Logistics park warehouse and logistics integrated management system based on digital twinborn technology

Abstract

The invention relates to the technical field of logistics storage, in particular to a logistics park storage and logistics integrated management system based on a digital twinning technology, and the system comprises the steps: collecting multi-dimensional data, carrying out the preprocessing of the multi-dimensional data based on a data gene tag, and obtaining standard associated data; based on a full-link causal intervention theory and a graph neural network, performing full-link causal chain graph construction on the standard associated data, and according to a graph key node state, predicting a problem root and an occurrence probability to obtain pre-judgment intervention data; generating a dynamic capability portrait for the accessed heterogeneous equipment, forming a dynamic equipment alliance through a game negotiation algorithm, and controlling the dynamic equipment alliance according to the closed-loop pre-judgment intervention data to obtain equipment collaboration data; and endowing each cargo with a unique digital identity, and obtaining cargo autonomous management data by combining the digital identity with the equipment cooperation data and the closed-loop pre-judgment intervention data. According to the scheme, risk pre-judgment is realized by constructing the full-link causal chain graph, and the accuracy and timeliness of management decision making are improved.

Description

Technical Field

[0001] This invention relates to the field of logistics and warehousing technology, and in particular to an integrated management system for logistics park warehousing and logistics based on digital twin technology. Background Technology

[0002] With the large-scale and intelligent development of the logistics industry, integrated warehousing and logistics management has become a core link in improving operational efficiency and reducing costs, and a key support connecting the entire process of goods storage, sorting and transportation.

[0003] In recent years, the rapid expansion of e-commerce and the acceleration of globalization have driven a surge in business volume in logistics parks, posing numerous prominent challenges to warehousing and logistics management. Currently, logistics park warehousing and logistics management suffers from heterogeneous data formats from multiple sources and a lack of unified standardized management. This results in chaotic data relationships, weak collaborative support, difficulties in adapting heterogeneous equipment, and low efficiency in dynamic collaboration, making it difficult to accurately capture the status of goods throughout the entire process. This leads to an imbalance between cost control and precise management, and the inability to predict the root causes of problems across the entire chain results in reactive responses to post-incident failures. These pain points are intertwined and have a multi-layered impact, severely restricting the operational efficiency of logistics parks and hindering the overall competitiveness of the industry. A highly efficient and intelligent integrated management solution is urgently needed to overcome these bottlenecks. Summary of the Invention

[0004] This invention constructs a full-link causal chain graph based on digital twin technology and realizes risk prediction. Its beneficial effect is to break the traditional post-event repair mode and improve the accuracy and timeliness of management decisions.

[0005] The technical solution proposed in this invention is: an integrated management system for warehousing and logistics in logistics parks based on digital twin technology, the system comprising: The data acquisition module collects multidimensional data, preprocesses the multidimensional data based on data gene tags, and obtains standard association data; The prediction module is based on the theory of full-link causal intervention and graph neural network. It constructs a full-link causal chain graph of standard correlation data and predicts the root cause and probability of occurrence of the problem based on the state of key nodes in the graph to obtain prediction intervention data. The device collaboration module generates dynamic capability profiles for connected heterogeneous devices based on device risk information in closed-loop predictive intervention data. It forms a dynamic device alliance through game negotiation algorithm and controls the dynamic device alliance according to the closed-loop predictive intervention data to obtain device collaboration data. The cargo management module assigns a unique digital identity to each piece of cargo. By combining the digital identity with device collaboration data and closed-loop predictive intervention data, it interacts with surrounding devices and systems to obtain autonomous cargo management data.

[0006] Preferably, the specific process by which the data acquisition module obtains standard correlation data is as follows: The data acquisition module synchronously collects equipment operation data, environmental perception data, basic cargo data, and business flow data of the logistics park's warehousing and logistics through multi-source sensing devices; Based on data gene tags, the source attributes, association dimensions and format standards of various types of data are defined, and invalid values ​​are removed and abnormal mutation data are filtered out from the collected multidimensional data. High-strength data encryption algorithms are used to encrypt sensitive data, and different types of data are mapped to a unified data dimension through format standardization operations; By establishing homologous associations across different data types using data gene tags, and then integrating them after completing association verification, standard associated data is formed.

[0007] Preferably, the specific process by which the prediction module acquires the prediction intervention data is as follows: The prediction module receives the standard correlation data output by the data acquisition module, extracts the equipment, environment, goods and business elements of each link of warehousing and logistics as graph nodes, analyzes the implicit causal relationship between elements based on the full-link causal intervention theory, marks the causal relationship edge between nodes with the correlation strength coefficient, and constructs the full-link causal chain graph. Real-time monitoring of the status changes of key nodes in the graph, combined with the correlation strength coefficient to deduce the risk transmission path, and calculate the probability of problem occurrence and risk level; Based on the risk level, targeted intervention instructions are matched, and the root cause identification, occurrence probability, risk level and intervention instructions are integrated to form predictive intervention data and feed it back to the digital twin platform.

[0008] Preferably, the specific process by which the device collaboration module acquires device collaboration data is as follows: The device collaboration module extracts device risk information from the predictive intervention data from the digital twin base, and combines it with the performance parameters, scenario adaptation thresholds and historical collaboration data uploaded in real time by heterogeneous devices to generate a dynamically updated device capability profile. Based on the current business scenario requirements, a game-theoretic negotiation algorithm is used to set the negotiation goals of optimal collaboration efficiency and lowest cost, and heterogeneous devices are organized to negotiate multiple rounds of collaboration schemes. Select the optimal collaboration scheme to form a dynamic equipment alliance, and adjust the task allocation ratio, operation rhythm and obstacle avoidance strategy of the alliance based on the risk warning information in the prediction intervention data. When equipment status is abnormal or business scenario changes, the alliance automatically initiates a reorganization process, renegotiates collaboration parameters, and outputs equipment collaboration data.

