Supply chain management method and device, computer equipment, computer readable storage medium and computer program product

Abstract

The application relates to a logistics supply chain management method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: in response to a logistics resource occupation request, acquiring real-time running data of logistics resource objects provided by a plurality of resource providers respectively, performing logistics resource allocation planning, and obtaining an initial logistics resource scheduling scheme corresponding to the logistics resource occupation request; acquiring feedback information returned by each resource provider for a logistics resource scheduling sub-scheme; adjusting the initial logistics resource scheduling scheme according to the feedback information returned by all the resource providers, and obtaining a target logistics resource scheduling scheme corresponding to the logistics resource occupation request; and performing scheduling control on the logistics resource objects according to the target logistics resource scheduling scheme. The method can improve the efficiency, feasibility and resource utilization rate of logistics resource scheduling.

Description

Technical Field

[0001] This application relates to computer data processing, and in particular to a logistics supply chain management method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] With the deep penetration of IoT and AI technologies into the supply chain, logistics resource scheduling methods are gradually evolving from traditional manual telephone coordination and email communication to system-assisted decision-making. Among these technologies, large-scale supply chain management platforms generally adopt a "centralized planning, instruction-based issuance" operating model. Operators aggregate resource inventory and status data from all participants through standardized interfaces, input this data into a centralized optimization engine, and after mathematical programming solutions, output a theoretically lowest-cost or most time-efficient solution.

[0003] In this model, the operator is regarded as a central decision-maker with complete information, while each resource provider only plays the role of an executor of instructions. This results in problems with poor efficiency, accuracy, and utilization of resources. Summary of the Invention

[0004] Therefore, it is necessary to provide a logistics supply chain management method, apparatus, computer equipment, computer-readable storage medium, and computer program product that can improve the efficiency of logistics resource scheduling and resource utilization, in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a logistics supply chain management method, including:

[0006] In response to requests for logistics resource utilization, real-time operational data of logistics resource objects provided by multiple resource providers are obtained.

[0007] Based on the real-time operating data, a logistics resource allocation plan is performed for the logistics resource occupation request to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0008] Send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers respectively;

[0009] Obtain feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints;

[0010] Based on the feedback information returned by all resource providers, the initial logistics resource scheduling scheme is adjusted to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0011] The logistics resource objects are scheduled and controlled according to the target logistics resource scheduling scheme.

[0012] Secondly, this application also provides a logistics supply chain management device, comprising:

[0013] The response module is used to respond to logistics resource occupancy requests and obtain real-time operating data of logistics resource objects provided by multiple resource providers.

[0014] The planning module is used to plan the allocation of logistics resources based on the real-time operating data in response to the logistics resource occupancy request, and to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupancy request.

[0015] The sending module is used to send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers respectively;

[0016] The acquisition module is used to acquire feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints;

[0017] The adjustment module is used to adjust the initial logistics resource scheduling scheme based on the feedback information returned by all resource providers, so as to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0018] The scheduling module is used to schedule and control the logistics resource objects according to the target logistics resource scheduling scheme.

[0019] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0020] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0021] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps included in any of the above-described method embodiments.

[0022] The aforementioned logistics supply chain management methods, devices, computer equipment, computer-readable storage media, and computer program products overcome the technical bottleneck of undetectable private constraints by enabling resource providers to perform feasibility assessments and cost calculations locally. Resource providers complete constraint verification and cost calculations locally, only returning feasibility status and pricing information. This ensures the physical executability of the solution without disclosing private data. Through a distributed decision-making paradigm of "offer-feedback-adjustment," which differs from centralized command systems, the feasibility and resource utilization of logistics resource scheduling solutions can be effectively improved. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a diagram illustrating the application environment of a logistics supply chain management method in one embodiment.

[0025] Figure 2 This is a flowchart illustrating a logistics supply chain management method in one embodiment;

[0026] Figure 3 This is a structural block diagram of a logistics supply chain management device in one embodiment;

[0027] Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0029] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0030] The logistics supply chain management method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located on the cloud or other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, drones, low-altitude aircraft, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0031] In one exemplary embodiment, such as Figure 2 As shown, a logistics supply chain management method is provided, which is applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0032] Step 202: In response to the logistics resource occupancy request, obtain real-time operating data of logistics resource objects provided by multiple resource providers.

[0033] The logistics resource occupancy request can be generated by an upstream system (such as an order management system, OMS) when a specific logistics task needs to be executed, and is used to request the occupancy of specific logistics resources. Specifically, the logistics resource occupancy request can be a structured data object, and the parameter information it carries can include at least one of the following: request ID (unique identifier), task description (such as "transport 100 items from warehouse A to distribution station in city B"), constraints (such as the latest delivery time "before 18:00 today", temperature control requirement "2-8°C"), and logistics task priority.

[0034] Typical triggering scenarios for logistics resource occupancy requests can include: order fulfillment scenarios (requiring inventory allocation after an e-commerce order is generated), inventory replenishment scenarios (requiring goods to be transferred from regional warehouses when warehouse inventory falls below a safety threshold), transportation capacity scheduling scenarios (requiring the use of special vehicles for sudden large-item orders), and anomaly handling scenarios (requiring emergency replacement of transportation capacity due to vehicle malfunctions of the original carrier). It is understandable that in these scenarios, the instantaneous availability of logistics resources can be highly dynamic; for example, a vehicle that appeared idle 5 minutes ago may now be occupied by other tasks. If these changes cannot be perceived in real time, the generated scheduling plan may be invalid by the time it is issued.

[0035] Logistics resource objects refer to specific physical entities or virtual resources that can be scheduled and controlled. In the embodiments of this application, they may include at least one of the following: Transportation vehicles: such as trucks and containers, identified by a vehicle ID and associated with IoT devices such as GPS and OBD. Storage space: such as specific shelves or storage areas in a warehouse, identified by a storage location ID and associated with RFID and temperature / humidity sensors. Sorting equipment: such as automated sorting lines, identified by an equipment ID and associated with a PLC controller. Loading and unloading equipment: such as forklifts and lifting platforms, identified by an equipment ID.

[0036] A resource provider refers to an independent entity or its agent system that actually owns and operates logistics resources (i.e., logistics resource objects) within the supply chain network. Examples include warehousing service providers responsible for regional warehouses and transportation companies that own fleets of vehicles. Each resource provider is registered with a unique identifier in the system.

[0037] Real-time operational data refers to technical parameters that reflect the current status of logistics resources, collected in real time through IoT devices, device controller APIs, or resource provider system interfaces. These parameters provide a physical basis for determining the availability of logistics resources. For example, for transport vehicles, real-time operational data may include GPS coordinates, speed, vehicle load, remaining fuel / battery charge, and OBD fault codes. For warehouse space, real-time operational data may include inventory quantity, storage location occupancy status (idle / occupied), and current ambient temperature and humidity. For sorting / loading equipment, real-time operational data may include equipment on / off status, current task queue, and estimated idle time.

[0038] Specifically, in this embodiment, the acquisition of real-time operational data can be performed by a cloud-based collaborative decision-making server. For example, the cloud server can proactively pull or receive the aforementioned real-time operational data pushed by various resource provider systems or their associated edge computing nodes through predefined, standardized data interfaces (such as REST APIs or MQTT topic subscriptions). Edge computing nodes can be deployed locally on the resource objects and are responsible for cleaning (filtering outliers), filtering (smoothing jitter), and formatting the raw sensor data to generate a unified and clean real-time operational data stream before reporting it, significantly reducing the data processing pressure and network latency in the cloud.

[0039] Considering that real-time operational data contains a large number of key status parameters that directly determine resource availability, such as "equipment fault codes," "warehouse occupancy flags," and "vehicle offline status," this embodiment differs from related technologies (such as Oracle SCM Cloud's API data integration) which rely entirely on a central server for static, one-way, and near real-time data aggregation and resource allocation. In this embodiment, the cloud server can perform rapid feasibility prediction based on these status parameters before generating the initial logistics resource scheduling plan. Obviously unusable resources can be directly removed from the candidate set, thereby reducing the search space for subsequent path planning and resource matching, reducing the cloud computing load, and improving the overall response speed.

[0040] Step 204: Based on the real-time operating data, perform logistics resource allocation planning for the logistics resource occupation request to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0041] The cloud server receives real-time operational data from all relevant resources and matches it against the constraints of the logistics resource occupancy requests. Specifically, the cloud server can run a pre-defined resource matching and path planning algorithm (e.g., a constraint-based graph search algorithm). This algorithm models the supply chain network as a graph, where nodes are logistics facilities (warehouses, distribution stations) and edges are transportation paths. The algorithm inputs are real-time operational data (reflecting the current state of nodes and edges) and task requests, and the output is one or more physically feasible preliminary solutions.

