A logistics scheduling platform for efficient supply chain distribution

Abstract

The application discloses a logistics scheduling platform for realizing efficient supply chain distribution, and relates to the technical field of intelligent logistics and supply chain management. The platform comprises: a data perception layer for collecting multi-source heterogeneous data of a whole supply chain in real time; a digital twin modeling layer for constructing a dynamic digital twin model comprising entity mapping, process simulation and situation deduction submodels; a multi-objective optimization engine for generating an initial scheduling scheme based on the model with total distribution cost, distribution timeliness and resource utilization rate as targets; a disturbance response module for monitoring disturbance events in real time, predicting the influence range and selecting to trigger local or global rescheduling; and an execution feedback module for feeding actual execution data to the model for online calibration. Through the closed-loop cooperation of dynamic digital twin and multi-objective optimization, the application realizes intelligent scheduling of supply chain distribution, effectively improves the distribution efficiency, resource utilization rate and system anti-disturbance ability, and is suitable for various logistics distribution scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent logistics and supply chain management technology, specifically to a logistics scheduling platform for achieving efficient supply chain distribution. Background Technology

[0002] Traditional logistics scheduling methods are mostly static solutions, making it difficult to respond in real time to multi-source heterogeneous data in supply chain operations. They also lack the ability to quickly perceive and accurately respond to disturbances such as traffic congestion and vehicle malfunctions, often requiring passive manual adjustments, which is inefficient. Furthermore, existing systems are mostly open-loop controls, lacking execution data feedback mechanisms and failing to continuously optimize model accuracy. In recent years, while digital twin technology has been attempted to be introduced into the logistics field, existing solutions still suffer from incomplete data collection, insufficient static updates of the twin model, difficulties in balancing multi-objective optimization, coarse disturbance responses (either over-adjustment or under-response), and a lack of closed-loop calibration. Therefore, this invention provides a logistics scheduling platform for achieving efficient supply chain distribution, aiming to address the aforementioned shortcomings. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a logistics scheduling platform for achieving efficient supply chain distribution.

[0004] To achieve the aforementioned objective, the technical solution of the present invention is implemented as follows: a logistics scheduling platform for achieving efficient supply chain distribution, comprising: The data perception layer is used to collect multi-source heterogeneous data from the entire supply chain in real time.

[0005] The digital twin modeling layer, connected to the data perception layer, is used to construct dynamic digital twin models based on multi-source heterogeneous data.

[0006] A multi-objective optimization engine, connected to the digital twin modeling layer, is used to generate an initial scheduling scheme based on a dynamic digital twin model.

[0007] The disturbance response module, connected to the multi-objective optimization engine, is used to monitor disturbance events in the delivery process in real time, predict the impact range of disturbance events based on a dynamic digital twin model, and trigger rescheduling based on the prediction results.

[0008] The execution feedback module, connected to the disturbance response module, is used to feed back the actual execution data to the digital twin modeling layer to update the dynamic digital twin model.

[0009] Preferably, the dynamic digital twin model includes an entity mapping sub-model for mapping logistics resource entities to digital space, a process simulation sub-model for dynamic simulation of the delivery process, and a situation inference sub-model for multi-branch prediction of system situation.

[0010] Preferably, the multi-objective optimization engine takes total delivery cost, delivery timeliness and resource utilization as optimization objectives, and uses an optimization strategy that combines constraint hierarchical processing, adaptive weight adjustment and multi-group parallel search to generate an initial scheduling scheme.

[0011] Preferably, the disturbance response module includes: The disturbance identification unit is used to identify the type and assess the level of disturbance events; Impact prediction unit, used to predict the propagation path and impact of disturbance events in the supply chain network based on dynamic digital twin model; The rescheduling decision unit is used to selectively trigger local or global rescheduling based on the predicted impact range.

[0012] Preferably, local rescheduling means: while keeping the scheduling scheme of unaffected delivery tasks unchanged, only the local delivery tasks and local logistics resources affected by the disturbance event are re-optimized; global rescheduling means: all unfinished delivery tasks and all available logistics resources are re-optimized.

