Data-driven food material supply and demand matching and peak shifting scheduling system

Abstract

The invention discloses a data-driven food material supply and demand matching and peak shifting scheduling system, and relates to the technical field of scheduling, high-risk orders are identified in advance and included in a pre-buffer area through resource permission evaluation, order precursor dependence mapping and scheduling vulnerability index calculation, and path rhythm disturbance or resource conflicts caused by the high-risk orders in formal scheduling are avoided; in the formal scheduling execution process, the system collects state information in real time and constructs a change index, the system stability is dynamically judged, buffer orders are released only when fluctuation is controllable, and a self-adaptive control mechanism of the preposed avoidance and execution stage of scheduling risks is formed. The problem that chain fluctuation amplification is difficult to deal with due to the fact that a traditional method only depends on static inventory snapshots and a round of decision is effectively solved, and the scheduling robustness of a system in a high-frequency fluctuation and complex coupling scene is improved.

Description

Technical Field

[0001] This invention relates to the field of scheduling technology, specifically to a data-driven food supply and demand matching and off-peak scheduling system. Background Technology

[0002] With the rapid development of bulk food procurement, instant delivery, and smart warehousing technologies, an increasing number of food supply chain management platforms are adopting data-driven supply-demand matching and off-peak scheduling mechanisms to improve allocation efficiency, alleviate delivery congestion, and reduce food spoilage. "Data-driven food supply-demand matching and off-peak scheduling" refers to a system that dynamically matches the material flow between the supply and demand sides using algorithmic models based on multi-source data such as order demand data, warehouse inventory data, real-time traffic information, and delivery route load. This automatically generates delivery scheduling plans and optimizes delivery times based on peak traffic periods, warehouse workload, and route density, avoiding systemic congestion caused by concentrated bursts of logistics resources. In such systems, demand forecasting models (such as time series analysis, regression, or neural networks) are typically used to predict the demand for different foods, while route planning algorithms (such as graph optimization based on traffic flow) are combined to execute off-peak delivery arrangements. The entire scheduling system aims for automation, using algorithmic decisions to optimize the scheduling of food delivery based on "when, from which warehouse, to which customer, and which route," thereby maximizing transportation efficiency, reducing vehicle congestion, and effectively compressing peak delivery load while meeting order timeliness requirements.

[0003] Most data-driven supply and demand scheduling platforms now possess relatively sophisticated algorithmic structures, enabling them to quickly process large-scale order data in typical scenarios, identify supply and demand matching relationships, and dynamically push off-peak delivery plans based on delivery timeliness requirements and traffic load information. The system typically uses a combination of "current inventory snapshot + predicted orders + route accessibility index" as a basis, completing a scheduling calculation within a fixed time period to generate delivery execution instructions and updating the status of various resources at a high-frequency pace. Some platforms have also introduced IoT-enabled cargo location identification terminals and GPS route monitoring devices, achieving real-time feedback on actual delivery status and enabling closer linkage between off-peak scheduling and execution.

[0004] However, in real-world operations, the food supply chain is not always in a static equilibrium. Especially with increased scheduling frequency, order density, and intensified resource competition, micro-fluctuations from the supply side—such as delivery delays, inventory changes, batch returns, and quantity adjustments—are occurring more frequently between scheduling nodes. Although these fluctuations are small in magnitude and often do not trigger alarm thresholds in traditional monitoring mechanisms, when they occur at the intersection of scheduling chains, they can easily amplify downstream order scheduling, causing problems such as continuous rollback of scheduling results, missed off-peak windows, and duplicate occupation of delivery routes, seriously disrupting the overall rhythm and resource stability of the system.

[0005] For example, suppose a warehouse system records 100 boxes of a certain type of vegetable in stock. The system needs to fulfill two off-peak delivery orders, one for 60 boxes and the other for 45 boxes. The system first generates a scheduling instruction for the first order and pre-allocates inventory. However, because the inventory status is not updated in time, the system incorrectly judges that there is sufficient remaining inventory when processing the second order, thus repeatedly allocating resources and causing execution failure. More seriously, the scheduling results of such erroneous instructions cause duplicate scheduling at the path level, resulting in overlapping path occupancy, vehicle resource conflicts, and delivery delays, ultimately disrupting the original off-peak rhythm and scheduling logic.

[0006] Because the current system lacks the ability to handle such "tiny perturbations" Scheduling cross The lack of early warning and modeling capabilities for the "path rollback" chain makes it difficult to proactively avoid or buffer problems at their initial stages. This technical gap not only reduces scheduling stability but also causes the system scheduling results to rely excessively on the "current static state" rather than on the controllability of the global dynamic structure. Summary of the Invention

[0007] The purpose of this invention is to solve the problems mentioned above and provide a data-driven food supply and demand matching and off-peak scheduling system.

[0008] This invention proposes a data-driven food supply and demand matching and off-peak scheduling system, the system comprising:

[0009] Allowance module: Collects information on the inventory status of ingredients in each warehouse, the number of times resources are referenced, and the scheduling confidence information, and calculates the allowance score for each resource unit based on the data;

[0010] Trailing Module: For each order to be scheduled, based on the resource availability score, a dependency graph is established between its corresponding resources, paths, and related orders, serving as the trailing graph for order fluctuations;

[0011] The segmentation module calculates the scheduling vulnerability index for each order based on the order fluctuation precursor trace map, and then divides the orders into formal scheduling orders and pre-buffered orders according to the scheduling vulnerability index.

[0012] Change Index Module: Prioritizes the execution of formally scheduled orders and collects information during the order execution process in real time to calculate the change index;

[0013] Scheduling module: Compares the change index with a preset threshold, and determines whether the pre-buffered order can be added to the subsequent formal scheduling based on the comparison result.

[0014] Optionally, the allowance module includes:

[0015] Parameter module: Collects the number of times each resource unit in the warehouse is currently referenced by orders waiting to be scheduled, recorded as the reference count; inventory status integrity parameters and scheduling confidence parameters of the resource unit;

[0016] Allowance value module: Multiply the inventory status integrity parameter by the scheduling confidence parameter as the numerator, add the reference count to the value 1 as the denominator, and divide the result as the allowance value of the resource unit;

[0017] Allowance Score Module: Normalizes the allowance values ​​of multiple resource units. By using max-min normalization, the allowance score of each resource unit is mapped to the range of 0-1. The normalization result is used as the allowance score of each resource unit.

