Elastic evaluation and reconstruction method for burst disturbance by full-link logistics network

Abstract

The invention belongs to the technical field of logistics management, supply chain optimization and resource scheduling, and particularly discloses and provides an elastic assessment and reconstruction method for burst disturbance of a full-link logistics network, which comprises the following steps: acquiring real-time service data, and constructing a physical state layer and virtual state layer coupled dual-state network model; mapping the physical operation parameters to a virtual state layer to generate an elastic potential energy field; attack defense game deduction is carried out in combination with the disturbance simulation parameters, and a game evaluation result containing a service retention index is generated; executing potential energy gradient offset, and generating a potential energy gradient field for guiding the flow direction of the task; analyzing the field extraction path guide vector, generating a reconstruction execution instruction in combination with task constraint information, and issuing the reconstruction execution instruction to a physical terminal through a communication gateway; and finally, performing closed-loop tuning on the evaluation parameters based on actual feedback data. According to the method, global risk identification and microscopic resource self-organization repair are realized, and the recovery efficiency and stability of the logistics network in a sudden change environment are improved.

Description

Technical Field

[0001] This invention belongs to the field of logistics management, supply chain optimization and resource scheduling technology, and relates to a method for elastic assessment and reconstruction of a full-link logistics network in response to sudden disturbances. Background Technology

[0002] As the core carrier system of the supply chain, the logistics network is a complex spatiotemporal system composed of nodes and routes. During the entire supply chain operation, resilience assessment refers to the network's ability to absorb disturbances, maintain core functions, and quickly recover to a stable state when subjected to shocks such as traffic disruptions, natural disasters, or sudden control measures. With the continuous expansion of modern logistics scale and the increasing demands for order fulfillment timeliness, how to measure the anti-interference capability of the entire supply chain in real time amidst complex network dynamic changes and execute intelligent path and resource reorganization has become a key research focus in the field of logistics resource optimization and scheduling.

[0003] Existing logistics network management solutions mostly employ redundant configurations based on fixed routes, reserving a certain proportion of spare capacity and warehousing inventory during the planning phase, and relying on static risk warning indicator systems for management. When emergencies occur, the central dispatch system typically obtains information about localized failures, then uses mathematical optimization models to reroute affected orders, or manual intervention to issue temporary dispatch instructions based on experience. Some technical methods also involve offline assessments of the vulnerability of specific areas through simulation to provide a reference for infrastructure layout.

[0004] The aforementioned existing technologies still have significant limitations when facing sudden, transient disturbances. Fixed paths and redundant configurations often incur additional costs in daily operations and are difficult to flexibly mitigate risks across the entire chain using static rules when subjected to unexpected shocks. Traditional management systems have low coupling between physical data and decision-making logic, making it impossible to dynamically adjust the risk weights of each link based on real-time traffic load and topology evolution. When disturbances trigger large-scale chain reactions, the central computing module often faces enormous computational pressure due to the lack of a global and guiding elastic model. This results in significant delays in the generated reconstruction instructions, hindering the self-organizing repair of resources, and limiting decision-making accuracy to the completeness of historical data. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution: a method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances, comprising: S1, acquiring real-time multi-source business data of the full-link logistics network to obtain network status data.

[0006] S2. Based on network state data, extract the topological relationships of physical entities and their corresponding processing capacity boundary parameters to construct a physical state layer, and synchronously generate a virtual state layer mirror that is linked with the state of the physical state layer in real time based on a digital mapping mechanism, thereby coupling to obtain a dual-state network model.

[0007] S3. Extract the real-time operating parameter set of each entity in the physical state layer and map it to the virtual state layer for weighted calculation to generate an elastic potential energy field for measuring the risk pressure of network elements.

[0008] S4. Obtain disturbance simulation parameters that characterize the sudden impact, combine them with the elastic potential energy field to conduct attack and defense game simulation, and generate game evaluation results that include business retention indicators.

[0009] S5. Based on the game evaluation results, perform gradient offset processing on the elastic potential energy values ​​in the elastic potential energy field to generate a potential energy gradient field that guides the flow of logistics tasks.

[0010] S6. Analyze the potential energy gradient field to extract the path guidance vector, and match it with the logistics task constraint information retrieved from the management system to generate a reconstruction execution instruction that includes path replanning instructions and resource scheduling instructions.

[0011] S7. The reconfiguration execution command is sent to each physical execution terminal in the end-to-end logistics network through the communication gateway.

[0012] (1) Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention realizes the real-time digital mapping of the operation status and elastic potential of the logistics network by constructing a dual-state network model with deep coupling between the physical state layer and the virtual state layer. This mechanism transforms the complex physical network fluctuations into a quantifiable elastic potential field, solving the problems of single evaluation dimensions and insufficient timeliness in traditional management methods, enabling managers to identify vulnerable nodes in the entire link from the dual dimensions of topology and real-time load, thereby improving the granularity of network risk identification.

[0013] (2) The attack-defense game theory deduction mechanism and potential gradient guidance strategy introduced in this invention improve the scientific nature and execution speed of reconstruction decisions. Compared with traditional experience adjustment or global re-optimization, this solution guides the task flow to migrate spontaneously to the robust region through the gradient descent principle and triggers resource cluster instructions at key locations, forming a distributed and centralized collaborative response mode, shortening the business recovery cycle after sudden disturbances and reducing the impact of disturbances on the overall operation status.

[0014] (3) The closed loop of reconstruction deviation assessment and online parameter optimization established in this invention enables the system to have self-evolution capabilities. By continuously comparing actual business retention indicators with predicted indicators, the assessment weights and strategy parameters can be dynamically adjusted, thereby ensuring that the assessment model after continuous iteration can better fit the complex real-world working conditions. This dynamic adjustment mechanism based on data feedback ensures the accuracy of reconstruction instructions and provides a solid technical guarantee for the long-term stability of the entire logistics network in a changing environment. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 As shown, the present invention proposes a method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances, including: S1, acquiring real-time multi-source business data of the full-link logistics network to obtain network status data.

