Order agriculture channel supply and demand intelligent matching method and system

Abstract

The application provides an order agricultural channel supply and demand intelligent matching method and system. The method comprises the following steps: acquiring planting base and purchasing merchant data, generating supply end production characteristics and demand end purchase characteristics, and establishing supply and demand matching relationship; generating supply and demand matching representation according to space-time adaptation degree, generating cross-category substitution elasticity from historical alternative purchase relationship; determining order urgency according to remaining period and category supply and demand state, and dividing time-effective priority order and global balanced order; screening local supply end for time-effective priority order and generating rapid matching result; determining candidate range for global balanced order according to substitution elasticity, constructing supply and demand distribution constraint model, adjusting constraint according to period supply and demand state, and solving global matching result; identifying resource conflicts of the two types of results, preferentially allocating according to order urgency, re-solving the constraint model based on the remaining resources, and generating channel supply and demand matching result. The application can improve the supply and demand matching efficiency, improve the matching accuracy, and enhance the adaptation ability to supply and demand changes.

Description

Technical Field

[0001] This application relates to the field of agricultural information technology, and in particular to a method and system for intelligent matching of supply and demand in contract farming channels. Background Technology

[0002] Contract farming is a business model in which buyers or processing enterprises pre-agree with planting bases on the types, quantities, quality, and delivery deadlines of agricultural products, and organize production and procurement accordingly. As the number of participating entities increases, platforms need to establish supply-demand matching between the projected output of multiple planting bases and multiple demand orders to ensure the allocation of channel resources and order fulfillment.

[0003] Existing supply and demand matching typically employs either fixed-region matching or global search matching. Fixed-region matching limits candidate suppliers to a predetermined geographical area, which can reduce the search scale, but the regional boundaries remain unchanged in the long term, making it difficult to utilize excess capacity in adjacent areas to fill local supply gaps. Global search matching performs combined calculations across all supply and demand entities, resulting in a wider candidate range, but as the number of planting bases and demand orders increases, the combined scale expands rapidly, easily leading to processing delays during peak harvest and delivery periods.

[0004] Furthermore, existing methods often employ fixed candidate ranges and static matching parameters, failing to switch between local rapid matching and global optimized matching based on remaining order fulfillment time. They also do not adequately consider transportation distance, fulfillment windows, and substitution relationships among similar product categories. When market supply and demand conditions or projected output change, existing matching constraints are difficult to adjust in a timely manner, easily leading to delayed responses to urgent orders, underutilization of cross-regional resources, and redundant allocation of supply resources. Therefore, existing technologies struggle to balance matching timeliness, accuracy, and dynamic adaptability. Summary of the Invention

[0005] In view of this, embodiments of this application provide a method and system for intelligent matching of supply and demand in order agriculture channels, in order to solve the problems of existing technologies, such as difficulty in balancing matching timeliness and accuracy, fixed candidate range, and insufficient adaptability to changes in supply and demand.

[0006] A first aspect of this application provides a method for intelligent matching of supply and demand in order-based agriculture channels, comprising: acquiring production data from planting bases and demand data from buyers, generating supply-side production characteristics and demand-side procurement characteristics, and establishing a supply-demand matching relationship; generating a supply-demand matching representation based on the degree of fit between fulfillment time and transportation space in the supply-demand matching relationship, and generating cross-category substitution elasticity based on the historical substitution procurement relationships of buyers; determining the urgency of orders based on the remaining fulfillment time of demand orders and the supply and demand status of target categories, and classifying demand orders into time-priority orders and globally balanced orders; and for time-priority orders, according to... The system filters local candidate suppliers based on spatial scope, fulfillment time, and supply capacity, and generates rapid matching results based on the adaptability of transportation space. For globally balanced orders, it determines the candidate range of the same category or the expansion range of related categories based on cross-category substitution elasticity. It constructs a supply and demand allocation constraint model based on supply and demand matching characteristics, adjusts the constraint strength of the supply and demand allocation constraint model according to the cyclical market supply and demand status, and solves to obtain the global matching results. It identifies supply resource conflicts between rapid matching results and global matching results, determines the priority allocation relationship according to the urgency of orders, and resolves the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

[0007] A second aspect of this application provides an intelligent supply and demand matching system for order-based agriculture channels, comprising: an acquisition module for acquiring production data from planting bases and demand data from buyers, generating supply-side production characteristics and demand-side procurement characteristics, and establishing a supply-demand matching relationship; a generation module for generating a supply-demand matching representation based on the fulfillment time and transportation space adaptability of the supply-demand matching relationship, and generating cross-category substitution elasticity based on the buyer's historical substitution procurement relationship; a segmentation module for determining the urgency of orders based on the remaining fulfillment time of demand orders and the supply and demand status of target categories, and segmenting demand orders into time-priority orders and globally balanced orders; and a filtering module for targeting time-priority orders. Priority orders are selected by filtering local candidate suppliers based on spatial scope, fulfillment time, and supply capacity, and generating rapid matching results based on the adaptability of transportation space. The construction module is used to determine the candidate range of the same category or the expansion range of related categories for globally balanced orders based on cross-category substitution elasticity, construct a supply and demand allocation constraint model based on supply and demand matching characteristics, adjust the constraint strength of the supply and demand allocation constraint model according to the cyclical market supply and demand status, and solve to obtain the global matching result. The re-solution module is used to identify supply resource conflicts between the rapid matching result and the global matching result, determine the priority allocation relationship according to the urgency of the order, and re-solution the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

[0008] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: By acquiring production data from planting bases and demand data from buyers, supply-side production characteristics and demand-side procurement characteristics are generated, and a supply-demand matching relationship is established. Supply-demand matching representations are generated based on the suitability of fulfillment time and transportation space in the supply-demand matching relationship. Cross-category substitution elasticity is generated based on the historical substitution relationships of buyers. The urgency of orders is determined based on the remaining fulfillment time of demand orders and the supply and demand status of target categories, classifying demand orders into time-priority orders and globally balanced orders. For time-priority orders, local candidate suppliers are screened according to spatial scope, fulfillment time, and supply capacity, and rapid matching results are generated based on transportation space suitability. For globally balanced orders, the candidate range of the same category or the expansion range of related categories is determined based on cross-category substitution elasticity. A supply-demand allocation constraint model is constructed based on the supply-demand matching representation. The constraint strength of the supply-demand allocation constraint model is adjusted according to the cyclical market supply and demand status, and the global matching result is obtained. Supply resource conflicts between the rapid matching result and the global matching result are identified. Priority allocation relationships are determined according to the urgency of orders, and the supply-demand allocation constraint model is resolved based on the remaining supply resources to generate channel supply-demand matching results. This application can improve the efficiency of supply and demand matching, enhance the accuracy of matching, and strengthen the ability to adapt to changes in supply and demand. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating the intelligent supply and demand matching method for order agriculture channels provided in the embodiments of this application; Figure 2 This is a schematic diagram of the intelligent supply and demand matching system for order-based agriculture provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0011] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0012] Existing supply and demand matching methods in contract farming channels typically employ fixed-region matching or global search matching. Fixed-region matching filters suppliers based on a preset geographical area, reducing the scale of a single matching operation, but the candidate boundaries are relatively fixed, making it difficult to utilize available supply resources outside the region. While global search matching can expand the scope of supply and demand combinations, the computational scale increases rapidly with the number of planting bases and demand orders, making it difficult to meet the timely response requirements of urgent orders. Furthermore, existing methods mostly use static matching parameters, failing to fully incorporate dynamic decision-making based on fulfillment time, transportation space, product substitution relationships, and market supply and demand changes.

[0013] Therefore, existing technologies suffer from several problems: difficulty in balancing matching timeliness and accuracy, lack of dynamic adjustment capability for candidate supply range, and insufficient adaptability of supply and demand matching results to market changes. In particular, in scenarios of concentrated harvesting, concentrated order delivery, or shortage of supply in certain product categories, fixed matching methods are prone to causing delays in responding to urgent orders, underutilization of cross-regional supply resources, ineffective use of alternative product categories, and duplicate allocation of the same supply resources.

[0014] To address the aforementioned technical challenges, this application generates supply-side production characteristics and demand-side procurement characteristics by acquiring production data from planting bases and demand data from buyers, establishing a supply-demand matching relationship. It generates supply-demand matching representations based on the suitability of fulfillment time and transportation space, and generates cross-category substitution elasticity based on historical substitution procurement relationships. Combining remaining fulfillment time and the supply and demand status of the target category, it determines the urgency of orders, classifying demand orders into time-priority orders and globally balanced orders. For time-priority orders, it filters local candidate suppliers and generates rapid matching results. For globally balanced orders, it determines the candidate range for the same category or the extended range for related categories, constructs a supply-demand allocation constraint model, and dynamically adjusts the constraint strength according to the cyclical market supply and demand status to obtain global matching results.

[0015] Furthermore, this application identifies supply resource conflicts between the rapid matching results and the global matching results, determines the priority allocation relationship according to the urgency of orders, and re-solves the supply and demand allocation constraint model based on the remaining supply resources to form the final channel supply and demand matching result. Through the above technical solution, it is possible to select local rapid matching or global equilibrium matching according to the timeliness requirements of orders, and dynamically adjust the matching scope and resource allocation constraints by combining cross-category substitution and market supply and demand status, thereby improving supply and demand matching efficiency, enhancing matching accuracy, reducing supply resource conflicts, and strengthening adaptability to supply and demand changes.

