller the preset reference average value should be.

[0059] The process of obtaining the construction distance includes: simulating the spatial distribution of floating demand points under multiple sets of construction distance conditions; setting different construction distance thresholds to test range coverage and regional scheduling set construction; and selecting the average value of the construction distance thresholds that meet user requirements for supply and demand forecast accuracy as the construction distance. In this embodiment, the higher the user's requirement for the completeness of the regional scheduling set construction, the larger the construction distance value.

[0060] Please see Figure 4As shown, it is a flowchart of an embodiment of the present invention for determining whether to perform interactive node analysis based on the floating ratio. For the regional scheduling set where the floating ratio is less than the preset floating ratio, the number of selected exchange nodes is determined as the number of base nodes. For the set of regions where the floating percentage is greater than or equal to the preset floating percentage, we denote it as the scheduling analysis set and determine the interaction node analysis to be performed. During the selection process, the selected number of exchange nodes are obtained sequentially within the selection area in descending order of the differential adjustment value.

[0061] In this embodiment of the invention, the process of obtaining the floating ratio includes: the floating ratio is determined based on the ratio of the number of floating demand points to the total number of demand points in the regional scheduling set. The floating ratio is positively correlated with the number of floating demand points and negatively correlated with the total number of demand points. In this embodiment, the floating ratio is the ratio of the number of floating demand points to the total number of demand points in the regional scheduling set. It can be understood that the floating ratio is used to quantify the relative proportion of unstable demand nodes within the regional scheduling set. The larger the floating ratio, the greater the likelihood that the regional scheduling set will experience an overall demand forecast deviation in the future supply and demand forecasting period.

[0062] The process of obtaining the preset floating percentage includes: simulating multiple sets of regional scheduling data under different floating percentage conditions; setting different floating percentage thresholds for interactive node analysis and trigger judgment tests; and selecting the average value of the floating percentage thresholds that meet user requirements for supply and demand forecast accuracy as the preset floating percentage. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset floating percentage.

[0063] Specifically, in the interaction node analysis, for the condition that the product maturity and stability of floating demand points in the scheduling analysis set is less than the preset product maturity and stability, the selection area of ​​the exchange node is determined as the coverage area of ​​the scheduling analysis set, and the confirmation criterion for the number of exchange nodes is the product maturity and stability. The exchange node is the demand point where the differential adjustment value within the stage prediction period is greater than the preset differential adjustment value.

[0064] In this embodiment of the invention, the process of obtaining the differential transfer value includes: the differential transfer value is determined based on the average transfer ratio in each stage of the forecast period, and the differential transfer value and the average transfer ratio are positively correlated. In this embodiment, the differential transfer value is the average transfer ratio in each stage of the forecast period, and the transfer ratio is the ratio of the cumulative transfer value of the demand point in the most recent complete stage of the forecast period to the sum of the new inventory in each detection time. Among them, the cumulative transfer value is the total weight of the target agricultural products transferred from the demand point to other demand points in the stage of the forecast period. The transfer records are obtained through big data. It can be understood that the differential transfer value is used to quantify the activity level and supply contribution level of the demand point as the source of product transfer in the historical stage of the forecast period. The larger the differential transfer value, the higher the proportion of the cumulative amount of products transferred by the demand point to other demand points in the historical period to its own new inventory, indicating that the demand point not only has a stable product inflow channel, but also has the capacity to radiate supply to the surrounding areas, and its node value in regional supply and demand scheduling is more significant.

[0065] The process of obtaining the preset differential adjustment value includes: simulating the historical adjustment data of demand points under multiple sets of different differential adjustment value conditions; setting different differential adjustment value thresholds for exchange node determination tests; and selecting the average value of the differential adjustment value thresholds that meet user requirements for supply and demand forecast accuracy as the preset differential adjustment value. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the preset differential adjustment value.