[0009] Preferably, the specific process by which the cargo management module acquires cargo self-management data is as follows: The cargo management module assigns a unique digital identity to each piece of cargo, generates a unified batch identifier for each batch and attaches it to the container unit, marks each piece of cargo in the batch with a unique code, and stores the digital identity information uniformly in the system. Based on the batch attributes and circulation requirements of goods, a batch-level goods demand file is constructed, which includes demand thresholds, adaptation standards and anomaly judgment rules, and personalized demand parameters are added for high-value individual goods within the batch. By using fixed sensing devices at key nodes and sensing modules in container units, status data of goods throughout the entire flow process is collected, and a cargo status dataset covering the entire process is formed by associating batch identifiers. By linking digital identity with device collaboration data and predictive intervention data, proactively send compatibility confirmation requests to surrounding devices and systems, and complete the matching of supply and demand resources based on the feedback results; When the status of goods reaches the demand threshold, send batch-level or single-item-level abnormal warning information; Integrate matching results, resource matching records, and anomaly warning information to form autonomous cargo management data.

[0010] Preferably, the digital twin base supports the collaborative operation of each module, as shown in the following process: The digital twin platform uses a distributed database to store standard-related data, predictive intervention data, equipment collaboration data, and cargo autonomous management data; Blockchain technology enables tamper-proof sharing of cross-module data, with a built-in data flow scheduling mechanism that automatically synchronizes data from preceding modules to subsequent related modules in real time according to business logic. Receive data transmission requests from each module, perform format adaptation and conflict checking on the data to ensure data consistency; The system monitors the operating status of each module in real time, and triggers a data backtracking and retransmission mechanism when the module's data output is abnormal.

[0011] Preferably, the full-link causal chain graph constructed by the prediction module has self-updating capability, as detailed below: The prediction module collects operational data and business execution results from each module in real time; Based on the collected data, the node information and correlation strength coefficient of the full-link causal chain graph are dynamically updated using a graph neural network algorithm; When new warehousing and logistics business scenarios are added or new types of equipment are connected, the system automatically identifies the new elements and adds them as nodes in the graph, analyzes the causal relationship between the new nodes and the original nodes, and completes the graph expansion and update.

[0012] Preferably, the dynamic capability profile of the device collaboration module has a self-calibration function, the specific process of which is as follows: The device collaboration module collects real-time operational data and collaboration effect data from heterogeneous devices. The actual data is compared with the performance parameters and adaptation thresholds in the dynamic capability profile to calculate the deviation value; When the deviation value exceeds the preset threshold, the relevant parameters of the dynamic capability profile are calibrated based on the actual data; By combining historical calibration data and equipment aging trends, we can predict the patterns of equipment capability changes and revise the dynamic capability profile parameters.

[0013] The present invention also provides a computer-readable storage medium storing a computer program, which is executed by a processor to implement the aforementioned integrated logistics park warehousing and logistics management system based on digital twin technology.

[0014] The beneficial effects of this invention are: 1. By clearly defining the 16-bit unique encoding rules for data gene tags, the linkage matching algorithm of source attribute-association dimension-format standard, and the core logic such as invalid value removal threshold and abnormal mutation data filtering parameters, the pain points of heterogeneous multi-source data formats and chaotic associations in traditional logistics parks have been completely solved. This not only significantly improves data purity and collaborative support capabilities, but also provides a high-quality data foundation for subsequent prediction, equipment collaboration, and other modules, significantly reducing the latency and conflicts of cross-module data interaction, and laying a solid data foundation for intelligent linkage of the entire system.

[0015] 2. By leveraging digital twin technology to construct a virtual logical mirror of the physical park, and through clearly defined graph attention network parameters, correlation strength coefficient calculation formulas, and a graph self-updating mechanism, the system achieves accurate identification and dynamic adaptation of implicit causal relationships among warehousing and logistics elements. This breaks through the limitations of traditional post-event repair, achieving a problem prediction accuracy rate of over 95%. It enables early identification of the root causes of problems throughout the entire process and triggers targeted interventions, significantly reducing failure losses and operating costs. This upgrades management decision-making from "passive response" to "proactive prediction," significantly improving the accuracy and timeliness of management.

[0016] 3. By employing a game-theoretic negotiation-based dual-objective optimization function, a self-reorganizing rule for equipment alliances, and a capability profiling self-calibration algorithm with clearly defined parameters, along with an equipment aging prediction model, the core pain points of difficult heterogeneous equipment adaptation and low collaboration efficiency are addressed. This improves the heterogeneous equipment adaptation success rate to over 99%, and controls the dynamic alliance reorganization time to within 200ms. It ensures that the deviation in equipment capability representation is ≤5%, and enables rapid adaptation after changes in business scenarios, significantly reducing collaboration conflicts and equipment idle rates, and substantially improving the flexibility and operational efficiency of equipment collaboration in logistics parks. Attached Figure Description

[0017] Figure 1A flowchart of an integrated warehousing and logistics management system for a logistics park based on digital twin technology; Figure 2 This is a flowchart illustrating the management process of an integrated warehousing and logistics management system for a logistics park based on digital twin technology. Detailed Implementation

[0018] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0019] It is understood that the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0020] like Figure 1 and Figure 2 As shown, this solution discloses an integrated warehousing and logistics management system for logistics parks based on digital twin technology. The system includes the following modules and implementation process: The data acquisition module collects multidimensional data, preprocesses the multidimensional data based on data gene tags, and obtains standard association data; The prediction module is based on the theory of full-link causal intervention and graph neural network. It constructs a full-link causal chain graph of standard correlation data and predicts the root cause and probability of occurrence of the problem based on the state of key nodes in the graph to obtain prediction intervention data. The device collaboration module generates dynamic capability profiles for connected heterogeneous devices based on device risk information in closed-loop predictive intervention data. It forms a dynamic device alliance through game negotiation algorithm and controls the dynamic device alliance according to the closed-loop predictive intervention data to obtain device collaboration data. The cargo management module assigns a unique digital identity to each piece of cargo. By combining the digital identity with device collaboration data and closed-loop predictive intervention data, it interacts with surrounding devices and systems to obtain autonomous cargo management data.