[0013] Preferably, the execution feedback module includes a deviation calculation unit and a model calibration unit. The deviation calculation unit is used to calculate the deviation value between the actual execution data and the scheduling scheme, and the model calibration unit is used to perform online calibration of the parameters of the dynamic digital twin model based on the deviation value.

[0014] The present invention also provides a supply chain distribution scheduling method using the above-mentioned logistics scheduling platform, comprising: Collect multi-source heterogeneous data from the entire supply chain; A dynamic digital twin model is constructed based on multi-source heterogeneous data; An initial scheduling scheme is generated based on a dynamic digital twin model; Perform delivery tasks and monitor disturbance events in real time; When a disturbance event is detected, the impact range of the disturbance event is predicted based on a dynamic digital twin model, and rescheduling is triggered based on the prediction results; The actual execution data is fed back to the dynamic digital twin model to update the model parameters.

[0015] Preferably, the dynamic digital twin model is built based on historical data and is updated on a rolling basis using real-time collected multi-source heterogeneous data, with the update frequency adaptively adjusted according to the data change rate.

[0016] Preferably, rescheduling includes: triggering local rescheduling when the impact range of the disturbance event is less than a preset threshold, and triggering global rescheduling when the impact range is greater than or equal to the preset threshold.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method.

[0018] The beneficial effects of this invention are reflected in: This invention achieves real-time collection and fusion of multi-source heterogeneous data across the entire supply chain through a data perception layer, providing a comprehensive and timely data foundation for scheduling decisions. The digital twin modeling layer dynamically constructs and continuously updates a three-layer twin model, including entity mapping, process simulation, and situational simulation, ensuring real-time synchronization between the model and the physical system. The multi-objective optimization engine employs constraint hierarchical design, adaptive weight adjustment, and multi-group parallel search to effectively balance delivery costs, timeliness, and resource utilization. The disturbance response module accurately predicts the impact range of disturbances based on the twin model and selectively triggers local or global rescheduling, maintaining solution stability while preventing disturbance propagation. The execution feedback module forms a closed-loop optimization through deviation calculation and online model calibration, continuously improving system performance. The model update frequency can also be adaptively adjusted according to the data change rate, balancing accuracy and computational efficiency. Attached Figure Description

[0019] In the attached diagram: Figure 1 This is a system architecture block diagram of the logistics scheduling platform for efficient supply chain distribution according to the present invention; Figure 2 This is a schematic diagram of the structure of the dynamic digital twin model of the present invention; Figure 3 This is a schematic diagram illustrating the workflow of the multi-objective optimization engine of the present invention; Figure 4 This is a structural block diagram of the disturbance response module of the present invention; Figure 5 This is a structural block diagram of the feedback module of the present invention; Figure 6 This is a flowchart illustrating the supply chain distribution scheduling method of the present invention; Figure 7 This is a schematic diagram of the switching decision process between local rescheduling and global rescheduling in this invention. Detailed Implementation

[0020] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the invention, and not all of them. Unless otherwise specified, the embodiments and features described in this application can be combined with each other. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] Example 1: Overall Architecture of the Logistics Dispatch Platform like Figure 1 As shown, the present invention provides a logistics scheduling platform for achieving efficient supply chain distribution, including a data perception layer, a digital twin modeling layer, a multi-objective optimization engine, a disturbance response module, and an execution feedback module.