[0018] Optionally, the tracking module includes:

[0019] Resource Dependency Edge Module: Based on the resource unit availability score, select a set of resource units with scores higher than a preset threshold as candidate scheduling resources for the target order, and establish a dependency edge between the target order and the selected resource units as a resource dependency edge;

[0020] Resource contention edge module: Performs reverse lookup on the selected resource unit to identify the resource unit currently referenced by other pending orders. If there is a shared resource unit, a resource contention connection edge is established between the two order nodes in the dependency graph as a resource contention edge.

[0021] Dependency graph module: Based on the target order's delivery destination, time requirements, and system path planning results, it generates the order's expected delivery path and connects the order node with the path node in the dependency graph to represent the path resources that the order scheduling will occupy;

[0022] Path Conflict Edge Module: Performs conflict detection on critical path nodes or main road segments in the generated delivery path, identifies whether there are other order delivery paths that overlap with it spatially or have scheduling overlap in time. If so, establishes conflict edges between path nodes in the graph as path conflict edges, traces back to the corresponding associated order nodes, and establishes indirect scheduling interference paths.

[0023] Order fluctuation precursor trace graph module: The final result is a multi-level dependency graph containing three types of nodes: order nodes, resource nodes, and path nodes, which serves as the order fluctuation precursor trace graph; the edge types in the graph include resource dependency edges, resource contention edges, and path conflict edges.

[0024] Optionally, the partitioning module includes:

[0025] Resource Dependency Impact Module: Extract the resource dependency edges corresponding to each order from the order fluctuation precursor trace map, assign different preset weights to each resource dependency edge, multiply each resource dependency edge by the allowance score of the corresponding resource unit, and use the sum of all multiplication results as the resource dependency influence degree.

[0026] Scheduling Vulnerability Index Module: Extracts the total number of resource contention edges between each order and other orders. The total number of path conflict edges for each order is used to calculate the scheduling vulnerability index for each order by adding the total number of resource contention edges to the total number of path conflict edges and the resource dependency impact.

[0027] Optionally, the partitioning module further includes:

[0028] The first partitioning module compares the scheduling vulnerability index of each order with a preset threshold. If the scheduling vulnerability index is not less than the preset threshold, the order is partitioned into a pre-buffered order.

[0029] The second partitioning module: If the scheduling vulnerability index is less than the preset threshold, the order will be partitioned into a formal scheduling order.

[0030] Optionally, the change index module includes:

[0031] The calculation module collects information in real time during the execution of formally scheduled orders, calculates the path backtracking instability index and the resource consumption change index, and adds the path backtracking instability index and the resource consumption change index to obtain the change index.

[0032] Optionally, the change index module further includes:

[0033] Sequence Module: During the execution of all formally scheduled orders, collect path segments of the actual delivery route for each formally scheduled order to form a continuous sequence of actual delivery route segments;

[0034] Path backtracking marker module: Constructs a path transition graph for each transition pair in a continuous sequence of segments. If path backtracking behavior exists, define a path backtracking flag function. , ; Indicates the first A path segment,

[0035] Disturbance strength module: for all transition pairs Calculate its path jump disturbance intensity The calculation formula is: In the formula, Indicates the spatial distance between segments. Time used for segment transitions;

[0036] Weighted jump perturbation value module: Multiply the backtracking flag by the perturbation intensity to obtain the weighted jump perturbation value;

[0037] Instability module: For each formal scheduling order, the average of all weighted jump disturbance values ​​is used as the path backtracking instability of the corresponding formal scheduling order;

[0038] Path backtracking instability index module: The average of the path backtracking instability of all formally scheduled orders is used as the path backtracking instability index.

[0039] Optionally, the change index module further includes:

[0040] Consumption Sequence Module: For all formal scheduling orders, resource consumption data of each formal scheduling order is collected in real time during the scheduling process to form the resource consumption sequence of each order;

[0041] Resource rhythm discreteness module: Calculates the resource rhythm discreteness function for the resource consumption sequence of each order. The calculation formula is as follows: ;in; For orders The first formal dispatch order in the The amount of resources consumed per unit of time;

[0042] Resource fluctuation tension spectrum module: for discrete functions After normalization, the resource fluctuation tension spectrum for each time unit is calculated. .

[0043] Optionally, the change index module further includes:

[0044] Resource Consumption Disturbance Offset Module: Calculates the resource consumption disturbance offset for each order. The calculation formula is as follows: ;

[0045] Disturbance density index module: This module sets the disturbance offset. Converted to per unit time per disturbance density exponent The calculation formula is as follows: ;

[0046] Resource Consumption Variation Index Module: Based on the disturbance density index of all formal scheduling orders. Calculate the resource consumption change index The calculation formula is: ,in, and These are the maximum and minimum values ​​of the disturbance density index among all formal scheduling orders, respectively.

[0047] Optionally, the scheduling module includes:

[0048] First scheduling module: If the change index is less than the preset threshold, the pre-buffered order can continue to be added to the subsequent formal scheduling;

[0049] Second scheduling module: If the change index is not less than the preset threshold, the pre-buffered order cannot be added to the subsequent formal scheduling and delivery is stopped.

[0050] The beneficial effects of this invention are:

[0051] This invention proposes a data-driven food supply and demand matching and off-peak scheduling system. It effectively identifies and addresses chain-reaction scheduling failures caused by factors such as micro-inventory fluctuations, resource reference conflicts, or path overlaps in the supply chain. Before issuing actual scheduling instructions, it pre-identifies potentially high-risk orders through resource availability assessment, order precursor dependency mapping, and scheduling vulnerability index calculation, and includes these orders in a pre-buffered scheduling zone to avoid disruptions to the overall system's path rhythm and inventory resources caused by the failed scheduling of highly vulnerable orders. Simultaneously, during formal scheduling, the system dynamically collects execution status and constructs a change index to assess the system's scheduling stability in real time. Only when system fluctuations are controllable are buffered orders gradually released, thus achieving proactive risk avoidance and adaptive stability control during the scheduling execution phase. This solves the problem in existing technologies where scheduling systems rely solely on static inventory snapshots and a single-round decision-making mechanism, failing to perceive the chain-reaction amplification effect of small fluctuations in a multi-order convergence structure, leading to path lock-up, missed off-peak windows, and resource allocation conflicts. This significantly improves the reliability and robustness of the data-driven scheduling system in high-frequency fluctuation and complex coupling scenarios. Attached Figure Description

[0052] The invention will now be further described with reference to the accompanying drawings.

[0053] Figure 1 A framework diagram of a data-driven food supply and demand matching and off-peak scheduling system;

[0054] Figure 2 This is a flowchart illustrating the implementation method of a data-driven food supply and demand matching and off-peak scheduling system. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all the works obtained by those skilled in the art without creative effort are not applicable. All embodiments are within the scope of protection of this invention.