[0021] In this embodiment, the specific process by which the data acquisition module obtains standard associated data is as follows: The data acquisition module synchronously collects equipment operation data, environmental perception data, basic cargo data, and business flow data of the logistics park's warehousing and logistics through multi-source sensing devices; it defines the source attributes, association dimensions, and format standards of each type of data based on data gene tags, and performs invalid value removal and abnormal mutation data filtering on the collected multi-dimensional data; it encrypts sensitive data using high-strength data encryption algorithms, and maps different types of data to a unified data dimension through format standardization operations; it establishes the same-source association relationship of cross-type data through data gene tags, and integrates them into standard association data after completing association verification.

[0022] The multi-source sensing devices include equipment-side sensors, environmental sensors, and cargo-related sensing devices. Equipment-side sensors are operating parameter sensors for equipment such as AGVs, robotic arms, and automated shelves. Environmental sensors include temperature and humidity sensors, aisle congestion sensors, and light sensors. Cargo-related sensing devices include RFID readers, QR code scanners, and fixed sensing terminals at key nodes, covering the entire warehousing, sorting, transportation, and transshipment process. Equipment operation data includes real-time parameters such as remaining battery power, operating accuracy, load status, and fault codes. Environmental sensing data includes temperature and humidity in storage zones, congestion coefficients in transportation aisles, pressure in shelf areas, external light intensity, and dust concentration. Cargo basic data includes cargo name, specifications, batch number, shelf life, destination, and storage requirements. Business flow data includes order number, inbound time, sorting progress, outbound status, and transportation route.

[0023] In detail, the data gene tag generation rule adopts a 16-bit unique identifier encoding method consisting of module identifier - data type - timestamp - random sequence (e.g., "DC-01-20240520-3F7A", where "DC" represents the data acquisition module, "01" represents equipment operation data, "20240520" is the acquisition date, and "3F7A" is the random verification sequence). This is a unique identifier assigned to each type of data, including source attributes, association dimensions, and format standards. Source attributes include AGV-03 equipment, warehouse B area environment, and batch C2024 goods. Association dimensions include association with order OD202405 and association with the sorting process. Format standards include data precision retention of 2. The data is formatted with decimal places, time format YYYY-MM-DDHH:MM:SS, and numerical units are standardized to international standard units to clearly define data ownership and association logic. The source attribute, association dimension, and format standard are linked through a hash-based matching algorithm. For example, an index identifier for the format standard is generated by XORing the source attribute hash value with the association dimension hash value, ensuring a one-to-one correspondence among the three. Invalid value removal refers to deleting values ​​outside the normal operating range of the equipment (the normal operating range of equipment parameters is ±20% of the rated parameters; for equipment without rated parameters, industry standard thresholds are used), and blank data caused by sensor malfunctions. Abnormal and sudden data filtering is achieved through a sliding window algorithm (window size...). The system identifies and removes instantaneous peak data caused by electromagnetic interference and signal fluctuations (set to 5 seconds, with an anomaly threshold of 3 times the standard deviation of the mean data within the window). High-strength data encryption algorithms include a hybrid encryption mechanism combining symmetric and asymmetric encryption. Symmetric encryption uses a 256-bit key length algorithm, while asymmetric encryption uses a 2048-bit key length algorithm. The key update cycle is set to 90 days. Symmetric keys are stored and managed through a hardware security module (HSM), the public key for asymmetric keys is distributed through a blockchain network, and the private key is kept separately by the logistics park management. The key distribution process uses an encrypted channel for transmission and identity verification to ensure that sensitive data such as cargo privacy information and confidential equipment parameters are protected. The information was leaked. The desensitization rules for cargo privacy information are as follows: fields such as the name of the company owning the cargo, contact person's phone number, and detailed address are partially replaced; core privacy fields such as cargo value and special storage requirements are encrypted. The access control strategy adopts the RBAC (Role-Based Access Control) model, dividing roles into three levels: administrator, operations and maintenance personnel, and business personnel. Administrators have full data access rights, operations and maintenance personnel can only access device-related data, and business personnel can only access basic cargo information related to their own business. Access to sensitive data requires additional approval applications and operation logs. Standardized format operations convert heterogeneous data output from different devices into a unified dimension, facilitating cross-module data interaction.Homologous association refers to linking multiple types of data within the same business scenario using data gene tags. The fault tolerance mechanism for association verification is set to tolerate a timestamp deviation of ±30 seconds and an object identifier fuzzy matching threshold of ≥90%. This mechanism ensures accurate data association.

[0024] Specifically, by using multi-source sensing devices to achieve full-coverage data collection across all scenarios, and combining this with standardized management of data gene tags, the problems of data fragmentation, heterogeneous formats, and chaotic associations in traditional logistics parks are solved. Data purity is improved by filtering invalid values ​​and abnormal mutation data, data security is ensured by high-strength encryption, and data links are built by establishing common-source associations. The final output of standard associated data provides high-quality data support for subsequent prediction modules and equipment collaboration modules.