[0022] The data perception layer is used to collect multi-source heterogeneous data from the entire supply chain in real time. The entire supply chain covers the complete logistics chain from raw material suppliers, manufacturing enterprises, warehousing centers, distribution centers to end customers. Multi-source heterogeneous data includes, but is not limited to: order data (order volume, order type, priority, time window, etc.), inventory data (inventory quantity, inventory location, inventory turnover rate, etc.), vehicle data (vehicle location, speed, load, status, fuel consumption, etc.), road condition data (real-time traffic flow, road conditions, traffic restriction information, etc.), weather data (rainfall, snowfall, temperature, wind speed, etc.), personnel data (driver status, working hours, etc.), and equipment data (loading and unloading equipment status, warehousing equipment operation data, etc.). The data perception layer uses various sensing devices such as IoT sensors, GPS positioning devices, RFID tags, vehicle terminals, cameras, and weather stations, as well as data interfaces with enterprise ERP systems, WMS systems, TMS systems, and other business systems to achieve real-time collection and unified aggregation of multi-source heterogeneous data. The data perception layer also has data preprocessing capabilities, including data cleaning, data alignment, data fusion, and standardization, which convert multi-source heterogeneous data into a standardized data stream in a unified format for use by upper-layer modules.

[0023] The digital twin modeling layer is connected to the data perception layer, and is used to construct a dynamic digital twin model based on multi-source heterogeneous data. For example... Figure 2 As shown, the dynamic digital twin model comprises three sub-models: (1) Entity mapping sub-model, used to realize one-to-one mapping between logistics resource entities and digital space. Logistics resource entities include delivery vehicles, warehousing facilities, loading and unloading equipment, delivery personnel, delivery routes, delivery nodes (warehouses, distribution centers, customer points), etc. The entity mapping sub-model establishes a corresponding digital twin for each physical entity, recording the entity's static attributes (such as vehicle model, load capacity, size specifications, etc.) and dynamic attributes (such as real-time location, operating status, remaining capacity, etc.), to realize real-time synchronization between physical space and digital space.

[0024] (2) Process simulation sub-model, used for dynamic simulation of the delivery process. Based on the entity state data provided by the entity mapping sub-model, the process simulation sub-model performs dynamic simulation of each link in the delivery process (order receiving, task allocation, route planning, loading operation, transportation and delivery, unloading and delivery, etc.), simulates the execution process of delivery tasks, and evaluates the system performance under different scheduling schemes.

[0025] (3) Situation simulation sub-model, used for multi-branch prediction of system situation. Based on the current system state and historical data, the situation simulation sub-model performs multi-branch prediction of the future situation of the supply chain distribution system, predicts the evolution trend of the system under different decision conditions, and provides forward-looking decision support for scheduling decisions and disturbance response.

[0026] The digital twin modeling layer constructs an initial digital twin model based on historical data, and uses multi-source heterogeneous data collected in real time by the data perception layer to continuously update the model. The update frequency is adaptively adjusted according to the data change rate: when the data change rate is high (such as during peak traffic hours or severe weather), the update frequency is increased to ensure model accuracy; when the data change rate is low (such as during off-peak hours at night), the update frequency is decreased to reduce computational resource consumption.

[0027] A multi-objective optimization engine is connected to the digital twin modeling layer to generate an initial scheduling scheme based on a dynamic digital twin model. For example... Figure 3 As shown, the multi-objective optimization engine has three main optimization objectives: total delivery cost, delivery timeliness, and resource utilization rate. Total delivery cost includes transportation costs (fuel costs, toll fees, vehicle depreciation, etc.), labor costs (driver salaries, loading and unloading personnel salaries, etc.), warehousing costs (inventory holding costs, warehousing facility usage costs, etc.), and penalty costs (delay penalties, breach of contract compensation, etc.). Delivery timeliness includes indicators such as order completion time, on-time delivery rate, and average delivery time. Resource utilization rate includes indicators such as vehicle loading rate, vehicle utilization rate, warehouse space utilization rate, and personnel utilization rate.

[0028] The multi-objective optimization engine employs an optimization strategy that combines constraint hierarchical processing, adaptive weight adjustment, and multi-group parallel search. Constraint layering: Constraints in delivery scheduling are layered according to their importance. The first layer consists of hard constraints, which must be met unconditionally, such as vehicle load constraints, volume constraints, time window constraints, and driving time regulations. The second layer consists of soft constraints, which can be relaxed under certain conditions, such as cost budget constraints and service priority constraints. By layering constraints, the objective function is optimized as much as possible while ensuring the feasibility of the solution.