[0019] In practical implementation, the system establishes data communication links with various physical execution terminals (such as automated sorting equipment and vehicle-mounted GPS terminals) and business management platforms (such as warehouse management systems (WMS) and transportation management systems (TMS)) in the entire logistics network through a communication gateway. The system acquires multi-source business data, including real-time throughput of nodes, equipment operating load, warehouse space occupancy rate, and spatiotemporal trajectory of transportation vehicles, according to a preset collection frequency or a real-time push mechanism triggered by business events.

[0020] To ensure data consistency in subsequent modeling, the system performs data cleaning and timestamp alignment on the acquired raw business data streams. Specifically, the system uses a preset threshold range to remove outliers caused by sensor malfunctions or communication jitter, and uses the system's unified time synchronization server as a benchmark to synchronize and correct the timestamps of business messages from different data sources. Through these preprocessing actions, the system transforms the loose raw data into structured network state data with a unified time base and standard field format, providing real-time and accurate data input for the subsequent construction of the physical state layer.

[0021] S2. Based on network state data, extract the topological relationships of physical entities and their corresponding processing capacity boundary parameters to construct a physical state layer, and synchronously generate a virtual state layer mirror that is linked with the state of the physical state layer in real time based on a digital mapping mechanism, thereby coupling to obtain a dual-state network model.

[0022] In a preferred embodiment, the dual-state network model includes the following construction steps: extracting the entity connection relationship matrix and processing capacity boundary parameters from the network state data, and constructing the physical state layer topology; Perform a graph isomorphism transformation on the physical state layer topology and generate a corresponding mirror mapping structure in the computing space to obtain a virtual state layer mirror; An asynchronous message communication bus is established between the physical state layer topology and the virtual state layer mirror to form a dual-state network model with real-time data synchronization capability.

[0023] In practical implementation, the system uses a pre-defined data adapter to parse the network state data obtained in step S1 into a structured dataset in a unified format, and maps it to a graph computation space composed of nodes and edges. Based on the mapping rule table between physical business indicators and virtual elastic variables, the system establishes a bidirectional data synchronization mechanism between the physical and virtual states. During the construction process, the system extracts entity attributes of key logistics nodes and channels from the structured data, including node identifiers, spatial coordinates, and rated capacity; based on the business flow relationships between nodes, it establishes an adjacency topology matrix containing directional and weight attributes.

[0024] The entity attributes in the engineering operation environment include the set hourly throughput safety range and the rated load range of the transport vehicles. The system acquires real-time load information for each node based on data transmitted back from the IoT gateway and aggregated from the business platform, constructing a physical state layer that represents the dynamic operation of the real world. Next, the system triggers a memory mirroring program, replicating the directed network topology of the physical state layer according to the principle of graph isomorphism, mapping and generating a virtual state layer with a corresponding topology. This computational action establishes a parallel abstract computing space to carry the elastic feature quantification data after filtering out unstructured data from the physical layer.

[0025] To complete this low-level numerical mapping, the system performs algebraic operations on each network element within the virtual state layer: In the formula, This represents the initial resilience potential mapping characteristic value assigned to the i-th network element in the virtual state layer. A larger value indicates higher resource redundancy and structural robustness of the node in its initial state. i is a unique sequence index of the network node, used to characterize the one-to-one mapping relationship between virtual state layer network elements and physical state layer entities; its value ranges from 1 to... , This represents the total number of network nodes. This represents the theoretical maximum design processing capacity boundary value of the physical state layer entity mapped to the i-th network element. It is set based on the device hardware parameters or historical peak throughput. For example, the maximum number of items that can be processed per hour for a certain sorting center is set to 5000 items / hour. This represents the actual load data value of the physical state layer entity mapped to the i-th network element at the current acquisition time, obtained based on real-time business flow data. For example, the current actual number of processed items is 2000 items / hour. This represents the global structural importance of the i-th network element in the topology; it is obtained by performing a betweenness centrality algorithm on the physical connection topology matrix and then performing normalization, with a value range of [0,1]. The value is set based on the core position of the node in the network path, such as the core hub node. Set to 0.85. and These are preset normalized weighting coefficients, whose values ​​are set according to the sensitivity of the logistics chain to load fluctuations, satisfying... For example, in load-sensitive networks, set , 0.3.

[0026] After calculation, a bidirectional real-time driving relationship based on asynchronous message queue middleware is established between the physical layer node storage area and the virtual layer computing area. This enables the business flow and state evolution data of the physical layer to push update messages within a preset communication delay range, driving the iterative update of the virtual layer's computing core. Simultaneously, the system packages the optimization guidance and evaluation prediction conclusions periodically generated by the virtual layer into standardized protocol instruction frames, which are then fed back to the lower-level physical layer scheduling engine system for real-time vehicle allocation and business flow restriction scheduling decisions.

[0027] This invention, through a system that sequentially executes entity extraction, graph topology mapping, attribute algebra calculation, and cross-layer communication bus deployment steps on collected network state data within a computing space, binds the real-world logistics operation with the digital space used for optimization algorithms. This achieves a closed-loop functional operation that provides real-time feedback on underlying anomalies and guidance from the upper layers. The aforementioned action flow structurally assembles the underlying data, ultimately generating a bi-state network model that carries a bilateral dynamic interlocking mechanism, serving as the chassis component supporting subsequent stress performance calculations for the entire logistics network.