[0016] The technical solution of this application will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0017] Figure 1 This is a flowchart illustrating the intelligent supply and demand matching method for order-based agriculture channels provided in this application embodiment. For example... Figure 1 As shown, the method may specifically include: S101: Obtain production data from planting bases and demand data from buyers, generate supply-side production characteristics and demand-side procurement characteristics, and establish supply-demand matching relationships; S102 generates a supply and demand matching representation based on the degree of fit between the fulfillment time and the degree of fit between the supply and demand matching relationship, and generates cross-category substitution elasticity based on the historical substitution procurement relationship of the acquirer; S103, based on the remaining time for fulfillment of demand orders and the supply and demand status of target categories, determine the urgency of orders and divide demand orders into time-priority orders and globally balanced orders; S104, for time-priority orders, filters local candidate suppliers according to spatial range, fulfillment time and supply capacity, and generates rapid matching results based on the adaptability of transportation space; S105, for globally balanced orders, determine the candidate range of the same category or the expansion range of related categories based on the cross-category substitution elasticity, construct a supply and demand allocation constraint model based on the supply and demand matching representation, adjust the constraint strength of the supply and demand allocation constraint model according to the supply and demand status of the cyclical market, and solve to obtain the global matching result; S106 identifies supply resource conflicts between the fast matching results and the global matching results, determines the priority allocation relationship according to the urgency of the orders, and resolves the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

[0018] In some embodiments, production data from planting bases and demand data from buyers are acquired to generate supply-side production characteristics and demand-side procurement characteristics, and to establish a supply-demand matching relationship, including: By associating current cycle data with historical production records according to the planting base identification, the projected supply and fluctuation range are generated based on the historical unit output distribution and current production scale, and the supply-side production characteristics are obtained by combining the planned harvest window and the location of the production area. Based on the demand order identifier, associated procurement constraints, and historical fulfillment records, generate demand-side procurement characteristics that include demand size, fulfillment window, quality boundary, and demand location; Based on product category compatibility, the production characteristics of the supply side and the procurement characteristics of the demand side are screened, and a supply and demand matching relationship is established based on the overlap of fulfillment windows, quality adaptability, and basic supply capacity.

[0019] Specifically, taking the channel matching of fresh tomato procurement orders on a certain order-based agriculture platform as an example, the platform accesses the production and procurement data of 3 planting bases and 2 buyers. Each planting base is configured with a unique base identifier on the platform. The current period data includes planting area, planting batch, variety type, planting date, field growth, current harvest progress, locked supply, planned harvest date, and base coordinates; historical production records include the annual yield per unit area of ​​the same variety, yield reduction rate at different growth stages, harvest window offset, and actual delivery volume.

[0020] The platform associates current cycle data with corresponding historical records according to the base identifier, and completes the execution time and unifies the units for abnormal and missing records, forming a base-level production data sequence. Each data record is simultaneously written with the collection time and data version, and the latest valid record within the same production cycle is used first when associating data. When there are multiple planting batches in the same base, the expected supply is calculated separately according to the batch identifier to avoid incorrect merging of different harvest windows and quality states.

[0021] In a specific example, Base A currently has a planting area of ​​80 mu (approximately 5.3 hectares). Over the past three years, the unit yield for the same season has mainly ranged between 4200 kg and 4700 kg per mu. Based on the current growth score, meteorological influence coefficient, and the proportion of harvested crops, the central value of the unit yield for this cycle is determined to be 4450 kg. According to historical yield percentiles, the projected supply is 356,000 kg, with a fluctuation range of -28,000 kg to +24,000 kg. Bases B and C are calculated using the same method to determine the projected supply and fluctuation range. For bases with records of continuous rainfall, disease, or temporary crop reduction, the platform adjusts the unit output distribution based on the corresponding impact level, rather than directly using the historical average.

[0022] Subsequently, the projected supply volume, lower limit of fluctuation, upper limit of fluctuation, planned harvest start and end dates, locked-in supply volume, and base latitude and longitude are combined to form the supply-side production characteristics. To avoid including unallocated output in the candidate supply capacity, the contracted quantity and the quantity retained by the base are also deducted from the projected supply volume to obtain the basic supply volume that can be used for this matching.

[0023] On the buyer's side, order categories, quantities, allowed number of batches, earliest arrival date, latest arrival date, specifications, maturity, single fruit weight, pesticide residue limits, packaging requirements, and warehouse coordinates are aggregated according to demand order identifiers, and linked to the corresponding buyer's performance records for the past 12 months. Historical performance records are used to identify the buyer's actual tolerance range for early delivery, late delivery, specification deviations, and batch delivery. For example, order D1 declares a demand of 120,000 kg, a performance window of September 18th to September 20th, quality requirements of Grade 1 fresh tomatoes, and a delivery location at a city distribution center. If historical records indicate that the buyer typically accepts early delivery of no more than one day but does not accept late delivery, the effective performance window for the order is adjusted to September 17th to September 20th. Demand-side procurement characteristics are jointly constituted by demand size, effective performance window, quality limits, demand location, and historical performance stability.

[0024] When establishing supply and demand matching relationships, suppliers with inconsistent product categories and no substitutes are first screened out based on the category compatibility relationships maintained by the platform. For fresh tomato orders, only batches from bases that meet the order requirements in terms of variety, intended use, maturity, and packaging are retained. Subsequently, the number of days overlapping between the planned harvest window and the effective fulfillment window is calculated, and the time required for post-harvest sorting, packaging, and transportation is combined to determine whether the supply batch can be delivered before the latest arrival date.

[0025] For example, if Base A's planned harvest window is from September 16th to September 19th, post-harvest processing requires 6 hours, and transportation is expected to take 8 hours, then there is an overlap in effective timeframes between Base A and Order D1. Regarding quality matching, the expected grade, specification range, and test results of the Base batch are compared item by item with the order's quality boundaries. If any mandatory quality indicator is not met, the corresponding combination is eliminated. Regarding basic supply capacity, the available supply quantity is compared with the order's minimum acceptable batch size. Bases with batches below the minimum acceptable batch size are not paired separately, but can participate in the combination as supplementary suppliers when joint supply is permitted.

[0026] Furthermore, after screening, the platform establishes a pairing record for each eligible base and order, recording the base identifier, order identifier, available supply quantity, time overlap, quality compatibility status, and transportation distance. For order D1, base A can supply 100,000 kg, and base B can supply 60,000 kg. Both meet the fulfillment window and quality boundaries, so a pairing relationship is established for each. Although base C has sufficient production, its planned harvest time is later than the latest delivery date of the order, so no pairing relationship is established.

[0027] Through the above implementation methods, scattered and heterogeneous production and procurement data can be transformed into supply-side production characteristics and demand-side procurement characteristics with unified semantics, and supply-demand combinations that are not feasible to fulfill can be eliminated in advance, providing an accurate data foundation for subsequent order urgency assessment, local rapid matching, and global resource allocation.

[0028] In some embodiments, a supply-demand matching representation is generated based on the degree of fit between fulfillment time and transportation space in the supply-demand matching relationship, and cross-category substitution elasticity is generated based on the acquirer's historical substitution procurement relationships, including: Calculate the degree of overlap between the planned harvest window in the production characteristics of the supply side and the fulfillment window in the procurement characteristics of the demand side, and determine the transportation distance based on the location of the production site and the location of the demand. The degree of overlap and transportation distance are unified in direction and normalized in scale. The corresponding weights are determined by combining the historical performance characteristics of the acquirers, and a supply and demand matching representation is generated. Extract the linkage changes in purchase volume between related product categories from the historical purchase records of the acquirer, identify the compensatory purchase changes in related product categories when the purchase of the target product category decreases, and generate cross-category substitution elasticity based on the ratio of the compensation magnitude to the change magnitude of the target product category.

[0029] Specifically, the supply-demand matching characterization is used to convert the time fulfillment rate and spatial transportation convenience in supply-demand pairing relationships into comparable numerical features. For each supply-demand pairing relationship, the planned harvest start and end times in the supply-side production characteristics and the fulfillment start and end times in the demand-side procurement characteristics are first read, and the estimated time for post-harvest sorting, packaging, and transportation is superimposed on the planned harvest time to obtain the available delivery time window. The intersection duration of the available delivery time window and the fulfillment window is calculated, and the proportion of the intersection duration to the fulfillment window duration is used as the time coverage rate; when the two windows do not intersect, a time deviation is generated based on the offset duration between the available delivery time and the nearest boundary of the fulfillment window.

[0030] Furthermore, combining time coverage and time deviation yields a performance time fit rating, where higher coverage and smaller deviation indicate a greater likelihood of on-time fulfillment of the supply and demand combination. For orders that allow for partial deliveries, the fit rating results for different time periods can be weighted and aggregated according to the quantity of each batch, avoiding judgment based solely on a single delivery time.

[0031] The suitability of transportation space is determined based on the feasible transportation routes between the production location and the demand location. The platform maps the base coordinates and receiving warehouse coordinates to the road transportation network, and calculates the transportation distance and estimated transportation time by combining road mileage, road grade, historical travel time, and cold chain transfer nodes. For areas lacking road network data, a spherical distance can be used to form a base distance, which is then corrected based on the ratio of historical actual transportation mileage to the spherical distance. Transportation distance is a negative indicator of higher suitability as the value decreases, while overlap is a positive indicator of higher suitability as the value increases; therefore, the execution direction needs to be consistent. The transportation distance can be converted into the difference between the upper limit of distance and the actual distance, or a distance decay function can be used to generate a spatial suitability value, and then interval normalization can be performed according to the maximum and minimum values ​​within the current matching batch, so that the suitability of fulfillment time and transportation space are mapped to a unified numerical range.

[0032] The weights are dynamically determined based on the historical fulfillment characteristics of the buyers. The platform statistically analyzes the acceptance rate, return rate, and fulfillment rate corresponding to delayed delivery, early delivery, long-distance transportation, and short-distance transportation in the buyers' historical orders. When historical records show that buyers are sensitive to delivery time deviations, the weight of fulfillment time adaptability is increased; when there is a significant correlation between fresh produce spoilage and transportation distance, the weight of transportation space adaptability is increased. After weighted aggregation of the two normalized adaptability values ​​and their corresponding weights, a supply-demand matching score is obtained. The time adaptability value, space adaptability value, weights, and comprehensive score can be combined to form a supply-demand matching representation to retain the multi-dimensional information required for subsequent constraint modeling. For buyers with insufficient historical sample size, the statistical weights of similar buyers can be used as initial weights, and gradually updated as more fulfillment data is added.