[0056] This invention provides a data-driven food supply and demand matching and off-peak scheduling system. See also... Figure 1 , Figure 1 This is a framework diagram of a data-driven food supply and demand matching and off-peak scheduling system provided in an embodiment of the present invention. The system includes...

[0057] Allowance module: Collects information on the inventory status of ingredients in each warehouse, the number of times resources are referenced, and the scheduling confidence information. Calculates the allowance score for each resource unit based on the data to characterize the allocatability stability of the resource in the current scheduling cycle.

[0058] The trailing graph module: For each order to be scheduled, based on the resource availability score, it establishes a dependency graph between the corresponding resources, paths, and related orders, which serves as the trailing graph for order fluctuations; to simulate the chain reaction that the execution of this order scheduling may have on other orders or path nodes;

[0059] The segmentation module calculates the scheduling vulnerability index for each order based on the order fluctuation precursor trace map, and then divides the orders into formal scheduling orders and pre-buffered orders according to the scheduling vulnerability index.

[0060] Change Index Module: Prioritizes the execution of formally scheduled orders and collects information during the order execution process in real time to calculate the change index;

[0061] Scheduling module: Compares the change index with a preset threshold, and determines whether the pre-buffered order can be added to the subsequent formal scheduling based on the comparison result.

[0062] Based on the data-driven food supply and demand matching and off-peak scheduling system provided by this invention, the system can effectively identify and address chain scheduling failures caused by factors such as micro-inventory fluctuations, resource reference conflicts, or path overlaps in the supply chain. Before the actual scheduling instructions are issued, potential high-risk orders are identified in advance through resource availability assessment, order precursor dependency mapping, and scheduling vulnerability index calculation. These orders are then included in the pre-buffered scheduling area to avoid disruptions to the overall system path rhythm and inventory resources caused by the failed scheduling behavior of highly vulnerable orders. Simultaneously, during the formal scheduling process, the system dynamically collects the execution status and constructs a change index to evaluate the scheduling stability of the system in real time. Only when system fluctuations are controllable are buffered orders gradually released. This achieves proactive avoidance of scheduling risks and adaptive stability control during the scheduling execution phase. It solves the problem in existing technologies where scheduling systems are based only on static inventory snapshots and a single-round decision-making mechanism, which cannot perceive the chain amplification effect of small fluctuations in a multi-order convergence structure, leading to path lock-up, missed off-peak windows, and resource allocation conflicts. This significantly improves the scheduling reliability and robustness of the data-driven scheduling system in high-frequency fluctuation and complex coupling scenarios.

[0063] In one implementation, see [link to implementation details]. Figure 2 , Figure 2 A flowchart illustrating the implementation method of a data-driven food supply and demand matching and off-peak scheduling system is provided:

[0064] S1: Collect the inventory status, resource reference count, and scheduling confidence information of ingredients in each warehouse, and calculate the allowance score of each resource unit based on the data to characterize the allocatability stability of the resource in the current scheduling cycle;

[0065] S2: For each order to be scheduled, based on the resource availability score, establish a dependency graph between its corresponding resources, paths, and related orders, as a precursor trace graph of order fluctuation; to simulate the possible chain effects of the execution of this order scheduling on other orders or path nodes.

[0066] S3: Based on the order fluctuation precursor trace map, calculate the scheduling vulnerability index for each order, and divide the orders into formal scheduling orders and pre-buffered orders according to the scheduling vulnerability index;

[0067] S4: Prioritize the execution of formally scheduled orders and collect information during the order execution process in real time to calculate the change index;

[0068] S5: Compare the change index with the preset threshold, and determine whether the pre-buffered order can be added to the subsequent formal scheduling based on the comparison result.

[0069] In one embodiment, the allowance module collects the inventory status of ingredients, the number of resource references, and scheduling confidence information in each warehouse, and calculates the allowance score for each resource unit based on the data, which is used to characterize the allocatability stability of the resource in the current scheduling cycle.

[0070] The parameter module collects the number of times each resource unit in the warehouse is currently referenced by orders awaiting scheduling, and records this as the reference count. Inventory status integrity parameters of resource units Used to characterize whether the resource unit is in an available, locked, or quality check state, where normal availability is indicated by... ; when there are partial defects or freezing When unavailable ; scheduling confidence parameters of resource units This reflects the real-time reliability of the resource's status information, calculated based on its status update frequency, IoT terminal reporting latency, or abnormal fluctuation frequency, with a value range of [value missing]. ;

[0071] The allowance value module: multiplies the inventory status integrity parameter by the scheduling confidence parameter as the numerator, adds the reference count to the value of 1 as the denominator, and divides the result to obtain the allowance of the resource unit. , Among them, the molecular part The denominator represents the degree of joint availability and information credibility of the resource unit. Characterizes the degree of competitive load on resources;

[0072] Allowance Score Module: Normalizes the allowance values ​​of multiple resource units. By using max-min normalization, the allowance score of each resource unit is mapped to the range of 0-1. The normalization result is used as the allowance score of each resource unit.

[0073] It should be noted that the aforementioned resource unit refers to a specific inventory object in the warehouse that can be individually identified and independently scheduled by the system. It is typically measured in units of the smallest allocatable package, such as "a box of lettuce," "a bag of rice," or "a pack of chicken legs." Each resource unit should have a unique identifier (such as a location number in the warehousing system, an RFID tag, a QR code, or a barcode). It serves as the most basic entity for the system to perform resource allocation, status analysis, and priority ranking within the scheduling cycle. For example, a box of potatoes numbered "W2-POTATO-038" is a typical resource unit. It is located in warehouse W2, its current status is "available," and it can be used by the system for scheduling. Among the three corresponding parameters, the reference count... The order pre-allocation module of the scheduling system can provide real-time statistics, indicating how many pending orders are currently pre-referencing this resource unit; inventory status integrity parameters. This can be directly provided by the warehouse management system. Its value is automatically assigned based on whether the resource unit is in a frozen, quality inspection, out-of-stock, or normal state. A value of 1 indicates normal availability; if it is under quality inspection and locked, it is set to 0.5 or another value less than 1 according to rules; and 0 indicates unavailability. (Scheduling confidence parameter) The confidence level is calculated by the system's IoT acquisition module, primarily based on a comprehensive evaluation of factors such as the update frequency of the resource unit's status information, the data reporting delay, and the degree of fluctuation in sensor status changes over a recent period. For example, if the status information of a resource unit has not been updated for a long time, or its reporting time is significantly delayed, or it fluctuates multiple times within a unit of time, the confidence level is... It will decrease, and conversely, it will approach 1. After these three parameters are collected, they are used to calculate the value using the formula. The initial allowance value of each resource unit is calculated, and then a normalization algorithm is used to standardize the allowance values ​​of all resource units. Finally, all scores are mapped to the range of 0 to 1, which serves as the basis for the system to determine the schedulable priority of each resource unit in the current scheduling cycle.