[0028] S3. Extract the real-time operating parameter set of each entity in the physical state layer and map it to the virtual state layer for weighted calculation to generate an elastic potential energy field for measuring the risk pressure of network elements.

[0029] In a preferred embodiment, the real-time operating parameter set of each entity in the physical state layer is extracted and mapped to the virtual state layer for weighted calculation to generate an elastic potential energy field for measuring the risk pressure of network elements. This includes: obtaining the passage time ratio, equipment load ratio and backup path distribution parameters within the geographical proximity range of each entity in the physical state layer to obtain the real-time operating parameter set. Normalization mapping is performed on the real-time running parameter set to obtain quantized feature vectors of different dimensions; By using preset weighting coefficients that characterize the degree of security sensitivity, each quantized feature vector is fused and accumulated to generate an elastic potential energy field mapped onto the virtual state layer image.

[0030] In practice, the system retrieves real-time operating parameters of each network element through the interface between the underlying sensors of the physical layer and the business system, constructing a real-time operating parameter set. This parameter set includes: Travel time ratio The coefficient of friction in a specific transportation section is determined by the latitude and longitude change rate transmitted back by the logistics trajectory tracking system. It is calculated by dividing the actual travel time of the section obtained by real-time monitoring by the nominal ideal travel time preset in that section.

[0031] Equipment load ratio Used to quantify the load saturation level of each physical warehouse or transit node, it is determined by the current real-time storage space occupancy rate of the warehouse execution system. It is calculated by dividing the length of the queue of parcels to be processed in the logistics processing pipeline by the maximum concurrent throughput limit of the rated design of the hardware equipment.

[0032] Backup path distribution parameters It is used to quantify the fault tolerance probability after node failure from two dimensions: topology and resource reserve. It is obtained by statistically analyzing the proportion of available parallel branches and the proportion of remaining throughput capacity within a specific geographical radiation range, and then performing a weighted fusion operation.

[0033] To unify physical indicators with different dimensions, the system performs linear normalization mapping on the above three parameters, transforming them into dimensionless characteristic values ​​located in the [0,1] interval. Subsequently, the normalized values ​​are substituted into a preset potential energy calculation rule for algebraic weighted calculation, as shown in the following formula: In the formula, This represents the calculated elastic potential energy value of a specific network element. and and The weighting coefficients configured for the system are determined by sensitivity analysis based on the impact weights of each dimension of the logistics link on the total timeliness, and satisfy the constraints. For example, in a load-sensitive link, set , , Due to path redundancy A higher value indicates a stronger ability of the node to withstand risks, hence the formula uses... Perform reverse logic processing, so that , and The positive growth of the values ​​indicates an increase in risk pressure. The system backfills the elastic potential energy values ​​of each network element into the topology matrix structure of the virtual state layer and refreshes it periodically according to a preset time step, thereby generating an elastic potential energy field that changes synchronously with the overall network situation, and quantifying the sudden pressure state of each link.

[0034] The purpose of this step in the project is to transform the scattered operational indicators into a continuous potential energy and pressure space. By organizing all the calculated elastic potential energy values ​​spatially according to their physical connections, a dynamic evaluation base map covering the entire network is formed, enabling real-time visibility and quantification of the sudden pressure state of each link.

[0035] In a further preferred embodiment, the distribution parameters of the backup pathways within the geographic proximity of the target node are obtained through the following steps: obtaining the geographic coordinates of the target node and extracting the preset detection radius for that node type; The number of alternative connected branches within the detection radius that lead to the same logical target and have a passage efficiency higher than a preset threshold is counted, and then divided by the theoretical total number of outgoing paths of the target node to obtain the channel richness feature. The real-time remaining throughput capacity of each alternative connecting branch is summarized and divided by the nominal design capacity of the target node to obtain the path redundancy ratio. By using preset weighting coefficients, channel richness features and path redundancy ratios are weighted and fused to generate backup path distribution parameters.

[0036] In practical implementation, the system performs geographic rasterization processing on the spatial topology associations of nodes in the end-to-end logistics network, thereby achieving automated collection and feature quantification of backup path distribution parameters. First, the system retrieves the latitude and longitude attribute coordinates stored in the topology database for each target network element within the virtual state layer. Using these coordinates as the center, the system sets a kilometer-level radius value as the geographic radiation range in the geographic information system. This radius value is dynamically configured within a certain kilometer range based on the node level. The system performs spatial intersection operations on the link detection data of the physical state layer to filter out paths located within the detection radius that lead to the same logical downstream target. Further, the system performs status verification on the filtered paths, marking paths whose ratio of real-time travel time to nominal ideal travel time is less than a preset threshold (e.g., 1.5) as alternative connecting branches. The system interacts with the business management system's data interface to obtain the nominal design capacity and real-time remaining throughput capacity of each alternative connecting branch.

[0037] To quantify the redundant load-bearing capacity of the surrounding structure after the failure of a target network element, the system uses the following dimensionless composite formula to solve for the distribution parameters of the backup path: In the formula, This represents the distributed parameters of the calculated alternative pathways. This represents the number of alternative connecting branches that are ready, based on real-time statistics. This represents the maximum theoretical total number of outgoing paths for the target network element in the original topology plan, determined by the node's logical link quota, for example, a value set to 5. By normalizing the quantitative dimension, the richness characteristics of the logical connections of the nodes are reflected. This represents the sum of the remaining available processing capacity of all identified alternative connectivity branches. This value is obtained by performing arithmetic summation on the real-time remaining throughput capacity of the corresponding lines in the physical state layer. The nominal total output throughput capacity of the target network element is determined by the rated parameters of the hardware device and is used to measure the potential acceptance ratio of the traffic of the node by surrounding resources. As a path redundancy ratio, it reflects the potential probability that surrounding paths will accept spillover traffic on a physical scale. and This represents the re-adjustment coefficient, preset based on the node's sensitivity to topological or capacity redundancy, to satisfy... And all are within the (0,1) interval, for example, set , .