[0033] Cross-category substitution elasticity reflects the likelihood that a buyer will shift its procurement demand to related categories when there is a shortage in the target category. The platform aligns historical procurement records according to the procurement cycle, extracting planned procurement volume, actual procurement volume, shortfall, and temporary additional purchases for each category, and eliminating common fluctuations caused by seasonal procurement changes and promotional activities. Subsequently, it identifies cycles in which the actual procurement volume of the target category decreases relative to the planned procurement volume, and determines whether there are compensatory additional purchases in related categories that exceed the normal fluctuation range within the same cycle.

[0034] For example, the basic substitution ratio can be determined by using the proportion of increased purchases in related categories to decreased purchases in the target category. This is then adjusted for credibility based on factors such as the frequency of occurrence, consistency rate, and fulfillment success rate, resulting in cross-category substitution elasticity. If the target category decreases while related categories do not increase, or if the purchase volumes of both categories decline in the same direction over a long period, a substitution relationship is not considered. To prevent a small number of unusual purchases from causing distortion in elasticity, the basic substitution ratio can be truncated at certain thresholds, and a preset quantity must be required for the effective substitution period to be reached.

[0035] To ensure consistency between the matching representation and the current operational status, only production, order, and road network data within the validity period are used during calculation. When the planned harvest window, fulfillment window, or transportation route is updated, the affected supply and demand matching relationships are recalculated without repeating the processing of all combinations. Each calculation is associated with the data version and calculation time, facilitating the tracing of the scoring source during subsequent resource conflict handling. For combinations with high time fit but abnormal transportation risks, multi-dimensional representations can be retained and risk markers can be set to prevent the comprehensive score from masking situations where individual constraints are not met.

[0036] For example, a buyer's fresh tomato order fulfillment window is from September 18th to September 20th. After post-harvest processing and transportation adjustments, Base A's delivery window is from the afternoon of September 17th to the afternoon of September 19th, resulting in a time coverage rate of 75% and a transportation route distance of 180 kilometers. Normalization yields a time fit value of 0.82 and a spatial fit value of 0.68. This buyer has historically been sensitive to delayed deliveries; therefore, a time weight of 0.65 and a spatial weight of 0.35 are applied, resulting in a comprehensive matching score of 0.77. Historical procurement records also show that in 6 out of 8 effective periods of reduced supply of ordinary fresh tomatoes, cherry tomato procurement was increased, with an average compensation amount accounting for 58% of the reduction in ordinary fresh tomatoes. After adjustments for directional consistency rate and fulfillment success rate, the cross-category substitution elasticity between the two categories is 0.49. This supply-demand matching characteristic and substitution elasticity can serve as inputs for subsequent order classification, candidate range expansion, and supply-demand allocation constraint construction, improving the comparability between different matching paths and the reliability of substitution decisions.

[0037] In some embodiments, the method further includes: Extract compensatory procurement relationships between different product categories based on the historical procurement records of the acquirers, and construct a relationship diagram of related product categories by combining the degree of fit of the quality attributes of the corresponding product categories; Using product categories as graph nodes and compensatory procurement relationships as graph connections, the substitution reliability of each graph connection is determined based on cross-category substitution elasticity and the degree of adaptation of quality attributes. When the global balanced orders meet the category expansion conditions, starting from the graph node corresponding to the target category, a restricted graph traversal is performed according to the substitution confidence level, and related categories are filtered according to the preset expansion level and cumulative substitution deviation to generate the related category expansion range.

[0038] Specifically, the product category relationship diagram is used to represent the alternative paths that can be actually accepted by buyers between different agricultural product categories. The platform first aggregates historical purchase orders, stockout records, temporary re-allocation records, and actual receipt results according to buyer identification, and aligns the planned purchase volume, actual purchase volume, and supplementary purchase volume for different categories according to the purchase cycle. For cycles in which the purchase volume of the target category is lower than the planned volume, it identifies whether there are additional purchases of other categories that exceed the normal fluctuation range, and determines whether a compensatory purchase relationship has been formed by combining the time of the additional purchase, the quantity of the additional purchase, and the subsequent performance status.

[0039] If the increase in purchases of related categories occurs after the decrease in purchases of the target category, and the compensation amount and the decrease amount have a stable ratio, then a directed relationship from the target category to the related category is established; if the two categories can be substituted for each other, then bidirectional relationships are established respectively. Synchronous changes caused only by promotions, seasonal changes, or centralized procurement activities are eliminated through periodic baseline correction and anomaly event marking.

[0040] The quality attribute fit degree is used to constrain the substitutability of different product categories in terms of use and delivery requirements. The platform establishes a category quality attribute vector, with attributes including edible use, processing use, maturity, specification grade, moisture content, storage and transportation resistance, packaging method, and testing boundaries. The quality attributes of the target product category and candidate related product categories are standardized and oriented, with mandatory attributes judged using consistency criteria and non-mandatory attributes calculated using difference degree. When the use category, core quality grade, or testing boundaries are incompatible, the quality attribute fit degree is recorded as non-substitutable; otherwise, a fit value is generated based on the differences in each attribute and the historical acceptance preferences of the purchaser. A related product category relationship graph is constructed by compensatory procurement relationships and quality attribute fit degrees. Graph nodes correspond to product categories, and graph connections record the substitution direction, historical compensation frequency, average compensation ratio, and quality fit value.

[0041] The reliability of substitution for each graph connection is determined jointly based on cross-category substitution elasticity and quality attribute fit. Cross-category substitution elasticity reflects the actual compensation capability during historical purchase volume changes, while quality attribute fit reflects the feasibility of meeting order constraints after substitution. The platform first normalizes these two indicators, then determines the reliability correction coefficient based on historical sample size, substitution direction consistency rate, and substitution fulfillment success rate. The reliability correction coefficient is increased when historical samples are sufficient and the return rate after substitution is low; it is decreased when samples are few or substitution results fluctuate significantly. The normalized substitution elasticity, quality fit value, and reliability correction coefficient are weighted and aggregated to obtain the reliability of substitution for the graph connection. For relationships with continuous returns, quality complaints, or fulfillment failures, a decay factor can be set to reduce the reliability of substitution with negative feedback.

[0042] Category expansion conditions are determined based on the supply-demand gap of the target category, the supply capacity of candidate products in the same category, and the order's allowable substitution status. When the supply of candidate products in the same category is insufficient to meet the minimum demand of a globally balanced order, and the order's procurement constraints allow the use of related categories, the category expansion condition is deemed met. The platform performs a restricted graph traversal starting from the target category's graph nodes, prioritizing adjacent nodes with higher substitution credibility. Preset expansion levels limit the traversal depth; for example, first-level expansion only allows direct substitution, while second-level expansion allows further expansion through an intermediate category. For each graph connection traversed, the cumulative substitution deviation is updated according to the substitution deviation of the connection. The substitution deviation can be obtained by subtracting the substitution credibility from 1.

[0043] If the cumulative substitution deviation exceeds a threshold, the quality attribute enters an irreplaceable state, or the expansion level reaches its upper limit, the traversal of the corresponding path is stopped. Retained nodes are sorted according to substitution reliability, cumulative substitution deviation, and available supply to generate the related category expansion range. To avoid loops caused by bidirectional connections during traversal, the platform sets path markers for visited nodes, preventing repeated visits to the same category within the same path. When the same related category can be reached by multiple paths, the path with the smaller cumulative substitution deviation and higher substitution reliability is retained. The related category expansion range also records the source path, the upper limit of the substitutable ratio, and the applicable quality boundary for subsequent use by the supply and demand allocation constraint model.

[0044] For example, a processing buyer has an 80,000 kg supply gap in its regular tomato order. The order allows for the substitution of processed tomatoes of the same purpose, but does not accept fresh cherry tomatoes. Historical records show that when the supply of regular tomatoes decreases, the average compensation ratio for processed tomatoes is 62%, the substitution fulfillment success rate is 91%, and the quality attribute fit value is 0.88. Therefore, the substitution reliability of the graph connection between the two is 0.84. Although there is a small amount of compensatory procurement between regular tomatoes and cherry tomatoes, their uses and specifications differ significantly, resulting in a quality attribute fit value of only 0.36, and the corresponding relationship is marked as low reliability. The platform performs a first-level graph traversal starting from the regular tomato node, including processed tomatoes in the related category expansion scope and excluding cherry tomatoes. If the supply of processed tomatoes is still insufficient, the sauce-making tomato node can continue to be accessed if the second-level expansion allows, but expansion stops when the cumulative substitution deviation exceeds 0.35.

[0045] Through the above implementation methods, historical substitution behavior, quality compatibility, and fulfillment feedback can be uniformly encoded into a computable graph structure. By controlling the expansion boundary of candidate categories through constrained graph traversal, the unconstrained expansion of the substitution range can be avoided. This improves the accuracy of related category selection, reduces the risk of returns and fulfillment issues caused by unsuitable substitutions, and enhances the ability of global matching to call upon available supply resources when the target category supply is insufficient.

[0046] In some embodiments, the urgency of an order is determined based on the remaining fulfillment time of the demand order and the supply and demand status of the target category, and the demand orders are divided into time-priority orders and globally balanced orders, including: Calculate the time interval between the start time of the fulfillment window for each demand order and the current time, and generate a basic urgency threshold based on the distribution of the time interval among the demand orders to be processed. The projected supply and order demand for the target product category are aggregated. The product category supply and demand status is generated based on the ratio of the two. The basic urgency threshold is dynamically adjusted according to the product category supply and demand status to obtain the urgency judgment threshold. The time interval is compared with the urgency threshold. If the time interval is not greater than the urgency threshold, the corresponding demand order is determined as a time-priority order; otherwise, the corresponding demand order is determined as a globally balanced order.

