Logistics platform coordination scheduling method and system based on service-oriented architecture

Abstract

The application discloses a logistics platform collaborative scheduling method and system based on a service-oriented architecture, relates to the technical field of logistics scheduling, and comprises the following steps: after a logistics platform receives a collaborative scheduling request, triggering a global situation awareness service to analyze the request and generating initial logistics element situation data; inputting the data into a resource field strength simulation model to calculate dynamic resource scarcity; generating a service grid dynamic arrangement instruction according to the comparison result of the scarcity and a preset threshold; activating a process disturbance channel and injecting a process disturbance factor; starting a steady-state resilience evaluation to calculate service structure compensation parameters; recombining microservices according to the parameters to form a compensated service grid; and importing the initial logistics element situation data into the grid to complete task decomposition and resource matching and generate a final scheduling scheme. The method can realize accurate matching of logistics resources and flexible adaptation of the scheduling process, and improve the stability and efficiency of collaborative scheduling.

Description

Technical Field

[0001] This invention belongs to the field of logistics scheduling technology, specifically a collaborative scheduling method and system for logistics platforms based on a service-oriented architecture. Background Technology

[0002] Currently, service-oriented architecture has been gradually applied to the field of collaborative scheduling in logistics platforms. In existing technologies, after receiving a collaborative scheduling request, the logistics platform typically parses the request to obtain relevant data on logistics elements, and then directly calls a pre-defined microservice module to decompose the scheduling task and match relevant resources to complete the collaborative scheduling process. In the resource determination stage, existing technologies mostly adopt static statistical methods, judging whether resources meet scheduling requirements by pre-setting fixed resource reserve thresholds, lacking dynamic perception and accurate calculation of real-time resource status.

[0003] In existing technical solutions, static resource determination methods cannot capture the dynamic changes of logistics resources in real time, making it difficult to accurately reflect the resource tension corresponding to different collaborative scheduling requests, which can easily lead to a disconnect between resource allocation and actual needs. At the same time, microservice calls under a service-oriented architecture are mostly fixed process orchestrations. When the status of logistics elements changes or process disturbances occur, the microservice structure cannot be flexibly adjusted, making it difficult to cope with various uncertainties in dynamic scheduling scenarios. This can easily lead to problems such as scheduling process blockage, resource waste, or scheduling failure, making it impossible to achieve efficient and stable collaborative scheduling. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art; To this end, the present invention proposes a collaborative scheduling method for logistics platforms based on a service-oriented architecture, comprising: When the logistics platform receives a collaborative scheduling request, it triggers a global situational awareness service to parse the collaborative scheduling request and generate initial logistics element situational data. The initial logistics element status data is input into a preset resource field strength simulation model to calculate the dynamic resource scarcity of the collaborative scheduling request. Based on the comparison between the dynamic resource scarcity and the preset resource scarcity threshold, a service mesh dynamic orchestration instruction is generated. In response to the service mesh dynamic orchestration command, the process disturbance channel of the service architecture is activated, and process disturbance factors are injected into the service architecture through the process disturbance channel; After injecting the process disturbance factor, the steady-state resilience assessment of the service architecture is initiated, and the service structure compensation parameters are calculated based on the results of the steady-state resilience assessment. Based on the service structure compensation parameters, the microservices in the service-oriented architecture are reorganized and configured to form a compensated service mesh; The initial logistics element status data is imported into the compensated service grid for task decomposition and resource matching to generate the final logistics collaborative scheduling scheme.

[0005] Furthermore, when the logistics platform receives a collaborative scheduling request, it triggers a global situational awareness service to parse the collaborative scheduling request and generate initial logistics element situational data, including: The global situational awareness service performs multi-level scanning of the collaborative scheduling request to identify the cargo feature clusters, transportation route chains, and time constraint units contained in the request. The identified cargo feature clusters are analyzed for aggregation and dispersion, the identified transportation route chains are verified for feasibility, and the identified time-constrained units are quantified for urgency. By integrating the aggregation and distribution analysis results of the cargo feature clusters, the feasibility verification results of the transportation route chain, and the quantified time constraint urgency, a temporary scheduling feature field is constructed. Monitor the state evolution trajectory of the temporary scheduling feature field within a specified decision period, and generate the initial logistics element status data based on the state evolution trajectory; The monitoring of the state evolution trajectory of the temporary scheduling feature field within a specified decision period includes: Within the specified decision period, snapshots of the temporary scheduling feature field are collected at equal time intervals; Record the instantaneous state of the cargo distribution index, route feasibility index, and timeliness urgency value in the temporary scheduling feature field at the time of each snapshot collection; By connecting the instantaneous states of all snapshot acquisition moments, a multi-dimensional state evolution trajectory surface is formed.

[0006] Furthermore, the initial logistics element situation data is input into a preset resource field strength simulation model to calculate the dynamic resource scarcity of the coordinated scheduling request, including: The resource field strength simulation model integrates historical resource load baseline and real-time resource fluctuation surface. The initial logistics element status data are input in parallel into the historical resource load baseline and the real-time resource fluctuation surface; In the historical resource load baseline, the degree of fit between the initial logistics element status data and the historical resource supply pattern is matched; Within the real-time resource fluctuation surface, the demand diffusion path of the initial logistics element status data in the logistics resource network is deduced; By aggregating the results of the fit degree and the deduction of the demand diffusion path, the dynamic resource scarcity is obtained through the scarcity calculation function; The process of deriving the demand diffusion path of the initial logistics element status data in the logistics resource network within the real-time resource fluctuation surface includes: The logistics resource network is abstracted as a flow diagram with capacitive and resistive characteristics; The initial logistics element status data is regarded as a resource demand flow injected from the demand occurrence node; Based on the capacity and resistance attributes of each side in the flow diagram, the congestion and dissipation of the resource demand flow along different channels are calculated to determine the core demand diffusion path and the effective demand intensity at its end.

[0007] Furthermore, based on the comparison result between the dynamic resource scarcity and the preset resource scarcity threshold, the service mesh dynamic orchestration instructions are generated, including: Define high scarcity thresholds and low scarcity thresholds; When the dynamic resource scarcity is higher than the high scarcity threshold, a centralized orchestration instruction for guiding service aggregation is generated. When the dynamic resource scarcity is lower than the low scarcity threshold, a discrete orchestration instruction for guiding service distribution is generated. When the dynamic resource scarcity is between the high scarcity threshold and the low scarcity threshold, a steady-state orchestration instruction to maintain the current service status is generated. The centralized orchestration instructions, discrete orchestration instructions, or steady-state orchestration instructions are collectively referred to as service grid dynamic orchestration instructions.