[0038] The system ultimately synchronizes the calculated backup path distribution parameters to the elastic potential field calculation module in real time via a high-speed data bus, serving as a quantitative factor characterizing the node's risk buffering capacity. This engineering action deeply couples the static geographical layout advantages with the dynamic availability of remaining resources, enabling the resilience assessment process to accurately reflect the node's structured fault tolerance capability when encountering sudden disturbances, and providing the necessary structural resilience data support for generating scientific reconfiguration instructions.

[0039] S4. Obtain disturbance simulation parameters that characterize the sudden impact, combine them with the elastic potential energy field to conduct attack and defense game simulation, and generate game evaluation results that include business retention indicators.

[0040] In a preferred embodiment, disturbance simulation parameters characterizing the sudden impact are obtained, and attack and defense game simulation is performed in combination with the elastic potential energy field to generate game evaluation results containing business retention indicators. This includes mapping the list of affected nodes, expected failure duration, and processing capacity loss ratio contained in the disturbance simulation parameters to the corresponding nodes in the physical state layer and converting them into corresponding processing capacity loss components to simulate the attacker's behavior.

[0041] Select a set of defense strategies from the reconstructed contingency plan library, which includes route detour ratios and emergency capacity compensation rules, as the defense strategy set; In a dual-state network model, the impact of the attacker's behavior on the elastic potential field is simulated, and each strategy in the set of defense strategies is applied iteratively. The business retention index after the application of each strategy is calculated, and the game evaluation results are summarized.

[0042] In practical implementation, the system simulates the instantaneous impact of disturbance sources on the network topology in a bi-state network model, thereby completing the quantitative ranking and vulnerability analysis of multiple alternative solutions in virtual space. First, the system retrieves a preset sudden disturbance event and parses it into disturbance simulation parameters including a list of affected nodes, expected failure duration, and the proportion of processing capacity loss. Then, the system extracts multiple sets of predefined defense strategies from the reconfiguration plan library. Each set of defense strategies represents a logical set covering path detour ratios and emergency capacity compensation rules.

[0043] The system performs simulations for each set of strategies, mapping real-time business flow data to the physical state layer and calculating the order flow trajectory under each strategy intervention. During this process, the system synchronously updates the elastic potential energy field values ​​in the virtual state layer. When load shifting causes changes in traffic on surrounding lines, the system calculates and updates the elastic potential energy values ​​of each network element in real time based on the ratio of the node's current load to its rated capacity.

[0044] The system iterates through and calculates the business retention indicators corresponding to all candidate defense strategies, and selects the strategy with the highest retention rate as the game evaluation result through numerical comparison. In this evaluation result, the system performs hierarchical quantitative extraction of affected nodes based on the dynamic evolution characteristics of the elastic potential energy field. Specifically, the system calculates the gradient change rate of the potential energy field of each node at the initial stage of physical impact and automatically compares it with a preset structural failure threshold. If the gradient change rate of a node exceeds the structural failure threshold, it indicates that the instantaneous load-bearing capacity of the node has exceeded its processing capacity boundary parameter, causing a precipitous drop in energy or direct physical paralysis. The system defines such nodes as primary weak points. At the same time, for the surrounding area that has not been directly paralyzed, the system tracks the diffusion trend of its energy over time and marks affected nodes whose potential energy change rate exceeds a preset fluctuation warning threshold as potential secondary weak points. Finally, this game evaluation result serves as the decision input for the downstream reconstruction execution stage, providing not only the optimal tactical path parameters but also a structured perspective from local damaged points to the global impact domain through hierarchical location of weak points.

[0045] In a further preferred embodiment, the business retention metric includes the following generation steps: collecting the expected delivery completion rate of tasks after simulating the application of various strategies, and obtaining the task completion time retention rate; The field stability parameters are obtained by statistically analyzing the standard deviation of potential energy fluctuations at all nodes in the elastic potential energy field over the simulation period. The retention rate upon task completion and the stability parameters are weighted and scored to generate business retention metrics.

[0046] In practical implementation, after completing simulations of different defense strategies, the system calculates business retention metrics by integrating performance data from the business side and pressure distribution data from the topology side, in order to rank the current game evaluation results. For each simulated path planning during the game evaluation process, the system retrieves the estimated delivery time of the generated logistics tasks and, combined with baseline delivery time data under normal conditions, calculates the task completion time retention rate. In engineering implementation, the task completion time retention rate characterizes the business system's ability to maintain promised timeliness under disturbances. Simultaneously, the system performs spatiotemporal standard deviation analysis on the elastic potential energy field in the virtual state layer affected by traffic redistribution. By extracting the dispersion evolution trend of the elastic potential energy values ​​of all network nodes, it calculates the field stability parameter. The field stability parameter measures whether large-scale task replanning leads to the severity of secondary pressure fluctuations; it is obtained by statistically analyzing the standard deviation of potential energy fluctuations of all active nodes within the simulation period and then normalizing it.

[0047] To balance business efficiency and network topology stability, the system uses the following calculation method to generate business retention metrics: In the formula, Represents business retention metrics. This represents the task completion time retention rate, which is obtained by the ratio of the number of on-time orders output by the simulator to the total number of orders. For example, the value is set to 0.85. The representative field stability parameter, namely the spatial standard deviation of the elastic potential energy values ​​of all network nodes, reflects the spatial distribution balance of pressure across the entire network. This represents the preset baseline coefficient for network topology complexity, which is set based on the network node size and the historical maximum allowable potential energy fluctuation amplitude, for example, a value set to 100. and This represents the preset evaluation weighting coefficients, set based on the business's sensitivity to structural security, and The basis for this setting is the business's ability to withstand structural risks: for trunk hub networks, increase... The weights, such as Prioritize ensuring the structure does not collapse; for the last-mile delivery network, increase... The weights, such as To prioritize ensuring timely delivery.