[0066] In this embodiment of the invention, the process of obtaining product maturity stability includes: the product maturity stability is determined based on the ratio of the average to the variance of the thermal respiration ratio measured by a respiration intensity meter for a randomly selected number of target agricultural products within the same preset time interval. The product maturity stability is positively correlated with the average thermal respiration ratio and negatively correlated with the variance of the thermal respiration ratio. In this embodiment, the product maturity stability is the ratio of the average thermal respiration ratio to the variance of a randomly selected number of target agricultural products within the same preset time interval. Specifically, for the selected target agricultural products, the thermal respiration ratio is the ratio of the heat generated by respiratory metabolism of the target agricultural product within the preset time interval, as measured by the respiration intensity meter, to a preset thermal respiration amount. It is understandable that the lower the product maturity stability, the more dispersed or generally higher the thermal respiration ratio of each target agricultural product is. This indicates that some target agricultural products are in the rapid post-ripening stage, with significant differences in respiratory metabolic activity. The uncertainty of losses caused by transportation and environmental changes is greater. At this time, the more active the thermal respiration metabolism of the target agricultural products, the more significant the differences between the extracted target agricultural products, and the shorter their schedulable time window. If the exchange node is set outside the regional scheduling set, the risk of loss will be significantly aggravated due to the extended cross-regional transportation time.

[0067] In this embodiment of the invention, the process of obtaining the preset thermal respiration rate includes: simulating the respiratory metabolism data of target agricultural products under multiple sets of different thermal respiration rate conditions; calculating the thermal respiration ratio and evaluating product maturity and stability by setting different thermal respiration rate thresholds; and selecting the average value of the thermal respiration rate thresholds that meet user needs in terms of supply and demand prediction accuracy as the preset thermal respiration rate. In this embodiment, the more sensitive the user is to the prediction accuracy and transportation losses of the target agricultural product, the smaller the value of the preset thermal respiration rate.

[0068] The process of setting the extraction quantity includes: simulating the respiratory metabolism of target agricultural products under multiple sets of extraction quantity conditions; setting different extraction quantity thresholds for product maturity and stability assessment tests; and selecting the average of the extraction quantity thresholds that meet user requirements for supply and demand forecast accuracy as the extraction quantity. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the larger the extraction quantity value.

[0069] The process of obtaining the preset time interval includes: simulating the respiratory metabolism data of target agricultural products under multiple time interval conditions; calculating the thermal respiration ratio and evaluating product maturity stability by setting different time interval thresholds; and selecting the average value of the time interval thresholds whose coefficient of variation of the thermal respiration ratio measurement meets user requirements as the preset time interval. The coefficient of variation of the thermal respiration ratio measurement is the ratio of the standard deviation to the average value of the thermal respiration ratio obtained from multiple measurements of the same batch of target agricultural products at the same time interval. The smaller the coefficient of variation, the higher the consistency of the respiration intensity measurement and the higher the measurement accuracy of the thermal respiration ratio. The greater the user's sensitivity to this indicator, the lower the requirement for the coefficient of variation when selecting the preset value using the coefficient of variation of the thermal respiration ratio measurement as an evaluation indicator. The higher the user's requirement for the accuracy of supply and demand forecasting, the smaller the value of the preset time interval.

[0070] The process of obtaining the preset product maturity stability includes: conducting simulations by collecting scheduling analysis datasets under multiple product maturity stability conditions; setting different product maturity stability thresholds for testing the exchange node acquisition method; and selecting the average value of the product maturity stability thresholds that meet user requirements for supply and demand forecast accuracy as the preset product maturity stability. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the smaller the value of the preset product maturity stability.

[0071] Specifically, for the condition that the product maturity stability of floating demand points in the scheduling analysis set is greater than or equal to the preset product maturity stability, the selection area for exchange nodes is determined to be the expanded overlapping area, and the confirmation criterion for the number of exchange nodes is the neighborhood connectivity density.

[0072] The process of obtaining the expanded overlapping region includes: the expanded overlapping region is composed of the enclosing overlapping region of a single scheduling analysis set and its adjacent regional scheduling sets. For a single scheduling analysis set and a single adjacent regional scheduling set, the enclosing overlapping region is the overlapping part of the enclosing region of the scheduling analysis set and the enclosing region of the adjacent single regional scheduling set. The enclosing region is the smallest rectangle that can contain a single regional scheduling set.