[0008] Furthermore, in response to the service mesh dynamic orchestration command, activating the process disturbance channel of the service architecture and injecting process disturbance factors into the service architecture through the process disturbance channel includes: The service-oriented architecture is pre-set with process disturbance channel entry points corresponding to different types of service mesh dynamic orchestration instructions; Based on the specific category of the service mesh dynamic orchestration instruction, the corresponding process disturbance channel entry is activated; In the activated process disturbance channel, a process disturbance factor with a specific dimension and intensity of action is generated based on the value of the dynamic resource scarcity. The generated process disturbance factor is injected into the core service dependency chain of the service architecture via the enabled process disturbance channel entry.

[0009] Furthermore, the step of initiating a steady-state resilience assessment of the service architecture after injecting the process disturbance factor, and calculating service structure compensation parameters based on the results of the steady-state resilience assessment, includes: Within the observation window after injecting the process disturbance factor, monitor the recovery trend of inter-service call latency and the regression tendency of service interface stability in the service-oriented architecture. Based on the recovery trend of the call latency and the regression tendency of the service interface stability, the steady-state resilience of the service architecture against internal disturbances is quantitatively evaluated. A steady-state resilience evaluation function is constructed, and the recovery trend of the call latency and the regression tendency of the service interface stability are used as input variables to calculate the steady-state resilience index. Based on the offset between the steady-state toughness index and the preset toughness reference band, the service structure compensation parameters are derived through the compensation parameter calculation model.

[0010] Furthermore, the microservices in the service-oriented architecture are reorganized and configured according to the service structure compensation parameters to form a compensated service mesh, including: The service structure compensation parameters are parsed into instance size compensation components and service link compensation components. The instance size compensation component is applied to elastically scale the number of deployment instances of the target microservice in the service-oriented architecture. The service link compensation component is applied to adjust the weight of the call links between microservices in the service architecture or to establish new backup links. After completing the compensation operations for instance size and service links, the reorganized service mesh is subjected to dependency closed-loop verification and conflict circuit breaking to form a stable and usable compensated service mesh.

[0011] Furthermore, the construction of the steady-state resilience evaluation function, using the recovery trend of the call latency and the regression tendency of the service interface stability as input variables, calculates the steady-state resilience index, including: Weights are assigned to the recovery trend of the call latency and to the regression tendency of the service interface stability; Substitute the weighted call delay recovery trend and the weighted service interface stability regression tendency into the steady-state resilience evaluation function that includes nonlinear transformation; The steady-state resilience evaluation function is used to calculate and output a quantitative value that characterizes the system's ability to maintain service stability, namely the steady-state resilience index.

[0012] Furthermore, the construction method of the preset resource field strength simulation model includes: Collect historical collaborative scheduling records of the logistics platform within a complete scheduling evaluation cycle, wherein the scheduling evaluation cycle includes at least one complete peak-off-peak cycle sequence of logistics business. The historical collaborative scheduling records are cleaned and structured to extract scheduling request features, available resource status, scheduling decision results, and scheduling execution efficiency indicators to form a model training sample set. From the model training sample set, the average resource load level under different resource types and different network regions is calculated based on the time window moving average method. The calculation results are then fitted according to the time dimension and the spatial dimension to construct the historical resource load baseline that represents the background situation of resource supply and demand. From the model training sample set, time series analysis techniques are used to separate the periodic trend component, seasonal component and random fluctuation component in the resource load data; Based on the separated random fluctuation components and combined with the real-time acquired resource dynamic signals, the instantaneous diffusion process of resource fluctuations is simulated through spatial interpolation and propagation models to construct the real-time resource fluctuation surface that characterizes the real-time uncertainty of resources. The historical resource load baseline is used as the static background field of the simulation model, and the real-time resource fluctuation surface is used as the dynamic superposition field of the simulation model. The field strength coupling algorithm is used to fuse the static background field and the real-time resource fluctuation surface to form the resource field strength simulation model. The field strength coupling algorithm defines the dynamic adjustment rules for the contribution weights of the static background field and the dynamic superposition field to the final simulation result under different time scales and event triggering conditions.

[0013] Furthermore, the present invention also includes a service-oriented architecture-based logistics platform collaborative scheduling system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the service-oriented architecture-based logistics platform collaborative scheduling method described above.

[0014] Compared with the prior art, the beneficial effects of the present invention are: By inputting initial logistics element status data into a preset resource field strength simulation model, the dynamic resource scarcity of collaborative scheduling requests is calculated. Compared with the traditional method of statically calculating resource reserves and comparing fixed thresholds, this method can capture the dynamic changes of logistics resources in real time, accurately quantify the resource tension corresponding to different collaborative scheduling requests, break the limitations of traditional static judgment, make resource judgment results more in line with actual scheduling scenarios, achieve accurate control of resource scarcity, avoid the disconnect between resource allocation and actual needs, reduce resource waste, and make resource matching more targeted.

[0015] In response to dynamic orchestration commands from the service mesh, the process disturbance channel of the service architecture is activated. Through this channel, process disturbance factors are injected into the service architecture. Then, based on the steady-state resilience assessment results after injection, service structure compensation parameters are calculated. Based on these parameters, the microservices in the service architecture are reorganized and configured to form a compensated service mesh. Task decomposition and resource matching are then carried out based on the compensated service mesh. Compared with the fixed microservice call process under the traditional service architecture, this approach can flexibly adjust the microservice structure when the status of logistics elements changes or process disturbances occur. It can quickly adapt to various uncertainties in dynamic scheduling scenarios, alleviate scheduling instability caused by process disturbances, ensure the smooth progress of collaborative scheduling processes, and improve the flexibility and adaptability of scheduling processes. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of the service-oriented architecture-based collaborative scheduling method for logistics platforms as described in this invention. Figure 2 A flowchart for generating initial logistics element status data; Figure 3 A flowchart for calculating dynamic resource scarcity; Figure 4 This is the curve showing the change in the steady-state toughness index; Figure 5 The final field strength simulation results are for various types of resources in East China. Detailed Implementation

[0017] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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] See Figure 1When the logistics platform receives a collaborative scheduling request, it triggers a global situational awareness service. This service parses the received request and generates initial logistics element situational data. This data is then input into a pre-defined resource field strength simulation model, which calculates the dynamic resource scarcity for the specific collaborative scheduling request. The system then compares the calculated dynamic resource scarcity with a pre-defined resource scarcity threshold and generates corresponding service mesh dynamic orchestration instructions based on the comparison results. The system responds to these instructions by activating a pre-defined process disturbance channel within the service-oriented architecture and injecting process disturbance factors into the core service dependency chain. After the process disturbance factor injection is complete, a steady-state resilience assessment of the service-oriented architecture is initiated, and service structure compensation parameters are calculated based on the assessment results. Using these compensation parameters, the system reorganizes and configures the instance size and service links of microservices within the service-oriented architecture, thereby forming a new compensated service mesh. The initial logistics element status data is imported into this compensated service grid, which then performs specific task decomposition and resource matching operations, and generates the final logistics collaborative scheduling scheme.