[0048] The system iterates in real-time and compares all candidate defense strategies based on the business retention metrics calculated using the above formula, ultimately selecting the strategy with the highest score as the optimal defense strategy. Because this evaluation system successfully couples macro-level business performance goals (reflected in task completion time retention rate) with micro-level topological pressure states (reflected in field stability parameters) to a unified decision-making dimension, the system does not require cumbersome switching of the underlying evaluation model. Instead, the system only needs to adjust the weighting coefficients as described above. and Through scenario-based dynamic adjustment, the optimal self-healing path can be accurately identified in complex network reconstruction games, ensuring both service timeliness and effective prevention of local network structural collapse. Ultimately, this optimal defense strategy and its corresponding evaluation parameters are incorporated into the overall game evaluation results.

[0049] In a further preferred embodiment, after generating the game evaluation result including business retention indicators, the method further includes: identifying the affected nodes in the game evaluation result whose potential energy change rate exceeds a preset fluctuation warning threshold, and defining them as secondary weak points. Search for the coordinates of idle resources within the preset coverage radius of secondary weak points, and generate resource activation requirement information; Based on the resource activation demand information, an early warning and standby command is issued to the terminal devices in the affected area.

[0050] In practical implementation, the system proactively controls the potential risk diffusion trend output from the game evaluation results, thereby completing the redundant configuration of emergency resources before the physical impact arrives. In the specific engineering implementation logic, the system first analyzes the elastic potential energy field evolution data mapped by the game evaluation results and calculates the potential energy change rate of each node within a preset time window. The system compares these change rates with a preset fluctuation warning threshold, identifying indirectly affected nodes (i.e., affected nodes) where stress surges due to traffic redirection and transfer cause the potential energy change rate to exceed the warning threshold, and uniformly marks these nodes as secondary vulnerable points. Subsequently, the system retrieves vacant transportation capacity and temporary storage space within the preset coverage radius of each secondary vulnerable point through a geographic information system interface, extracts their current latitude and longitude coordinates and readiness status, and obtains a set of backup resource distributions. For each backup resource in the set, the system calculates the activation weight parameter between it and the target secondary vulnerable point. .

[0051] To ensure the balance of decision-making logic across dimensions, the system uses the following dimensionless activation evaluation formula for quantitative judgment: In the formula, This represents the activation weight parameter for a specific backup resource, used to determine the priority of resource allocation. This represents the maximum processing increment that the backup resource can provide per unit time. It is obtained based on the rated hardware configuration or load parameters of the backup resource read by the system, for example, the value is set to 500 pieces / hour. The rate at which new packages are predicted to flow into the target's secondary weak points in the game simulation is determined by the flow change value output by the simulator, for example, a value set to 2000 packages / hour. It is used to characterize the resource's ability to meet load gaps, and its value is between [0,1]. This represents the preset maximum road search distance, serving as a boundary condition and normalization benchmark for selecting backup resources; for example, the value is set to 50 kilometers. The measured road distance between the current location of the backup resource and the target's secondary weak point is obtained through a geographic information system interface and meets the following requirements. For example, the value is set to 15 kilometers. The remaining distance percentage reflects the geographical proximity of resources. and This represents the preset weighted distribution coefficient, set according to the business's preference for "processing capacity" and "response distance," with a value within the range [0,1] and satisfying the following conditions: For example, set to , .

[0052] The system compares the calculated activation weight parameters with a preset response threshold. When the calculated activation weight parameters exceed the preset response threshold, the system generates a control message containing a task number and warning level, and sends it to the corresponding execution terminal via the communication network. Upon receiving the message, the execution terminal switches its execution state, such as putting sorting equipment into warm-up mode or transport vehicles into standby mode, to complete the pre-configuration of emergency resources. This engineering action forms a proactive defense barrier in the entire logistics network, ensuring that secondary weak points have pre-set hedging capabilities the moment they sense upstream pressure fluctuations, achieving the engineering effect of using digital predictive information to exchange for physical pressure resistance reserves.

[0053] S5. Based on the game evaluation results, perform gradient offset processing on the elastic potential energy values ​​in the elastic potential energy field to generate a potential energy gradient field that guides the flow of logistics tasks.

[0054] In a preferred embodiment, generating a potential energy gradient field that guides the flow of logistics tasks includes: analyzing the game evaluation results, extracting the load surge component of the affected node within the prediction time window, and combining it with the corresponding processing capacity boundary parameters in the physical state layer to calculate the potential energy offset increment that reflects the node congestion risk trend. The potential energy offset increment is superimposed on the current elastic potential energy value of the corresponding node in the virtual state layer image, and the preset nonlinear mapping function is called to perform numerical correction on the potential energy distribution of the entire network, so as to obtain a corrected elastic potential energy distribution with congestion warning characteristics.

[0055] For the modified elastic potential energy distribution, the potential energy spatial derivative between adjacent nodes on each topological path is calculated, the gradient guidance vector representing the flow of logistics tasks to the optimal descent path is extracted, and a potential energy gradient field covering the entire link and having the function of task migration guidance is generated.