[0029] Adaptive weight adjustment: For the three optimization objectives of total delivery cost, delivery timeliness, and resource utilization, the weight coefficients of each objective are dynamically adjusted based on the real-time operational status of the supply chain and business priorities. For example, the weight of delivery timeliness is increased during peak promotional seasons, the weight of total delivery cost is increased when cost pressures are high, and the weight of resource utilization is increased when resources are scarce. Through adaptive weight adjustment, a dynamic balance is achieved among multiple objectives.

[0030] Multi-population parallel search: A multi-population parallel evolutionary algorithm is employed for optimization. The population is divided into multiple subpopulations, each employing different search strategies (such as different mutation operators, crossover operators, or different objective preferences). Each subpopulation evolves in parallel and exchanges superior individuals through a migration operator, improving convergence speed while maintaining global search capability. The multi-objective optimization engine ultimately outputs a set of Pareto optimal solutions, allowing decision-makers to select the final scheduling scheme based on actual preferences.

[0031] Example 2: Disturbance Response Module like Figure 4 As shown, the disturbance response module is connected to the multi-objective optimization engine and includes a disturbance identification unit, an impact prediction unit, and a rescheduling decision unit.

[0032] The disturbance identification unit is used to identify and assess the level of disturbance events. Disturbance events include, but are not limited to: traffic disturbances (traffic accidents, road closures, traffic congestion, etc.), vehicle disturbances (vehicle malfunctions, vehicle delays, vehicle deviations from routes, etc.), order disturbances (new orders, canceled orders, order changes, etc.), environmental disturbances (severe weather, natural disasters, etc.), personnel disturbances (driver absences, personnel injuries, etc.), and equipment disturbances (loading and unloading equipment malfunctions, warehousing equipment malfunctions, etc.). The disturbance identification unit monitors the data stream transmitted from the data perception layer in real time, uses a rule engine or anomaly detection algorithm to identify the occurrence of disturbance events, and classifies and assesses the level of disturbance events (e.g., minor, moderate, severe, and major).

[0033] The impact prediction unit is used to predict the propagation path and extent of the impact of disturbance events in the supply chain network based on a dynamic digital twin model. The impact prediction unit calls the situational inference sub-model of the digital twin modeling layer, takes the disturbance event as input, and performs multi-branch situational inference in the digital twin model to predict the propagation path of the disturbance event in the supply chain network (e.g., which orders will be delayed due to a vehicle malfunction, and how the delay will propagate to downstream links) and the extent of the impact (e.g., the number of affected orders, the number of affected delivery nodes, the amount of increased costs, and the amount of time delay).

[0034] The rescheduling decision unit is used to selectively trigger local or global rescheduling based on the predicted impact range. For example... Figure 7 As shown, the rescheduling decision unit compares the range of influence affecting the prediction unit's output with a preset threshold: When the impact of a disturbance event is less than a preset threshold (e.g., affecting only a single vehicle or a small number of orders at a single delivery node), a local rescheduling is triggered. This involves re-optimizing only the local delivery tasks and logistics resources affected by the disturbance event, while maintaining the existing scheduling scheme for unaffected delivery tasks. Local rescheduling re-solves the problem locally by invoking a multi-objective optimization engine, resulting in low computational cost, fast response time, and the ability to maintain the stability of the overall scheduling scheme.

[0035] When the impact of a disturbance event is greater than or equal to a preset threshold (such as affecting multiple delivery nodes, multiple delivery routes, or a large number of orders), a global rescheduling is triggered: all unfinished delivery tasks and all available logistics resources are re-optimized. The global rescheduling calls the multi-objective optimization engine to solve the problem globally, generating a completely new scheduling scheme to ensure the overall optimal operation of the supply chain under major disturbances.