[0047] Specifically, order urgency reflects the required matching response speed for orders at the current processing time. The platform reads the start time of the fulfillment window for each order using a unified time base and calculates the time interval between the start time of the fulfillment window and the current time. For orders that have already entered the fulfillment window, the time interval is recorded as 0 and marked as immediate processing. For orders whose fulfillment window start time is earlier than the current time and have not yet been matched, the overdue duration can be further recorded to increase the processing priority within the same type of order. To avoid using a fixed number of days threshold for different processing batches, which may not be suitable for the concentration of orders, the platform sorts the currently pending orders by time interval from smallest to largest, calculates the quantile position of each time interval in the order set, and generates a basic urgency threshold by combining the number of orders, the fulfillment window width, and the historical average processing time.

[0048] The basic urgency threshold is used to define time boundaries requiring rapid response. Specifically, a target quantile value in the time interval distribution can be selected as the initial boundary, and adjusted based on the processing load of the current order batch. When the number of pending orders is large or the number of new orders per unit time increases significantly, the basic urgency threshold should be appropriately shortened to limit the number of orders entering the time-priority processing channel. When the number of orders is small and computing resources are sufficient, the basic urgency threshold can be appropriately extended to allow more orders nearing the fulfillment window to enter rapid matching earlier. The basic urgency threshold can also be constrained by historical fulfillment times to ensure that the shortest processing time required from matching completion and supply confirmation to delivery arrival is not compressed.

[0049] The category supply and demand status reflects the sufficiency of supply for the target category within the current matching cycle. The platform aggregates the projected supply from all suppliers for the target category, deducting locked supply, unallocated quantities, and safety reserves to obtain the effective projected supply available for current matching. Simultaneously, it aggregates the demand for all pending orders within the same category and adjusts this demand based on order cancellation probabilities, historical actual acceptance rates, and completed allocations to obtain the effective order demand. The ratio of the effective projected supply to the effective order demand serves as the baseline supply-demand ratio. When this ratio exceeds a preset sufficiency threshold, the target category is considered to have sufficient supply; when the ratio is within the equilibrium range, the target category is considered to be in a supply-demand balance; and when the ratio is below the shortage threshold, the target category is considered to be in a supply-tight state. To reduce misjudgments caused by fluctuations in single production forecasts, a conservative supply-demand ratio can be calculated by combining the lower limit of projected supply fluctuations. The category supply and demand status is then determined jointly based on the baseline and conservative supply-demand ratios.

[0050] The platform dynamically adjusts the basic urgency threshold based on the supply and demand status of each product category. When supply is ample, local rapid matching easily finds suppliers that meet the requirements, allowing the urgency threshold to be appropriately increased, enabling orders nearing their fulfillment window to be prioritized for timely processing. When supply is tight, it's necessary to reserve space for global coordination and cross-regional allocation for more orders; therefore, the urgency threshold is reduced, classifying only orders with short remaining time and unable to wait for global optimization as time-priority orders, while the remaining orders are retained as globally balanced orders. When supply and demand are balanced, the basic urgency threshold can remain unchanged or be slightly adjusted based on expected supply fluctuations. The adjusted urgency threshold is stored in association with the corresponding target product category, processing cycle, and data version for subsequent tracking of order classification criteria.

[0051] Furthermore, after generating the thresholds, the platform compares the time interval of each demand order with the urgency threshold of the corresponding target category. If the time interval is no greater than the urgency threshold, the order is identified as a time-priority order and enters the local candidate supplier screening process. If the time interval is greater than the urgency threshold, the order is identified as a globally balanced order and enters the supply and demand allocation constraint solution process within the same or related categories. For orders containing multiple delivery batches, the time interval can be calculated based on the fulfillment window of the earliest incomplete batch to avoid later batches masking current urgent delivery needs. If the order fulfillment window, demand volume, or category supply and demand status changes, the platform recalculates the urgency of the corresponding order and allows orders to dynamically switch between two processing channels.

[0052] For example, at 8:00 AM on September 15th, there are 20 pending tomato orders. The 30th percentile of the time interval distribution is 48 hours. Considering the historical average time required for matching and confirming supply is 12 hours, the basic urgency threshold is set at 48 hours. The current effective estimated supply of tomatoes is 520,000 kg, and the effective order demand is 650,000 kg, resulting in a basic supply-demand ratio of 0.80. A conservative supply-demand ratio calculated based on the lower limit of the estimated supply fluctuation is 0.72, indicating a tight supply situation. The basic urgency threshold is then shortened to 36 hours. Order D1's fulfillment window begins at 12:00 PM on September 16th, with a time interval of 28 hours, thus classifying it as a time-priority order. Order D2's fulfillment window begins at 8:00 AM on September 18th, with a time interval of 72 hours, thus classifying it as a globally balanced order. If subsequent new supply increases the supply-demand ratio to 1.15, the urgency threshold can be expanded to 54 hours, and some near-term orders will then be moved to time-priority processing.

[0053] Through the above implementation methods, urgency determination boundaries can be dynamically formed based on order time distribution and supply and demand changes in target categories, avoiding order classification distortion caused by using fixed time limits. This prioritizes the response speed of orders nearing fulfillment while reserving global optimization space for orders with sufficient remaining time, improving the accuracy of order diversion, matching processing efficiency, and adaptability of supply and demand resource allocation.

[0054] In some embodiments, for time-sensitive orders, local candidate suppliers are filtered based on spatial range, fulfillment time, and supply capacity, and rapid matching results are generated based on the adaptability of transportation space, including: Using the demand location of time-sensitive orders as the search center, the dynamic search range is determined based on the historical transportation range of the corresponding purchaser and the urgency of the order, and the supply side within the dynamic search range is filtered. The supply-side production characteristics are matched with the demand-side procurement characteristics of time-sensitive orders, and the available supply is verified. Suppliers that do not meet the preset fulfillment conditions are eliminated, forming a partial pool of candidate suppliers. Based on the transportation distance from local candidate suppliers to the demand location, a transportation space fit degree is generated. Local candidate suppliers are sorted according to the transportation space fit degree. During the sequential allocation process, the remaining supply and remaining demand of time-priority orders of each supplier are updated to generate a fast matching result.

[0055] Specifically, quick matching is used to rapidly identify suppliers capable of delivering on time from the vicinity of the demand location when there is a short remaining time for fulfillment. The platform first reads the demand location, urgency level, target category, and buyer identifier of time-priority orders, and then forms a historical transportation range based on the actual transportation mileage, transportation time, on-time delivery rate, and transportation loss rate of the corresponding buyer's historical orders. The historical transportation range can be represented by the quantile value of the transportation distance corresponding to orders with high fulfillment success rates, and this quantile value is used as the basic search radius. For buyers who lack historical transportation records, the transportation statistics of buyers in the same region and category can be used to generate an initial range, which is then gradually updated after subsequent fulfillment.

[0056] The dynamic search range adjusts the basic search radius based on order urgency. The more urgent the order, the shorter the available transportation time, and the smaller the search radius becomes, reducing the chance of distant candidate suppliers entering the quick matching process. When orders still retain some transportation buffer, the search radius can be appropriately expanded within historically acceptable limits. The platform can also combine road traffic conditions, cold chain vehicle availability, and estimated loading and unloading time to convert the geometric radius into an reachable time range. Specifically, starting from the demand location, the maximum path range that can be reached after deducting sorting, packaging, and acceptance time from the remaining fulfillment time is calculated along the road network, and the intersection of the historical transportation range and the reachable time range is determined as the dynamic search range. For areas with traffic restrictions, suppliers are screened using actual traversable paths.

[0057] Furthermore, after obtaining the dynamic search range, the platform retrieves suppliers whose production locations are within the range and whose product categories are compatible with the target category, and performs a fulfillment window adaptation check. During the check, the planned harvest window, post-harvest processing time, available loading time slots, and estimated transportation time are read to calculate the earliest and latest arrival times for each supplier. If there is a valid overlap between the available delivery time window and the order fulfillment window, and the latest arrival time is not later than the order's allowable boundary, the time condition is deemed met. For orders that allow for partial delivery, the arrival time can be calculated separately for each harvest batch, and the quantity that can arrive within the fulfillment window can be accumulated. If the supplier's estimated arrival time exceeds the boundary, or the harvest plan has not yet been determined, the corresponding supplier is removed.

[0058] Available supply verification is used to confirm that candidate suppliers have actual allocable capacity. Based on the projected supply, the platform deducts locked supply, reserved supply, confirmed losses, and other high-priority order occupancy to obtain the current remaining supply. If the remaining supply reaches the minimum order quantity and the quality boundary, packaging capacity, and loading capacity meet the requirements, it is retained as a partial candidate supplier. If a single supplier is insufficient to cover all demand but reaches the minimum order quantity, multiple supplier combinations can be used. Suppliers with remaining supply below the minimum order quantity or conflicting loading times are directly excluded to avoid generating unexecutable allocation relationships.

[0059] The spatial suitability of transportation is generated based on the road transportation distance from the candidate supplier to the demand location, the estimated transportation time, and the historical on-time performance rate. The platform first converts the transportation distance and transportation time into positive indicators, with higher values ​​indicating a higher suitability. Then, it combines this with the historical fulfillment stability of the corresponding route to form a spatial suitability value. Suppliers with shorter road distances, less traffic fluctuations, and higher historical on-time performance rates are given higher ranking positions. To avoid small-volume nodes frequently consuming processing resources due to ranking solely by distance, suppliers with larger supply volumes and shorter loading preparation times can be prioritized when spatial suitability is similar. The ranking results form a local candidate sequence for subsequent order allocation.

[0060] During sequential allocation, the platform reads the remaining supply from the supply side according to the local candidate sequence, using the remaining demand of time-priority orders as the current allocation limit. After each allocation, the platform immediately deducts the remaining supply from the corresponding supply side and the remaining demand of the order, while recording the allocated quantity, estimated arrival time, and transportation route. Allocation stops when the remaining demand of the order drops to 0. If a gap still exists after all candidate supply sides have been processed, the gap is passed to subsequent conflict coordination or global supplementary allocation processes. If multiple time-priority orders simultaneously occupy the same supply side, sequential allocation can be performed according to the urgency of the orders from highest to lowest, and the available supply for subsequent orders is re-verified after each deduction.