[0042] In this embodiment, the cloud server leverages its global data advantage to view the observable real-time status (location, inventory, availability) of all resource providers, thereby constructing a macroscopically reasonable and logically consistent solution framework. It is understandable that without the cloud server generating an initial solution, allowing resource providers to freely offer and propose solutions would lead to combinatorial explosion and negotiation divergence. The initial logistics resource scheduling solution provides a common and reasonable anchor point for multi-agent negotiation, compressing the infinite-dimensional negotiation space into a finite-dimensional solution adjustment space, effectively reducing the complexity and number of rounds of negotiation.

[0043] For example, the cloud server maps the real-time operational data it acquires to a pre-established supply chain network topology. Specifically, facilities such as warehouses, distribution centers, transit stations, and ports are modeled as graph nodes, and transportation routes such as highways, railways, sea freight, and air freight are modeled as graph edges. Each node and edge is attached with a dynamic attribute vector extracted from the real-time operational data. For example, node attributes include: [current inventory, available storage capacity, outbound capacity, inbound capacity, equipment availability]; edge attributes include: [estimated travel time, real-time congestion index, road condition level, carbon emission coefficient]. This results in a weighted, directed, and dynamically updated hypergraph model.

[0044] The natural language or semi-structured task description in the logistics resource occupancy request is then converted into a formal constraint satisfaction problem. For example, "Transport 100 items from warehouse A to the distribution station in city B, requiring delivery before 18:00 today, temperature controlled between 2-8°C" is parsed as follows: Source constraint: Outbound node = warehouse A, SKU = XX, quantity = 100; Time constraint: Arrival time ≤ 18:00 today; Quality constraint: Temperature throughout the journey ∈ [2,8]°C; Route constraint: Allowed transportation methods (e.g., air freight is prohibited); Cost objective: Minimize the overall cost (or meet the user's preset cost ceiling) while satisfying the above constraints.

[0045] Based on the aforementioned dynamic graph model and constraints, heuristic graph search algorithms (such as A* variants, bidirectional search, and hierarchical graph shrinking) or constraint programming solvers can be used to search for all physically and technically feasible resource combinations and path solutions in a vast combinatorial space. The "feasibility" criterion here is strict technical feasibility, not commercial feasibility. For example, the algorithm will determine: Is the current inventory of warehouse A ≥ 100 items? Does a transportation route exist from warehouse A to city B? Does this route support temperature-controlled transportation? Is vehicle Y idle and temperature-controlled? The searched feasible solutions are then sorted according to a preset initial sorting rule (such as prioritizing the solution with the lowest estimated cost, highest historical fulfillment rate, or lowest carbon emissions) to generate at least one initial logistics resource scheduling scheme.

[0046] Step 206: Send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers.

[0047] The process involves using a cloud server to segment the initial global logistics resource scheduling plan according to the resource providers for each sub-task within the plan, generating a "logistics resource scheduling sub-plan" for each resource provider. A logistics resource scheduling sub-plan is a task offer sent to a specific resource provider, detailing the portion of the task they are advised to undertake. For example, a sub-plan sent to carrier X might include: "Please drive vehicle Y from location P1 to P2 at time T." The cloud server can then send the sub-plans to the service endpoints connected to each resource provider via secure network communication protocols (such as HTTPS and WSS).

[0048] In this embodiment, dividing the global initial logistics resource scheduling plan into sub-plans and distributing them is not a simple communication action, but a distributed offloading of decision-making power and computational tasks. Specifically, the global initial logistics resource scheduling plan contains task assignment information for all participants. If the complete plan were sent directly to each resource provider, it could lead to the leakage of commercially sensitive information (e.g., carrier X could see which tasks and prices carrier Y was assigned). By dividing and distributing the plan, each resource provider can only know the task offers relevant to itself, while remaining invisible to the task content, pricing, and constraints of other parties. This architecture design, which minimizes information exposure, protects the commercial privacy of all participants and eliminates the trust crisis caused by information asymmetry.

[0049] In a multi-agent negotiation mechanism, the agents (resource providers) are not always actively negotiating. The act of issuing sub-solutions explicitly activates the decision-making process of the corresponding agent in the time dimension, setting clear negotiation topics (acceptance / rejection / counter-offer) and response deadlines. This allows multiple rounds of negotiation to proceed synchronously, orderly, and predictably, rather than asynchronously and disorderly message exchange. Compared to the resource scheduling mode in related technologies where the central server directly issues execution instructions, this application's embodiment optimizes the "scheduling instruction" into an "offer" and upgrades the "executor" into a "negotiator," providing a foundation for the operation of subsequent collaborative mechanisms.

[0050] Step 208: Obtain feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints.

[0051] Resource usage constraints refer to a set of technical restrictions maintained internally by the resource provider and dynamically updated to determine whether its logistics resources can undertake specific tasks within a specific time window. These rules can be embedded in the resource provider's local management system and exist in the form of database tables, configuration files, or business rule engines.

[0052] Correspondingly, resource cost information may include a parameterized mathematical model used by the resource provider to calculate the economic cost required to perform a specific logistics task. This model reflects characteristics such as the resource provider's operational efficiency, asset structure, procurement agreements, and energy consumption.

[0053] Feedback information refers to the structured decision result data packet returned by the resource provider to the cloud server after completing the local feasibility assessment and cost calculation. Specifically, the feedback information includes the resource provider's unique identifier, such as the carrier ID "CARRIER_7890"; the corresponding sub-scheme ID, such as "SUBPLAN_20250211_001"; the feasibility assessment results, including status fields (accepted / rejected / counter-offer), rejection reason code (if rejected), and a list of specific constraints that caused the conflict (if rejected); and quotation information (if accepted or counter-offered), including the total price, unit price, cost breakdown details (optional, used to improve interpretability), and quotation validity period.

[0054] After receiving the logistics resource scheduling sub-plan, the resource provider's local system activates the constraint solving engine. It extracts the assigned resource object ID from the sub-plan, such as "Vehicle Y," and queries the local resource status database for the object's real-time status record, including current location, current task, remaining capacity, and equipment health status. It then performs a time overlap detection between the task time interval required by the sub-plan and the time axis of the resource object's already assigned tasks. If overlapping intervals exist, a time conflict is identified. For example, if Vehicle Y has been assigned another transportation task from 14:00 to 16:00, and the sub-plan requires execution from 14:30 to 17:30, the two overlap from 14:30 to 16:00, thus constituting a conflict.

[0055] If the task involves resource consumption, such as warehouse capacity usage, load capacity usage, or picking time usage, the consumption required by the sub-solution is compared with the current remaining capacity of the resource object. If the consumption exceeds the remaining capacity, it is determined to be a capacity conflict. For example, if the sub-solution requires the outbound shipment of 100 items, and warehouse A currently has a maximum of 50 picking manpower per hour that can be processed, and 50 orders are already being processed, the remaining capacity is 0, thus a conflict is determined.

[0056] The task type tags in the sub-solution, such as "cold chain," "dangerous goods," "cross-border," and "large items," are matched with the capability attribute tags of the resource object. If the resource object lacks a necessary tag, or its current status is "faulty," "offline," or "under maintenance," it is considered incompatible. For example, if the sub-solution requires temperature-controlled transportation at 2-8°C throughout, but vehicle Y is not equipped with a refrigerated box, it is considered incompatible. If the sub-solution specifies operators, such as drivers, forklift operators, or pickers, the qualification tags required for the task are compared with the personnel's skill certificate database. If the qualification is missing or the certificate has expired, it is considered a skill conflict. For example, if the sub-solution specifies driver D008 to perform dangerous goods transportation tasks, but D008's dangerous goods transportation qualification expired last month, it is considered a conflict.

[0057] The sub-solution is deemed "feasible" only if all the above checks pass. If any check fails, the system generates an "infeasible" feedback, which may include a structured rejection reason, such as "Time conflict: Vehicle Y has already been assigned a task," "Insufficient capacity: Warehouse A's picking manpower is saturated," or "Lack of qualifications: Driver D008 does not have hazardous materials qualifications." These structured rejection reasons will serve as the direct basis for subsequent solution adjustments.

[0058] Step 210: Based on the feedback information returned by all resource providers, adjust the initial logistics resource scheduling scheme to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0059] All feedback information is collected via a cloud server. If all feedback is "accepted," it indicates that the initial logistics resource scheduling plan has reached an agreement on cost and feasibility, and can be directly implemented as the target logistics resource scheduling plan.

[0060] When cloud servers generate initial logistics resource scheduling plans, they rely on observable real-time operational data, including resource location, inventory, and workload status. However, cloud servers cannot access the private constraints and actual costs of each resource provider. For example, a driver may have worked continuously for four hours and require a mandatory 30-minute rest, or the carrier's actual execution costs may exceed industry standards due to recent oil price increases or aging vehicles. Therefore, while the initial plan may be reasonable in terms of "physical feasibility," it often deviates in terms of "economic feasibility." Furthermore, each resource provider is an independent stakeholder whose decision-making objective is to maximize their own interests, rather than minimizing global costs. Although the initial plan may be mathematically "globally optimal," distributing the cooperative benefits deserved by each party often fails to satisfy all participants. One party may feel its contribution is underestimated or its benefits are being appropriated, thus refusing to accept the plan or offering a price deviating from the actual cost. This conflict of objectives cannot be resolved through centralized optimization and requires a negotiation mechanism to seek a game equilibrium.