[0056] In practical implementation, the system maps the predicted pressure obtained from game theory back to the virtual potential energy space, thereby constructing a digital force field with spontaneous guidance, achieving a logical leap from "passive evaluation" to "active guidance." The system first analyzes the game evaluation results, identifies the task accumulation trends of primary and secondary weaknesses in the future prediction time window through time series analysis, and calculates the difference between the predicted peak load and the current baseline load to extract the predicted load surge. The system compares this surge with the processing capacity boundary parameters of nodes in the physical state layer, calculates the difference or overflow ratio to quantify the node's overload level, and combines it with a preset energy conversion coefficient to convert this overload level into a potential energy offset increment used to simulate "virtual congestion."

[0057] To ensure that the generated gradient field can smoothly guide logistics tasks to low-risk areas, the system uses the following offset correction formula to numerically update the virtual state layer image: In the formula, This represents the corrected elastic potential energy value of the i-th network element after gradient offset processing. This represents the current real-time elastic potential value of the i-th network element. This represents the predicted load surge of the i-th network element under the influence of the disturbance in the game evaluation results. This represents the theoretical maximum design processing capacity boundary value of the physical state layer entity mapped to the i-th network element. This represents the actual load data value of the physical state layer entity mapped to the i-th network element at the current acquisition time. Represents the remaining carrying space of the i-th network element. This is a preset gradient sensitivity adjustment coefficient used to control the driving force of the offset on model reconstruction. For a preset, extremely small positive constant, such as 10 −6 This formula uses a logarithmic function to ensure that when a node approaches its capacity limit, its corrected elastic potential energy value will rise sharply, thus logically forming a "high potential energy barrier".

[0058] Subsequently, for node pairs (i,j) with logical connections across the entire network, the system calculates their spatial derivative vectors based on the modified elastic potential energy distribution to obtain the potential energy gradient field: In the formula, This represents the potential gradient vector between node i and node j. This represents the corrected elastic potential energy value of the j-th network element after gradient offset processing. This represents the logical topological distance or physical path length between two nodes. This represents the unit direction vector pointing from node i to node j.

[0059] Through the aforementioned algebraic operations, the system generates a potential energy gradient field covering the entire link in the virtual state layer mirror. This gradient field not only reflects the current physical load but also incorporates the risk evolution trend after game theory deduction. The significance of this engineering action lies in the fact that it provides a clear "pressure reduction" guide for path replanning in subsequent steps, enabling logistics tasks to spontaneously avoid high-load areas and migrate to high-redundancy areas like fluids. Thus, it achieves self-organized reconstruction of network resources under local disturbances without relying on large-scale global recalculation.

[0060] S6. Analyze the potential energy gradient field to extract the path guidance vector, and match it with the logistics task constraint information retrieved from the management system to generate a reconstruction execution instruction that includes path replanning instructions and resource scheduling instructions.

[0061] In a preferred embodiment, the potential energy gradient field is parsed to extract the path guidance vector, and the logistics task constraint information retrieved from the management system is matched to generate a reconstruction execution instruction containing path replanning instructions and resource scheduling instructions, including: extracting the potential energy decrease vector of each connecting edge in the potential energy gradient field to obtain the path reconstruction guidance direction; By combining the business delivery time boundary values ​​in the logistics task constraint information, the feasibility of the path reconstruction guidance direction is filtered, and an initial version of the replanning path is generated. The node processing capacity load of the initial replanning path is verified. Under the premise of meeting the processing capacity boundary parameters in the dual-state network model, the reconstructing execution instructions are generated by combining them.

[0062] In practical implementation, the system extracts the spatial vector difference in the potential energy gradient field and transforms it into path evaluation indicators, thereby automatically generating traffic scheduling strategies while meeting business timeliness constraints. The system first executes a multi-dimensional analytical algorithm on the potential energy gradient field currently stored in the in-memory database. Utilizing the difference in elastic potential energy values ​​between adjacent network elements in the spatial coordinate system, it calculates the gradient direction of each connecting edge and identifies high-elasticity potential energy regions under high load pressure and low-elasticity potential energy regions with resource absorption capabilities. In an engineering environment, the system's trigger condition is the detection of a potential energy mutation value exceeding a preset safety tolerance threshold due to a sudden disturbance. The system then reads the preset task constraints for each logistics task to be forwarded. These constraints, contained in the system data frame, include the maximum allowable delivery time and the maximum allowable transshipment cost.

[0063] To balance the length of disaster avoidance paths with the quality of business performance during the reconstruction process, the system uses the following guiding function to calculate the optimal path: In the formula, The generated reconstruction guidance weight score represents the lower the overall resistance of the path, and the value is a positive real number. The modified elastic potential energy value representing the target node in the candidate path is calculated in step S5 and is used to characterize the predicted risk pressure level of the node, for example, the value is set to 75.5. This represents the highest potential energy benchmark value of the entire logistics network under normal conditions. It is set based on the peak potential energy in the network's historical operating data, for example, the value is set to 100. This reflects the relative risk level of the target node. The expected total transit time required for the simulated new route is calculated from the route length and the average speed of the vehicle, for example, a value set to 18 hours. This represents the maximum remaining available time for the task as stipulated in the service level agreement, i.e., the business delivery time boundary value, for example, a value set to 24 hours. It reflects the proportion of time consumed to the remaining available time. and The weighted sensitivity coefficient retrieved by the system has a value in the range [0,1] and satisfies the following conditions: The setting is based on the urgency of the task: for urgent tasks, increase... The weights, such as For routine tasks, a balanced selection of weights is used, such as... , .

[0064] The system iterates through and calculates the reconstruction guidance weight score of all candidate directions, selects the direction with the lowest value (i.e., the least resistance), plans a new diversion ratio and transfer node for the logistics task, and generates path switching and resource scheduling instructions containing detailed transfer coordinates and operation frequency.