[0061] For example, a buyer's tomato order D1 requires 120,000 kg, with only 30 hours remaining before the fulfillment window begins. Historical transportation records indicate a high on-time fulfillment rate within a 220 km radius. Considering current road conditions and loading / unloading times, the dynamic search range is determined to be 180 km. Within this range, there are three bases: Base A, Base B, and Base C. Base A is expected to supply 70,000 kg at a distance of 90 km; Base B is expected to supply 50,000 kg at a distance of 140 km; Base C is 160 km away, but its expected arrival time after the planned harvest is later than the fulfillment window, so it is excluded. Bases A and B both meet the quality, minimum order quantity, and loading requirements, with transportation space fit values ​​of 0.91 and 0.78, respectively. The platform first allocates 70,000 kg to base A, then updates the remaining demand of the order to 50,000 kg, and then allocates 50,000 kg to base B, generating a quick matching result that includes the two supply sides, the corresponding allocation quantities, the transportation routes, and the expected arrival times.

[0062] Through the above implementation methods, the search boundaries can be dynamically defined based on the urgency of the order and actual availability, and the candidate set can be quickly reduced through layer-by-layer verification adapted to fulfillment time, available supply, and transportation space. This reduces the computational load for matching urgent orders, minimizes the number of non-fulfillable suppliers entering the allocation results, and improves the response speed, execution feasibility, and supply resource utilization efficiency of rapid matching.

[0063] In some embodiments, the candidate range of the same category or the expansion range of related categories is determined based on the cross-category substitution elasticity; a supply and demand allocation constraint model is constructed based on the supply and demand matching characterization; the constraint strength of the supply and demand allocation constraint model is adjusted according to the cyclical market supply and demand status; and the global matching result is obtained by solving the model, including: The cross-category substitution elasticity corresponding to the globally balanced order is compared with the substitution judgment threshold. When the cross-category substitution elasticity does not reach the substitution judgment threshold, a candidate range of the same category is constructed within the target category. When the cross-category substitution elasticity reaches the substitution judgment threshold, the candidate range of the same category is expanded to related categories based on historical substitution procurement relationships to obtain the extended range of related categories. Using the allocation amount of candidate supply to demand orders as decision variables, a matching target is constructed based on the supply and demand matching characteristics and cross-category substitution elasticity. In combination with the supply available quantity boundary, the minimum quantity of demand orders to be satisfied and the fulfillment feasibility conditions, a supply and demand allocation constraint model is generated. It aggregates the total supply and demand within a continuous trading cycle, generates the supply and demand status of the cycle market, and adjusts the constraint strength between demand guarantee constraints and supply utilization constraints based on the supply and demand status of the cycle market. The adjusted supply and demand allocation constraint model is solved iteratively. When the change in the matching target satisfies the convergence condition, the global matching result containing the supply and demand pairing relationship and the corresponding allocation amount is output.

[0064] Specifically, global matching is used to uniformly allocate resources across multiple supply sides and multiple demand orders when there is relatively sufficient remaining time for order fulfillment. The platform first reads the target category, allowed substitution status, and cross-category substitution elasticity corresponding to the globally balanced order, and compares the cross-category substitution elasticity with the substitution judgment threshold. The substitution judgment threshold can be determined based on the acquirer's historical substitution fulfillment success rate, category quality differences, and the effective substitution sample size. When the cross-category substitution elasticity is lower than the threshold, only suppliers in the target category that meet the fulfillment conditions are retained, forming a candidate range within the same category; when the cross-category substitution elasticity reaches the threshold, related categories with stable substitution directions, compatible quality attributes, and fulfillment success rates are selected according to historical substitution procurement relationships, and the corresponding suppliers are added to the candidate set, forming an expanded range of related categories. For orders that explicitly prohibit substitution or where there is a mandatory quality conflict in related categories, the candidate range within the same category is maintained even if the substitution elasticity reaches the threshold.

[0065] When constructing the supply and demand allocation constraint model, the allocation amount from candidate suppliers to each demand order is used as the decision variable, and a supply and demand matching representation is associated with each candidate supply and demand combination. The matching objective can be formed by weighting the fulfillment time fit value, transportation space fit value, supply stability, and order fulfillment degree. For related category combinations, substitution benefits or substitution penalties are also set based on cross-category substitution elasticity, so that combinations with higher substitution elasticity are more likely to be adopted, while combinations with lower substitution elasticity only participate in allocation when resources of the same category are insufficient.

[0066] In some examples, the supply availability boundary is used to limit the cumulative quantity allocated to all orders from the same supplier to no more than the current available supply; the minimum order fulfillment quantity is used to ensure that selected orders meet at least the minimum receiving batch size or minimum fulfillment ratio; and the fulfillment feasibility condition is used to exclude combinations that cannot be completed within the fulfillment window for receiving, sorting, transporting, and accepting. For orders that allow for partial delivery, allocation variables and time boundaries can be set separately for each delivery batch, and then aggregated to form the total order fulfillment quantity.

[0067] The cyclical market supply and demand status is generated based on the total effective supply, total effective demand, supply fluctuations, and unmet demand over multiple consecutive trading cycles. The platform can calculate the supply-demand ratio, the number of cycles the gap lasts, and the inventory digestion rate, and classify the market status into tight supply, balanced supply and demand, and loose supply. When supply is tight, the weight of demand guarantee constraints is increased to prioritize ensuring more orders reach the minimum fulfillment quantity, while reducing the requirements of supply utilization constraints on compressing the remaining quantity, allowing for the retention of some supply buffers. When supply is loose, the weight of supply utilization constraints is increased to reduce idle available resources and allow for expansion of allocation within the feasible range of fulfillment. When supply and demand are balanced, both types of constraints maintain their basic strength. The adjustment of constraint strength is achieved by modifying the penalty coefficient and constraint relaxation boundary in the matching objective, and the adjustment range of a single cycle is limited to avoid drastic changes in matching results due to short-term fluctuations in market status.

[0068] During iterative solution processing, the platform first generates an initial allocation solution based on the supply-demand matching representation. Then, under the premise of satisfying the supply boundary, minimum fulfillment quantity, and feasibility conditions, it adjusts the allocation quantity of different supply-demand combinations round by round. After each round of solution processing, the matching target value is calculated, and the change range of the target value in two consecutive rounds is compared. When the change range is lower than the convergence threshold and all constraints are satisfied, the iteration ends. If convergence is not achieved after reaching the maximum number of iterations, the candidate solution with the smallest constraint violation and the best target value is retained, and the orders that need further coordination are marked. During the solution process, an upper limit can be set on the number of orders undertaken by the same supplier to avoid excessive splitting that leads to complexity in loading and delivery execution. Finally, the global matching result is output, including supplier identifier, demand order identifier, allocation category, allocation quantity, expected arrival time, and substitution relationship marker.

[0069] For example, there are currently two orders for regular tomatoes, D2 and D3, with demand quantities of 100,000 kg and 80,000 kg respectively, and both have more than 72 hours remaining to fulfillment time. The effective supply of regular tomatoes of the same category is 130,000 kg, which cannot cover all demand. The cross-category substitution elasticity for D2 is 0.18, which is lower than the substitution threshold of 0.40, so the candidate range is only constructed within the supply side of regular tomatoes. The substitution elasticity for D3 corresponding to processed tomatoes is 0.62, and its quality attributes meet the requirements for processing purposes, so the supply side of processed tomatoes is included in the related category expansion range. The supply-demand ratios for three consecutive trading cycles are 0.84, 0.79, and 0.81, respectively, indicating a tight supply in the market, and the platform increases the minimum fulfillment constraint. After solving the problem, the regular tomato supply side allocates 90,000 kg to D2 and 40,000 kg to D3, while the processed tomato supply side supplements the allocation of 40,000 kg to D3, and both orders reach the minimum fulfillment ratio. After two consecutive rounds of matching target change is less than 0.5%, the corresponding global matching result is output.

[0070] Through the above implementation methods, the range of candidate product categories can be dynamically controlled based on actual substitution capabilities, and time, space, supply capacity, and substitution relationships can be uniformly incorporated into the constraint solution. Simultaneously, the balance between order fulfillment and resource utilization can be adjusted according to cyclical market supply and demand changes. This improves the executability of the global matching results and the rationality of resource allocation, reduces order gaps caused by target product shortages, and enhances the matching model's adaptability to market supply and demand fluctuations.

[0071] In some embodiments, the method further includes: Multiple supply disturbance scenarios are constructed based on the expected supply volume and fluctuation range of each candidate supply side. The candidate matching results corresponding to each supply disturbance scenario are solved by using the supply and demand allocation constraint model. Compare the demand satisfaction status, supply remaining status, and transportation space adaptability of candidate matching results under different supply disturbance scenarios to generate a matching stability characterization. Based on the stability characterization of the matching, robust control parameters including supply reserve and robustness penalty parameters are generated. The supply and demand allocation constraint model is updated using robust control parameters, and candidate matching results that meet the preset stability conditions across supply disturbance scenarios are determined as global matching results.

[0072] Specifically, the supply disturbance scenario simulates changes in available supply caused by production forecast errors, weather changes, pests and diseases, harvesting progress deviations, and temporary contract fulfillment. The platform reads the expected supply, lower limit of fluctuation, upper limit of fluctuation, and source of fluctuation for each candidate supplier, uses the expected supply as the baseline supply state, and generates multiple scenario supply quantities based on the historical forecast error distribution within the fluctuation range. Scenarios can include a baseline scenario, a general production reduction scenario, a partial production reduction scenario, and a production increase scenario. The general production reduction scenario synchronously compresses the supply according to the lower limit of fluctuation for each supplier; the partial production reduction scenario only adjusts some suppliers affected by the same weather region or production risk; and the production increase scenario is used to verify whether the newly added supply resources can be reasonably absorbed. For related suppliers, joint disturbance relationships can be set according to region, category, and production cycle to avoid treating related production reduction risks in the same region as independent events.