[0073] The process of obtaining neighborhood connectivity density includes: neighborhood connectivity density is determined based on the average number of travel constraint paths for each floating demand point in the scheduling analysis set, and the neighborhood connectivity density is positively correlated with the average number of travel constraint paths. In this embodiment, the average number of travel constraint paths for each floating demand point in the scheduling analysis set is used. A floating demand point in a single adjacent regional scheduling set that is closest to the center point of the area enclosed by the regional scheduling set is selected as a path access point. For a single floating demand point, the travel constraint path is a traversable path whose path length ratio between it and the path access point is less than or equal to a preset path length ratio. The path length ratio is the ratio of the traversable path length to the distance between the floating demand point and the path access point. It can be understood that neighborhood connectivity density is used to quantify the spatial connectivity convenience between the scheduling analysis set and its adjacent regional scheduling sets. The higher the neighborhood connectivity density, the more short-distance traversable paths exist between the floating demand points in the scheduling analysis set and the path access points in the adjacent regions. The richer the physical channels for product scheduling between this region and its surroundings, and the higher the spatial accessibility of cross-regional scheduling.

[0074] The process of determining the number of exchange nodes includes: if the product maturity stability of a floating demand point is greater than or equal to a preset product maturity stability, the number of exchange nodes selected is the product of the number of base nodes and the path adjustment coefficient, where the path adjustment coefficient is the ratio of the neighborhood connectivity density to the preset neighborhood connectivity density; if the product maturity stability of a floating demand point in the regional scheduling set is less than the preset product maturity stability, the number of nodes selected is the product of the number of base nodes and the maturity adjustment coefficient, where the maturity adjustment coefficient is the ratio of the preset product maturity stability to the product maturity stability. It is worth noting that the number of nodes selected has a maximum value, which is the number of exchange nodes in the selected region.

[0075] The process of setting the number of basic nodes includes: conducting simulations by collecting regional scheduling data and exchanging node scheduling data under multiple conditions with different numbers of basic nodes; setting different thresholds for the number of basic nodes for interactive node analysis and testing; and selecting the average value of the thresholds for the number of basic nodes whose supply and demand forecast accuracy meets user requirements as the number of basic nodes. In this embodiment, the higher the user's requirement for supply and demand forecast accuracy, the larger the value of the number of basic nodes.

[0076] The process of obtaining the preset path length percentage includes: simulating the travel paths of demand points under multiple sets of different path length percentage conditions; calculating and testing the neighborhood connectivity density by setting different path length percentage thresholds; and selecting the average value of the path length percentage thresholds that meet user requirements for supply and demand prediction accuracy as the preset path length percentage. In this embodiment, the higher the user's requirement for supply and demand prediction accuracy, the smaller the value of the preset path length percentage.

[0077] The process of obtaining the preset neighborhood connectivity density includes: simulating the process by collecting neighboring region data from multiple sets of scheduling analysis datasets under different neighborhood connectivity density conditions; conducting tests to determine the selection region for exchange nodes by setting different neighborhood connectivity density thresholds; and selecting the average value of the neighborhood connectivity density thresholds that meet user requirements for supply and demand prediction accuracy as the preset neighborhood connectivity density. In this embodiment, the more sensitive the user is to transportation losses, the smaller the value of the preset neighborhood connectivity density.