[0061] Specifically, based on the problem types represented in the feedback information, the following technical elements in the initial logistics resource scheduling plan can be adjusted in a single point or multiple points: Task assignment adjustment. When the original assignor reports "infeasible," or its quotation is significantly higher than the quotations of other resource providers in the candidate set, the system will replace the relevant task with another resource provider. For example, the trunk transportation task can be changed from "Carrier X" to "Carrier Y." Resource allocation adjustment. When the specific resource object originally assigned reports a conflict, such as the designated vehicle being occupied or the designated warehouse space being full, but other available resources exist under the same resource provider, the system will replace the task with other resource objects under the same provider.

[0062] When the simulated profit distribution results do not meet the consensus conditions, the system adjusts the bid-sharing ratio or total price allocation scheme of each participant. For example, it increases the unit freight rate of carrier X and correspondingly reduces the outbound storage costs of the warehousing party, making the cooperative surplus obtained by all parties more balanced.

[0063] Finally, each resource provider returning an "accept" or "counter-offer" state is modeled as a cooperative game agent. Each agent has the following attributes: a policy space, i.e., accepting the current offer, rejecting the current offer, and making a counter-offer; a private valuation function, representing the true value of performing a specific task for the agent, which can be understood as the sum of the cost negative and the benefit; and a bid, i.e., the price fed back in step 208, which is a noisy observation of the private valuation function, and the bid is usually not lower than the cost. Understandably, the cloud server does not attempt to obtain the specific form of the private valuation function, but only uses the received bid as the decision input.

[0064] Step 212: Perform scheduling control on the logistics resource object according to the target logistics resource scheduling scheme.

[0065] Once the target logistics resource scheduling plan is determined, it is converted into a series of specific, executable control instructions through a cloud server.

[0066] Control commands can be technical action commands that can be directly parsed and executed by equipment or systems. For example: sending an API call to the warehouse WMS to "Create outbound order, item XX, quantity 100, target vehicle Y"; sending an instruction to the vehicle TMS and onboard terminal to "Plan route: from A to B, estimated arrival time"; sending a control signal to the sorting line PLC to "Start at time T, process order batch Z". These control commands are routed to the corresponding logistics resource objects or their control systems via their respective communication protocols. After receiving the commands, these systems drive physical equipment (such as AGVs, conveyor belts, and vehicles) to perform specific operations, thereby completing the physical flow of the logistics task.

[0067] This application implements data preprocessing and local emergency response through edge computing nodes, combined with efficient cloud-based collaborative algorithms, reducing the decision-making latency of traditional centralized systems from minutes or even hours to seconds or minutes, meeting the real-time requirements of modern logistics. Through a multi-agent automatic negotiation mechanism, the cross-enterprise resource matching and negotiation process is automated, solving the "information silo" problem and reducing most of the manual coordination work in logistics resource scheduling.

[0068] In some embodiments, adjusting the initial logistics resource scheduling scheme based on the feedback information returned by all resource providers to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupancy request includes:

[0069] Each of the resource providers is modeled as an agent in a cooperative game. Based on the feedback information corresponding to each agent, the marginal contribution value of each agent to the total collaborative benefit brought about by the execution of the initial logistics resource scheduling scheme is calculated.

[0070] Based on the marginal contribution value, a profit distribution simulation is performed among multiple agents to obtain the profit distribution simulation result, and it is determined whether the profit distribution simulation result meets the preset consensus conditions.

[0071] If the consensus conditions are met, the current logistics resource scheduling scheme will be determined as the target logistics resource scheduling scheme.

[0072] If the consensus conditions are not met, at least one of the task assignment, resource allocation, or cost sharing clauses in the current logistics resource scheduling scheme shall be modified according to the profit distribution simulation results to obtain the logistics resource scheduling scheme for the next round of negotiation.

[0073] Send the logistics resource scheduling sub-schemes associated with each of the intelligent agents in the next round of negotiated logistics resource scheduling scheme to the corresponding intelligent agents to obtain the new round of feedback information returned by the intelligent agents;

[0074] Return to the step of calculating the marginal contribution value, and use the logistics resource scheduling scheme of the next round of negotiation as the new logistics resource scheduling scheme of the current round for a new round of evaluation, until the profit distribution simulation result of the current round meets the consensus condition or the negotiation round reaches the preset upper limit.

[0075] Each resource provider that returns an "accept" or "counter-offer" status is modeled as a cooperative game agent. Resource providers that respond with "rejection" are not included in the agent set for the current negotiation round because they have explicitly stated they will not participate in this cooperation; however, their reasons for rejection will be recorded and used for subsequent scheme adjustments.

[0076] Each agent i in this embodiment is abstracted as a decision entity with the following mathematical properties: a policy space. Feasible policies for agent i in the current round include: accepting the current solution, rejecting the current solution, and proposing a counter-offer. The counter-offer is expressed through the offer field in the feedback information. A private valuation function, denoted as v_i(task), quantifies the true value of performing a specific task for agent i, which can be understood as the benefit of performing the task minus the cost incurred. v_i(task) is agent i's core trade secret and cannot be directly observed by the cloud server. And the offer, denoted as p_i, is the offer field in the feedback information. The offer is a noisy observation of the private valuation function v_i(task). Under the rational offer assumption, p_i ≥ v_i(task), meaning the offer is not lower than the true cost. Although the cloud server cannot know the specific value of v_i(task), it can perform subsequent collaborative benefit evaluation and allocation calculations based on p_i.

[0077] After completing the agent modeling, the cloud server evaluates the total collaborative revenue of the current round's logistics resource scheduling scheme S. The total collaborative revenue V(S) is defined as: the overall surplus generated if all agents cooperate to execute the logistics task according to scheme S. Its calculation formula is: V(S) = Total budget paid by the customer to complete the logistics task - ∑(Actual cost of each agent executing its assigned sub-task);

[0078] Since the actual cost v_i(task) of each agent is unobservable private information, this embodiment uses the feedback quote p_i as an upper bound estimate of the actual cost to calculate the observable total collaborative revenue V_obs(S): V_obs(S) = total budget - ∑p_i; V_obs(S) represents whether, at the current quote level, if the task is executed according to scheme S, there is still a surplus after the budget paid by the customer covers the quotes of all participants. If V_obs(S) > 0, it indicates that the scheme has positive collaborative revenue, and this surplus can be distributed among the agents as a "cooperation bonus" to incentivize cooperation. If V_obs(S) ≤ 0, it indicates that the scheme is economically infeasible—the scheme needs to be adjusted to reduce the total quote, or an increase in budget needs to be requested from upstream.

[0079] To fairly distribute the total collaborative benefit V_obs(S), it is necessary to quantify the contribution of each agent to the overall cooperation. In this embodiment, the Shapley value from cooperative game theory is used as the contribution metric. For agent i, its Shapley value is defined as: the average marginal contribution of agent i across all possible cooperative alliance orders. The specific calculation formula is as follows: the Shapley value is equal to the ratio of the factorial of all subsets S that do not contain agent i, the factorial of the remaining agents, and the factorial of the total number of agents, multiplied by the marginal benefit increment brought by agent i joining alliance S, and finally summed over all subsets S.

[0080] Where N is the set of all agents, S is any subset of N that does not contain agent i, i.e., a possible "partial cooperative alliance", V(S) is the total cooperative benefit that can be generated by the cooperation of agents in subset S, and V(S ∪ {i}) minus V(S) is the marginal benefit increment brought by agent i joining alliance S.

[0081] Based on the calculated Shapley value, the total cooperative revenue is virtually allocated among the agents. After allocation, agent i's total revenue equals its negative bid plus the Shapley value allocation it receives. If the bid is the cost, then this revenue value is its cooperative surplus.

[0082] The consensus condition is defined as the simultaneous fulfillment of the following two conditions: The simulated profit distribution result for each agent is no less than its quoted price, i.e., at least breaking even or achieving a positive surplus. If any agent is in a loss-making state after the distribution, it means the solution is unattractive to it, and negotiation cannot reach an agreement. Furthermore, the simulated profit distribution results should not be too disparate; the ratio of the worst-performing agent's profit to the best-performing agent's profit should be greater than a preset fairness threshold, typically set between 0.3 and 0.5. This threshold can be dynamically adjusted according to the application scenario and industry practices.