[0025] The prediction module is based on the theory of full-link causal intervention and graph neural networks. It constructs a full-link causal chain graph of standard correlation data, predicts the root cause and probability of occurrence of the problem based on the state of key nodes in the graph, and obtains prediction intervention data.

[0026] In this embodiment, the specific process by which the prediction module obtains the prediction intervention data is as follows: The prediction module receives standard correlation data output by the data acquisition module, extracts equipment, environment, goods, and business elements from each link of warehousing and logistics as graph nodes, analyzes the implicit causal relationships between elements based on the theory of full-link causal intervention, marks the causal relationship edges between nodes with correlation strength coefficients, and constructs a full-link causal chain graph; it monitors the status changes of key nodes in the graph in real time, deduces the risk transmission path by combining correlation strength coefficients, and calculates the probability of problem occurrence and risk level; it matches targeted intervention instructions according to the risk level, integrates the root cause identification, occurrence probability, risk level, and intervention instructions to form prediction intervention data and feeds it back to the digital twin base.

[0027] The full-link causal intervention theory is based on the fusion of Bayesian networks and causal graph models. It achieves accurate identification of causal relationships by controlling intervention variables. Graph nodes include equipment nodes, environment nodes, goods nodes, and business nodes. Equipment nodes include AGV-03, robotic arm-05, and shelf-12; environment nodes include temperature and humidity in warehouse B area and the congestion status of transportation channels; goods nodes include batch C2024 goods and high-value single item G202405; and business nodes include inbound sorting, cross-border transportation, and outbound verification. Implicit causal relationships refer to the necessary connections between elements that are not apparent. For example, continuously exceeding the temperature and humidity standard in warehouse B area → spoilage of batch C2024 goods, insufficient power in AGV-03 → decreased transportation efficiency → delay in the outbound shipment of order OD202405. Unlike the correlation relationships in traditional data statistics, implicit causal relationship identification uses L1 regularization to suppress overfitting and mutual information testing to eliminate spurious correlations (correlation with mutual information values ​​< 0.1 is judged as spurious correlation). The correlation strength coefficient is a 0-1 range value calculated based on historical and real-time data. It adopts a modified Pearson correlation coefficient combined with the causal contribution calculation formula: Correlation strength coefficient = (0.6 × Pearson correlation coefficient) + (0.4 × causal contribution). The causal contribution is determined by the intervention experiment method. The closer the coefficient is to 1, the more significant the causal relationship.

[0028] In detail, the full-link causal intervention theory is used to analyze the causal transmission logic between elements, avoiding the interference of pseudo-correlation in traditional correlation analysis. The graph neural network model architecture adopts a graph attention network, with specific parameters set as follows: the input layer dimension is consistent with the feature dimension of the standard correlation data (default 64 dimensions, supporting dynamic adaptation), the hidden layer has 3 layers (128 hidden units in the first layer, 64 in the second layer, and 32 in the third layer), the number of attention heads is 4, the output layer dimension corresponds to the predicted value of the correlation strength coefficient of the graph node, the activation function is LeakyReLU, the dropout probability is set to 0.2 to prevent overfitting, and the weight of key causal correlation is strengthened through the multi-head attention mechanism to map the standard correlation data into a graph structure of nodes and edges, and the correlation strength coefficient of node attributes and edges is optimized through iterative training. Key node status monitoring refers to real-time tracking of the changes in the core parameters corresponding to the nodes, and triggering risk inference when the parameters deviate from the normal range. Risk transmission path inference is based on the correlation strength coefficient starting from the abnormal node, traversing the upstream and downstream correlation nodes to form a complete risk propagation link.

[0029] Specifically, the probability of a problem occurring is calculated using a weighted formula: Probability of Problem Occurrence = Weight Coefficient 1 × Risk Transmission Path Length + Weight Coefficient 2 × Mean of Correlation Strength Coefficient + Weight Coefficient 3 × Historical Failure Frequency. The risk transmission path length is normalized to 0-1, the mean correlation strength coefficient is the average of the correlation strength coefficients between nodes, the historical failure frequency is normalized to 0-1, and the weight coefficients are set as: Weight Coefficient 1 = 0.3, Weight Coefficient 2 = 0.4, Weight Coefficient 3 = 0.3. The thresholds for risk level classification are determined based on the logistics park's failure statistics for the past three years and industry safety standards. High risk corresponds to a problem occurrence probability ≥ 80%, medium risk corresponds to a problem occurrence probability of 30%-80%, and low risk corresponds to a problem occurrence probability < 30%. Targeted intervention instructions are matched according to the risk level: high risk triggers emergency intervention, medium risk triggers parameter adjustment, and low risk triggers early warning.

[0030] In this embodiment, the deep correlation of warehousing and logistics elements is realized through the construction of a full-link causal chain graph, breaking the limitations of traditional point-to-point monitoring; the risk transmission path deduction and probability calculation realize the accurate location and early prediction of the root cause of the problem, and change the problem handling mode from post-remediation to pre-intervention.

[0031] The device collaboration module generates dynamic capability profiles for connected heterogeneous devices based on device risk information in closed-loop predictive intervention data. It forms a dynamic device alliance through game negotiation algorithm and controls the dynamic device alliance according to the closed-loop predictive intervention data to obtain device collaboration data.