[0074] The trailing graph module: For each order to be scheduled, based on the resource availability score, it establishes a dependency graph between the corresponding resources, paths, and related orders, which serves as the trailing graph for order fluctuations; to simulate the chain reaction that the execution of this order scheduling may have on other orders or path nodes;

[0075] The tracking module includes the following components:

[0076] Resource Dependency Edge Module: Based on the resource unit availability score, select a set of resource units with scores higher than a preset threshold as candidate scheduling resources for the target order, and establish a dependency edge between the target order and the selected resource units as a resource dependency edge to represent the direct occupation relationship of the order scheduling on the resource.

[0077] Resource contention edge module: Performs a reverse query on the selected resource unit to identify the resource unit currently referenced by other orders to be scheduled. If there is a shared resource unit, a resource contention connection edge is established between the two order nodes in the dependency graph as a resource contention edge, which is used to represent the order-level conflict that may be caused by this resource scheduling behavior.

[0078] Dependency graph module: Based on the target order's delivery destination, time requirements, and system path planning results, it generates the order's expected delivery path and connects the order node with the path node in the dependency graph to represent the path resources that the order scheduling will occupy;

[0079] Path Conflict Edge Module: Performs conflict detection on critical path nodes or main road segments in the generated delivery path, identifies whether there are other order delivery paths that overlap with it spatially or have scheduling overlap in time. If so, establishes conflict edges between path nodes in the graph as path conflict edges, traces back to the corresponding associated order nodes, and establishes indirect scheduling interference paths.

[0080] Order Fluctuation Precursor Trace Graph Module: Ultimately, a multi-level dependency graph containing three types of nodes—order nodes, resource nodes, and path nodes—is formed as the order fluctuation precursor trace graph. The edge types in the graph include resource dependency edges, resource contention edges, and path conflict edges, which are used to fully represent the chain reaction that a target order may cause to other orders or path resources after it is scheduled for execution, providing an input structure for subsequent scheduling vulnerability index calculation.

[0081] It should be noted that in the above steps, in order to accurately simulate the potential impact of a scheduled order on other orders or path nodes after actual scheduling execution, the system constructs a multi-level dependency graph called the "Order Fluctuation Precursor Trace Graph" based on the resources, paths, and potential scheduling conflicts that the order depends on. The construction process includes the following aspects: First, based on the calculated resource unit allowance score, the system selects a group of high-reliability resources with scores higher than a preset threshold as candidate scheduling resources for the target order. Resource dependency edges are established between the order node and these resource nodes in the graph structure, indicating that if the order is scheduled, it will directly occupy the resource unit. Next, the system performs a reverse indexing on these selected resource units to find other scheduled orders that also reference these resources. If a resource unit r5 is found to be referenced by both order O1 and order O3, a resource contention edge is constructed between O1 and O3 in the graph, indicating... Orders have resource conflicts at the scheduling level. Then, based on the target order's delivery destination, time window, and route planning results, the system generates its expected delivery route. For example, if path P1 includes nodes N1-N2-N5, the system establishes path occupancy edges between order nodes and path nodes in the graph, indicating that this path will be used during scheduling. Furthermore, the system performs conflict detection on key nodes in the path, determining whether node N2 of path P1 spatially overlaps or conflicts with node N2 of path P3 used by other orders O4. If this is confirmed, a path conflict edge is established for node N2 in the graph, and backtracking is performed from N2 to nodes O1 and O4 respectively, thus constructing an indirect scheduling interference path chain, forming a multi-layered influence structure. Similarly, when multiple orders have explicit or implicit relationships in resource references or path usage, the system clearly expresses these relationships using edges in a graph structure. This results in a complex directed graph structure containing three types of nodes: order nodes, resource nodes, and path nodes, as well as four types of edges: resource dependency edges, resource contention edges, path occupancy edges, and path conflict edges. This serves as the precursor trace graph for order fluctuations, used for subsequent assessment of the order's scheduling vulnerability index. For example, it determines whether the failure of scheduling O1 will affect the scheduling execution of O3 and O4 through shared resource r5 or path N2, thereby identifying in advance which orders might experience chain failure risks due to this minor scheduling fluctuation. This provides a structured basis for the system to make reasonable buffering, adjustments, or rearrangements before officially issuing scheduling instructions.

[0082] In one implementation, the module is divided as follows: based on the order fluctuation precursor trace map, a scheduling vulnerability index is calculated for each order to measure the sensitivity of the order to the overall stability of the system under the current scheduling environment, and the orders are divided into formal scheduling orders and pre-buffered orders accordingly.

[0083] In one implementation, a scheduling vulnerability index is calculated for each order based on the order fluctuation precursor trace map; the partitioning module includes:

[0084] Resource Dependency Impact Module: Extract the resource dependency edges corresponding to each order from the order fluctuation precursor trace map, assign different preset weights to each resource dependency edge, multiply each resource dependency edge by the allowance score of the corresponding resource unit, and use the sum of all multiplication results as the resource dependency influence degree.

[0085] Scheduling Vulnerability Index Module: Extracts the total number of resource contention edges between each order and other orders. The total number of path conflict edges for each order is used to calculate the scheduling vulnerability index for each order by adding the total number of resource contention edges to the total number of path conflict edges and the resource dependency impact.

[0086] It should be noted that in the above steps, the process of calculating the scheduling vulnerability index first extracts resource dependency edges, resource contention edges, and path conflict edges related to each order from the order fluctuation precursor trace graph. Specifically, resource dependency edges represent the resource occupancy relationship required during order scheduling. The system assigns different weights to these dependency edges based on the importance and availability of each resource unit. For example, an order may require a cold chain food resource, and if this resource is unavailable during scheduling, it will have a significant impact on scheduling. Therefore, the system assigns a high weight to this resource dependency edge. Next, the system multiplies the weight of each resource dependency edge by the allowance score of that resource unit to calculate the resource dependency impact of each order, that is, the degree of dependence on resources during order scheduling. Immediately afterward, the system extracts resource contention edges and path conflict edges related to the order. Resource contention edges indicate situations where different orders share the same resource, while path conflict edges indicate conflicts or overlaps between different orders on the delivery path. For example, if order A and order B both require the same storage space in a warehouse, this will trigger resource contention; if the delivery paths of order A and order C overlap, then a path conflict will occur. Next, the system counts the number of resource contention edges and path conflict edges for each order and adds them to the resource dependency impact to obtain the scheduling vulnerability index for each order. A higher scheduling vulnerability index indicates that the order is more susceptible to changes in other orders or resources under the current scheduling environment. Based on the scheduling vulnerability index value, the system classifies orders into formally scheduled orders and pre-buffered orders. Orders with lower scheduling vulnerability indices are prioritized for scheduling, while orders with higher indices are placed in a buffer state, awaiting system adjustments before scheduling. This method, by quantifying the scheduling vulnerability of each order, provides an effective basis for optimizing resource allocation, path scheduling, and preventing systemic conflicts.