[0019] In one embodiment of the present invention, see [reference] Figure 2 The global situational awareness service performs multi-level scanning of collaborative scheduling requests, identifying the cargo feature clusters, transportation route chains, and time-constraint units inherent in the requests. It performs aggregation and dispersion analysis on the identified cargo feature clusters, feasibility verification on the identified transportation route chains, and urgency quantification on the identified time-constraint units. By integrating the aggregation and dispersion analysis results of cargo feature clusters, the feasibility verification results of transportation route chains, and the quantified urgency of time constraints, a temporary scheduling feature field is constructed. The system monitors the state evolution trajectory of this temporary scheduling feature field within a specified decision-making period and generates initial logistics element situational data based on the monitored state evolution trajectory. Specifically, the monitoring of the state evolution trajectory involves taking snapshots of the temporary scheduling feature field at equal time intervals within the specified decision-making period. The system records the instantaneous state of the cargo aggregation and dispersion index, route feasibility index, and time-constraint value in the temporary scheduling feature field at each snapshot acquisition time. By concatenating these instantaneous states at all snapshot acquisition times, a multi-dimensional state evolution trajectory surface is formed.

[0020] In practice, the process of generating initial logistics element situational data is executed by the global situational awareness service. This service is triggered when the logistics platform receives a collaborative scheduling request. The global situational awareness service performs a multi-level scan of the request, identifying cargo feature clusters, transportation route chains, and time-bound constraint units within the request. Taking an emergency medical cold chain transportation collaborative scheduling request from Beijing to Guangzhou as an example, the request includes drug information, temperature control requirements, cargo volume and weight, coordinates of the originating warehouse, coordinates of the destination hospital, and the required delivery deadline. Through multi-level scanning, the global situational awareness service identifies cargo feature clusters representing aggregated information on drug type, temperature, volume, and weight; identifies the transportation route chain connecting the Beijing warehouse node, transit city nodes, and the Guangzhou hospital node; and identifies the time-bound constraint unit centered on the delivery deadline.

[0021] In some embodiments, the global situational awareness service performs aggregation and dispersion analysis on identified cargo feature clusters, feasibility verification on identified transportation route chains, and urgency quantification on identified time-constrained units. Aggregation and dispersion analysis, for cargo feature clusters, calculates the degree of aggregation of goods at the shipping location and the expected distribution difficulty at the destination and potential transit points, forming a quantitative aggregation and dispersion index. Feasibility verification, for transportation route chains, combines real-time traffic network status, weather information, and traffic control notices to evaluate the traversability and efficiency of each segment of the route chain in the current and predicted time periods, generating a route feasibility index. Urgency quantification, for time-constrained units, calculates a time-constrained urgency value that dynamically changes over time based on the time difference between the current time and the required delivery deadline, combined with the length and complexity of the route chain. In a specific implementation, by integrating the aggregation and dispersion analysis results of cargo feature clusters, the feasibility verification results of transportation route chains, and the quantified time-constrained urgency, the global situational awareness service constructs a temporary scheduling feature field. This temporary scheduling feature field is a multi-dimensional data space used to comprehensively characterize the real-time scheduling environment state of a single collaborative scheduling request.

[0022] In practical implementation, the global situational awareness service monitors the state evolution trajectory of the temporary scheduling feature field within a specified decision period and generates initial logistics element situational data based on the state evolution trajectory. The process of monitoring the state evolution trajectory involves the global situational awareness service taking snapshots of the temporary scheduling feature field at equal time intervals within the specified decision period. It can be understood that the length of the specified decision period can be configured according to the complexity of the collaborative scheduling request; for example, for the aforementioned emergency medical transportation, the specified decision period might be set to 30 minutes after the request is received. The global situational awareness service records the instantaneous state of the cargo distribution index, route feasibility index, and timeliness urgency value in the temporary scheduling feature field at each snapshot collection. By concatenating the instantaneous states at all snapshot collection times, a multi-dimensional state evolution trajectory surface is formed. This surface describes the continuous changes in the three dimensions of cargo distribution, route feasibility, and timeliness urgency within the specified decision period. Optionally, the data from the state evolution trajectory surface can be used for subsequent analysis; the state evolution trajectory surface itself is structured and time-seriesd initial logistics element situational data. In some embodiments, the mathematical expression of the state evolution trajectory surface can be a surface function describing the evolution of multidimensional states over time, expressed by the formula:

[0023] in: Indicates at a point in time The state evolution trajectory surface coordinates Indicates at a point in time The instantaneous status of the collected cargo distribution indicators Indicates at a point in time The instantaneous state of the feasibility index of the collected path. Indicates at a point in time The instantaneous state of the collected timeliness and urgency values. The value range covers all equally spaced snapshot collection times from the start to the end of the specified decision period.

[0024] In one embodiment of the present invention, see [reference] Figure 3The process of calculating dynamic resource scarcity is accomplished by a pre-defined resource field strength simulation model, which integrates historical resource load baselines and real-time resource fluctuation surfaces. Initial logistics element status data is input in parallel to both the historical resource load baseline and the real-time resource fluctuation surface. In the historical resource load baseline, the model matches the initial logistics element status data with the historical resource supply pattern. In the real-time resource fluctuation surface, the model extrapolates the demand diffusion path of the initial logistics element status data within the logistics resource network. The system aggregates the results of the matching degree and the demand diffusion path extrapolation, and obtains the dynamic resource scarcity through a pre-defined scarcity calculation function. Specifically, the process of extrapolating the demand diffusion path involves abstracting the logistics resource network as a flow graph with capacitive and resistive characteristics. The initial logistics element status data is considered as a resource demand flow injected from the demand occurrence node. Based on the capacity and resistance attributes of each edge in the flow graph, the congestion and dissipation of this resource demand flow as it diffuses along different channels are calculated, thereby determining the core demand diffusion path and the effective demand intensity at its endpoint.