[0032] In this embodiment, the specific process by which the device collaboration module acquires device collaboration data is as follows: The device collaboration module extracts device risk information from the predictive intervention data from the digital twin base, and combines it with the performance parameters, scenario adaptation thresholds, and historical collaboration data uploaded in real time by heterogeneous devices to generate dynamically updated device capability profiles. Based on the current business scenario requirements, it sets the negotiation goals of optimal collaboration efficiency and lowest cost through game theory negotiation algorithms, and organizes heterogeneous devices to conduct multiple rounds of collaboration scheme negotiations. The optimal collaboration scheme is selected to form a dynamic device alliance, and the task allocation ratio, operation rhythm, and obstacle avoidance strategy of the alliance are adjusted based on the risk warning information in the predictive intervention data. When the device status is abnormal or the business scenario changes, the alliance automatically starts the reorganization process, renegotiates the collaboration parameters, and outputs device collaboration data.

[0033] The heterogeneous equipment includes different types and manufacturers of equipment such as AGVs, robotic arms, automated shelves, transport vehicles, and sorting equipment. Equipment risk information is extracted from the predictive intervention data, including equipment failure risk, performance degradation risk, and compatibility risk. The dynamic capability profile includes real-time equipment performance parameters, scenario adaptation thresholds, and historical collaboration data. Real-time performance parameters include AGV remaining battery power, robotic arm grasping accuracy, and vehicle load capacity. Scenario adaptation thresholds include low temperature tolerance range, explosion-proof rating, and load limit. Historical collaboration data includes collaboration success rate with other equipment, task completion time, and number of conflicts. The profile is updated every 10 seconds. Data transmission between modules uses the RESTful API interface protocol, with the interface format being JSON, including data identifier, timestamp, data content, and verification code fields.

[0034] In detail, the optimization function of the negotiation objective is as follows: Minimum collaboration cost = Weight coefficient 1 × Collaboration time + Weight coefficient 2 × Energy cost + Weight coefficient 3 × Collaboration conflict rate, where collaboration time is the total time for the devices to complete the collaboration task, energy cost is the total energy consumption cost of the devices during the collaboration process, and the collaboration conflict rate is the proportion of task conflicts between devices during the collaboration process. The weight coefficients are set as Weight coefficient 1 = 0.5, Weight coefficient 2 = 0.3, and Weight coefficient 3 = 0.2. The collaboration efficiency is quantified by 1 / minimum collaboration cost. The game-theoretic negotiation algorithm takes optimal collaboration efficiency and minimum cost as its dual objective functions, organizes relevant heterogeneous devices to submit collaboration schemes, and eliminates schemes with low adaptability and high cost through multiple rounds of iterative negotiation. The solution is to select the globally optimal solution. A dynamic device alliance refers to a combination of devices temporarily formed to adapt to specific business scenarios. The devices in the alliance do not have a fixed master-slave relationship and achieve collaboration through real-time data interaction. The test conditions for the reorganization time of a dynamic device alliance are: accessing 10 heterogeneous devices, network latency ≤ 50ms, business scenario switching command response time ≤ 10ms, and under these conditions, the alliance reorganization time ≤ 200ms. Alliance collaboration parameters include task allocation ratio, running rhythm, and obstacle avoidance strategy. For example, the task allocation ratio is 70% handled by AGV and 30% by human assistance. The running rhythm is such that the AGV transport speed and the robotic arm grasping frequency are matched at a ratio of 1:2. The obstacle avoidance strategy is such as setting priority avoidance rules based on the risk of channel congestion.

[0035] Specifically, abnormal equipment status includes equipment malfunctions, depleted power, and performance parameters exceeding limits. Business scenario switching includes switching from inbound to outbound scenarios and from regular cargo transportation to cold chain cargo transportation. When the above situations occur, the alliance automatically dissolves and restarts the negotiation process, forming a new alliance based on the updated equipment capability profile and scenario requirements. Equipment collaboration data includes an alliance composition list, collaboration parameters of each device, task execution progress, and collaboration conflict records, providing a basis for equipment resource adaptation for the cargo management module. The real-time synchronization mechanism between the digital twin base and the equipment collaboration module adopts a publish-subscribe model (MQTT protocol) with a synchronization frequency of 50ms / time to ensure data real-time performance.

[0036] In this embodiment, the standardized representation of heterogeneous device capabilities is achieved through dynamic capability profiling, the game negotiation algorithm solves the problem of insufficient flexibility in the adaptation of traditional fixed protocols, and the self-forming and self-reorganizing capabilities of dynamic alliances greatly improve the device collaboration efficiency in complex scenarios.

[0037] The cargo management module assigns a unique digital identity to each piece of cargo. By combining the digital identity with device collaboration data and closed-loop predictive intervention data, it interacts with surrounding devices and systems to obtain autonomous cargo management data.

[0038] In this embodiment, the specific process by which the cargo management module acquires cargo autonomous management data is as follows: The cargo management module assigns a unique digital identity to each piece of cargo, generates a unified batch identifier on a batch basis and attaches it to the container unit, and marks each piece of cargo within a batch with a unique code. The digital identity information is uniformly stored in the system. Based on the cargo batch attributes and circulation requirements, a batch-level cargo demand file is constructed, which includes demand thresholds, adaptation standards, and anomaly judgment rules. Personalized demand parameters are added to high-value individual goods within the batch. Through fixed sensing devices at key nodes and sensing modules in the container unit, status data of the entire cargo circulation process is collected, and a cargo status dataset covering the entire process is formed by associating it with the batch identifier. Through the data of the digital identity-associated devices and the predictive intervention data, an adaptation confirmation request is actively sent to surrounding devices and the system, and the supply and demand resource matching is completed based on the feedback results. When the cargo status reaches the demand threshold, batch-level or individual-level anomaly warning information is sent. The matching results, resource matching records, and anomaly warning information are integrated to form cargo autonomous management data.