[0083] If the simulation results of the current plan's revenue distribution simultaneously meet the above consensus conditions, then a consensus is considered reached through negotiation, and the current plan becomes the target logistics resource scheduling plan. Conversely, if the consensus conditions are not met, the system needs to generate an adjusted plan. The adjustment strategy selection follows this priority order: First priority: Replacement strategy for those with excessively high bids. The system identifies the agent whose bid deviates most from the industry benchmark cost and attempts to select a replacement with a better bid from the candidate set. For example, if carrier X's bid is 30% higher than carrier Y's, and carrier Y still has the capacity to handle tasks, the system will reassign the task to carrier Y. Second priority: Task reallocation strategy. If an agent reports "infeasible," or its revenue distribution simulation results are negative and cannot be improved through price negotiation, then all or part of its assigned tasks are reallocated to other agents. During reallocation, priority is given to candidates with high historical fulfillment rates and low current loads. Third priority: Time buffer strategy. If the conflict stems from overlapping time windows, the system attempts to fine-tune the task time, typically in 15- or 30-minute increments, shifting it forward or backward within the customer-allowed time window, and then re-initiating a quote request to the same agent. The fourth priority is task splitting and merging strategies. If a single resource provider cannot independently handle the entire batch of tasks, the system attempts to split the tasks into smaller, more granular subtasks; if the total cost after splitting is significantly higher than merging the packages, the system attempts to merge multiple subtasks into a single task. Task splitting and merging require the generation of a new solution framework.

[0084] The adjusted new plan is used as the current plan for the next round of negotiation. Steps 206 to 210 of the aforementioned embodiments are repeated until any of the following termination conditions are met: Termination Condition 1: Consensus Reached. The simulation results of the current plan's benefit distribution satisfy the two consensus conditions of individual rationality and fairness. The system outputs the current plan as the target logistics resource scheduling plan. Termination Condition 2: Round Limit. The number of negotiation rounds reaches the preset maximum value, usually set to 3 to 5 rounds. The system automatically selects the plan with the best overall benefit from historical plans, i.e., the plan with the highest total collaborative benefit or the highest sum of satisfaction among all parties, as the target logistics resource scheduling plan. Termination Condition 3: Time Deadline. As the latest decision-making time for the task approaches, the system forcibly outputs the plan with the highest overall benefit in the current round that has passed feasibility verification as the target logistics resource scheduling plan, ensuring that the task is not delayed.

[0085] In some embodiments, the logistics resource objects include at least one of transport vehicles, storage space, sorting equipment, and loading and unloading machinery;

[0086] The step of obtaining feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme includes:

[0087] Each resource provider compares the logistics task requirements in the logistics resource scheduling sub-scheme with its own current resource usage constraints to determine whether there are any constraint conflicts; wherein, the resource usage constraints include at least one of time window constraints, resource capacity constraints, equipment status constraints, and personnel skill constraints.

[0088] If no constraint conflict exists, the total resource cost required to execute the logistics resource scheduling sub-plan is calculated by each of the resource providers based on the resource cost information; wherein, the resource cost information includes at least one of fixed cost allocation, variable cost rate, and dynamic adjustment coefficient;

[0089] Each of the resource providers generates its own feedback information, which includes at least one of the following: total price, unit price, or conditional commitments, based on the total resource cost and the resource provider's pricing strategy.

[0090] The term "transport vehicle" refers to motorized equipment used for the spatial movement of goods, including but not limited to box trucks, container trucks, refrigerated transport vehicles, flatbed trailers, and new energy logistics vehicles. Each transport vehicle is assigned a unique vehicle ID in the system and maintains a connection with the system through an onboard IoT terminal. "Storage space" refers to physical locations within a warehouse that can be used to store goods, including but not limited to heavy-duty racks, light-duty racks, flow racks, floor stacking areas, and automated storage and retrieval systems (AS / RS). Each storage space is assigned a unique location ID in the system, and its status is monitored by fixed or mobile sensing devices. "Sorting equipment" refers to equipment used for the automated classification of goods by destination, category, etc., including but not limited to cross-belt sorters, swing wheel sorters, sliding block sorters, and stacker cranes in automated storage and retrieval systems (AS / RS). Each sorting piece of equipment is assigned a unique equipment ID in the system and interfaces with the system through an industrial fieldbus or OPC UA interface. Loading and unloading equipment refers to auxiliary equipment used for loading, unloading, handling, and palletizing goods, including but not limited to counterbalance forklifts, reach trucks, electric pallet trucks, automated guided vehicles (AGVs), and palletizing robots. Each loading and unloading device is assigned a unique device ID in the system and communicates with the system via an industrial wireless network. The above four types of logistics resource objects together constitute the logistics resource pool scheduled in this application embodiment. Each type of resource object has a state space that can be perceived in real time and a command interface that can be remotely controlled.

[0091] After receiving the logistics resource scheduling sub-plan from the cloud server, the resource provider initiates an atomic decision transaction in its local system. This transaction consists of three sequentially executed sub-steps: constraint verification, cost calculation, and quotation generation. The following provides a detailed explanation of this process with specific technical parameters.

[0092] The goal of the constraint verification sub-step is to determine whether the received logistics resource scheduling sub-scheme is technically feasible at the current time and under the current state.

[0093] The resource provider's local system first parses the following fields from the sub-scheme's data structure: Resource object identifier: such as "Vehicle VIN code: LSVCD49J5KN123456", "Storage location ID: WH-A-12-03"; Task type label: such as "Cold chain transportation", "Dangerous goods", "Oversized", "Urgent"; Task time window: such as start time "2026-02-12 14:00:00", end time "2026-02-12 17:30:00"; Resource consumption: such as "Load capacity 850kg", "Storage capacity 12 pallets", "Picking time 2.5 hours"; Qualification requirements: such as "Dangerous goods transportation qualification certificate", "High-value goods operation certification".

[0094] Based on the parsed resource object identifier, query the real-time status record of the object in the local resource status database. The status record can contain at least the following dimensions: Time dimension: the task timeline of the resource object, storing the allocated task time windows in the form of a list of intervals. Perform interval overlap detection between the task time intervals required by the sub-scheme and the existing timeline. The detection algorithm uses a red-black tree-based time interval index, with a time complexity of O(log n). If non-zero overlapping intervals exist, it is determined to be a time window conflict, and the start and end times of the conflicting interval are recorded.

[0095] Capacity dimension: The current remaining capacity vector of the resource object. The resource consumption required by the sub-scheme is compared with the current remaining capacity item by item. If any consumption exceeds the remaining capacity, it is determined to be a capacity conflict, and the excess consumption item and its current remaining value are recorded.

[0096] State Dimension: The current running state enumeration value of the resource object. The state space includes: "Idle", "Busy", "Fault", "Offline", "Maintenance", "Charging", etc. If the current state of the resource object is not "Idle", and the interval between the task start time required by the sub-scheme and the current time is less than the expected remaining duration of the state, it is determined to be a state incompatibility, and the current state and the expected recovery time are recorded.

[0097] Capability Dimension: The set of capability attribute tags for the resource object. The task type tags in the sub-solution are matched against the capability tag set of the resource object. If the task type tag set is not a subset of the capability tag set, it is determined that a capability is missing, and the specific missing capability item is recorded.

[0098] Qualification Dimension: If a sub-solution specifies a particular operator, the system queries that operator's qualification certificate database. Each certificate includes the certificate type, issuing authority, and validity period. The system compares the current date with the certificate's expiration date. If the certificate has expired or is about to expire during the task execution period, it is determined that the qualification is invalid, and the invalid certificate information is recorded.

[0099] If all the above dimensions pass the verification, the sub-solution is deemed feasible. If any verification fails, the decision-making process is immediately terminated, and infeasibility feedback is generated. The structured data packet of infeasibility feedback includes: resource provider identifier, sub-solution identifier, feasibility status "REJECT", rejection reason main code (such as "TIME_CONFLICT", "CAPACITY_SHORTAGE", "SKILL_MISSING"), and specific conflict details (such as "Vehicle VIN:LSVCD49J5KN123456 already has a task between 14:00 and 16:00").

[0100] Correspondingly, if the constraint verification sub-step determines that the sub-solution is feasible, the system proceeds to the cost accounting sub-step. The objective of this sub-step is to calculate the total economic cost required to execute the sub-solution.

[0101] The cost accounting sub-step calls the resource cost information database. This database is the resource provider's core business data asset, stored locally in the form of a parametric model, and will never be released externally. The resource cost information database contains the following four types of model parameters:

[0102] Fixed cost allocation parameters: Fixed costs refer to costs that are not directly proportional to the intensity of task execution but need to be covered in long-term operations. Typical fixed costs include: vehicle depreciation, warehouse rent, equipment depreciation, insurance premiums, and management personnel salaries.

[0103] In the cost accounting model, fixed costs are quantified using either the time-based allocation method or the frequency-based allocation method. For transportation vehicles, the fixed cost allocation rate is expressed in "yuan / minute," calculated as: vehicle purchase cost ÷ depreciation period ÷ annual working days ÷ daily working hours ÷ 60. For warehouse space, the fixed cost allocation rate is expressed in "yuan / pallet / day" or "yuan / item / day," calculated as: monthly warehouse rent ÷ average monthly inventory turnover ÷ single pallet capacity. The estimated execution time of tasks is extracted from sub-schemes, or the task duration is read from spatiotemporal planning information, and multiplied by the corresponding fixed cost allocation rate to obtain the fixed cost allocation amount.