[0036] Through this hierarchical response mechanism, the present invention achieves a precise response to disturbance events, which avoids frequent changes in the scheme and system instability caused by small disturbances triggering large adjustments, and also prevents the disturbance from spreading and losses from being amplified due to insufficient response under large disturbances.

[0037] Example 3: Execution Feedback Module like Figure 5 As shown, the execution feedback module is connected to the disturbance response module, and includes a deviation calculation unit and a model calibration unit.

[0038] The deviation calculation unit is used to calculate the deviation between the actual execution data and the scheduling plan. During the delivery task execution, the execution feedback module continuously receives actual execution data from the data perception layer (such as actual vehicle arrival time, actual driving route, actual load, actual unloading time, etc.), compares the actual execution data with the corresponding planned data in the scheduling plan item by item, and calculates various deviation values, including time deviation (the difference between planned arrival time and actual arrival time), route deviation (the difference between planned route and actual driving route), load deviation (the difference between planned load and actual load), and cost deviation (the difference between planned cost and actual cost).

[0039] The model calibration unit is used to calibrate the parameters of the dynamic digital twin model online based on deviation values. The unit analyzes the statistical characteristics and distribution patterns of various deviation values ​​to identify systematic factors (such as inaccurate model parameters or invalid assumptions) and random factors (such as occasional traffic events) that cause deviations. For deviations caused by systematic factors, the unit uses Bayesian inference and online learning methods to calibrate and update the corresponding parameters of the digital twin model, enabling the model to more accurately reflect the real operational characteristics of the supply chain. For deviations caused by random factors, the unit records them as noise data and does not use them for adjusting model parameters.

[0040] By implementing the closed-loop feedback mechanism of the feedback module, the dynamic digital twin model can be continuously optimized as running data accumulates, and the system performance can be continuously improved.

[0041] Example 4: Supply Chain Distribution Scheduling Method like Figure 6 As shown, the present invention also provides a supply chain distribution scheduling method using the above-mentioned logistics scheduling platform, comprising the following steps: Step S1: Collect multi-source heterogeneous data from the entire supply chain. Collect multi-source heterogeneous data from the entire supply chain in real time through the data perception layer.

[0042] Step S2: Construct a dynamic digital twin model based on multi-source heterogeneous data. The digital twin modeling layer constructs a dynamic digital twin model based on the collected multi-source heterogeneous data. The model is initially built based on historical data and is continuously updated using real-time collected multi-source heterogeneous data. The update frequency is adaptively adjusted according to the data change rate.

[0043] Step S3: Generate an initial scheduling scheme based on the dynamic digital twin model. The multi-objective optimization engine, based on the dynamic digital twin model, uses total delivery cost, delivery timeliness, and resource utilization as optimization objectives. It generates an initial scheduling scheme by combining constraint hierarchical processing, adaptive weight adjustment, and multi-group parallel search.

[0044] Step S4: Execute the delivery task and monitor disturbance events in real time. The initial scheduling plan is sent to the execution end to start the delivery task, while the disturbance response module monitors disturbance events in real time during the delivery execution process.

[0045] Step S5: When a disturbance event is detected, the impact range of the disturbance event is predicted based on the dynamic digital twin model, and rescheduling is triggered according to the prediction results. Specifically, this includes: the disturbance identification unit identifies the type and assesses the level of the disturbance event; the impact prediction unit predicts the propagation path and degree of impact of the disturbance event in the supply chain network based on the dynamic digital twin model; the rescheduling decision unit selectively triggers local or global rescheduling according to the predicted impact range—local rescheduling is triggered when the impact range is less than a preset threshold, and global rescheduling is triggered when the impact range is greater than or equal to the preset threshold.

[0046] Step S6: Feed the actual execution data back to the dynamic digital twin model to update the model parameters. The execution feedback module continuously collects the actual execution data, calculates the deviation between the actual execution data and the scheduling plan, and performs online calibration of the parameters of the dynamic digital twin model based on the deviation value.