[0087] It's important to note that the key advantage of calculating the scheduling vulnerability index using the above method lies in its ability to comprehensively consider the impact of multiple factors, such as resource dependence, resource competition, and path conflicts, on order scheduling stability, rather than relying solely on a single factor. Specifically, the weights and admissibility scores of resource dependence edges consider the importance and current availability of each resource unit, accurately reflecting the sensitivity of order scheduling to resource occupancy. This is especially true for critical resources (such as cold chain ingredients and limited storage space), whose fluctuations have a far greater impact on the system than other ordinary resources. By multiplying the resource dependence edge score by the resource's admissibility score, the resource's impact can be quantified, allowing the scheduling system to prioritize highly dependent resources and respond to their potential fluctuations. On the other hand, the calculation of resource competition edges and path conflict edges further integrates the interrelationships between orders, identifying conflicts between orders sharing resources and paths. This provides a collaborative consideration for multi-order scheduling, avoiding scheduling failures caused by multiple orders sharing limited resources or paths. By incorporating the number of resource-competing edges and path-conflicting edges into the calculation, the competitive relationships between orders and the conflict risks on paths can be quantified. This allows the system to more accurately identify high-conflict, high-risk orders, prioritizing their processing or postponing their scheduling. Therefore, calculating the scheduling vulnerability index in this way not only comprehensively assesses the stability of orders under the current scheduling environment but also effectively avoids the limitations of relying on single factors, such as simply focusing on insufficient inventory or path congestion. This improves the overall system's scheduling robustness and decision-making accuracy. This multi-factor, comprehensive approach can identify potential scheduling risks in advance, effectively prevent systemic failures and scheduling inefficiencies, improve resource utilization and scheduling efficiency, and ultimately enhance the reliability of the food supply and demand matching and off-peak scheduling system.

[0088] In one embodiment, and based thereon, orders are divided into formally scheduled orders and pre-buffered orders, the division module further includes:

[0089] The first partitioning module compares the scheduling vulnerability index of each order with a preset threshold. If the scheduling vulnerability index is not less than the preset threshold, the order is partitioned into a pre-buffered order.

[0090] The second partitioning module: If the scheduling vulnerability index is less than the preset threshold, the order will be partitioned into a formal scheduling order.

[0091] It's important to note that comparing the scheduling vulnerability index of each order with a preset threshold essentially quantifies the risk of scheduling failure to determine the scheduling priority and handling method for each order. If the scheduling vulnerability index of an order is not less than the preset threshold, the order is classified as a pre-buffered order. This means the system judges that the order has a high risk of scheduling failure, potentially affecting the normal scheduling of other orders or paths due to resource conflicts, path contention, or other factors. Therefore, it needs to be delayed or adjusted during the scheduling process, and scheduling will only proceed after the system determines that the situation is stable. For example, a catering supplier's order requires the delivery of a batch of ingredients within a specific time period. This batch of ingredients requires warehouse resources shared with several other high-priority orders, and the warehouse resource has a low availability score, indicating poor resource availability. In this case, even if the order itself is urgent, the system will postpone its scheduling based on its high FVI value to avoid large-scale delivery failures caused by resource contention. If the scheduling vulnerability index is less than the preset threshold, the order is classified as a formal scheduling order. This means the system considers the order to have a small impact on the scheduling system and a low scheduling risk, and it can be processed and executed according to the normal process. The advantage of this approach is that the system can prioritize scheduling low-risk orders based on risk assessment, while pre-buffering high-risk orders prevents a chain reaction caused by the failure of multiple highly vulnerable orders, thus ensuring the stability and efficient operation of the overall system. In this way, the system effectively allocates resources, ensuring that critical orders are executed first in an environment with less resource contention and stable scheduling, while high-risk orders are deferred through adaptive scheduling. This optimizes the overall scheduling process, reduces the overall scheduling failure rate caused by minor fluctuations or contention, and thereby improves the system's reliability and robustness under high load conditions.

[0092] In one embodiment, the change index module: prioritizes the execution of formally scheduled orders and collects information during the order execution process in real time to calculate the change index; the change index module includes:

[0093] The calculation module collects information in real time during the execution of formally scheduled orders, calculates the path backtracking instability index and the resource consumption change index, and adds the path backtracking instability index and the resource consumption change index to obtain the change index.

[0094] In one implementation, the change index module further includes: the calculation steps for the path backtracking instability index are as follows:

[0095] Sequence Module: During the execution of all formally scheduled orders, collect path segments of the actual delivery route for each formally scheduled order to form a continuous sequence of actual delivery route segments; In the formula, This indicates the number of the formal dispatch order. Indicates the first Each path segment (e.g., from warehouse to sorting point, from sorting point to distribution station, etc.). Indicates the total number of path segments;

[0096] Path backtracking marker module: Constructs a path transition graph for each transition pair in a continuous sequence of segments. If path backtracking behavior exists, define a path backtracking flag function. , If from arrive If spatial rollback behavior exists (i.e., rollback / return / path cancellation occurs, i.e., the jump "violates the normal direction"), it is recorded as 1;

[0097] Disturbance strength module: for all transition pairs Calculate its path jump disturbance intensity The calculation formula is: In the formula, Indicates the spatial distance between segments. Time used for segment transitions;

[0098] Weighted jump perturbation value module: backtracking flag With disturbance intensity Multiply to obtain the weighted jump perturbation value. The calculation formula is: ;

[0099] Instability Module: For each formal scheduling order, the average of all weighted jump disturbance values ​​is used as the path backtracking instability of the corresponding formal scheduling order; the calculation formula is: In the formula, Indicates the first The instability of path backtracking for a formal dispatch order;

[0100] Path backtracking instability index module: The average of the path backtracking instability of all formally scheduled orders is used as the path backtracking instability index.