[0104] Variable cost rate parameters: Variable costs refer to direct consumption costs that are directly proportional to the intensity of task execution. Typical variable costs include: fuel / electricity costs, road and bridge tolls, driver / operation labor costs, packaging materials, equipment maintenance materials, etc. In the cost accounting model, variable costs are calculated using the rate multiplied by the usage. Each variable cost element has a corresponding unit of measurement and rate: Fuel cost: Rate "Yuan / km", usage is the planned mileage of the sub-scheme. Mileage data is obtained by calling a third-party route planning API. This API takes the origin, destination, waypoints, and vehicle type as input and returns the optimal route length (km). If real-time traffic data is enabled, the mileage will be converted into equivalent mileage considering the congestion factor. Electricity cost: Rate "Yuan / kWh", usage is the estimated electricity consumption of the sub-scheme. Electricity consumption is estimated based on historical operational energy consumption data, and usage is the estimated operation labor hours of the sub-scheme. Operation labor hours are estimated based on historical operational efficiency statistics. Road and bridge tolls: The toll rate is "yuan / km" (it may be a fixed fee for specific road sections), and the usage is the mileage of the toll road sections involved in the planned driving mileage of the sub-scheme. The total variable cost can be obtained by multiplying the usage of each variable cost element by the corresponding toll rate and summing them up.

[0105] Dynamic Adjustment Coefficient: The dynamic adjustment coefficient is a cost adjustment factor for resource providers to cope with real-time environmental fluctuations and task-specific characteristics. This coefficient is a floating-point number greater than or equal to 1.0, weighted and synthesized from the following sub-factors: Traffic Congestion Factor: The system obtains the real-time congestion index of the task path through the real-time traffic API. This factor reflects the increased fuel consumption, accelerated vehicle wear, and increased driver fatigue caused by congestion. Task Urgency Factor: The system reads the task priority label from the sub-schemes. If the label is "urgent" or "expedited," or the task response time (from receiving the sub-scheme to the task start time) is shorter than a preset threshold (e.g., 2 hours), the urgency factor is 1.3~1.8. This factor reflects the additional costs incurred by resource providers, such as interrupting existing work plans, adjusting schedules, and paying overtime. Equipment Load Factor: The system queries the current task queue length and average utilization rate of the resource object. When the equipment utilization rate exceeds 80%, the load factor is 1.1~1.3. This factor reflects the risk premium of increased equipment failure risk, increased maintenance costs, and decreased service quality under high load conditions. The formula for synthesizing the dynamic adjustment coefficient is: dynamic_factor = max(1.0, w1×congestion factor + w2×emergency factor + w3×load factor). Each weight coefficient can be dynamically adjusted according to the resource provider's operation strategy.

[0106] Opportunity Cost Coefficient: Opportunity cost refers to the quantitative estimate of other potential benefits forgone by undertaking this task. The calculation of the opportunity cost coefficient is based on customer value rating and capacity scarcity. The resource provider's customer relationship management system maintains a long-term cooperation value rating for each customer, with rating dimensions including: historical cooperation years, average annual order amount, timely payment rate, complaint rate, etc. Requests from high-value customers should be prioritized even if the profit margin is lower, because the "opportunity cost" of forgone is higher. Capacity scarcity is calculated as the proportion of the resource provider's current remaining capacity to its total capacity. The lower the remaining capacity, the higher the capacity scarcity, and the greater the marginal opportunity cost of undertaking low-profit tasks. The formula for calculating the opportunity cost coefficient can be: opportunity_cost = base_opportunity_rate × (1 - customer value coefficient) × capacity scarcity. Where, base_opportunity_rate is the resource provider's preset opportunity cost benchmark value, the customer value coefficient is normalized to the [0,1] range, and the coefficient for high-value customers approaches 1 (low opportunity cost).

[0107] The complete calculation process for cost synthesis: Calculate the cost before adjustment: cost_before_adjustment = fixed_cost + variable_cost;

[0108] Apply dynamic adjustment: cost_adjusted = cost_before_adjustment × dynamic_factor;

[0109] Additional opportunity cost: total_cost = cost_adjusted + opportunity_cost;

[0110] Apply the target profit margin: quoted_price = total_cost × (1 + target_margin).

[0111] Among them, target_margin is the target profit margin preset by the resource provision policy for this task type and customer level, and its value range is usually from 0.05 to 0.30.

[0112] After the cost accounting sub-step outputs the total cost (total_cost), the system proceeds to the quotation generation sub-step. The goal of this sub-step is to transform internally calculated costs into externally disclosed, structured, and standardized quotation information. Specifically, the system selects the quotation expression format based on the task characteristics of the sub-scheme. For tasks with clearly defined quantities and distances, a total price quotation is used; for scenarios with uncertain quantities or long-term framework agreements, a unit price quotation is used, such as "yuan / ton·km", "yuan / piece", or "yuan / pallet"; for complex scenarios, both total and unit prices can be provided simultaneously. To improve the interpretability of subsequent negotiations, the system can selectively include detailed cost breakdowns in the quotation information. These cost breakdowns do not disclose specific rate parameters and consumption quantities, only the amount and item name, providing both decision-making transparency and protecting the core cost model from reverse engineering. A validity period timestamp is set for the quotation based on the task type and resource status. For routine tasks, the validity period is typically set to 15 minutes to 1 hour; for urgent tasks, the validity period can be shortened to 5 minutes. The validity period of the offer ensures that the offers from all parties will not become invalid over time during subsequent rounds of negotiations.

[0113] Optionally, resource providers may selectively include service commitment clauses in their quotations, such as "on-time arrival rate ≥ 98%", "full-process temperature control records are traceable", and "damage rate ≤ 0.1%". These service commitments are technical guarantees from the resource provider regarding the quality of their services and will be tracked and assessed after the task is completed.

[0114] After completing the above operations, the system will encapsulate the quotation information into a structured feedback data packet and send it to the cloud server through a secure API interface or message queue.

[0115] In some embodiments, the method is executed on a cloud server; the logistics resource object is associated with a locally deployed edge computing node; the edge computing node is communicatively connected to the cloud server; the step of obtaining real-time operating data of logistics resource objects managed by multiple resource providers in response to a logistics resource occupancy request includes:

[0116] The raw operational data collected by IoT devices connected through the edge computing node is cleaned to obtain the real-time operational data stream.

[0117] The edge computing node uploads the real-time operational data stream to the cloud server; wherein, if the edge computing node detects that the real-time operational data meets the preset local emergency rules, the edge computing node generates and executes local control instructions for the associated logistics resource object according to the local emergency rules.

[0118] The cloud server can be a standalone physical server, a cluster or distributed system of multiple servers, or a cloud server instance providing cloud computing services. Edge computing nodes are dedicated computing devices deployed locally on the logistics resources, adaptable to industrial environments. Edge computing nodes maintain a persistent connection to the cloud server via wired Ethernet (1000BASE-T) or industrial wireless networks (5G, Wi-Fi 6, industrial WLAN). The deployment of edge computing nodes follows the principle of proximity to the data source. For warehousing resources, edge computing nodes are deployed near the warehouse's low-voltage electrical room or wireless access point; for transport vehicles, edge computing nodes are deployed in the cab as in-vehicle tablets or industrial computers; for sorting equipment, edge computing nodes are integrated into the equipment control cabinet as embedded modules.

[0119] IoT devices are primary instruments and actuators attached to logistics resources to sense the state of the physical world. IoT devices connect to edge computing nodes via wired or wireless means and report raw sampled data at intervals ranging from milliseconds to seconds.

[0120] The IoT device types involved in this application embodiment may include: GPS / BeiDou positioning modules, OBD-II vehicle diagnostic terminals, RFID readers, temperature and humidity sensors, PLC status registers, power monitoring modules, etc. Edge computing nodes collect raw operational data through the interfaces of the various IoT devices they connect to, either in a polling or interrupt manner.

[0121] Considering that raw sensor data inevitably contains noise, missing values, and anomalous jumps, edge computing nodes perform the following cleaning algorithms to transform the raw data into reliable, continuous state signals: Outlier detection: For measurements with physical upper and lower limits (e.g., temperature -40°C to 80°C, voltage 9V to 16V), a threshold truncation method is used to mark data points outside the reasonable range as invalid and fill them with the previous valid value or moving average. For measurements with statistical regularities (e.g., vehicle speed, engine speed), the 3σ principle is used to calculate the mean and standard deviation of the most recent N sampling points, and data points deviating from the mean by more than 3 times the standard deviation are discarded.

[0122] Missing value imputation: For sampling interval discontinuities caused by network packet loss or device sleep mode, linear interpolation or spline interpolation is used to fill in the estimated values ​​of missing moments on the time axis, ensuring the equal time interval characteristics of the output data stream. Additionally, unit unification and dimension normalization: Data from different sources and with different dimensions are converted to standard units. For example, Fahrenheit temperatures are converted to Celsius, miles per hour speeds are converted to kilometers per hour speeds, and percentage-expressed state of charge is converted to floating-point numbers in the 0-1 range.