[0073] Furthermore, after generating the scenario supply quantities, the platform writes each scenario into the supply and demand allocation constraint model, updates the supply-side available supply boundary, and keeps the demand orders, fulfillment windows, candidate range, and basic matching parameters unchanged. The model uses the allocation quantity of candidate suppliers to demand orders as the decision variable, and solves independently under each supply disturbance scenario to obtain the corresponding candidate matching results. Each candidate matching result records the order fulfillment quantity, minimum fulfillment ratio, supply-side allocation quantity, remaining supply quantity, transportation space adaptation value, and related category substitution quantity. If a supply and demand combination loses fulfillment feasibility in some scenarios due to insufficient supply, the combination is marked as a disturbance-sensitive relationship.

[0074] Matching stability is used to compare the reliability of the same candidate matching scheme under different supply states. Demand fulfillment status can be generated based on the order fulfillment ratio, the number of unfulfilled orders, and the minimum demand guarantee status; supply surplus status can be generated based on the remaining quantity of each supplier, the degree of over-concentration in allocation, and the safety margin gap; transportation space adaptability can be calculated by statistically analyzing the average transportation distance, the proportion of low-adaptability combinations, and the magnitude of changes in transportation routes under each scenario. The platform performs unified and normalized processing on the above indicators and calculates the mean, range, and volatility coefficient between different scenarios. Matching stability is high when the order fulfillment ratio remains consistently high, the remaining supply status changes little, and the transportation space adaptability is stable; if a slight reduction in production causes multiple orders to fall below the minimum fulfillment quantity, or if allocation relationships frequently shift to distant suppliers, then matching stability is low.

[0075] The platform can also identify key supply chains that lead to decreased stability. Specifically, it compares changes in order fulfillment status before and after removing a single supply chain to calculate the degree of supply dependence. Supply chains with high dependence and expected large fluctuations in supply volume are marked as vulnerable supply nodes. For cases where multiple orders are concentrated on the same vulnerable supply node, the risk contribution value of the corresponding allocation relationship is increased, and the risk contribution is written into the matching stability representation. This distinguishes between general supply fluctuations and key fluctuations that may trigger widespread order gaps.

[0076] When generating robust control parameters based on the matching stability characterization, the platform first determines the supply reserve for each supply side. The supply reserve can be calculated jointly based on the expected supply fluctuation range, supply dependence, and order urgency. Supply sides with larger fluctuations or those jointly depended on by multiple orders retain higher reserves, while supply sides with stable supply and sufficient alternative sources retain lower reserves. The supply reserve is deducted from the model's allocable upper limit to cope with subsequent production reductions or temporary adjustments. A robustness penalty parameter is used to reduce the model's tendency to select combinations sensitive to disturbances; its value can be determined based on order fulfillment fluctuations, transportation adaptability fluctuations, and the risk contribution of key supply sides. The more sensitive the disturbance, the higher the penalty for allocating that combination.

[0077] Subsequently, the platform uses supply reserves to correct the supply-side available supply boundary and adds robustness penalty parameters to the matching objective, resolving the supply-demand allocation constraint model. The updated model no longer simply pursues the highest matching value in the baseline scenario, but balances order fulfillment, transportation space adaptability, and cross-scenario stability. For the re-obtained matching results, the platform performs replay verification under various supply disturbance scenarios. If the minimum order fulfillment ratio, supply safety margin, and average transportation space adaptability all meet stability conditions in all scenarios or scenarios with a preset ratio, the corresponding result is determined as the global matching result; otherwise, the reserve amount or robustness penalty parameter of vulnerable supply nodes is further increased until the iteration limit is reached.

[0078] For example, the combined demand for regular tomatoes, D2 and D3, is 180,000 kg. The estimated supply from candidate bases A, B, and C are 90,000 kg, 70,000 kg, and 50,000 kg, respectively, with fluctuation ranges of ±20,000 kg, ±10,000 kg, and ±5,000 kg. The platform constructs a baseline scenario, a scenario where base A's production decreases by 20,000 kg, a scenario where base A and base B jointly reduce production, and a scenario where base C increases production. The baseline model concentrates the 90,000 kg demand at base A. However, in the scenario where base A reduces production, the minimum fulfillment rate for D2 drops to 70%, indicating that the allocation result is overly dependent on base A. Based on this, the platform identifies base A as a vulnerable supply node, sets a 15,000 kg supply reserve for it, and increases the robustness penalty parameter for the concentrated allocation combination at base A. After resolving the problem, bases A, B, and C are assigned to demands of 75,000 kg, 65,000 kg, and 40,000 kg, respectively. Scenario replay shows that even if base A reduces production, the two orders can still be fulfilled at a minimum ratio using the remaining capacity of bases B and C, and the average transport space fit value does not fall below the stability threshold. Therefore, this solution is determined as the global matching result.

[0079] The above implementation method allows for the early incorporation of uncertainties in projected supply into the global matching process. It identifies allocation relationships and key supply nodes sensitive to production reduction disturbances and adjusts resource allocation through supply reserves and robustness penalty parameters. This reduces the risk of large-scale order gaps caused by fluctuations in a single supply segment and improves the stability, fulfillment reliability, and continuous execution capability of the global matching results under different supply conditions.

[0080] In some embodiments, supply resource conflicts between fast matching results and global matching results are identified, priority allocation relationships are determined according to order urgency, and the supply and demand allocation constraint model is re-solved based on remaining supply resources to generate channel supply and demand matching results, including: Based on the allocation quantities in the quick matching results and global matching results associated with the supplier identifier, compare the cumulative allocation quantity of the same supplier with the available supply quantity boundary to identify supply resource conflicts caused by duplicate occupation or over-allocation. For conflicting demand orders, compare the urgency of the orders, prioritize the allocation relationship corresponding to the time-sensitive orders, and deduct the corresponding supply quantity from the supply side to generate the remaining supply resources; The candidate range, decision variable boundaries, and supply-demand matching representation of the global equilibrium order are updated based on the remaining supply resources. The updated results are then written into the supply-demand allocation constraint model and solved again. The system aggregates the retained fast matching results and the re-solved global matching results, verifies the cumulative allocation of each supplier and the fulfillment status of each demand order, and generates channel supply and demand matching results.

[0081] Specifically, supply resource conflict refers to situations where both quick matching and global matching results simultaneously occupy the same available production capacity on the same supply side, causing the cumulative allocation to exceed the available supply limit, or the same batch of production being repeatedly associated with different demand orders. After the platform generates the two types of matching results, it establishes a unified resource index based on the supply side identifier, production batch identifier, and available time period, mapping each allocation record to the corresponding supply resource unit. The supply resource unit can be jointly determined by the base, category, production batch, and planned harvest window to avoid incorrect merging of different batches from the same base. The platform summarizes the allocation of each resource unit in the quick matching and global matching results and compares it with the available supply limit after deducting the locked quantity, safety reserve quantity, and expected loss quantity.

[0082] When the cumulative allocation exceeds the available supply limit, an over-allocation conflict is identified. Even when the same resource unit is occupied by two different types of results but the cumulative allocation has not yet exceeded the limit, a duplicate allocation conflict can still be determined based on loading time, harvesting capacity, and packaging capacity. For example, if the same batch is sufficient in quantity, but two orders require loading to be completed within the same time period, and the base's single-period processing capacity can only support one order, this is also marked as a conflict. Conflict records include the supplier identifier, conflicting batch, related orders, respective allocation quantities, conflict quantity, and conflict time period, forming a set of resources to be coordinated according to the scope of the conflict.

[0083] For conflicting demand orders, the platform reads the order urgency, fulfillment window start time, minimum fulfillment quantity, and currently allocated quantity to determine priority allocation. In principle, priority is given to retaining the fast-matching relationships corresponding to time-sensitive orders to avoid fulfillment delays caused by re-matching. If multiple time-sensitive orders conflict, their urgency, remaining fulfillment time, and minimum fulfillment gap are compared sequentially, prioritizing orders with higher urgency and shorter remaining time. For orders with the same urgency, priority can be determined based on the demand fulfillment increment generated after unit resource allocation, ensuring that limited supply is prioritized for orders that can meet the minimum fulfillment conditions. Once the priority allocation relationship is determined, the corresponding allocated quantity is locked and deducted from the available supply on the supply side to obtain the remaining supply resources.

[0084] Remaining supply resources are re-statusd according to supplier identifiers and production batches, recording remaining supply quantity, available loading time slots, remaining packaging capacity, and the range of fulfillable orders. If a supplier's remaining supply quantity after priority allocation is lower than the minimum receiving batch size for the globally balanced order, it is removed from the corresponding candidate range; if it still has partial allocation capacity, the upper limit of the decision variable is updated. For supply-demand combinations that became invalid due to resource deduction in the original global matching results, the fulfillment time suitability and transportation space suitability are recalculated, and the supply-demand matching representation is corrected based on the remaining supply quantity. If the target category resources are insufficient, the related category expansion range can be re-invoked under the condition that the order allows substitution to supplement the candidate set with alternative suppliers.

[0085] The platform incorporates the updated candidate range, decision variable boundaries, supply-demand matching representation, and remaining supply resources into the supply-demand allocation constraint model. During re-solution, locked fast matching relationships are treated as fixed allocation conditions and no longer subject to adjustment; the allocation quantity corresponding to the globally balanced order is treated as a variable to be optimized and reconfigured within the remaining resource range. The model continues to satisfy the supply quantity boundary, minimum order fulfillment quantity, fulfillment feasibility conditions, and category substitution constraints, and can set adjustment penalties for significant re-allocations caused by conflicts, reducing unnecessary supply-side switching. After the solution is completed, the regenerated global matching results are compared with the original results to confirm that all allocation quantities come from remaining supply resources and that there is no new duplicate allocation.