[0101] It should be noted that in the calculation of the path backtracking instability index, the data involved in each step is mainly obtained through the real-time order execution trajectory acquisition module in the scheduling system. Specifically, this includes the following methods: During the execution of each formally scheduled order, its delivery path is recorded in real time by the system's built-in task tracking module based on data sources such as vehicle positioning terminals (e.g., GPS), logistics node check-in records (e.g., QR code check-in), and execution feedback logs from the path planning module. The system constructs a sequence of actual delivery path segments for each order in timestamp order. Each segment It includes basic information such as the coordinates of the start and end points, start and end times, and delivery task numbers; when constructing the path transition map, the system automatically identifies the continuous transition relationships between path segments and calculates each pair of transition paths using the spatial coordinates of the segments (the actual distance can be calculated by the map engine). Simultaneously extract from the task execution log This refers to the time spent transitioning between two segments; the determination of path backtracking behavior is achieved by comparing the direction vectors of adjacent path segments to determine if there is "reversal" or "path overlap and reverse direction" behavior. Backtracking markers are obtained by comparing the angle change analysis of geographic coordinates with the expected travel direction of the scheduling system. Ultimately, all jump disturbance values ​​and backtracking behavior labels are automatically generated by the system in the background and recorded in the order execution log as input data for subsequent calculation of path backtracking instability. Therefore, the complete indicator data collection and calculation process can be achieved without manual intervention.

[0102] It should be noted that the Path Backtracking Instability Index is an indicator used to measure the overall intensity of "abnormal path jump behavior" in the actual execution of all formally scheduled orders in the current scheduling system. Its core is to identify whether abnormal behaviors such as "path backtracking, turning back, cancellation, and retracing" frequently occur in the delivery path, deviating from the established scheduling plan. It is combined with the disturbance intensity of each jump for comprehensive evaluation. The larger the index, the more severe the "delivery disturbance phenomenon" at the path execution level, and the worse the stability. In data-driven food supply and demand matching and off-peak scheduling systems, such abnormal path behavior often reflects two problems: first, the irrationality of delivery resource scheduling, such as multiple orders competing for the same path or resource scheduling mismatch leading to frequent interruptions and adjustments; second, sudden changes in external execution conditions, such as traffic control, station congestion, and node failure. These factors can directly cause the delivery path to deviate from the original plan during execution, resulting in backtracking or detour behavior, disrupting the continuity and coherence of the overall system scheduling. For example, if an order was originally supposed to be executed according to "warehouse..." Sorting point The "delivery station" process is sequential, but in practice, situations arise where orders "return from the sorting point to the warehouse" and then restart, creating a clear path backtracking behavior. This not only delays the order's execution time but may also consume critical path resources that should be allocated to other orders, leading to a chain reaction of scheduling conflicts. Therefore, when the path backtracking instability index rises, it indicates that the current officially scheduled order has revealed strong systemic instability signals during execution. If pre-buffered orders are rashly allowed to enter the main scheduling process at this time, it may further exacerbate resource conflicts, path congestion, and node degradation, causing the scheduling system to fall into a vicious cycle. Therefore, it is necessary to dynamically sense the system's state through this index. If the index is higher than the threshold, the buffered state of pre-buffered orders should be kept unchanged until the system spontaneously or through regulation recovers to a relatively stable state, thereby ensuring that the entire supply and demand matching system remains controllable and recoverable under high-pressure loads.

[0103] The greatest advantage of calculating the path backtracking instability index using the above method is that it can deeply characterize the structural deviations of formally scheduled orders in the actual delivery process from the execution behavior level, rather than relying solely on superficial features such as traditional path errors, travel time fluctuations, or the number of abnormal statistics. This method first captures "path structural disturbances" that are not easily detected by time or distance errors by constructing a path jump map and explicitly identifying directional deviations (i.e., backtracking behavior), avoiding the hidden problems that are easily missed by relying solely on mileage or time differences. Second, it uses the product of a backtracking label function and the disturbance intensity to form a "weighted disturbance value" with direction awareness, and constructs the disturbance intensity by the ratio of spatial distance between consecutive jumps to the jump time, effectively avoiding the dimensional imbalance problem caused by independent calculation of distance or time in traditional methods, making the index more physically interpretable. Furthermore, by extracting the weighted disturbance mean within each order as the path backtracking index... By tracing instability and further averaging all orders at the system level to form a comprehensive indicator, this method not only preserves the impact of individual-level anomalies but also achieves steady-state awareness at the system level, ensuring that local extreme behaviors are not masked by the overall trend. More importantly, this method completely avoids weighted summation and the introduction of unknown factors. The calculation process is traceable, the parameters are clear, and it has engineering application value. At the same time, by defining a clear relationship between path segment pairs and time and distance combinations, it can be calculated in real time in the actual system based on trajectory data, path planning records, and real-time feedback information from the scheduling platform, facilitating embedded deployment and dynamic monitoring. Compared with conventional stability judgments based on error statistics or path similarity, this method has stronger structural identification capabilities, interference differentiation capabilities, and behavioral anomaly explicitness capabilities. It truly extracts the inherent volatility of the scheduling system state from the path execution mechanism, making it a highly innovative, stable, and interpretable path anomaly assessment method.

[0104] In one implementation, the change index module further includes: the calculation steps for the resource consumption change index are as follows:

[0105] Resource Consumption Sequence Module: For all formal scheduling orders, resource consumption data for each formal scheduling order is collected in real time during the scheduling process, forming a resource consumption sequence for each order. ,in For the first The total scheduling time step for a formal scheduling order. For orders The first formal dispatch order in the The quantity of resources consumed per unit of time (can be a physical unit, such as pallets, man-hours, number of vehicles, etc.);

[0106] Resource rhythm discreteness module: resource consumption sequence for each order. Computational resource rhythm discrete function This represents the degree of jump in resource consumption between adjacent time units, and its calculation formula is: ;in, For the first The fluctuation in resource consumption at any given moment;

[0107] Resource fluctuation tension spectrum module: for discrete functions After normalization, the resource fluctuation tension spectrum for each time unit is calculated. The calculation formula is: ;

[0108] Resource Consumption Disturbance Offset Module: Calculates the resource consumption disturbance offset for each order. The calculation formula is as follows: ;in, Calculate the perturbation jump behavior across time periods to express the long-term shift in the rhythm of resource consumption;

[0109] Disturbance density index module: This module sets the disturbance offset. Converted to per unit time per disturbance density exponent The calculation formula is as follows: ,in, Used for normalization to avoid division by zero errors and to keep the result in the [0,1] range;

[0110] Resource Consumption Variation Index Module: Based on the disturbance density index of all formal scheduling orders. Calculate the resource consumption change index The calculation formula is: ,in, and These are the maximum and minimum values ​​of the disturbance density index in all formal scheduling orders, used to measure the extreme degree of resource consumption fluctuation in the system, ensuring that the final output is in the range [0,1].