[0047] Example 5: Computer-readable storage medium The present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the above-described supply chain distribution scheduling method. The computer-readable storage medium includes, but is not limited to, various media capable of storing program code, such as ROM, RAM, magnetic disk, optical disk, USB flash drive, and portable hard drive.

[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A logistics scheduling platform for achieving efficient supply chain distribution, characterized in that, include: The data perception layer is used to collect multi-source heterogeneous data from the entire supply chain in real time. A digital twin modeling layer, connected to the data perception layer, is used to construct a dynamic digital twin model based on the multi-source heterogeneous data; A multi-objective optimization engine, connected to the digital twin modeling layer, is used to generate an initial scheduling scheme based on the dynamic digital twin model; The disturbance response module, connected to the multi-objective optimization engine, is used to monitor disturbance events in the delivery process in real time, predict the impact range of the disturbance events based on the dynamic digital twin model, and trigger rescheduling based on the prediction results. An execution feedback module, connected to the disturbance response module, is used to feed back actual execution data to the digital twin modeling layer to update the dynamic digital twin model.

2. The logistics scheduling platform for achieving efficient supply chain distribution according to claim 1, characterized in that, The dynamic digital twin model includes an entity mapping sub-model for mapping logistics resource entities to digital space, a process simulation sub-model for dynamic simulation of the delivery process, and a situation inference sub-model for multi-branch prediction of system situation.

3. The logistics scheduling platform for achieving efficient supply chain distribution according to claim 1, characterized in that, The multi-objective optimization engine optimizes the total delivery cost, delivery time, and resource utilization rate, and uses an optimization strategy that combines constraint hierarchical processing, adaptive weight adjustment, and multi-group parallel search to generate the initial scheduling scheme.

4. The logistics scheduling platform for achieving efficient supply chain distribution according to claim 1, characterized in that, The disturbance response module includes: A disturbance identification unit is used to identify the type and assess the level of the disturbance event; The impact prediction unit is used to predict the propagation path and impact extent of the disturbance event in the supply chain network based on the dynamic digital twin model. The rescheduling decision unit is used to selectively trigger local or global rescheduling based on the predicted impact range.

5. The logistics scheduling platform for achieving efficient supply chain distribution according to claim 4, characterized in that, The local rescheduling refers to: re-optimizing only the local delivery tasks and local logistics resources affected by the disturbance event while keeping the scheduling scheme of the unaffected delivery tasks unchanged; the global rescheduling refers to: re-optimizing all incomplete delivery tasks and all available logistics resources.

6. The logistics scheduling platform for achieving efficient supply chain distribution according to claim 1, characterized in that, The execution feedback module includes a deviation calculation unit and a model calibration unit. The deviation calculation unit is used to calculate the deviation value between the actual execution data and the scheduling scheme. The model calibration unit is used to perform online calibration of the parameters of the dynamic digital twin model based on the deviation value.

7. A supply chain distribution scheduling method using the logistics scheduling platform described in any one of claims 1 to 6, characterized in that, include: Collect multi-source heterogeneous data from the entire supply chain; A dynamic digital twin model is constructed based on the multi-source heterogeneous data; An initial scheduling scheme is generated based on the aforementioned dynamic digital twin model; Perform delivery tasks and monitor disturbance events in real time; When a disturbance event is detected, the impact range of the disturbance event is predicted based on the dynamic digital twin model, and rescheduling is triggered according to the prediction result; The actual execution data is fed back to the dynamic digital twin model to update the model parameters.

8. The supply chain distribution scheduling method according to claim 7, characterized in that, The dynamic digital twin model is built based on historical data and is updated on a rolling basis using the multi-source heterogeneous data collected in real time. The update frequency is adaptively adjusted according to the data change rate.

9. The supply chain distribution scheduling method according to claim 7, characterized in that, The rescheduling includes: triggering local rescheduling when the impact range of the disturbance event is less than a preset threshold, and triggering global rescheduling when the impact range is greater than or equal to the preset threshold.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 7 to 9.

Images