[0078] Specifically, the time period corresponding to each stage prediction cycle is obtained. If the current time is not in the first time period of the stage prediction cycle and the time interval between the current time and the start time of the first time period is less than the preset start time, then the next stage prediction cycle closest to the current time is determined as the supply and demand prediction cycle. At the beginning of the supply and demand forecasting period, supply and demand forecasts are made for each demand point. Subsequently, a supply and demand forecast is made after each time period. These supply and demand forecasts are as follows: Obtain the prediction baseline parameters of each demand point in each regional scheduling set, and input them into the supply and demand prediction model to obtain the supply and demand tension information combination of each floating demand point within the prediction window period. The duration of the prediction window period is the duration of a single time period. According to the order of predicted tension values ​​from largest to smallest, the target agricultural products are retrieved from the floating demand points and exchange nodes whose predicted tension values ​​are greater than the preset predicted tension values. When retrieving, the exchange node with the closest center distance to the floating demand point is selected to retrieve the target agricultural products. The retrieval of the target agricultural products can be, but is not limited to, transporting the target agricultural products of the exchange node with the predicted tension value to the floating demand point.

[0079] In this embodiment of the invention, the process of setting the preset start time includes: simulating supply and demand forecast time series data under multiple sets of different start time conditions; setting different start time thresholds to conduct supply and demand forecast cycle determination tests; and selecting the average value of the start time thresholds whose supply and demand forecast accuracy meets user requirements as the preset start time. In this embodiment, the higher the user's requirement for the timeliness of supply and demand forecasting, the smaller the preset start time value.

[0080] In this embodiment of the invention, obtaining the preset predicted tension value includes: simulating supply and demand data of floating demand points under multiple sets of different predicted tension value conditions; setting different predicted tension value thresholds for trigger judgment tests; and selecting the average value of the predicted tension value thresholds whose supply and demand prediction accuracy meets user requirements as the preset predicted tension value. In this embodiment, the higher the user's requirement for supply and demand prediction accuracy, the smaller the preset predicted tension value.

[0081] The prediction baseline parameters include, but are not limited to, the demand characteristic parameters, attribute parameters, and environmental characteristic parameters of the collected demand points. The demand characteristic parameters include the weekday, holiday, and promotion day identifiers of the corresponding dates of the detection time in each stage of the prediction period within the historical observation period, the new inventory quantity, and the remaining inventory quantity of each demand point. The attribute parameters include at least the ID of the demand point, the ID of the scheduling set to which it belongs, the geographical coordinates, and the number of travel constraint paths between it and other demand points in the same scheduling set. The environmental characteristic parameters include at least the weather temperature, precipitation, and the market price of the target agricultural product corresponding to each target time period.

[0082] The supply and demand tension information combination is the prediction result of a single floating demand point output by the supply and demand forecasting model. The result includes the ID of the floating demand point, the ID of the regional scheduling set to which it belongs, and the predicted demand and predicted supply within the forecast window. The predicted demand is the total weight of the target agricultural products of the floating demand point that is reduced due to direct sales or other sales methods in the most recent forecast window, as predicted by the supply and demand forecasting model. The predicted supply is the total weight of the target agricultural products received by the demand point in the most recent forecast window, as predicted by the supply and demand forecasting model.

[0083] In this embodiment of the invention, the process of obtaining the predicted supply shortage value includes: the predicted supply shortage value is determined based on the difference between the predicted demand and the predicted supply in the supply and demand tension information combination of a single floating demand point in the regional scheduling set, and the predicted supply shortage value is positively correlated with the difference between the predicted demand and the predicted supply. In this embodiment, the predicted supply shortage value is the difference between the predicted demand and the predicted supply in the supply and demand tension information combination of a single floating demand point in the regional scheduling set. It can be understood that the predicted supply shortage value is used to quantify the degree of supply and demand gap of a single floating demand point within the forecast window period. The larger the predicted supply shortage value, the higher the risk of product shortage at that floating demand point within the forecast period, and the stronger the urgency of the scheduling response.