[0025] In practice, the calculation of dynamic resource scarcity is accomplished by a pre-defined resource field strength simulation model. This model integrates historical resource load baselines and real-time resource fluctuation surfaces. After initial logistics element status data is generated, it is input in parallel to both the historical resource load baseline and the real-time resource fluctuation surface. The resource field strength simulation model matches the initial logistics element status data with historical resource supply patterns within the historical resource load baseline. Within the real-time resource fluctuation surface, it extrapolates the demand diffusion path of the initial logistics element status data within the logistics resource network. It can be understood that the historical resource load baseline reflects the long-term average utilization level of resources, while the real-time resource fluctuation surface captures immediate fluctuations and uncertainties. The parallel processing of these two aspects can balance steady-state patterns and dynamic disturbances. The system aggregates the results of the matching degree and demand diffusion path extrapolation, and obtains the final dynamic resource scarcity value through a pre-defined scarcity calculation function.

[0026] In some embodiments, the specific steps for deducing demand diffusion paths in a real-time resource fluctuation surface include abstracting the entire logistics resource network as a flow graph with capacity and resistance characteristics. In the flow graph, logistics hubs, distribution centers, and transportation trunk lines are abstracted as nodes and edges. Each edge is assigned a capacity attribute and a resistance attribute. The capacity attribute represents the maximum resource throughput capacity of the path, and the resistance attribute represents the difficulty or cost coefficient of resources passing through the path. The initial logistics element status data is regarded as a resource demand flow injected from the demand occurrence node, which corresponds to the starting location of the coordinated scheduling request, such as Beijing. In specific implementations, based on the capacity and resistance attributes of each edge in the flow graph, the congestion and dissipation of the resource demand flow as it diffuses along different channels are calculated. Optionally, this calculation can simulate the diversion, attenuation, and superposition effects of resource demand during multi-path transmission in the network. Through calculation, the resource field strength simulation model can determine several core demand diffusion paths and the effective demand intensity at their ends. The effective demand intensity reflects the actual demand pressure reaching a specific area in the network after transmission and dissipation through the network. Taking a pharmaceutical cold chain transportation request from Beijing to Guangzhou as an example, initial logistics element status data is injected from the Beijing node as the resource demand flow. The model deduces two main diffusion paths: Path A is "Beijing-Zhengzhou-Guangzhou", and Path B is "Beijing-Wuhan-Guangzhou". At the data comparison level, the real-time capacity of the "Zhengzhou-Guangzhou" segment in Path A is reduced and resistance is increased due to weather conditions, showing severe congestion. Path B has sufficient capacity in each segment but longer distances, showing significant dissipation. After aggregate calculation, the effective demand intensity at the end of Path A may decrease due to high congestion, while the effective demand intensity at the end of Path B is reduced due to high dissipation. Finally, the model combines the end-point intensities of the two paths to arrive at an overall projection result.

[0027] In some embodiments, the scarcity calculation function for calculating dynamic resource scarcity takes as input the fit of historical resource load baselines and the effective demand intensity extrapolated from real-time resource fluctuations, and outputs the dynamic resource scarcity. The formula is expressed as:

[0028] in: This represents the calculated dynamic resource scarcity. This represents the scarcity calculation function, which can take the form of a weighted aggregation or a nonlinear mapping function. This represents the degree of fit obtained from matching the historical resource load baseline. The higher the value, the better the current request matches the historical abundant resource pattern. The representative obtained through deduction The intensity of effective demand at the end of the core demand diffusion path, This represents the total number of identified core demand diffusion paths. Representing the The weighting coefficients for the core demand diffusion paths can be set based on the path's importance or reliability within the network. The formula is derived through a function. The degree of fit that reflects historical patterns Weighted effective demand intensity, reflecting real-time network pressure A comprehensive calculation is performed to generate a dynamic resource scarcity index that simultaneously reflects both long-term resource scarcity trends and short-term fluctuation pressures. .

[0029] In one embodiment of the present invention, the process of generating service mesh dynamic orchestration instructions involves threshold comparison, and the system predefines a high scarcity threshold and a low scarcity threshold. When the dynamic resource scarcity is higher than the defined high scarcity threshold, the system generates a centralized orchestration instruction to guide service aggregation. When the dynamic resource scarcity is lower than the defined low scarcity threshold, the system generates a discrete orchestration instruction to guide service distribution. When the dynamic resource scarcity is between the high scarcity threshold and the low scarcity threshold, the system generates a steady-state orchestration instruction to maintain the current service status. Centralized orchestration instructions, discrete orchestration instructions, or steady-state orchestration instructions are collectively referred to as service mesh dynamic orchestration instructions. In response to the generated service mesh dynamic orchestration instructions, the system activates a process disturbance channel. The service architecture predefines process disturbance channel entry points corresponding to different categories of service mesh dynamic orchestration instructions. The system enables the corresponding process disturbance channel entry point according to the specific category of the service mesh dynamic orchestration instruction. In the enabled process disturbance channel, a process disturbance factor with a specific dimension and intensity of action is generated based on the value of dynamic resource scarcity. The generated process disturbance factors are injected into the core service dependency chain of the service-oriented architecture through the enabled process disturbance channel entry.

[0030] In practice, the process of generating dynamic service mesh orchestration instructions is based on a comparison between dynamic resource scarcity and preset thresholds. The system predefines high scarcity and low scarcity thresholds. The dynamic resource scarcity value is compared with the high and low scarcity thresholds, and corresponding service mesh dynamic orchestration instructions are generated based on the comparison results. When the dynamic resource scarcity is higher than the high scarcity threshold, the system generates intensive orchestration instructions to guide service aggregation. When the dynamic resource scarcity is lower than the low scarcity threshold, the system generates discrete orchestration instructions to guide service distribution. When the dynamic resource scarcity is between the high and low scarcity thresholds, the system generates stable orchestration instructions to maintain the current service status. Intensive orchestration instructions, discrete orchestration instructions, and stable orchestration instructions are collectively referred to as service mesh dynamic orchestration instructions. Taking the scenario of pharmaceutical cold chain transportation from Beijing to Guangzhou as an example, assuming that the calculated dynamic resource scarcity is 0.85, the preset high scarcity threshold is 0.7, and the low scarcity threshold is 0.3, since the dynamic resource scarcity of 0.85 is higher than the high scarcity threshold of 0.7, the system will generate an intensive orchestration instruction.

[0031] In some embodiments, the generation logic of service mesh dynamic orchestration instructions can be represented by a function mapping, as shown in the formula:

[0032] in: The instruction code represents the dynamic orchestration instructions for the generated service mesh. The centralized orchestration instructions, discrete orchestration instructions, and steady-state orchestration instructions correspond to different instruction code values. This represents a mapping function from resource scarcity to instruction code. A specific numerical value representing the scarcity of dynamic resources. This represents the specific value of the preset high scarcity threshold. This represents the specific value of the preset low scarcity threshold. Mapping function. The logic is: if ,but Corresponding discretization orchestration instruction code; if ,but Corresponding discretization orchestration instruction code; if ,but Corresponding to the steady-state orchestration instruction code.