[0086] During the aggregation phase, the retained quick matching results and the re-solved global matching results are merged according to the demand order identifier to form a matching set to be verified. The platform then aggregates the cumulative allocation volume of each supplier to confirm that it does not exceed the available supply limit and checks whether the loading time, harvesting capacity, and packaging capacity meet the execution requirements. Simultaneously, the cumulative fulfillment volume of each demand order is calculated to determine whether the minimum fulfillment volume or target fulfillment ratio has been reached. For orders that are still not fulfilled, the remaining demand gap and the reason for the gap are recorded; for orders that are supplemented using related product categories, the substitute product category, substitution ratio, and quality compatibility status are recorded. After verification, the channel supply and demand matching results are output, and each allocation relationship is associated with the corresponding supply batch and fulfillment window.

[0087] For example, base A can supply 100,000 kg of regular tomatoes in a certain batch. The quick matching result allocates 70,000 kg to the time-priority order D1, while the global matching result allocates another 50,000 kg to order D2, bringing the total allocation to 120,000 kg, creating a 20,000 kg over-limit conflict. D1 has only 26 hours left before the fulfillment window begins, while D2 has 80 hours remaining. Therefore, the platform retains the 70,000 kg allocation for D1 and updates the remaining supply from base A to 30,000 kg. Since the minimum acceptable batch size for D2 is 40,000 kg, base A is removed from D2's candidate range. The platform then re-selects base B and processing tomato base C, with base B supplying 30,000 kg and processing tomato base C providing an additional 20,000 kg. After the re-resolution, D2 is allocated 30,000 kg by base B and 20,000 kg by base C, achieving the minimum fulfillment requirement. The final verification showed that bases A, B, and C did not over-allocate, and D1 and D2 both met the corresponding performance requirements.

[0088] Through the above implementation methods, after local rapid matching and global optimized matching are generated in parallel, supply resources can be uniformly indexed, conflict identified, and prioritized for coordination. Remaining resources can then be used to re-optimize global allocation. This avoids duplicate or over-allocated supply batches, ensures the fulfillment priority of urgent orders, and improves the resource consistency, execution feasibility, and overall configuration efficiency of channel supply and demand matching results.

[0089] The following are system embodiments of this application, which can be used to execute the method embodiments of this application. For details not disclosed in the system embodiments of this application, please refer to the method embodiments of this application.

[0090] Figure 2 This is a schematic diagram of the intelligent supply and demand matching system for order-based agriculture provided in this application embodiment. For example... Figure 2 As shown, the system includes: The acquisition module 201 is used to acquire production data from planting bases and demand data from buyers, generate supply-side production characteristics and demand-side procurement characteristics, and establish supply-demand matching relationships. The generation module 202 is used to generate a supply and demand matching representation based on the degree of fit between the fulfillment time and the degree of fit between the supply and demand matching relationship, and to generate cross-category substitution elasticity based on the historical substitution procurement relationship of the acquirer. The segmentation module 203 is used to determine the urgency of orders based on the remaining time for fulfillment of demand orders and the supply and demand status of target categories, and to segment demand orders into time-priority orders and globally balanced orders. The filtering module 204 is used to filter local candidate suppliers for time-priority orders according to spatial range, fulfillment time and supply capacity, and generate fast matching results based on the adaptability of transportation space; Module 205 is used to determine the candidate range of the same category or the expansion range of related categories based on the cross-category substitution elasticity for globally balanced orders, construct a supply and demand allocation constraint model based on the supply and demand matching characterization, adjust the constraint strength of the supply and demand allocation constraint model according to the supply and demand status of the cyclical market, and solve to obtain the global matching result. The re-decomposition module 206 is used to identify supply resource conflicts between the fast matching results and the global matching results, determine the priority allocation relationship according to the urgency of the orders, and re-decompose the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

[0091] In some embodiments, Figure 2 The acquisition module 201 associates current cycle data with historical production records according to the planting base identifier, generates the expected supply and fluctuation range based on historical unit output distribution and current production scale, and obtains supply-side production characteristics by combining the planned harvest window and production location; it associates procurement constraints and historical fulfillment records according to the demand order identifier, and generates demand-side procurement characteristics including demand scale, fulfillment window, quality boundary and demand location; it filters supply-side production characteristics and demand-side procurement characteristics based on category compatibility, and establishes supply-demand matching relationship based on the overlap of fulfillment windows, quality adaptability and basic supply capacity.

[0092] In some embodiments, Figure 2 The generation module 202 calculates the degree of overlap between the planned harvest window in the production characteristics of the supply side and the fulfillment window in the procurement characteristics of the demand side, and determines the transportation distance based on the location of the production site and the location of the demand. It unifies the direction and normalizes the scale of the overlap degree and transportation distance, determines the corresponding weights in combination with the historical fulfillment characteristics of the purchaser, and generates a supply and demand matching representation. It extracts the linkage changes in the procurement volume between related categories from the historical procurement records of the purchaser, identifies the compensatory procurement changes of related categories when the procurement of the target category decreases, and generates cross-category substitution elasticity based on the ratio of the compensation magnitude to the change magnitude of the target category.

[0093] In some embodiments, Figure 2 The generation module 202 extracts compensatory procurement relationships between different product categories based on the historical procurement records of the acquirer, and constructs a related product category relationship graph by combining the quality attribute adaptability of the corresponding product categories; using product categories as graph nodes and compensatory procurement relationships as graph connections, the substitution reliability of each graph connection is determined based on the cross-category substitution elasticity and the quality attribute adaptability; when the globally balanced order meets the category expansion conditions, starting from the graph node corresponding to the target product category, a restricted graph traversal is performed according to the substitution reliability, and related product categories are filtered according to the preset expansion level and cumulative substitution deviation to generate the related product category expansion range.

[0094] In some embodiments, Figure 2The partitioning module 203 calculates the time interval between the start time of the fulfillment window for each demand order and the current time, and generates a basic urgency threshold based on the distribution of the time interval among the pending demand orders; it aggregates the expected supply and order demand for the target category, generates the category supply and demand status based on the ratio of the two, and dynamically adjusts the basic urgency threshold according to the category supply and demand status to obtain the urgency judgment threshold; it compares the time interval with the urgency judgment threshold, and if the time interval is not greater than the urgency judgment threshold, the corresponding demand order is determined as a time-priority order; otherwise, the corresponding demand order is determined as a globally balanced order.

[0095] In some embodiments, Figure 2 The filtering module 204 uses the demand location of time-priority orders as the search center, determines the dynamic search range based on the historical transportation range of the corresponding purchaser and the urgency of the order, and filters the suppliers within the dynamic search range; it matches the production characteristics of the suppliers with the procurement characteristics of the demand side of time-priority orders to the fulfillment window and verifies the available supply, and eliminates suppliers that do not meet the preset fulfillment conditions to form local candidate suppliers; it generates the transportation space fit degree based on the transportation distance from the local candidate suppliers to the demand location, sorts the local candidate suppliers according to the transportation space fit degree, and updates the remaining supply and remaining demand of time-priority orders of each supplier during the sequential allocation process to generate a fast matching result.

[0096] In some embodiments, Figure 2 The construction module 205 compares the cross-category substitution elasticity corresponding to the global equilibrium order with the substitution judgment threshold. When the cross-category substitution elasticity does not reach the substitution judgment threshold, a candidate range of the same category is constructed within the target category. When the cross-category substitution elasticity reaches the substitution judgment threshold, the candidate range of the same category is expanded to related categories based on historical substitution procurement relationships, resulting in an expanded range of related categories. Using the allocation amount from the candidate supply side to the demand order as the decision variable, a matching target is constructed based on the supply and demand matching characteristics and the cross-category substitution elasticity. Combined with the supply side's available supply boundary, the minimum fulfillment amount of the demand order, and the feasibility of performance, a supply and demand allocation constraint model is generated. The total supply and total demand within the continuous transaction cycle are aggregated to generate the cyclical market supply and demand status. The constraint strength between the demand guarantee constraint and the supply utilization constraint is adjusted according to the cyclical market supply and demand status. The adjusted supply and demand allocation constraint model is iteratively solved. When the change magnitude of the matching target meets the convergence condition, a global matching result containing the supply and demand pairing relationship and the corresponding allocation amount is output.

[0097] In some embodiments, Figure 2The construction module 205 constructs multiple supply disturbance scenarios based on the expected supply and fluctuation range of each candidate supplier. It uses the supply and demand allocation constraint model to solve the candidate matching results corresponding to each supply disturbance scenario. It compares the demand satisfaction state, supply remaining state and transportation space adaptability of the candidate matching results under different supply disturbance scenarios to generate a matching stability characterization. Based on the matching stability characterization, it generates robust control parameters including supply reservation and robustness penalty parameters. It uses the robust control parameters to update the supply and demand allocation constraint model and determines the candidate matching results that meet the preset stability conditions across supply disturbance scenarios as the global matching results.

[0098] In some embodiments, Figure 2 The resolution module 206 associates the allocation quantities in the fast matching results and global matching results with the supplier identifier, compares the cumulative allocation quantity of the same supplier with the available supply quantity boundary, and identifies supply resource conflicts caused by duplicate occupation or over-allocation; for conflicting demand orders, it compares the urgency of the orders, prioritizes the allocation relationship corresponding to the time-priority orders, and deducts the available supply quantity of the corresponding supplier to generate remaining supply resources; based on the remaining supply resources, it updates the candidate range, decision variable boundary, and supply-demand matching representation of the global equilibrium orders, writes the updated results into the supply-demand allocation constraint model, and resolves it; it combines the retained fast matching results with the resolved global matching results, verifies the cumulative allocation quantity of each supplier and the satisfaction status of each demand order, and generates channel supply-demand matching results.

[0099] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0100] Figure 3 This is a schematic diagram of the electronic device 3 provided in an embodiment of this application. Figure 3 As shown, the electronic device 3 of this embodiment includes: a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, it implements the steps in the various method embodiments described above. Alternatively, when the processor 301 executes the computer program 303, it implements the functions of each module / unit in the various system embodiments described above.

[0101] Electronic device 3 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 3 may include, but is not limited to, processor 301 and memory 302. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or different components.

[0102] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0103] The memory 302 can be an internal storage unit of the electronic device 3, such as a hard disk or memory of the electronic device 3. The memory 302 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the electronic device 3. The memory 302 can also include both internal and external storage units of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device.