[0084] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for predicting and analyzing the supply and demand of agricultural products based on multidimensional data, characterized in that, include: Based on the demand stability of the target agricultural products within the supply and demand forecast coverage area, determine whether the demand point is a floating demand point, and based on the local clustering density of the floating demand point, determine whether to adjust the establishment method of the regional scheduling set from point attribute acquisition to stage density acquisition. In the process of obtaining the phase density, it is determined whether the sub-regions within the supply and demand forecast coverage area are clustered sub-regions based on the floating density, and the initial regional scheduling set is determined based on the phase density change value of the clustered sub-regions. The discrete attribution benchmark is determined based on the distribution dispersion of the discrete sub-regions for merging, and the regional scheduling set is obtained. In the point attribute acquisition process, the range of demand floating points is covered based on the stage demand difference value to obtain the regional scheduling set; To determine whether to perform interaction node analysis based on the floating proportion in the regional scheduling set, the selection criteria for the region and the number of regions to be selected are determined based on the product maturity and stability of the floating demand points. Supply and demand forecasting is performed on the regional scheduling set to obtain the supply and demand tension information combination of floating demand points, and the predicted tension value of floating demand points is used to retrieve the information from the exchange nodes in sequence.

2. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 1, characterized in that, The point where the demand stability of the target agricultural product is less than the preset demand stability during the stage forecast period is denoted as the floating demand point. The phased prediction period consists of consecutive historical target time periods.

3. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 2, characterized in that, For the condition that the local clustering density of floating demand points is greater than or equal to the preset local clustering density, the method for establishing the regional scheduling set is changed from point attribute acquisition to stage density acquisition. The local cluster density is the ratio of the dense quantity range to the supply and demand forecast coverage area.

4. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 3, characterized in that, The supply and demand forecast coverage area is uniformly divided to obtain several sub-regions with the same area. Sub-regions with floating density greater than the preset floating density are recorded as clustered sub-regions. The phase coverage analysis is performed sequentially according to the demand point density of the clustered sub-regions from large to small, and the sub-regions whose phase density change value is less than the preset phase density change value are constructed as the initial region scheduling set. Subregions that did not participate in the construction of the initial region scheduling set are denoted as discrete subregions.

5. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 4, characterized in that, For the condition that the dispersion of the discrete sub-region is greater than or equal to the preset dispersion, the discrete attribution benchmark is determined to be the center distance ratio; For the condition that the dispersion of the discrete sub-regions is less than the preset dispersion, the discrete attribution benchmark is determined to be the initial floating quantity ratio; Each discrete sub-region is merged into the initial region scheduling set with the largest discrete attribution benchmark to obtain the region set.

6. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 2, characterized in that, For situations where the local clustering density of floating demand points is less than the preset local clustering density, the method for establishing the regional scheduling set is determined to be point attribute acquisition, including: Arrange the demand fluctuation points in descending order of the stage demand difference value to obtain the order list; Following the order of establishment, the range of demand floating points is covered sequentially, and the demand points that meet the range coverage conditions are constructed into a regional scheduling set.

7. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 5 or 6, characterized in that, For the set of regions where the floating percentage is greater than or equal to the preset floating percentage, we denote it as the scheduling analysis set and determine the interaction node analysis to be performed. The selected number of exchange nodes are obtained sequentially within the selected area in descending order of the differential adjustment value.

8. The method for predicting and analyzing the supply and demand of agricultural products based on multidimensional data according to claim 7, characterized in that, In the interaction node analysis, for the condition that the product maturity and stability of floating demand points in the scheduling analysis set is less than the preset product maturity and stability, the selection area of ​​the exchange node is determined as the coverage area of ​​the scheduling analysis set, and the confirmation criterion for the number of exchange nodes is the product maturity and stability.

9. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 8, characterized in that, For the condition that the product maturity stability of floating demand points in the scheduling analysis set is greater than or equal to the preset product maturity stability, the selection area for exchange nodes is determined to be the expanded overlapping area, and the basis for confirming the number of exchange nodes is the neighborhood connectivity density.

10. The method for predicting and analyzing agricultural product supply and demand based on multidimensional data according to claim 8 or 9, characterized in that, The forecasting baseline parameters of each regional scheduling set are obtained to predict the supply and demand of each demand point to obtain the supply and demand tension information combination of each floating demand point. The target agricultural products are retrieved in order of the predicted tension value from large to small, and the floating demand points and exchange nodes with predicted tension values ​​greater than the preset predicted tension value are retrieved in turn.

Images