[0039] The digital identity hierarchy and batch identification association mechanism adopts a parent-child ID mapping method, with the batch identifier as the parent ID and the individual item code as the child ID. The child ID format is "parent ID-serial number" (e.g., "B202405-003"), ensuring the hierarchical association between batches and individual items. The unique digital identity adopts a hierarchical model combining batch identifiers and individual item codes. The batch identifier is a QR code or RFID tag attached to container units such as turnover boxes and pallets. The individual item code is marked using low-cost methods such as inkjet printing and labeling. The digital identity information includes batch number, individual item ID, order of origin, manufacturer, circulation requirements, etc., and is uniformly stored in a distributed database. The container units include cargo carriers such as turnover boxes, pallets, and containers. Low-cost sensing modules, such as temperature and humidity sensors and pressure sensors, are integrated only in the container units, eliminating the need to configure chips for each item, reducing hardware costs by more than 60%. The supply and demand resource matching adopts the Hungarian algorithm, which solves the optimal matching scheme by constructing a supply and demand cost matrix.

[0040] In detail, the batch attributes of goods include the type of goods, physical characteristics, and circulation requirements. The type of goods includes cold chain goods, dangerous goods, and irregularly shaped goods. The physical characteristics include fragile, moisture-sensitive, and high-temperature resistant. The circulation requirements include transportation temperature, stacking height, and time limits. The batch-level goods demand profile includes demand thresholds, adaptation standards, and anomaly judgment rules. The demand thresholds include cold chain goods with a temperature ≤5℃ and a stacking height ≤3 layers. The adaptation standards include adaptation to low-temperature resistant AGVs and explosion-proof storage environments. The anomaly judgment rules include a temperature exceeding the standard for 5 minutes as an anomaly. The personalized demand parameters for high-value single goods include individual temperature and humidity thresholds, shock resistance levels, and priority outbound permissions.

[0041] Specifically, key node fixed sensing devices are deployed at core locations such as the inbound port, storage racks, sorting tables, outbound ports, and transportation transfer stations to collect status data of goods at these nodes; container unit sensing modules collect dynamic status data of goods during transportation and storage; the goods status dataset is linked to key node data and dynamic data through batch identifiers to achieve two-level traceability of batches and individual items, ensuring that the status of each item is traceable; the causal relationship between equipment capability profiles and dynamic alliances is as follows: based on the equipment capability profiles, the compatibility similarity between equipment is calculated (using the cosine similarity algorithm), and equipment with a similarity ≥ 85% is included in the candidate alliance pool, and then the optimal combination is selected through a game negotiation algorithm.

[0042] In detail, the compatibility confirmation request refers to the resource matching request sent by the goods to surrounding devices after the goods have coordinated data and predicted intervention data through digital identity association devices; the supply and demand resource matching automatically selects the optimal device resources based on the compatibility results fed back by the devices.

[0043] Specifically, anomaly alerts are divided into batch-level and single-item-level alerts. A batch-level alert is triggered when the status of most goods in a batch reaches the required threshold, while a single-item-level alert is triggered when the status of a high-value single item is abnormal. The alert information is notified to the management personnel through system pop-ups, SMS, APP push, and other means. The cargo self-management data includes a list of compatible equipment, resource matching time, anomaly alert type, alert processing result, and full-process status trajectory, realizing the transformation of cargo from passive tracking to proactive management.

[0044] The digital twin platform supports the collaborative operation of various modules, providing underlying support for data flow and interaction across the entire system.

[0045] In this embodiment, the specific process by which the digital twin base supports the coordinated operation of each module is as follows: The digital twin platform uses a distributed database to store standard related data, predictive intervention data, equipment collaboration data, and cargo autonomous management data. It achieves tamper-proof sharing of cross-module data through blockchain technology, and has a built-in data flow scheduling mechanism that automatically synchronizes data from preceding modules to subsequent related modules in real time according to business logic. It receives data transmission requests from each module, performs format adaptation and conflict verification on the data to ensure data consistency, and monitors the operating status of each module in real time. When a module's data output is abnormal, it triggers a data backtracking and retransmission mechanism.

[0046] The distributed database supports high-concurrency read / write operations and persistent data storage, with a concurrency level of ≥100,000 times / second, ensuring the storage security and access efficiency of massive amounts of data. The blockchain deployment mode adopts a consortium blockchain, with nodes jointly maintained by the logistics park management, equipment suppliers, and third-party regulatory agencies. The granularity of data on-chain is key metadata + raw data hash value. The raw data is stored in the distributed database, and the on-chain data includes data identifiers, generation timestamps, associated modules, and hash values, balancing security and storage efficiency. Blockchain technology is used to record data flow trajectories, modification records, and access logs, achieving data immutability and full traceability, ensuring data credibility. The data flow scheduling mechanism is based on the business logic of data collection → prediction → equipment collaboration → cargo management, automatically planning data transmission paths to ensure that the output data of the preceding modules is synchronized to the subsequent associated modules in real time, with a data flow latency of ≤30ms.

[0047] In detail, format adaptation converts heterogeneous data output from different modules into a unified data format, facilitating cross-module identification; the specific algorithm for data conflict verification uses hash value comparison + version number control, assigning a unique version number to each piece of data, retaining the latest version number data and recording historical versions in case of conflict; the priority judgment rule for multimodal perception data conflicts is: device status data > environmental perception data > business flow data > cargo basic data, ensuring the accuracy of core data; conflict verification refers to verifying the consistency of the identifier of transmitted data. When a conflict is detected, a data tracing mechanism is triggered to locate the root cause of the conflict and automatically correct it or prompt manual intervention; module operation status monitoring includes monitoring data output frequency, data integrity, and parameter rationality. When data interruption, missing data, or anomalies occur, a data backtracking mechanism is triggered to send a retransmission request to the corresponding module to ensure data transmission continuity.