[0033] In practical implementation, in response to the generated service mesh dynamic orchestration instructions, the system activates the process disturbance channels of the service-oriented architecture. The service-oriented architecture pre-defines process disturbance channel entry points corresponding to different categories of service mesh dynamic orchestration instructions. The system activates the corresponding process disturbance channel entry point based on the specific category of the service mesh dynamic orchestration instruction. For example, for intensive orchestration instructions, an entry point named "Service Aggregation Disturbance Channel" is activated; for discrete orchestration instructions, an entry point named "Service Distribution Disturbance Channel" is activated; and for steady-state orchestration instructions, an entry point named "Service Steady-State Maintenance Channel" is activated. Within the activated process disturbance channels, the system generates process disturbance factors with specific dimensions and strengths based on the dynamic resource scarcity value. It can be understood that the dimensions of the process disturbance factors may include service call topology adjustments, service instance load weight adjustments, and inter-service communication protocol parameter adjustments, etc. The strength of the process disturbance factors is positively correlated with the degree to which the dynamic resource scarcity deviates from the threshold. The generated process disturbance factors are then injected into the core service dependency chain of the service-oriented architecture via the activated process disturbance channel entry points. Optionally, in the example of pharmaceutical cold chain transportation, the centralized orchestration instruction triggers the activation of the "service aggregation disturbance channel". Based on the dynamic resource scarcity of 0.85, a process disturbance factor with the function dimension of "merging similar service instances" and the intensity of "high" is generated, and this process disturbance factor is injected into the core service dependency chain related to path planning and temperature control monitoring.

[0034] In one embodiment of the present invention, after injecting the process disturbance factor, the system initiates a steady-state resilience assessment. Specifically, within an observation window after the injection of the process disturbance factor, the system monitors the recovery trend of inter-service call latency and the regression tendency of service interface stability in the service-oriented architecture. Based on the recovery trend of call latency and the regression tendency of service interface stability, the system quantitatively assesses the steady-state resilience of the service-oriented architecture against this internal disturbance. A steady-state resilience assessment function is constructed, using the recovery trend of call latency and the regression tendency of service interface stability as input variables to calculate a steady-state resilience index. A weight is assigned to the recovery trend of call latency, and another weight is assigned to the regression tendency of service interface stability. The weighted recovery trend of call latency and the weighted regression tendency of service interface stability are substituted into a steady-state resilience assessment function containing nonlinear transformation for calculation, outputting a quantitative value characterizing the system's ability to maintain service stability, i.e., the steady-state resilience index. Based on the calculated steady-state resilience index and the offset of a preset resilience baseline, service structure compensation parameters are derived through a compensation parameter calculation model. Based on service structure compensation parameters, the system reorganizes and configures microservices in the service-oriented architecture. First, it parses the service structure compensation parameters into instance size compensation components and service link compensation components. The instance size compensation component is applied to elastically scale the number of deployed instances of the target microservice in the service-oriented architecture. The service link compensation component is applied to adjust the weights of the call links between microservices in the service-oriented architecture or to establish new backup links. After completing the instance size and service link compensation operations, dependency loop verification and conflict circuit breaking are performed on the reorganized service mesh, ultimately forming a stable and usable compensated service mesh.

[0035] In practical implementation, after injecting the process disturbance factor into the core service dependency chain of the service-oriented architecture, the system initiates a steady-state resilience assessment of the service-oriented architecture. This assessment is conducted within a pre-defined observation window after the injection of the process disturbance factor. Within this window, the system monitors the recovery trend of inter-service call latency and the regression tendency of service interface stability within the service-oriented architecture. The recovery trend of call latency can be characterized by recording the average call latency value at multiple time points, and the regression tendency of service interface stability can be characterized by recording the service interface success response rate at multiple time points. Taking a pharmaceutical cold chain transportation service scenario where a "merging similar service instances" process disturbance factor is injected due to high resource scarcity as an example, the initial injection of the process disturbance factor may lead to the reconstruction of some service call paths, causing an increase in call latency and fluctuations in interface response. Within the observation window after the injection of the process disturbance factor, the system continuously collects data to monitor the direction and rate of change in latency and stability.

[0036] In some embodiments, the monitored call latency recovery trend and service interface stability regression tendency data can be recorded in a table for subsequent quantitative evaluation. See Table 1 for an example of monitoring data.

[0037] Table 1: Monitoring data on microservice call latency and service interface stability within the observation window

[0038] In practical implementation, based on the recovery trend of call latency and the regression tendency of service interface stability, the system quantitatively evaluates the steady-state resilience of the service architecture against this internal disturbance. A steady-state resilience evaluation function is constructed, using the recovery trend of call latency and the regression tendency of service interface stability as input variables. The steady-state resilience index is obtained through the calculation of the steady-state resilience evaluation function. The steady-state resilience evaluation function assigns one weight to the recovery trend of call latency and another weight to the regression tendency of service interface stability. It can be understood that the specific values ​​of the weights can be configured according to the different sensitivities of the business to latency or stability. Substituting the weighted recovery trend of call latency and the weighted regression tendency of service interface stability into the steady-state resilience evaluation function, which includes nonlinear transformation, the steady-state resilience evaluation function outputs a quantitative value characterizing the system's ability to maintain service stability, namely the steady-state resilience index. The formula is expressed as:

[0039] in: This represents the calculated steady-state resilience index. It represents an aggregation function that includes a nonlinear transformation, such as a composite form of an exponentially decaying function. This represents the weight assigned to the delayed recovery trend. Represents the trend of call delay recovery and time The calculated function is used to quantify the speed and extent of delayed recovery. This represents the weight assigned to the service interface based on its stability regression tendency. This represents a tendency for service interface stability regression. and time The calculated function is used to quantify the speed and extent of stability regression. This represents the observation time starting after the injection of the process disturbance factor. (Function) For weighted and Nonlinear aggregation is performed to derive the steady-state toughness index. .

[0040] Based on the offset between the steady-state resilience index and the preset resilience benchmark band, the system derives service structure compensation parameters through a compensation parameter calculation model. This model can be a mapping relationship based on rules or linear regression. For example, when the steady-state resilience index is below the lower limit of the resilience benchmark band, a larger positive compensation parameter is derived; when the index is within the benchmark band, a compensation parameter close to zero is derived; and when the index is above the upper limit, a negative compensation parameter may be derived. In the example of pharmaceutical cold chain transportation, assuming the steady-state resilience index calculated from monitoring data is 0.65, and the preset resilience benchmark band is [0.7, 0.9], and the index of 0.65 is below the lower limit of the benchmark band of 0.7, the compensation parameter calculation model will derive a positive service structure compensation parameter, indicating that the robustness of the service structure needs to be enhanced.