[0065] Meanwhile, to further enhance throughput flexibility in low-elasticity potential energy areas, the system extracts the real-time coordinates and status of mobile assets such as idle vehicles and mobile loading / unloading teams in the vicinity of the physical state layer based on the convergence direction of the potential energy gradient field. The system then executes a resource organization algorithm on these assets, generating a control sequence that guides their rapid movement to key low-elasticity potential energy area coordinates; this sequence is the temporary resource cluster instruction. In this engineering action, the system forces assets to physically cluster together by issuing binary encapsulated instructions containing geofence boundary parameters and expected arrival time requirements to cope with surging transfer traffic.

[0066] The system ultimately encapsulates and combines the generated path switching and resource scheduling instructions, as well as temporary resource cluster instructions, according to a specific message sequence header, to generate reconfiguration execution instructions that can be parsed and executed by the automated warehouse management system, transportation execution system, and mobile terminals of each execution node. This step, by establishing a rigid mapping from potential energy gradient changes to physical asset flows, achieves efficient conversion from digital assessment results to on-site operational instructions, ensuring that the logistics network can spontaneously reconfigure its topology and compensate for capacity under disturbance conditions based on the principle of minimum potential energy.

[0067] S7. The reconfiguration execution command is sent to each physical execution terminal in the end-to-end logistics network through the communication gateway.

[0068] In a preferred embodiment, after the reconfiguration execution instruction is sent to each physical execution terminal in the end-to-end logistics network through a communication gateway, the method further includes: starting a monitoring process to obtain real-time operation messages from each execution terminal and obtaining actual business retention indicators. Calculate the deviation ratio between the actual business retention index and the predicted business retention index in the game evaluation results, and generate the reconstruction deviation evaluation index. Based on the parameter correction mechanism triggered by the reconstruction deviation evaluation index, the weighting coefficients used in calculating the elastic potential energy field are automatically optimized.

[0069] In practice, after issuing the reconfiguration execution command and following a preset task observation period, the system monitors the actual execution effect by collecting real-time job feedback messages from each physical node across the entire link. This allows the acquisition of post-execution network status data to assess the deviation rate of the reconfiguration decision. During project execution, the observation period is typically set between five and thirty minutes, depending on the average transfer granularity of the logistics link. Based on the newly acquired network status data, the system updates the bi-state network model and calculates the actual elastic potential energy distribution. Following the same calculation and quantification logic as the game theory evaluation phase, it calculates the actual business retention index reflecting the current true operational capability of the physical network. The system then retrieves the game theory evaluation results generated when the disturbance occurred from the in-memory database, extracts the corresponding predicted business retention index, and performs a difference comparison calculation between the two.

[0070] To quantify the effectiveness of the reconstruction scheme in real-world environments and identify overfitting or underfitting states of the prediction model, the system uses the following reconstruction bias evaluation formula for analysis: In the formula, The generated reconstruction deviation evaluation index is used to quantify the degree of fit between the prediction model and the actual working conditions. Its value is a positive real number, for example, a value set to 0.15. This represents the actual business retention metric. It is obtained by processing the actual observation and recovery data fed back by each execution terminal according to the calculation formula in step S4. The value range is [0,1]. For example, the value is set to 0.82. The actual observation and recovery data includes the actual number of successful orders, the actual transit time, etc. The value represents the predicted business retention metric, which is the predicted value output for the selected strategy in the game deduction stage in step S4. The value range is (0,1], for example, the value is set to 0.90. This represents the absolute dispersion between the predicted and actual values, expressed by dividing by... A dimensionless value reflecting the proportion of missing prediction accuracy is obtained.

[0071] The system combines the obtained reconstruction deviation evaluation index with the full dataset after execution and outputs a reconstruction quality assessment report using automated template generation technology. This report serves as a key variable in the system background to trigger optimization tasks. When the reconstruction deviation evaluation index exceeds a preset tolerance threshold, the system automatically activates the feedback learning module. This module performs online optimization based on gradient correction on the weighting coefficients used to generate elastic potential energy values ​​or the strategy parameters in game theory simulations, based on the abnormal potential energy mutation points recorded in the reconstruction quality assessment report. At the engineering level, this manifests as the system performing micro-step offset processing on the sensitivity operators of specific elements in the elastic potential energy field, thereby ensuring that the dual-state network model can generate predictive feedback that more closely matches actual physical conditions when facing similar sudden disturbance events. This closed-loop calibration mechanism, by establishing a negative feedback loop from actual deviation to parameter evolution, achieves a gradual enhancement of logistics resilience assessment capabilities, ensuring that the continuously optimized reconstruction scheme can accurately offset uncertainties in the real-world environment.

[0072] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for resilient assessment and reconstruction of a full-chain logistics network to sudden disturbances, characterized in that, include: S1. Obtain real-time multi-source business data of the entire logistics network to obtain network status data; S2. Based on network state data, extract the topological relationships of physical entities and their corresponding processing capacity boundary parameters to construct a physical state layer, and synchronously generate a virtual state layer mirror that is linked with the state of the physical state layer in real time based on a digital mapping mechanism, thereby coupling to obtain a dual-state network model. S3. Extract the real-time operating parameter set of each entity in the physical state layer and map it to the virtual state layer for weighted calculation to generate an elastic potential energy field for measuring the risk pressure of network elements. S4. Obtain disturbance simulation parameters that characterize the sudden impact, combine them with the elastic potential energy field to conduct attack and defense game simulation, and generate game evaluation results that include business retention indicators. S5. Based on the game evaluation results, perform gradient shift processing on the elastic potential energy values ​​in the elastic potential energy field to generate a potential energy gradient field that guides the flow of logistics tasks. S6. Analyze the potential energy gradient field to extract the path guidance vector, and match it with the logistics task constraint information retrieved from the management system to generate a reconstruction execution instruction containing path replanning instructions and resource scheduling instructions. S7. The reconfiguration execution command is sent to each physical execution terminal in the end-to-end logistics network through the communication gateway.

2. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 1, characterized in that, The construction steps of the dual-state network model are as follows: Extract the entity connection matrix and processing capacity boundary parameters from the network state data to construct the physical state layer topology; Perform a graph isomorphism transformation on the physical state layer topology and generate a corresponding mirror mapping structure in the computing space to obtain a virtual state layer mirror; An asynchronous message communication bus is established between the physical state layer topology and the virtual state layer mirror to form a dual-state network model with real-time data synchronization capability.

3. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 1, characterized in that, Extract the real-time operating parameter set of each entity in the physical state layer and map it to the virtual state layer for weighted calculation to generate an elastic potential energy field for measuring the risk pressure of network elements, including: The real-time operating parameter set is obtained by acquiring the passage time ratio, equipment load ratio, and backup path distribution parameters within the geographical vicinity of the target node for each entity in the physical state layer. Normalization mapping is performed on the real-time running parameter set to obtain quantized feature vectors of different dimensions; By using preset weighting coefficients that characterize the degree of security sensitivity, each quantized feature vector is fused and accumulated to generate an elastic potential energy field mapped onto the virtual state layer image.

4. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 3, characterized in that, The distribution parameters of alternative routes within the geographic proximity of the target node are obtained through the following steps: Obtain the geographic coordinates of the target node and extract the preset detection radius for that node type; The number of alternative connected branches within the detection radius that lead to the same logical target and have a passage efficiency higher than a preset threshold is counted, and then divided by the theoretical total number of outgoing paths of the target node to obtain the channel richness feature. The real-time remaining throughput capacity of each alternative connecting branch is summarized and divided by the nominal design capacity of the target node to obtain the path redundancy ratio. By using preset weighting coefficients, channel richness features and path redundancy ratios are weighted and fused to generate backup path distribution parameters.

5. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 1, characterized in that, Obtain disturbance simulation parameters characterizing sudden impacts, combine them with elastic potential energy fields to conduct attack and defense game theory simulations, and generate game theory evaluation results including business retention indicators, including: The list of affected nodes, the expected failure duration, and the proportion of processing capacity loss contained in the disturbance simulation parameters are mapped to the corresponding nodes in the physical state layer and transformed into corresponding processing capacity loss components to simulate the attacker's behavior. Select a set of defense strategies from the reconstructed contingency plan library, which includes route detour ratios and emergency capacity compensation rules, as the defense strategy set; In a dual-state network model, the impact of the attacker's behavior on the elastic potential field is simulated, and each strategy in the set of defense strategies is applied iteratively. The business retention index after the application of each strategy is calculated, and the game evaluation results are summarized.

6. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 5, characterized in that, The business retention metrics include the following generation steps: The expected delivery and completion rate of tasks after simulating the application of various strategies is obtained to achieve the task completion time retention rate. The field stability parameters are obtained by statistically analyzing the standard deviation of potential energy fluctuations at all nodes in the elastic potential energy field over the simulation period. The retention rate upon task completion and the stability parameters are weighted and scored to generate business retention metrics.

7. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 5, characterized in that, After generating the game theory evaluation results that include business retention metrics, it also includes: The affected nodes in the game evaluation results whose potential energy change rate exceeds the preset fluctuation warning threshold are defined as secondary weak points. Search for the coordinates of idle resources within the preset coverage radius of secondary weak points, and generate resource activation requirement information; Based on the resource activation demand information, an early warning and standby command is issued to the terminal devices in the affected area.

8. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 7, characterized in that, Generate a potential energy gradient field to guide the flow of logistics tasks, including: The game evaluation results are analyzed to extract the load surge component of the affected node within the prediction time window, and combined with the corresponding processing capacity boundary parameters in the physical state layer, the potential energy offset increment reflecting the node congestion risk trend is calculated. The potential energy offset increment is superimposed on the current elastic potential energy value of the corresponding node in the virtual state layer image, and the preset nonlinear mapping function is called to perform numerical correction on the potential energy distribution of the entire network, so as to obtain a corrected elastic potential energy distribution with congestion warning characteristics. For the modified elastic potential energy distribution, the potential energy spatial derivative between adjacent nodes on each topological path is calculated, the gradient guidance vector representing the flow of logistics tasks to the optimal descent path is extracted, and a potential energy gradient field covering the entire link and having the function of task migration guidance is generated.

9. The method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 1, characterized in that, The potential gradient field is analyzed to extract the path guidance vector, which is then matched with the logistics task constraint information retrieved from the management system to generate a reconstruction execution instruction containing path replanning instructions and resource scheduling instructions, including: Extract the potential energy decrease vector of each connecting edge in the potential energy gradient field to obtain the path reconstruction guidance direction; By combining the business delivery time boundary values ​​in the logistics task constraint information, the feasibility of the path reconstruction guidance direction is filtered, and an initial version of the replanning path is generated. The node processing capacity load of the initial replanning path is verified. Under the premise of meeting the processing capacity boundary parameters in the dual-state network model, the reconstructing execution instructions are generated by combining them.

10. A method for elastic assessment and reconstruction of a full-link logistics network to sudden disturbances according to claim 1, characterized in that, After the reconfiguration execution instructions are sent to each physical execution terminal in the end-to-end logistics network via the communication gateway, the process also includes: The monitoring process is initiated to obtain real-time job messages from each execution terminal, thereby obtaining actual business retention metrics. Calculate the deviation ratio between the actual business retention index and the predicted business retention index in the game evaluation results, and generate the reconstruction deviation evaluation index. Based on the parameter correction mechanism triggered by the reconstruction deviation evaluation index, the weighting coefficients used in calculating the elastic potential energy field are automatically optimized.

Images