[0048] The prediction module constructs a full-link causal chain graph with self-updating capability, ensuring the graph's adaptability to business changes.

[0049] In this embodiment, the specific process of the full-link causal chain graph self-updating of the prediction module is as follows: The prediction module collects operational data and business execution results from each module in real time. Based on the collected data, it dynamically updates the node information and correlation strength coefficients of the full-link causal chain graph using a graph neural network algorithm. When a new warehousing and logistics business scenario is added or a new type of equipment is connected, the module automatically identifies the new elements and adds them as graph nodes, analyzes the causal relationship between the new nodes and the original nodes, and completes the graph expansion and update.

[0050] The collected updated data includes collaboration effect data of the equipment collaboration module, abnormal handling result data of the cargo management module, order completion data of the business system, and manual intervention records; node information updates include node attribute supplementation and node status correction. For example, adding low temperature adaptability attribute to equipment nodes and updating the flow progress status of cargo nodes are examples of node attribute supplementation. The correlation strength coefficient update is based on the new data and is recalculated to optimize the quantitative representation accuracy of causal relationships.

[0051] In detail, the new business scenarios include cross-border cargo warehousing, special transportation of dangerous goods, and multi-warehouse coordinated scheduling, while the new equipment includes four-way shuttle vehicles, intelligent sorting robots, and unmanned delivery vehicles. The identification of new elements is achieved through attribute matching of data gene tags, which automatically determines the type of new elements (equipment / environment / goods / business) and adds them as graph nodes. The implicit causal relationship between new nodes and existing nodes is analyzed through the theory of full-link causal intervention, and associated edges are added and the initial association strength coefficient is labeled to complete the graph expansion.

[0052] In this embodiment, the dynamic updating and expansion of the full-link causal chain graph ensures that the graph always adapts to changes in logistics park business scenarios and equipment iterations, avoiding a decrease in prediction accuracy due to business expansion or equipment upgrades, and further consolidating the core advantage of pre-intervention.

[0053] The dynamic capability profile of the equipment collaboration module has a self-calibration function to ensure the accuracy of equipment capability representation.

[0054] In this embodiment, the specific process of self-calibration of the dynamic capability profile of the device collaboration module is as follows: The device collaboration module collects real-time operational data and collaboration effect data of heterogeneous devices; compares the actual data with the performance parameters and adaptation thresholds in the dynamic capability profile to calculate the deviation value; when the deviation value exceeds the preset threshold, it calibrates the relevant parameters of the dynamic capability profile based on the actual data; and combines historical calibration data and device aging trends to predict the pattern of device capability changes and correct the dynamic capability profile parameters.

[0055] The actual operating data includes the actual grasping accuracy, transportation speed, energy consumption, and frequency of failures of the equipment; the collaborative effect data includes the task completion time, the number of collaborative conflicts, and the success rate of adaptation with other equipment; the deviation value is the percentage difference between the actual data and the profile parameters. For example, if the robotic arm profile grasping accuracy is 99% and the actual operating accuracy is 93%, the deviation value is 6%, and the preset threshold is usually set to 5%; the weight allocation rule of the weighted average algorithm is: real-time data weight 0.7, historical data of the past 7 days weight 0.2, and historical data of the past 30 days weight 0.1, to ensure the sensitivity of the profile to the real-time status of the equipment.

[0056] In detail, parameter calibration employs a weighted average algorithm, combined with the reliability of actual data (such as sensor accuracy and data acquisition frequency) to correct the profile parameters. The confidence evaluation criteria for parameter calibration are as follows: when the deviation between the calibrated parameters and the actual data is ≤3%, the confidence level is judged as excellent; when the deviation is 3%-5%, the confidence level is judged as good; when the deviation is >5%, a secondary calibration is triggered. The input features of the equipment aging prediction model include equipment runtime, maintenance frequency, failure frequency, ambient temperature and humidity, and load intensity. A gradient boosting tree model is used to construct the prediction model, and the model hyperparameters are set as follows: The training rate is 0.05, the maximum tree depth is 6 layers, the number of decision trees is 100, the minimum number of splits per sample is 20, and the minimum number of leaf nodes per sample is 10. The training data uses historical data of the equipment's runtime, maintenance records, failure frequency, environmental temperature and humidity, and load intensity over the past 3 years. The data sample size is no less than 100,000 records, and the ratio of training set, validation set, and test set is 7:2:1. The model training uses 5-fold cross-validation to ensure generalization ability and predict the change pattern of equipment capabilities over time, such as the AGV's battery life decreasing by 3% per year. The profile parameters are corrected in advance to avoid collaborative adaptation deviations caused by equipment aging.

[0057] The processes described above with reference to the flowcharts in the embodiments disclosed in this invention can be implemented as computer software programs. The embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wire segments, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical fibers, RF, etc., or any suitable combination thereof.

[0058] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0059] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functions and structural principles of the present invention have been shown and explained in the embodiments. Without departing from the stated principles, the implementation of the present invention may have any variations or modifications.

Claims

1. A logistics park warehousing and logistics integrated management system based on digital twin technology, characterized in that: The system includes: The data acquisition module collects multidimensional data, preprocesses the multidimensional data based on data gene tags, and obtains standard association data; The prediction module is based on the theory of full-link causal intervention and graph neural network. It constructs a full-link causal chain graph of standard correlation data and predicts the root cause and probability of occurrence of the problem based on the state of key nodes in the graph to obtain prediction intervention data. The device collaboration module generates dynamic capability profiles for connected heterogeneous devices based on device risk information in closed-loop predictive intervention data. It forms a dynamic device alliance through game negotiation algorithm and controls the dynamic device alliance according to the closed-loop predictive intervention data to obtain device collaboration data. The cargo management module assigns a unique digital identity to each piece of cargo. By combining the digital identity with device collaboration data and closed-loop predictive intervention data, it interacts with surrounding devices and systems to obtain autonomous cargo management data.

2. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 1, characterized in that, The specific process by which the data acquisition module obtains standard correlation data is as follows: The data acquisition module synchronously collects equipment operation data, environmental perception data, basic cargo data, and business flow data of the logistics park's warehousing and logistics through multi-source sensing devices; Based on data gene tags, the source attributes, association dimensions and format standards of various types of data are defined, and invalid values ​​are removed and abnormal mutation data are filtered out from the collected multidimensional data. High-strength data encryption algorithms are used to encrypt sensitive data, and different types of data are mapped to a unified data dimension through format standardization operations; By establishing homologous associations across different data types using data gene tags, and then integrating them after completing association verification, standard associated data is formed.

3. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 2, characterized in that, The specific process by which the prediction module obtains prediction intervention data is as follows: The prediction module receives the standard correlation data output by the data acquisition module, extracts the equipment, environment, goods and business elements of each link of warehousing and logistics as graph nodes, analyzes the implicit causal relationship between elements based on the full-link causal intervention theory, marks the causal relationship edge between nodes with the correlation strength coefficient, and constructs the full-link causal chain graph. Real-time monitoring of the status changes of key nodes in the graph, combined with the correlation strength coefficient to deduce the risk transmission path, and calculate the probability of problem occurrence and risk level; Based on the risk level, targeted intervention instructions are matched, and the root cause identification, occurrence probability, risk level and intervention instructions are integrated to form predictive intervention data and feed it back to the digital twin platform.

4. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 3, characterized in that, The specific process by which the device collaboration module acquires device collaboration data is as follows: The device collaboration module extracts device risk information from the predictive intervention data from the digital twin base, and combines it with the performance parameters, scenario adaptation thresholds and historical collaboration data uploaded in real time by heterogeneous devices to generate a dynamically updated device capability profile. Based on the current business scenario requirements, a game-theoretic negotiation algorithm is used to set the negotiation goals of optimal collaboration efficiency and lowest cost, and heterogeneous devices are organized to negotiate multiple rounds of collaboration solutions. Select the optimal collaboration scheme to form a dynamic equipment alliance, and adjust the task allocation ratio, operation rhythm and obstacle avoidance strategy of the alliance based on the risk warning information in the prediction intervention data. When equipment status is abnormal or business scenario changes, the alliance automatically initiates a reorganization process, renegotiates collaboration parameters, and outputs equipment collaboration data.

5. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 4, characterized in that, The specific process by which the cargo management module obtains cargo self-management data is as follows: The cargo management module assigns a unique digital identity to each piece of cargo, generates a unified batch identifier for each batch and attaches it to the container unit, marks each piece of cargo in the batch with a unique code, and stores the digital identity information uniformly in the system. Based on the batch attributes and circulation requirements of goods, a batch-level goods demand file is constructed, which includes demand thresholds, adaptation standards and anomaly judgment rules, and personalized demand parameters are added for high-value individual goods within the batch. By using fixed sensing devices at key nodes and sensing modules in container units, status data of goods throughout the entire flow process is collected, and a cargo status dataset covering the entire process is formed by associating batch identifiers. By linking digital identity with device collaboration data and predictive intervention data, proactively send compatibility confirmation requests to surrounding devices and systems, and complete the matching of supply and demand resources based on the feedback results; When the status of goods reaches the demand threshold, send batch-level or single-item-level abnormal warning information; Integrate matching results, resource matching records, and anomaly warning information to form autonomous cargo management data.

6. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 5, characterized in that, The digital twin base supports the collaborative operation of each module, as shown in the following process: The digital twin platform uses a distributed database to store standard-related data, predictive intervention data, equipment collaboration data, and cargo autonomous management data; Blockchain technology enables tamper-proof sharing of cross-module data, with a built-in data flow scheduling mechanism that automatically synchronizes data from preceding modules to subsequent related modules in real time according to business logic. Receive data transmission requests from each module, perform format adaptation and conflict checking on the data to ensure data consistency; The system monitors the operating status of each module in real time, and triggers a data backtracking and retransmission mechanism when the module's data output is abnormal.

7. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 6, characterized in that, The full-link causal chain graph constructed by the prediction module has self-updating capability, and the specific process is as follows: The prediction module collects operational data and business execution results from each module in real time; Based on the collected data, the node information and correlation strength coefficient of the full-link causal chain graph are dynamically updated using a graph neural network algorithm; When new warehousing and logistics business scenarios are added or new types of equipment are connected, the system automatically identifies the new elements and adds them as nodes in the graph, analyzes the causal relationship between the new nodes and the original nodes, and completes the graph expansion and update.

8. The integrated logistics park warehousing and logistics management system based on digital twin technology according to claim 7, characterized in that, The dynamic capability profile of the device collaboration module has a self-calibration function, the specific process of which is as follows: The device collaboration module collects real-time operational data and collaboration effect data from heterogeneous devices. The actual data is compared with the performance parameters and adaptation thresholds in the dynamic capability profile to calculate the deviation value; When the deviation value exceeds the preset threshold, the relevant parameters of the dynamic capability profile are calibrated based on the actual data; By combining historical calibration data and equipment aging trends, we can predict the patterns of equipment capability changes and revise the dynamic capability profile parameters.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is executed by a processor to implement the integrated logistics park warehousing and logistics management system based on digital twin technology as described in any one of claims 1-8.

Images