[0041] In some embodiments, microservices in a service-oriented architecture are reorganized and configured based on service structure compensation parameters. The system first parses the service structure compensation parameters into instance size compensation components and service link compensation components. The instance size compensation component is a value indicating the direction and magnitude of elastic scaling of the number of microservice deployment instances. The service link compensation component is a value indicating whether weight adjustments or the establishment of new backup links are needed for inter-microservice call links. The instance size compensation component is applied to elastically scale the number of deployment instances of the target microservice in the service-oriented architecture. For example, if the instance size compensation component is positive, the number of running instances of the critical microservice is increased. The service link compensation component is applied to adjust the weights of inter-microservice call links in the service-oriented architecture or establish new backup links. For example, the weight of call links pointing to currently heavily loaded instances is reduced, while a new backup link pointing to a backup service instance is established for critical service calls. After completing the compensation operations for instance scale and service links, the system performs dependency closed-loop verification and conflict circuit breaking on the reorganized service mesh. Dependency closed-loop verification ensures that there are no circular dependencies in the new service call relationships, and conflict circuit breaking avoids service call conflicts caused by configuration changes, ultimately forming a stable and usable compensated service mesh.

[0042] See Figure 4In steady-state resilience assessment, the system quantifies the service-oriented architecture's ability to withstand internal disturbances by monitoring changes in the steady-state resilience index within the observation window after the injection of process disturbance factors. Specifically, the steady-state resilience index is a weighted aggregation of the microservice call latency recovery trend and the service interface stability regression tendency. Its value shows an evolutionary trajectory of gradually recovering from a low level after the initial disturbance over the observation time. The resilience baseline is defined as the interval [0.7, 0.9], serving as a reference threshold for evaluating the system's steady-state performance. At the beginning of the observation window (0 seconds), the steady-state resilience index is 0.59, significantly lower than the lower limit of the baseline, indicating a significant decrease in resilience after the injection of process disturbances. As time progresses, the index gradually recovers, reaching 0.735, 0.785, 0.813, and 0.826 at 5 seconds, 10 seconds, 15 seconds, and 20 seconds, respectively, but ultimately not fully entering the baseline interval at the end of the observation window (20 seconds). This resilience index recovery process provides a core basis for deriving service structure compensation parameters: when the index is below the lower limit of the baseline band, the system will derive positive compensation parameters, and enhance the robustness of the service mesh by elastically scaling the scale of microservice instances and adjusting the weight of service call links, thereby driving the steady-state resilience index back to the baseline band range.

[0043] In one embodiment of the present invention, historical collaborative scheduling records of a logistics platform are collected within a complete scheduling evaluation cycle, which includes at least one complete peak-off-peak cycle sequence of logistics operations. These historical collaborative scheduling records are cleaned and structured to extract scheduling request features, available resource status, scheduling decision results, and scheduling execution efficiency indicators, forming a model training sample set. From the model training sample set, the average resource load level under different resource types and different network regions is calculated based on the time window moving average method. The calculation results are then fitted along the time and spatial dimensions to construct a historical resource load baseline characterizing the resource supply and demand situation. From the model training sample set, time series analysis techniques are used to separate the periodic trend component, seasonal component, and random fluctuation component from the resource load data. Based on the separated random fluctuation component and combined with real-time collected resource dynamic signals, a spatial interpolation and propagation model is used to simulate the instantaneous diffusion process of resource fluctuations, constructing a real-time resource fluctuation surface characterizing the real-time uncertainty of resources. The established historical resource load baseline is used as the static background field of the simulation model, and the real-time resource fluctuation surface is used as the dynamic superposition field of the simulation model. The two are fused by a field-strength coupling algorithm to form the final resource field-strength simulation model. The field-strength coupling algorithm defines the dynamic adjustment rules for the contribution weights of the static background field and the dynamic superposition field to the final simulation results under different time scales and event triggering conditions.

[0044] In practical implementation, the construction of the pre-set resource field strength simulation model begins with data acquisition. Historical collaborative scheduling records from the logistics platform are collected within a complete scheduling evaluation cycle. This evaluation cycle needs to include at least one complete peak-off-peak cycle sequence of logistics operations. For example, for a nationwide logistics platform, the complete scheduling evaluation cycle can be set as a past full lunar year. This cycle typically includes the freight peak before and after the Spring Festival, the e-commerce promotion peak, and the daily off-peak period, thus ensuring that historical data covers typical fluctuation patterns of resource demand. In practice, the collected historical collaborative scheduling records undergo cleaning and structuring processing. This includes removing invalid records, standardizing data formats, and repairing outliers. Scheduling request features, available resource status, scheduling decision results, and scheduling execution efficiency indicators are extracted from the records. The extracted data items collectively form the model training sample set used for model training and construction. Each sample in the model training sample set is associated with the original scheduling timestamp and geographical region label.

[0045] In some embodiments, a historical resource load baseline is constructed from the model training sample set. The construction process is based on a time window moving average method to calculate the average resource load level under different resource types and different network regions. It can be understood that resource types may include refrigerated trucks, ordinary trucks, drones, etc., and network regions may be divided according to geographical regions such as North China, East China, and South China, or according to the radiation range of logistics hubs. For each "resource type-network region" combination, the moving average of its resource load rate is calculated in the time dimension, and the size of the moving window can be set according to the business rhythm. The calculated average resource load level is fitted in both the time and spatial dimensions. For example, a continuous function about time and spatial coordinates can be established using a surface fitting method. This fitted function or data field represents the historical resource load baseline, which reflects the periodic and trend-like average load state of resources over a long period of history.

[0046] In practical implementation, a real-time resource fluctuation surface is constructed from the same model training sample set. The construction process utilizes time series analysis techniques to separate the periodic trend component, seasonal component, and random fluctuation component from the resource load data. For example, the historical resource load rate time series of a certain vehicle type in a specific region is decomposed to separate the periodic component reflecting weekday / weekend cycles, the seasonal component reflecting quarterly changes, and the random fluctuation component that cannot be explained by any pattern. Based on the separated random fluctuation component, combined with real-time acquired resource dynamic signals, the instantaneous diffusion process of resource fluctuations is simulated through spatial interpolation and a propagation model. Real-time acquired resource dynamic signals can include sudden traffic control information, severe weather warnings, and announcements of large-scale events. Spatial interpolation is used to estimate the fluctuation intensity at other locations within the entire geographical area based on the fluctuation intensity at a known point (signal origin point); the propagation model defines the rules for the attenuation of fluctuation intensity with distance or network path. Through simulation, a real-time resource fluctuation surface characterizing the real-time uncertainty of resources is finally constructed. This real-time resource fluctuation surface is a dynamically updated data field reflecting the spatial distribution of short-term random disturbances.