[0111] It should be noted that the data involved in the calculation of the aforementioned resource consumption variation index is acquired through the real-time monitoring and data acquisition module of the scheduling system. First, the system collects resource consumption data for each formally scheduled order in real time from sensors and information systems at various warehouses, distribution points, and logistics nodes. This data includes resource usage within each time unit, such as the number of pallets consumed, man-hours, and vehicles. The data is fed back in real time through IoT devices (such as RFID tags, barcode scanners, and GPS positioning) to ensure that the resource consumption of each order is accurately recorded within each time unit. Next, the system constructs a resource consumption sequence based on this data and calculates the resource consumption fluctuation at each moment. By comparing the consumption differences between adjacent moments, a discrete function is obtained, and then normalization is performed to obtain a resource fluctuation tension spectrum, which reflects the intensity of resource consumption fluctuations. Subsequently, the system calculates the disturbance offset, measuring the long-term changes in the resource consumption rhythm to identify unstable resource consumption behaviors. Finally, by statistically analyzing the disturbance density index of all formal scheduling orders, the resource consumption variation index of the entire system is calculated. These data and calculation processes rely on the real-time resource consumption monitoring and data acquisition module within the system to ensure that all calculations are based on accurate and real-time operational data, so as to dynamically adjust the scheduling strategy.

[0112] It should be noted that the resource consumption variation index is used to measure the degree of fluctuation in resource consumption of formally scheduled orders during the scheduling process. It analyzes the fluctuation and instability of resource consumption across different time units by tracking the changes in resource usage for each order in real time during the scheduling process, reflecting... Predictability and stability of resource scheduling. A higher Resource Consumption Variation Index (RCFI) indicates more drastic fluctuations in resource consumption within the system, suggesting significant volatility or inconsistency during scheduling. This instability often leads to errors in scheduling decisions or path backtracking. For example, if the resource consumption of an order changes drastically from one time period to the next, it may indicate problems in resource allocation, such as over-reliance on certain resources, uneven resource distribution, or scheduling conflicts. In this case, resource management may be disrupted, leading to low scheduling efficiency, path backtracking, and unnecessary resource waste. A higher RCFI indicates greater instability in the scheduling system, potentially impacting path and resource utilization efficiency, making the system more vulnerable. A higher RCFI indicates greater system volatility risk and poorer predictability and control during scheduling. Continuing to process pre-buffered orders in this situation could lead to resource allocation conflicts, path congestion, order delays, and ultimately, system-wide scheduling collapse or a significant drop in efficiency. Therefore, to prevent system instability from affecting subsequent scheduling tasks, it is necessary to pause or delay the scheduling of pre-buffered orders when the volatility index is high, ensuring the stability and efficient operation of the system. For example, in a specific scenario, if a scheduling system forcibly schedules multiple orders during a period of high volatility, it may lead to excessive overlap of multiple delivery routes, extremely uneven resource consumption, and thus serious route backtracking. Ultimately, this will not only affect the timely completion of the current order but also the smooth execution of subsequent orders.

[0113] It's important to note that the advantage of calculating the resource consumption variation index using the above method is its ability to dynamically and comprehensively reflect the volatility of resource consumption, rather than relying solely on a single average consumption amount or simple time-series errors. Traditional methods may only focus on the magnitude of resource consumption changes or deviations in consumption volume, neglecting the rhythm and volatility of consumption behavior. This approach may fail to effectively capture high-frequency fluctuations or changes in consumption patterns within a short period. The method described above, by calculating the discrete function, volatility tension spectrum, and disturbance offset, delves into the resource consumption changes of each order over time, identifying small short-term fluctuations and unstable trends—potential early signals of system instability. Furthermore, this method incorporates cross-time-period disturbance jump behavior, effectively assessing the existence of periodic fluctuations or unstable trends in resource consumption by measuring the long-term offset of the resource consumption rhythm. This helps the system identify potential scheduling risks in a timely manner. Unlike traditional methods, the above method considers not only changes in absolute consumption volume but also the rhythm and sudden fluctuations of these changes, thus more closely reflecting real-world scheduling execution and more accurately reflecting the system's resource consumption instability. A high resource consumption variation index indicates significant fluctuations in the system, which may lead to scheduling conflicts, path backtracking, and other issues. Therefore, the system's scheduling strategy can be adjusted promptly to avoid executing more orders under high-risk conditions, thereby preventing excessive resource contention or scheduling failures. In summary, using this method to calculate the resource consumption variation index provides the scheduling system with more accurate dynamic awareness, optimizes resource allocation, and improves system robustness and scheduling efficiency.

[0114] In one embodiment, the scheduling module compares the change index with a preset threshold and determines whether a pre-buffered order can be added to the subsequent formal scheduling based on the comparison result; the scheduling module includes:

[0115] First scheduling module: If the change index is less than the preset threshold, the pre-buffered order can continue to be added to the subsequent formal scheduling;

[0116] Second scheduling module: If the change index is not less than the preset threshold, the pre-buffered order cannot be added to the subsequent formal scheduling and delivery is stopped.

[0117] It's important to note that the change index is used to measure system stability and potential risks during the current scheduling process. A low change index indicates that the system is operating smoothly during the current scheduling process, with minimal fluctuations in resource consumption, infrequent path backtracking, and good overall stability. In this case, pre-buffered orders can be allowed to proceed and added to subsequent formal scheduling. For example, suppose a food delivery system experiences a large number of orders and congested delivery routes during peak hours. However, by calculating the change index, the system determines that the current delivery routes and resource consumption are stable, without significant fluctuations or anomalies, so new orders can continue to be scheduled. Conversely, a high change index indicates increased system instability during the current scheduling process, such as frequent path backtracking and drastic fluctuations in resource consumption. This could lead to system overload or scheduling failures. In this case, the system will decide to stop adding new pre-buffered orders to avoid increasing the scheduling burden or causing path conflicts, excessive resource usage, and other problems, thereby ensuring the smooth completion of existing orders and maintaining system stability. In summary, by comparing the change index with a preset threshold, the system's capacity and stability can be assessed in real time, ensuring efficient and orderly scheduling and preventing the system from executing more orders in an unstable state, thus ensuring timely completion of delivery tasks. If the change index is less than or equal to the threshold, the system is considered to be in a stable state, and pre-buffered orders are released for formal scheduling execution; if the change index is greater than the threshold, the pre-buffered order status is maintained until the system meets the release conditions again.