[0123] The cleaned data remains a raw sequence of sampled points, unsuitable for direct transmission to the cloud for high-level decision-making. Edge computing nodes perform feature extraction, compressing the time-series data into a structured state vector. The cleaned, formatted, and compressed real-time running data stream is periodically reported to the cloud server's data access endpoint via a pre-established secure communication channel between the edge computing nodes and the cloud server, using MQTT, AMQP, or HTTP / 2 protocols.

[0124] In this embodiment, the edge computing node is not merely a data acquisition agent for the cloud server, but also a local control unit with independent decision-making capabilities. When a specific condition is detected to meet a preset local emergency rule, the edge computing node actively interrupts or bypasses the current negotiation process with the cloud server and directly executes local control instructions. The local emergency rule is a deterministic decision-making logic expressed using the Event-Condition-Action (ECA) paradigm. Each emergency rule can consist of the following three elements: Event: The signal source that triggers the rule evaluation. Events can be: timed events (e.g., "evaluate every 30 seconds"), data events (e.g., "receive new temperature and humidity sampling values"), status events (e.g., "device status register changes from 0x01 to 0x04 (fault)"), and external events (e.g., "receive emergency mode activation instruction from the cloud server"), etc.

[0125] In this embodiment, when a local emergency rule is triggered, one or more control commands are sent to the controlled device, the local operating parameters of the edge node (such as reporting cycle and alarm threshold) are modified, and an emergency event notification is sent to the cloud server (optional in bypass / interruption mode). To prevent the cloud server from conflicting with local emergency commands without the cloud server's knowledge, the edge computing node locks the logistics resource object after issuing the emergency command. During the lock period, the edge computing node still reports real-time operating data normally, but attaches a lock status identifier (such as "emergency_lock": true, "lock_reason": "low_battery") to the data packet. After receiving the data with the lock identifier, the cloud server will suspend the allocation of new tasks to the resource until the lock status is explicitly released by the edge computing node. Finally, the edge computing node records the complete lifecycle of the emergency event as a structured log, including: event ID, trigger time, rule ID, real-time operating data snapshot at the trigger time, executed command sequence, device response code, and lock duration. This log is persistently stored locally on the edge node and synchronously reported to the audit log system of the cloud server.

[0126] In this embodiment, when a local emergency rule is triggered, the edge computing node actively switches to emergency mode, no longer waiting for decision instructions from the cloud server. For cloud tasks that have been issued but not yet executed, the edge computing node can choose to interrupt (cancel the task) or bypass (temporarily ignore and continue executing local emergency instructions) according to the priority of the emergency rule. Simultaneously or after executing local emergency instructions, the edge computing node asynchronously sends an emergency event notification to the cloud server. Upon receiving the notification, the cloud server updates the resource status in its maintained digital twin model and adaptively adjusts the affected global scheduling scheme.

[0127] In some embodiments, the real-time operational data includes external environmental data, which includes at least one of real-time traffic flow data and weather warning data; the step of planning logistics resource allocation based on the real-time operational data in response to the logistics resource occupancy request includes:

[0128] Real-time operational data containing the external environment data is input into a pre-trained risk prediction model to obtain the risk interference probability related to the logistics resource occupation request; wherein, the risk prediction model is a long short-term memory network model; the long short-term memory network model is used to predict the supply chain interruption risk within a future preset time window based on historical operational data and real-time external environment data;

[0129] Based on the aforementioned risk interference probability, the expected travel time of relevant logistics paths or the availability and reliability of logistics nodes in the initial logistics resource scheduling scheme are adjusted using a weighted average.

[0130] Real-time traffic flow data refers to spatiotemporal dynamic data reflecting the current traffic conditions of the road network. Specifically, it can include: Road congestion index: quantifying the degree of congestion on a road segment using a value of 0-10 or 0-100; Average traffic speed: the average speed of vehicles passing through a road segment at the current moment, in kilometers per hour; Event information: the start and end locations, impact range, and estimated duration of emergencies such as traffic accidents, road construction, traffic control, and temporary road closures.

[0131] Meteorological warning data refers to forecast and real-time information reflecting the potential adverse impacts of weather phenomena on logistics operations. Meteorological warning data may include at least one of the following parameters: precipitation type and intensity; visibility; wind speed and direction; temperature and humidity; and meteorological disaster warning.

[0132] Long Short-Term Memory (LSTM) neural networks are a variant of recurrent neural networks specifically designed to process time-series data, effectively capturing long-term dependencies and solving the gradient vanishing problem. The input layer of the risk prediction model can receive a concatenation of the following three types of feature vectors: Historical operational data sequences: These refer to the actual operational performance of task types, routes, and nodes that are the same as or similar to the current logistics resource occupancy request over a past period. This data sequence is extracted from a time-series database on a cloud server. Real-time external environment data: This refers to traffic flow, weather warnings, and other data acquired at the current moment or the most recent moment (within the past 15 minutes). This type of feature is input in the form of a "current frame" and does not constitute a time series. Static attribute features refer to the inherent attributes of tasks, paths, and nodes that do not change over time or have a long change period. This type of feature is input in the form of auxiliary vectors to help the model distinguish different scenarios. Examples of static attribute features include: path length (km), road grade (highway / national highway / provincial highway / urban road), whether it crosses provinces, node type (distribution center / forward warehouse / delivery station), and node processing capacity level.

[0133] The hidden layer of the Long Short-Term Memory (LSTM) neural network in this embodiment consists of two stacked LSTM units, each containing 128 memory units. This two-layer stacked structure can effectively capture short-term fluctuations (such as real-time traffic congestion) while also remembering periodic patterns (such as morning rush hour, evening rush hour, and holiday effects).

[0134] The output sequence h_t^1 of the first LSTM layer is used as the input sequence x_t^2 of the second LSTM layer. After two layers of temporal feature extraction, the hidden state h_T^2 of the last time step is taken as the input of the fully connected layer. The output layer is a fully connected layer followed by a sigmoid activation function, which maps the hidden state vector to a scalar value between 0 and 1, which is the risk interference probability. The physical meaning of the risk interference probability is: at the current moment and under the current environmental conditions, the probability of a perceptible delay or interruption exceeding the normal fluctuation range when executing the planned path or using the logistics node of the logistics resource occupation request. During model training, events in the historical task execution data where "the actual arrival time is later than the planned arrival time by more than a threshold (e.g., 30 minutes)" or "the task is marked as abnormally interrupted" are labeled as positive samples (risk = 1), and events that are completed on time and without abnormalities are labeled as negative samples (risk = 0). The loss function is binary cross-entropy, the optimizer is Adam, and the initial learning rate is set to 0.001.

[0135] The output of the risk prediction model is used as a weighted adjustment factor in the resource allocation planning algorithm, rather than simply presented to the user as an "alarm signal." In the resource matching and route planning algorithm of step 204, the edge weights of the path are usually set as the historical average travel time of the path or the real-time estimated travel time based on real-time traffic data. The edge weights are adjusted as follows: Adjusted travel time = Original estimated travel time × (1 + α × Risk interference probability), where α is the risk aversion coefficient, with a default value of 0.5, which can be dynamically configured according to user preferences or task priorities. The larger α is, the higher the degree of risk avoidance of the system. For tasks involving warehousing nodes, transit nodes, and delivery stations, the resource matching algorithm of step 204 needs to evaluate the availability and reliability of the nodes. The availability and reliability of the nodes are adjusted as follows: Adjusted node reliability = Original reliability score × (1 - β × Risk interference probability); where β is the reliability decay coefficient, with a default value of 0.3. The original reliability score is derived from static or quasi-static indicators such as the node's historical fulfillment rate and equipment integrity rate, and the value ranges from 0 to 1.

[0136] In some embodiments, the step of scheduling and controlling the logistics resource object according to the target logistics resource scheduling scheme includes:

[0137] The target logistics resource scheduling scheme and the decision-making basis for the corresponding scheduling scheme are displayed in the preset interactive interface; the decision-making basis for the scheduling scheme includes cost-benefit comparison and / or reasons for scheme selection obtained from the feedback information of each resource provider;

[0138] The interactive interface receives confirmation or adjustment commands for the target logistics resource scheduling scheme.

[0139] The target logistics resource scheduling scheme is modified according to the adjustment instruction.

[0140] The visual interactive platform is a front-end application system independent of cloud servers and edge computing nodes, deployed in a browser / server architecture or a mobile client / server architecture. Cost-benefit comparison information is used to show users why the current solution was chosen over other alternatives. Specifically, key cost components can be extracted from the quotations obtained from various resource providers, and the target solution can be compared side-by-side with at least one alternative. Comparison metrics include: Total Cost: A numerical comparison of the total price of the target solution and the total price of the alternatives. Cost Breakdown: The total cost is broken down into categories such as transportation costs, warehousing costs, and operating costs, presented in a stacked bar chart showing the differences in different cost categories for each solution. Unit Cost: For tasks charged by piece, by ton, or by kilometer, a comparison of the unit price for each solution is displayed.