[0047] The established historical resource load baseline is used as the static background field of the simulation model, and the real-time resource fluctuation surface is used as the dynamic superimposed field. A field strength coupling algorithm is used to fuse the historical resource load baseline and the real-time resource fluctuation surface to form the final resource field strength simulation model. The field strength coupling algorithm defines the dynamic adjustment rules for the contribution weights of the static background field and the dynamic superimposed field to the final simulation results under different time scales and event triggering conditions. For example, during normal stable periods, the static background field has a higher weight; when the intensity of the real-time resource dynamic signal exceeds a certain threshold, the weight of the dynamic superimposed field is increased. The formula is expressed as:

[0048] in: Represents the position and time The final simulated output value of the resource field strength. This represents the value of the historical resource load baseline at the corresponding location and time, i.e., the static background field. This represents the value of the real-time resource fluctuation surface at the corresponding location and time, i.e., the dynamic overlay field. This represents dynamically adjusted weights; it relates to event-triggered variables. and time The function takes values ​​between 0 and 1. Optional, the function... The specific form can be a piecewise function or a smooth S-shaped function, ensuring that the weights can transition smoothly or quickly according to the rules.

[0049] See Figure 5In the final field strength simulation results for various resource types in East China, the dynamic evolution of resource field strength values ​​(load rate) can be analyzed through the time series curves of three types of transportation resources (ordinary trucks, refrigerated trucks, and drones). From the overall characteristics of the simulation curves, the field strength values ​​of all three resource types exhibit significant periodic fluctuations and phased peaks, which highly coincides with the typical "peak-off-peak" scheduling cycle in logistics operations. In the first 15 weeks of the simulation period, the field strength value of ordinary trucks (solid line) first reached a secondary peak of approximately 0.58, then quickly fell back to a trough of 0.27; the field strength changes of refrigerated trucks (dashed line) and drones (dotted dashed line) followed the same trend, but the peak values ​​were slightly lower, approximately 0.54 and 0.56 respectively, while the trough values ​​dropped to 0.23 and 0.26 respectively. The fluctuations in this stage mainly reflect the dominant role of the historical resource load baseline (static background field) on resource supply and demand, reflecting the average load level under the long-term business model. During the simulation period from weeks 15 to 25, the field strength values ​​of all three resource types exhibited more dramatic fluctuations, reaching their highest peaks throughout the entire cycle (approximately 0.65 for ordinary trucks, 0.63 for refrigerated trucks, and 0.66 for drones), before plummeting to their lowest points around week 25 (approximately 0.10 for ordinary trucks, 0.17 for refrigerated trucks, and 0.16 for drones). This phenomenon can be attributed to strong disturbances in the real-time resource fluctuation surface (dynamically superimposed field), such as sudden traffic control, severe weather, or large-scale promotional events, which cause resource demand to expand or contract rapidly in a short period, thus significantly amplifying the fluctuation amplitude of the field strength. During the simulation phase from weeks 25 to 40, the resource field strength values ​​oscillated repeatedly within a low range (0.1 to 0.4), exhibiting a typical "disturbance-recovery" process. The curve shape in this stage reflects the process by which the service-oriented architecture, after injecting process disturbance factors, gradually pulls the resource field strength back into a controllable range through steady-state resilience assessment and service structure compensation. Among them, the field strength recovery speed of drones was slightly faster than that of ordinary trucks and refrigerated trucks, demonstrating their flexibility advantage in responding to sudden disturbances. In the last 10 weeks of the simulation (weeks 40 to 52), the field strength values ​​of the three types of resources rebounded synchronously again, forming a secondary fluctuation cycle with a peak value between 0.3 and 0.4, eventually stabilizing. The curve convergence trend in this stage verifies the effectiveness of the compensated service grid in task decomposition and resource matching, enabling the resource field strength to return to a relatively stable state after experiencing multiple disturbances. The simulation results clearly reveal how the historical resource load baseline and the real-time resource fluctuation surface work together to influence the dynamic evolution of resource field strength under the service-oriented architecture logistics collaborative scheduling framework, and how the dynamic orchestration and compensation mechanism of the service grid effectively smooths out disturbances and maintains system steady state.

[0050] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A collaborative scheduling method for logistics platforms based on a service-oriented architecture, characterized in that, The method includes: When the logistics platform receives a collaborative scheduling request, it triggers a global situational awareness service to parse the collaborative scheduling request and generate initial logistics element situational data. The initial logistics element status data is input into a preset resource field strength simulation model to calculate the dynamic resource scarcity of the collaborative scheduling request. Based on the comparison between the dynamic resource scarcity and the preset resource scarcity threshold, a service mesh dynamic orchestration instruction is generated. In response to the service mesh dynamic orchestration command, the process disturbance channel of the service architecture is activated, and process disturbance factors are injected into the service architecture through the process disturbance channel; After injecting the process disturbance factor, the steady-state resilience assessment of the service architecture is initiated, and the service structure compensation parameters are calculated based on the results of the steady-state resilience assessment. Based on the service structure compensation parameters, the microservices in the service-oriented architecture are reorganized and configured to form a compensated service mesh; The initial logistics element status data is imported into the compensated service grid for task decomposition and resource matching to generate the final logistics collaborative scheduling scheme.

2. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 1, characterized in that, When the logistics platform receives a collaborative scheduling request, it triggers a global situational awareness service to parse the collaborative scheduling request and generate initial logistics element situational data, including: The global situational awareness service performs multi-level scanning of the collaborative scheduling request to identify the cargo feature clusters, transportation route chains, and time constraint units contained in the request. The identified cargo feature clusters are analyzed for aggregation and dispersion, the identified transportation route chains are verified for feasibility, and the identified time-constrained units are quantified for urgency. By integrating the aggregation and distribution analysis results of the cargo feature clusters, the feasibility verification results of the transportation route chain, and the quantified time constraint urgency, a temporary scheduling feature field is constructed. Monitor the state evolution trajectory of the temporary scheduling feature field within a specified decision period, and generate the initial logistics element status data based on the state evolution trajectory; The monitoring of the state evolution trajectory of the temporary scheduling feature field within a specified decision period includes: Within the specified decision period, snapshots of the temporary scheduling feature field are collected at equal time intervals; Record the instantaneous state of the cargo distribution index, route feasibility index, and timeliness urgency value in the temporary scheduling feature field at the time of each snapshot collection; By connecting the instantaneous states of all snapshot acquisition moments, a multi-dimensional state evolution trajectory surface is formed.

3. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 2, characterized in that, The initial logistics element situation data is input into a preset resource field strength simulation model to calculate the dynamic resource scarcity of the coordinated scheduling request, including: The resource field strength simulation model integrates historical resource load baseline and real-time resource fluctuation surface. The initial logistics element status data are input in parallel into the historical resource load baseline and the real-time resource fluctuation surface; In the historical resource load baseline, the degree of fit between the initial logistics element status data and the historical resource supply pattern is matched; Within the real-time resource fluctuation surface, the demand diffusion path of the initial logistics element status data in the logistics resource network is deduced; By aggregating the results of the fit degree and the deduction of the demand diffusion path, the dynamic resource scarcity is obtained through the scarcity calculation function; The process of deriving the demand diffusion path of the initial logistics element status data in the logistics resource network within the real-time resource fluctuation surface includes: The logistics resource network is abstracted as a flow diagram with capacitive and resistive characteristics; The initial logistics element status data is regarded as a resource demand flow injected from the demand occurrence node; Based on the capacity and resistance attributes of each side in the flow diagram, the congestion and dissipation of the resource demand flow along different channels are calculated to determine the core demand diffusion path and the effective demand intensity at its end.

4. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 3, characterized in that, Based on the comparison between the dynamic resource scarcity and the preset resource scarcity threshold, the service mesh dynamic orchestration instructions are generated, including: Define high scarcity thresholds and low scarcity thresholds; When the dynamic resource scarcity is higher than the high scarcity threshold, a centralized orchestration instruction for guiding service aggregation is generated. When the dynamic resource scarcity is lower than the low scarcity threshold, a discrete orchestration instruction for guiding service distribution is generated. When the dynamic resource scarcity is between the high scarcity threshold and the low scarcity threshold, a steady-state orchestration instruction to maintain the current service status is generated. The centralized orchestration instructions, discrete orchestration instructions, or steady-state orchestration instructions are collectively referred to as service grid dynamic orchestration instructions.

5. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 4, characterized in that, In response to the service mesh dynamic orchestration command, activating the process disturbance channel of the service architecture and injecting process disturbance factors into the service architecture through the process disturbance channel includes: The service-oriented architecture is pre-set with process disturbance channel entry points corresponding to different types of service mesh dynamic orchestration instructions; Based on the specific category of the service mesh dynamic orchestration instruction, the corresponding process disturbance channel entry is activated; In the activated process disturbance channel, a process disturbance factor with a specific dimension and intensity of action is generated based on the value of the dynamic resource scarcity. The generated process disturbance factor is injected into the core service dependency chain of the service architecture via the enabled process disturbance channel entry.

6. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 5, characterized in that, The step of initiating a steady-state resilience assessment of the service architecture after injecting the process disturbance factor, and calculating service structure compensation parameters based on the results of the steady-state resilience assessment, includes: Within the observation window after injecting the process disturbance factor, monitor the recovery trend of inter-service call latency and the regression tendency of service interface stability in the service-oriented architecture. Based on the recovery trend of the call latency and the regression tendency of the service interface stability, the steady-state resilience of the service architecture against internal disturbances is quantitatively evaluated. A steady-state resilience evaluation function is constructed, and the recovery trend of the call latency and the regression tendency of the service interface stability are used as input variables to calculate the steady-state resilience index. Based on the offset between the steady-state toughness index and the preset toughness reference band, the service structure compensation parameters are derived through the compensation parameter calculation model.

7. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 6, characterized in that, Based on the service structure compensation parameters, the microservices in the service-oriented architecture are reorganized and configured to form a compensated service mesh, including: The service structure compensation parameters are parsed into instance size compensation components and service link compensation components. The instance size compensation component is applied to elastically scale the number of deployment instances of the target microservice in the service-oriented architecture. The service link compensation component is applied to adjust the weight of the call links between microservices in the service architecture or to establish new backup links. After completing the compensation operations for instance size and service links, the reorganized service mesh is subjected to dependency closed-loop verification and conflict circuit breaking to form a stable and usable compensated service mesh.

8. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 7, characterized in that, The steady-state resilience assessment function is constructed by taking the recovery trend of the call latency and the regression tendency of the service interface stability as input variables, and calculating the steady-state resilience index, including: Weights are assigned to the recovery trend of the call latency and to the regression tendency of the service interface stability; Substitute the weighted call delay recovery trend and the weighted service interface stability regression tendency into the steady-state resilience evaluation function that includes nonlinear transformation; The steady-state resilience evaluation function is used to calculate and output a quantitative value that characterizes the system's ability to maintain service stability, namely the steady-state resilience index.

9. The collaborative scheduling method for a logistics platform based on a service-oriented architecture as described in claim 8, characterized in that, The construction methods of the preset resource field strength simulation model include: Collect historical collaborative scheduling records of the logistics platform within a complete scheduling evaluation cycle, wherein the scheduling evaluation cycle includes at least one complete peak-off-peak cycle sequence of logistics business. The historical collaborative scheduling records are cleaned and structured to extract scheduling request features, available resource status, scheduling decision results, and scheduling execution efficiency indicators to form a model training sample set. From the model training sample set, the average resource load level under different resource types and different network regions is calculated based on the time window moving average method. The calculation results are then fitted according to the time dimension and the spatial dimension to construct the historical resource load baseline that represents the background situation of resource supply and demand. From the model training sample set, time series analysis techniques are used to separate the periodic trend component, seasonal component and random fluctuation component in the resource load data; Based on the separated random fluctuation components and combined with the real-time acquired resource dynamic signals, the instantaneous diffusion process of resource fluctuations is simulated through spatial interpolation and propagation models to construct the real-time resource fluctuation surface that characterizes the real-time uncertainty of resources. The historical resource load baseline is used as the static background field of the simulation model, and the real-time resource fluctuation surface is used as the dynamic superposition field of the simulation model. The field strength coupling algorithm is used to fuse the static background field and the real-time resource fluctuation surface to form the resource field strength simulation model. The field strength coupling algorithm defines the dynamic adjustment rules for the contribution weights of the static background field and the dynamic superposition field to the final simulation result under different time scales and event triggering conditions.

10. A service-oriented architecture-based logistics platform collaborative scheduling system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the service-oriented architecture-based logistics platform collaborative scheduling method as described in any one of claims 1 to 9.

Images