[0118] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the scope of the claims.

Claims

1. A data-driven food material supply-demand matching and peak-shifting scheduling system, characterized in that, The system comprises Promise degree module: collect the inventory state, resource reference times and scheduling confidence information of each resource unit in the warehouse, and calculate the promise degree score of each resource unit according to the data; Drive trace module: for each to-be-scheduled order, a dependency graph is established between the corresponding resource, path and associated order according to the resource promise degree score, which is used as the order fluctuation precursor trace graph; Division module: based on the order fluctuation precursor trace graph, the scheduling vulnerability index of each order is calculated, and the orders are divided into formal scheduling orders and pre-buffering orders according to the scheduling vulnerability index; Change index module: the formal scheduling orders are executed preferentially, and the change index is calculated in real time by collecting information during the execution of the orders; Scheduling module: compare the change index with the preset threshold value, and judge whether the pre-buffering order can be added to the subsequent formal scheduling according to the comparison result.

2. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that The promise degree module comprises: Parameter module: for each resource unit in the warehouse, the number of to-be-scheduled orders that reference the resource unit is collected, which is denoted as reference times; the inventory state integrity parameter of the resource unit and the scheduling confidence parameter of the resource unit are collected; Promise degree value module: the inventory state integrity parameter is multiplied by the scheduling confidence parameter as the numerator, the value 1 is added to the reference times as the denominator, and the result of the division is taken as the promise degree value of the resource unit; Promise degree score module: the promise degree values of multiple resource units are normalized, the promise degree scores of the resource units are mapped to the interval of 0-1 by using the max-min normalization method, and the normalized result is taken as the promise degree score of each resource unit.

3. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that The drive trace module comprises: Resource dependency edge module: based on the resource unit promise degree score, a group of resource units with scores higher than a preset threshold value are selected as candidate scheduling resources of the target order, and a dependency edge is established between the target order and the selected resource units, which is used as a resource dependency edge; Resource competition edge module: the selected resource units are reversely queried to identify the resource units that are currently referenced by other to-be-scheduled orders simultaneously, if there is a shared resource unit, a resource competition connection edge between the two order nodes is established in the dependency relationship graph, which is used as a resource competition edge; Dependency relationship graph module: based on the delivery destination, time limit and system path planning result of the target order, a predicted delivery path of the order is generated, and the order node and the path node are connected in the dependency relationship graph to represent the path resources occupied by the order scheduling; Path conflict edge module: the key path nodes or main road sections in the generated delivery path are detected for conflict, whether the delivery paths of other orders overlap with it in space or scheduling overlaps in time is identified, if there is, a conflict edge between the path nodes is established in the graph, which is used as a path conflict edge, and the corresponding associated order nodes are traced back to establish an indirect scheduling interference path; Order fluctuation precursor trace graph module: a multi-level dependency relationship graph containing order nodes, resource nodes and path nodes is finally formed, which is used as the order fluctuation precursor trace graph; the types of edges in the graph include resource dependency edges, resource competition edges and path conflict edges.

4. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that The division module comprises: Resource dependency influence degree module: The resource dependency edges corresponding to each order are extracted in the order fluctuation precursor track diagram, different preset weights are assigned to each resource dependency edge, each resource dependency edge is multiplied by the promised degree score of the corresponding resource unit, and the sum of all multiplication results is taken as the resource dependency influence degree; The scheduling vulnerability index module extracts the total number of resource competition edges corresponding to each order and other orders The total number of path conflict edges of each order, the total number of resource competition edges, the total number of path conflict edges, and the resource dependence influence degree are added to obtain the scheduling vulnerability index of each order.

5. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that The division module further comprises: The first division module divides each order into a pre-buffer order if the scheduling vulnerability index of the order is not less than a preset threshold value. The second division module divides each order into a formal scheduling order if the scheduling vulnerability index of the order is less than the preset threshold value.

6. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that The change index module comprises: The calculation module collects information in the execution process of the formal scheduling order in real time, calculates a path backtracking instability index and a resource consumption change index, adds the path backtracking instability index and the resource consumption change index, and obtains the change index.

7. The data-driven food material supply-demand matching and peak shifting scheduling system according to claim 6. characterized in that The change index module further comprises: The sequence module collects path segments of an actual delivery path of each formal scheduling order in the execution process of all the formal scheduling orders, and forms a continuous segment sequence of the actual delivery path. path backtrack marker module: construct path jump profile, for each jump pair in the sequence of consecutive segments define path backtrack marker function if path backtrack behavior exists , ; denotes the th path segment, Disturbance intensity module: for all hop pairs , calculate its path hop disturbance intensity , the formula is: ; in which, denotes the spatial distance between fragments, is the time used for fragment hopping; The weighted jump disturbance value module multiplies the backtracking mark and the disturbance intensity to obtain a weighted jump disturbance value. The instability module takes an average of all the weighted jump disturbance values as a path backtracking instability of the corresponding formal scheduling order. The path backtracking instability index module takes an average of the path backtracking instabilities of all the formal scheduling orders as a path backtracking instability index.

8. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 6. characterized in that The change index module further comprises: The consumption sequence module collects resource consumption data of each formal scheduling order in the scheduling process in real time, and forms a resource consumption sequence of each order. Resource pace variability module: a resource pace variability function is calculated for each order's resource consumption sequence with the formula: ; wherein is the order number of resources consumed by the formal schedule order in the time unit; Resource fluctuation tension spectrum module: to the discrete function Normalization processing is performed, and resource fluctuation tension spectrum of each time unit is calculated .

9. The data-driven food material supply-demand matching and peak shifting scheduling system according to claim 8. characterized in that The change index module further comprises: resource consumption disturbance offset module: calculates a resource consumption disturbance offset for each order The formula is: ; Disturbance density index module: converts the disturbance offset into a disturbance density index per unit time with the formula: ; Resource consumption variation index module: the disturbance density index of all official dispatch orders calculating the resource consumption variation index , the formula of which is: wherein, and are the maximum and minimum of the disturbance density index of all official dispatch orders, respectively.

10. The data-driven food material supply-demand matching and peak-shaving scheduling system according to claim 1. characterized in that, The scheduling module comprises: The first scheduling module allows the pre-buffer order to continue to join subsequent formal scheduling if the change index is less than a preset threshold value. The second scheduling module does not allow the pre-buffer order to join subsequent formal scheduling, and stops delivery if the change index is not less than the preset threshold value.

Images