[0141] The rationale for the chosen solution explains to the user which rules or constraints the algorithm ultimately selected. Specifically, key decision nodes can be extracted from the negotiation log and presented in natural language or structured tags.

[0142] Correspondingly, risk warning information is used to alert users to identified potential risks that the current plan may face during execution. Specifically, the probability of risk interference and the sources of risk related to the current plan can be extracted from the risk prediction model and presented in the form of a risk heatmap or warning icons. Risk warning information includes at least: path risk, such as the real-time congestion index of a road segment exceeding a threshold, or the weather warning's impact area overlapping with the path. Node risk, such as the current queuing time of a warehouse exceeding the historical P90 percentile, or a short-term increase in equipment failure rate. Carrier risk, such as a recent downward trend in the carrier's fulfillment rate, or the existence of unresolved complaint records. Revenue distribution information is used to show users whether the distribution of the cooperative surplus among the parties during the multi-agent negotiation process is fair and reasonable.

[0143] The interaction process in this embodiment may include: When a user first enters the interactive interface, the system displays a panoramic overview view of the proposed solution by default. This view centers on a three-dimensional supply chain network topology map, highlighting the paths, nodes, and carriers planned by the target solution. A summary of the decision-making basis is displayed in card format on the right or bottom of the view. After the user clicks the "Compare Alternative Solutions" button, the system enters a solution comparison view. This view displays the target solution and the top 2-3 alternative solutions simultaneously in a parallel table or parallel card format. Each solution card may include key indicators such as: carrier name, total price, estimated delivery time, risk level, and historical fulfillment rate, and uses color coding (green / yellow / red) to visually represent the advantages and disadvantages of each solution.

[0144] Users can switch the currently selected alternative to the target alternative with a single click by clicking the "Set as Target" button on the alternative card. Upon receiving the switch command, the system immediately updates the highlighted path and decision basis summary on the interface, and sends a confirmation of the alternative change to the cloud server.

[0145] Optionally, when a user clicks on a transportation task card, the system displays a dropdown list of all available candidate carriers for that route, including each candidate's real-time quote, historical fulfillment rate, and current remaining capacity. When the user clicks the "Switch" button, the system immediately changes the task assignment to the user-selected carrier and recalculates the total cost and estimated delivery time in real time. If the user uses a slider or time selector to shift the start time of a task forward or backward, the system continuously monitors whether the new time window conflicts with constraints reported by the resource provider. If a conflict is found, a red warning is displayed to the user. The user can also change the priority of a task from "Standard" to "Urgent" via a dropdown menu. Based on the new priority label, the system re-calls step 208 to obtain the urgent quote from the resource provider and updates the total cost.

[0146] In some embodiments, the method further includes:

[0147] After completing the scheduling control according to the target logistics resource scheduling scheme, the actual operational efficiency data of the logistics resource object is collected;

[0148] The actual operational efficiency data is compared with the expected efficiency indicators of the target logistics resource scheduling scheme to generate optimization deviation data;

[0149] Using the optimization deviation data, adaptive parameter or model updates are performed on at least one of the following: the planning model used for logistics resource allocation planning, the resource cost information estimation model of each resource provider, and the consensus judgment rule used when adjusting the initial logistics resource scheduling scheme.

[0150] The actual operational efficiency data refers to quantifiable performance indicators generated by logistics resources during task execution, rather than planned values ​​estimated during the decision-making stage. Examples include: Actual travel time: the actual travel time of a vehicle from the originating warehouse to the destination warehouse / distribution station, in minutes; Actual mileage: the total length of the actual route traveled by the vehicle, in kilometers; On-time status: whether delivery was completed within the customer's required time window; Data source: comparison of arrival timestamps with task constraints; Abnormal event records: whether traffic accidents, vehicle malfunctions, route deviations, customer refusal, or other abnormalities occurred during execution; Actual operation time: the actual time taken from receiving the outbound instruction to completing picking, packing, and handover, in minutes; Inventory accuracy: whether the system's inventory deduction after task completion matches the actual inventory count; Damage / misdelivery records: events such as damaged goods, shortages, or incorrect delivery of goods during task execution.

[0151] Optionally, actual processing efficiency may also be included: the actual number of sorted items or the tonnage handled per unit time, expressed as items / hour or tons / hour. Data source: sorting line PLC counter, equipment SCADA system. Equipment downtime: the total downtime, maintenance, and debugging time of equipment due to malfunctions during task execution. Data source: equipment status log, maintenance work order system, etc.

[0152] The collected actual operational efficiency data is compared item by item with the expected efficiency indicators recorded in the target logistics resource scheduling plan to generate efficiency deviation data. Efficiency deviation data is not a simple binary evaluation of "good or bad," but a structured numerical vector that can be directly consumed by downstream optimization modules. Specifically, different deviation calculation methods are used for different types of efficiency indicators: for example, for continuous numerical indicators applicable to travel time, mileage, operating time, fuel consumption, and electricity consumption, the deviation value = actual value - expected value; a positive deviation value indicates that the actual performance is worse than expected (delay, overspending), and a negative deviation value indicates that the actual performance is better than expected (ahead of schedule, cost savings).

[0153] Correspondingly, for ratio-based indicators such as on-time performance, inventory accuracy, and equipment availability, the deviation rate is calculated as follows: Deviation Rate = (Actual Value - Expected Value) / Expected Value; the deviation rate is normalized to the range [-1, +∞), where 0 indicates complete compliance with expectations, -0.1 indicates 10% better than expected, and +0.2 indicates 20% worse than expected. Optionally, count-based indicators such as abnormal events, damage records, and complaint work orders can also be included. Deviation Count = Actual Number of Occurrences - Expected Number of Occurrences; the expected number of occurrences is usually set to 0, and the deviation count is the total number of actual abnormal events.

[0154] The planning model parameters refer to the basic cost coefficients and time estimation models upon which the resource matching and route planning algorithms in the preceding steps rely. Specifically, the edge weights of a route are typically set as the historical average travel time of that route or the instantaneous estimated travel time based on real-time traffic data. When the system continuously observes that the actual travel time of a certain route systematically deviates from the estimated time, the time baseline value of that route needs to be calibrated. The calibration algorithm can use the following exponentially weighted moving average:

[0155] T_new = λ × T_actual + (1 - λ) × T_old;

[0156] Where T_actual is the actual travel time for this task, T_old is the baseline travel time before calibration, and λ is the learning rate, with a default value of 0.3. The larger λ is, the more sensitive the system is to new samples; the smaller λ is, the more fully the system retains historical statistics.

[0157] Correspondingly, when conducting cost accounting, resource providers need to estimate parameters such as picking time and loading / unloading time based on historical operational efficiency statistics. If feedback data reveals that the actual operational efficiency of a warehouse is significantly lower than its historical statistical value, the efficiency coefficient of that warehouse needs to be adjusted downwards. The calibration algorithm uses the quantile tracking method: the system maintains the P50 (median) and P90 (90th quantile) of the actual operational efficiency of each warehouse over the past 30 days. When the picking efficiency of three consecutive tasks is lower than P50, the system lowers the estimated efficiency coefficient of that warehouse in step 204 from P50 to P40; when the picking efficiency of three consecutive tasks is higher than P90, the system raises the estimated efficiency coefficient from P50 to P80.

[0158] Resource cost information models are parameterized mathematical models used by resource providers to calculate the economic costs required to execute tasks, including fixed cost allocation rates, variable cost rates, dynamic adjustment coefficients, and opportunity cost coefficients. These models are typically maintained locally by the resource provider and cannot be directly modified on cloud servers. However, when the system detects a persistent mismatch between a resource provider's pricing competitiveness and its actual performance quality through feedback data, it can indirectly guide the optimization of its cost model. Specifically, in the strategy of replacing overpriced bidders, the system dynamically lowers the price deviation threshold for carrier X from 30% to 25%, making it easier to identify as a "replacement target." This allows carrier X to perceive the market signal that "overpriced bids lead to lost orders" from the increased number of replacements, prompting it to proactively adjust its internal cost accounting parameters (such as lowering the target profit margin and optimizing the dynamic adjustment coefficient) to improve its pricing competitiveness. This is an indirect learning mechanism based on market feedback, rather than a direct intrusion into a proprietary model.

[0159] Consensus judgment rules refer to meta-parameters used to determine whether a negotiation has reached an agreement, such as fairness thresholds, individual rationality conditions, and round limits. These rules do not belong to any specific task but rather are a global strategy-level configuration affecting all negotiation tasks. By analyzing macro-indicators such as the success rate, round distribution, and human intervention rate of historical negotiation tasks, the consensus judgment rules can undergo monthly or quarterly strategy evolution. Optionally, regarding adjustments to negotiation rounds, if it is found that the improvement in the effectiveness of the solution is less than 1% in the 4th and 5th rounds of negotiation for multiple consecutive tasks, it is determined that the marginal benefit of continuing iteration is insufficient to cover the time cost. K_max can be automatically reduced from 5 to 4, and the success rate of subsequent tasks can be observed.