[0104] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0105] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a readable storage medium (e.g., a computer-readable storage medium). Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which may be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable storage medium may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0106] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for intelligent matching of supply and demand in contract farming channels, characterized in that, include: Acquire production data from planting bases and demand data from buyers to generate supply-side production characteristics and demand-side procurement characteristics, and establish supply-demand matching relationships; A supply and demand matching representation is generated based on the degree of fit between the fulfillment time and the degree of fit between the transportation space of the supply and demand matching relationship, and a cross-category substitution elasticity is generated based on the historical substitution procurement relationship of the acquirer. The urgency of orders is determined based on the remaining time for fulfillment of demand orders and the supply and demand status of target categories, and the demand orders are divided into time-priority orders and globally balanced orders. For the time-priority orders, local candidate suppliers are filtered according to spatial range, delivery time and supply capacity, and a fast matching result is generated based on the adaptability of transportation space. For the global equilibrium order, the candidate range of the same category or the expansion range of related categories is determined according to the cross-category substitution elasticity. A supply and demand allocation constraint model is constructed based on the supply and demand matching characterization. The constraint strength of the supply and demand allocation constraint model is adjusted according to the supply and demand status of the cyclical market, and the global matching result is obtained by solving the problem. Identify supply resource conflicts between the fast matching results and the global matching results, determine priority allocation relationships according to the urgency of the orders, and re-solve the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

2. The method according to claim 1, characterized in that, The process of acquiring production data from planting bases and demand data from buyers, generating supply-side production characteristics and demand-side procurement characteristics, and establishing supply-demand matching relationships includes: By associating current cycle data with historical production records according to the planting base identification, the projected supply and fluctuation range are generated based on the historical unit output distribution and current production scale, and the supply-side production characteristics are obtained by combining the planned harvest window and the location of the production area. Based on the demand order identifier, associated procurement constraints, and historical fulfillment records, generate the demand-side procurement characteristics, including demand size, fulfillment window, quality boundary, and demand location. The supply-side production characteristics and demand-side procurement characteristics are screened based on product category compatibility, and the supply-demand matching relationship is established based on the overlap of fulfillment windows, quality adaptability, and basic supply capacity.

3. The method according to claim 1, characterized in that, The process of generating a supply-demand matching representation based on the degree of fit in fulfillment time and transportation space of the supply-demand matching relationship, and generating cross-category substitution elasticity based on the buyer's historical substitution procurement relationships, includes: Calculate the degree of overlap between the planned harvest window in the supply-side production characteristics and the fulfillment window in the demand-side procurement characteristics, and determine the transportation distance based on the production location and the demand location; The degree of overlap and transportation distance are unified in direction and normalized in scale. The corresponding weights are determined by combining the historical performance characteristics of the acquirers, and the supply and demand matching representation is generated. Extract the linkage changes in purchase volume between related product categories from the historical purchase records of the acquirer, identify the compensatory purchase changes in related product categories when the purchase of the target product category decreases, and generate the cross-category substitution elasticity based on the ratio of the compensation magnitude to the change magnitude of the target product category.

4. The method according to claim 3, characterized in that, Also includes: Based on the historical purchasing records of the acquirer, the compensatory purchasing relationships between different product categories are extracted, and a relationship diagram of related product categories is constructed by combining the matching degree of the quality attributes of the corresponding product categories. Using product categories as graph nodes and the compensatory procurement relationships as graph connections, the substitution reliability of each graph connection is determined based on the cross-category substitution elasticity and the quality attribute adaptability. When the global balanced order meets the category expansion condition, starting from the graph node corresponding to the target category, a restricted graph traversal is performed according to the substitution confidence level, and related categories are filtered according to the preset expansion level and cumulative substitution deviation to generate the related category expansion range.

5. The method according to claim 1, characterized in that, The urgency of orders is determined based on the remaining fulfillment time of demand orders and the supply and demand status of target categories. These demand orders are then categorized into time-priority orders and globally balanced orders, including: Calculate the time interval between the start time of the fulfillment window for each demand order and the current time, and generate a basic urgency threshold based on the distribution of the time interval among the demand orders to be processed. The projected supply and order demand for the target product category are collected, and the product category supply and demand status is generated based on the ratio of the two. The basic urgency threshold is dynamically adjusted according to the product category supply and demand status to obtain the urgency determination threshold. The time interval is compared with the urgency determination threshold. If the time interval is not greater than the urgency determination threshold, the corresponding demand order is determined as the time-priority order; otherwise, the corresponding demand order is determined as the globally balanced order.

6. The method according to claim 1, characterized in that, For the time-priority orders, the process involves filtering local candidate suppliers based on spatial range, fulfillment time, and supply capacity, and generating rapid matching results based on the suitability of transportation space, including: Using the demand location of the time-priority orders as the search center, a dynamic search range is determined based on the historical transportation range of the corresponding purchaser and the urgency of the orders, and suppliers located within the dynamic search range are selected. The supply-side production characteristics are matched with the demand-side procurement characteristics of the time-priority orders, and the available supply is verified. Supply-side suppliers that do not meet the preset fulfillment conditions are eliminated, forming the partial candidate supply-side suppliers. Based on the transportation distance from the local candidate suppliers to the demand location, a transportation space fit degree is generated. The local candidate suppliers are sorted according to the transportation space fit degree. During the sequential allocation process, the remaining supply of each supplier and the remaining demand of the time-priority order are updated to generate the fast matching result.

7. The method according to claim 1, characterized in that, The process involves determining the candidate range for the same category or the expansion range for related categories based on the cross-category substitution elasticity, constructing a supply and demand allocation constraint model based on the supply and demand matching representation, adjusting the constraint strength of the supply and demand allocation constraint model according to the cyclical market supply and demand status, and solving for the global matching result, including: The cross-category substitution elasticity corresponding to the globally balanced order is compared with the substitution judgment threshold; when the cross-category substitution elasticity does not reach the substitution judgment threshold, a candidate range of the same category is constructed within the target category; when the cross-category substitution elasticity reaches the substitution judgment threshold, the candidate range of the same category is expanded to related categories based on historical substitution procurement relationships to obtain the extended range of related categories. Using the allocation amount of candidate supply to demand orders as decision variables, a matching target is constructed based on the supply and demand matching characterization and the cross-category substitution elasticity. The supply and demand allocation constraint model is generated by combining the supply available quantity boundary, the minimum quantity of demand orders to be satisfied, and the fulfillment feasibility conditions. The total supply and demand within a continuous trading cycle are aggregated to generate the supply and demand status of the cycle market, and the constraint strength between demand guarantee constraints and supply utilization constraints is adjusted based on the supply and demand status of the cycle market. The adjusted supply and demand allocation constraint model is iteratively solved, and when the change magnitude of the matching target satisfies the convergence condition, the global matching result containing the supply and demand pairing relationship and the corresponding allocation amount is output.

8. The method according to claim 7, characterized in that, Also includes: Multiple supply disturbance scenarios are constructed based on the expected supply volume and fluctuation range of each candidate supplier, and the candidate matching results corresponding to each supply disturbance scenario are solved using the supply and demand allocation constraint model. Compare the demand satisfaction status, supply remaining status, and transportation space adaptability of the candidate matching results under different supply disturbance scenarios to generate a matching stability characterization; Based on the matching stability characterization, robust control parameters including supply reserve and robustness penalty parameters are generated. The supply and demand allocation constraint model is updated using the robust control parameters, and candidate matching results that meet the preset stability conditions across supply disturbance scenarios are determined as the global matching results.

9. The method according to claim 1, characterized in that, The process of identifying supply resource conflicts between the fast matching results and the global matching results, determining priority allocation relationships according to the urgency of orders, and resolving the supply and demand allocation constraint model based on remaining supply resources to generate channel supply and demand matching results includes: By associating the allocation amounts in the fast matching results and the global matching results with the supplier identifier, the cumulative allocation amount of the same supplier is compared with the available supply boundary to identify supply resource conflicts caused by duplicate occupation or over-allocation. For conflicting demand orders, compare the urgency of the orders, prioritize the allocation relationship corresponding to the time-priority orders, and deduct the available supply from the corresponding supply side to generate remaining supply resources; The candidate range, decision variable boundary, and supply-demand matching representation of the global equilibrium order are updated based on the remaining supply resources. The updated results are then written into the supply-demand allocation constraint model and solved again. The retained fast matching results and the re-solved global matching results are combined to verify the cumulative allocation of each supplier and the fulfillment status of each demand order, and the channel supply and demand matching results are generated.

10. A smart matching system for supply and demand in contract farming channels, characterized in that, include: The acquisition module is used to acquire production data from planting bases and demand data from buyers, generate supply-side production characteristics and demand-side procurement characteristics, and establish supply-demand matching relationships. The generation module is used to generate a supply and demand matching representation based on the degree of fit between the fulfillment time and the degree of fit between the transportation space of the supply and demand matching relationship, and to generate cross-category substitution elasticity based on the historical substitution procurement relationship of the acquirer; The segmentation module is used to determine the urgency of orders based on the remaining time for fulfillment of demand orders and the supply and demand status of target categories, and to segment the demand orders into time-priority orders and globally balanced orders. The filtering module is used to filter local candidate suppliers for the time-priority orders according to spatial range, fulfillment time and supply capacity, and generate fast matching results based on the degree of adaptability of transportation space; The module is used to determine the candidate range of the same category or the expansion range of related categories based on the cross-category substitution elasticity for the global equilibrium order, construct a supply and demand allocation constraint model based on the supply and demand matching characterization, adjust the constraint strength of the supply and demand allocation constraint model according to the supply and demand status of the cyclical market, and solve to obtain the global matching result. The re-decomposition module is used to identify supply resource conflicts between the fast matching results and the global matching results, determine the priority allocation relationship according to the urgency of the orders, and re-decompose the supply and demand allocation constraint model based on the remaining supply resources to generate channel supply and demand matching results.

Images