[0160] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0161] Based on the same inventive concept, this application also provides a logistics supply chain management device for implementing the logistics supply chain management method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the logistics supply chain management device provided below can be found in the limitations of the logistics supply chain management method described above, and will not be repeated here.

[0162] In one exemplary embodiment, such as Figure 3 As shown, a logistics supply chain management device 300 is provided, comprising:

[0163] Response module 302 is used to respond to logistics resource occupancy requests and obtain real-time operating data of logistics resource objects provided by multiple resource providers.

[0164] Planning module 304 is used to plan the allocation of logistics resources based on the real-time operating data in response to the logistics resource occupancy request, and to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupancy request.

[0165] The sending module 306 is used to send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers respectively;

[0166] The acquisition module 308 is used to acquire feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints;

[0167] The adjustment module 310 is used to adjust the initial logistics resource scheduling scheme according to the feedback information returned by all resource providers, so as to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request.

[0168] The scheduling module 312 is used to schedule and control the logistics resource object according to the target logistics resource scheduling scheme.

[0169] Each module in the aforementioned logistics supply chain management device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0170] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a logistics supply chain management method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0171] Those skilled in the art will understand that Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0172] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps included in any of the foregoing method embodiments.

[0173] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps included in any of the foregoing method embodiments.

[0174] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps included in any of the foregoing method embodiments.

[0175] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0176] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0177] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0178] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A logistics supply chain management method, characterized in that, The method is executed on a cloud server; Logistics resource objects are associated with locally deployed edge computing nodes; The edge computing node is communicatively connected to the cloud server; the method includes: In response to requests for logistics resource allocation, real-time operational data of logistics resource objects provided by multiple resource providers are obtained, including: The raw operational data collected by IoT devices connected through the edge computing node is cleaned to obtain the real-time operational data stream. The edge computing node uploads the real-time operational data stream to the cloud server; wherein, if the edge computing node detects that the real-time operational data meets the preset local emergency rules, the edge computing node generates and executes local control instructions for the associated logistics resource object according to the local emergency rules; Based on the real-time operating data, a logistics resource allocation plan is performed for the logistics resource occupation request to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupation request. Send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers respectively; Obtain feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints; Based on the feedback information returned by all resource providers, the initial logistics resource scheduling scheme is adjusted to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request. This includes: modeling each resource provider as an agent in a cooperative game, and calculating the marginal contribution value of each agent to the total collaborative benefit brought about by executing the initial logistics resource scheduling scheme based on the feedback information corresponding to each agent. Based on the marginal contribution value, a profit distribution simulation is performed among multiple agents to obtain the profit distribution simulation result, and it is determined whether the profit distribution simulation result meets the preset consensus conditions. If the consensus conditions are met, the current logistics resource scheduling scheme will be determined as the target logistics resource scheduling scheme. If the consensus conditions are not met, at least one of the task assignment, resource allocation, or cost sharing clauses in the current logistics resource scheduling scheme shall be modified according to the profit distribution simulation results to obtain the logistics resource scheduling scheme for the next round of negotiation. Send the logistics resource scheduling sub-schemes associated with each of the intelligent agents in the next round of negotiated logistics resource scheduling scheme to the corresponding intelligent agents to obtain the new round of feedback information returned by the intelligent agents; Return to the step of calculating the marginal contribution value, and use the logistics resource scheduling scheme of the next round of negotiation as the new logistics resource scheduling scheme of the current round for a new round of evaluation, until the profit distribution simulation result of the current round meets the consensus condition or the negotiation round reaches the preset upper limit; The logistics resource objects are scheduled and controlled according to the target logistics resource scheduling scheme.

2. The method according to claim 1, characterized in that, The logistics resources include at least one of the following: transport vehicles, storage space, sorting equipment, and loading and unloading equipment. The step of obtaining feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme includes: Each resource provider compares the logistics task requirements in the logistics resource scheduling sub-scheme with its own current resource usage constraints to determine whether there are any constraint conflicts; wherein, the resource usage constraints include at least one of time window constraints, resource capacity constraints, equipment status constraints, and personnel skill constraints. If no constraint conflict exists, the total resource cost required to execute the logistics resource scheduling sub-plan is calculated by each of the resource providers based on the resource cost information; wherein, the resource cost information includes at least one of fixed cost allocation, variable cost rate, and dynamic adjustment coefficient; Each of the resource providers generates its own feedback information, which includes at least one of the following: total price, unit price, or conditional commitments, based on the total resource cost and the resource provider's pricing strategy.

3. The method according to claim 1, characterized in that, The real-time operational data includes external environmental data, which includes at least one of real-time traffic flow data and weather warning data; the step of planning logistics resource allocation based on the real-time operational data in response to the logistics resource occupancy request includes: Real-time operational data containing the external environment data is input into a pre-trained risk prediction model to obtain the risk interference probability related to the logistics resource occupation request; wherein, the risk prediction model is a long short-term memory network model; the long short-term memory network model is used to predict the supply chain interruption risk within a future preset time window based on historical operational data and real-time external environment data; Based on the aforementioned risk interference probability, the expected travel time of relevant logistics paths or the availability and reliability of logistics nodes in the initial logistics resource scheduling scheme are adjusted using a weighted average.

4. The method according to claim 1, characterized in that, The step of scheduling and controlling the logistics resource object according to the target logistics resource scheduling scheme includes: The target logistics resource scheduling scheme and the decision-making basis for the corresponding scheduling scheme are displayed in the preset interactive interface; the decision-making basis for the scheduling scheme includes cost-benefit comparison and / or reasons for scheme selection obtained from the feedback information of each resource provider; The interactive interface receives confirmation or adjustment commands for the target logistics resource scheduling scheme. The target logistics resource scheduling scheme is modified according to the adjustment instruction.

5. A logistics supply chain management device, characterized in that, The device operates on a cloud server; Logistics resource objects are associated with locally deployed edge computing nodes; The edge computing node is communicatively connected to the cloud server; the device includes: The response module is used to respond to logistics resource occupancy requests and obtain real-time operational data of logistics resource objects provided by multiple resource providers, including: The raw operational data collected by IoT devices connected through the edge computing node is cleaned to obtain the real-time operational data stream. The edge computing node uploads the real-time operational data stream to the cloud server; wherein, if the edge computing node detects that the real-time operational data meets the preset local emergency rules, the edge computing node generates and executes local control instructions for the associated logistics resource object according to the local emergency rules; The planning module is used to plan the allocation of logistics resources based on the real-time operating data in response to the logistics resource occupancy request, and to obtain the initial logistics resource scheduling scheme corresponding to the logistics resource occupancy request. The sending module is used to send the logistics resource scheduling sub-schemes associated with each of the resource providers in the initial logistics resource scheduling scheme to the corresponding resource providers respectively; The acquisition module is used to acquire feedback information returned by each of the resource providers for the logistics resource scheduling sub-scheme; wherein, the feedback information is obtained by each of the resource providers through feasibility assessment and resource scheduling cost calculation of the received logistics resource scheduling sub-scheme based on their own resource cost information and resource usage constraints; The adjustment module is used to adjust the initial logistics resource scheduling scheme according to the feedback information returned by all resource providers, so as to obtain the target logistics resource scheduling scheme corresponding to the logistics resource occupation request. The adjustment module includes: modeling each resource provider as an agent in a cooperative game, and calculating the marginal contribution value of each agent to the total collaborative benefit brought about by the execution of the initial logistics resource scheduling scheme based on the feedback information corresponding to each agent. Based on the marginal contribution value, a profit distribution simulation is performed among multiple agents to obtain the profit distribution simulation result, and it is determined whether the profit distribution simulation result meets the preset consensus conditions. If the consensus conditions are met, the current logistics resource scheduling scheme will be determined as the target logistics resource scheduling scheme. If the consensus conditions are not met, at least one of the task assignment, resource allocation, or cost sharing clauses in the current logistics resource scheduling scheme shall be modified according to the profit distribution simulation results to obtain the logistics resource scheduling scheme for the next round of negotiation. Send the logistics resource scheduling sub-schemes associated with each of the intelligent agents in the next round of negotiated logistics resource scheduling scheme to the corresponding intelligent agents to obtain the new round of feedback information returned by the intelligent agents; Return to the step of calculating the marginal contribution value, and use the logistics resource scheduling scheme of the next round of negotiation as the new logistics resource scheduling scheme of the current round for a new round of evaluation, until the profit distribution simulation result of the current round meets the consensus condition or the negotiation round reaches the preset upper limit; The scheduling module is used to schedule and control the logistics resource objects according to the target logistics resource scheduling scheme.

6. A computer device comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images