Intelligent bill management system capable of being operated offline

Abstract

The invention relates to the technical field of Internet of Things and edge computing, in particular to an intelligent bill management system capable of being operated offline. And the edge calculation synchronization module confirms the right of the offline bill service data by using a hash chain and locally stores the offline bill service data to ensure the integrity of the transaction. And the edge calculation risk perception module verifies the data in real time and outputs an edge business risk index. And the flow decision module integrates risks, geography and timeliness, drives offline service flow path deduction, and generates an intelligent strategy of resource optimization. And the field execution module drives the logistics entity according to the strategy to complete bill transaction processing. When the network is reconstructed, the central module cooperatively processes data, the global consistency of the central data is guaranteed through service verification and a global consistency arbitration mechanism, continuous and efficient management of offline services is achieved, and logistics operation intelligence is accelerated.

Description

Technical Field

[0001] This invention relates to the fields of Internet of Things and edge computing technology, specifically to an intelligent invoice management system that can operate offline. Background Technology

[0002] In edge environments such as logistics and warehousing where network connectivity is unreliable, offline transaction data lacks real-time verification and distributed ownership confirmation. Traditional local caching cannot technically guarantee the tamper-proof and non-repudiation sequence of offline transaction data. This lack of data credibility directly prevents central management from effectively auditing and tracking offline operations, leading to inconsistencies and operational risks, and severely restricting the efficiency of enterprise operations and management.

[0003] In complex business processes, the lack of intelligent collaboration in task resource allocation and risk management leads to inefficiency. Traditional logistics operation models rely on manual or simple preset rules for task allocation (such as picking and verification). This model cannot efficiently integrate multi-dimensional management indicators such as the inherent risks of business transactions, timeliness priorities, and the real-time load of available resources (personnel / equipment). Therefore, it is difficult to quickly generate optimal operational strategies that balance efficiency and risk control in edge environments, resulting in low logistics operation efficiency and low resource utilization.

[0004] The challenge of global inventory management arising from asynchronous aggregation of edge-central data: After network recovery, edge systems need to aggregate offline data streams from different devices and locations to a central database. Due to operational differences and duplicate transactions, traditional simple synchronization strategies are prone to causing global inconsistencies in management data such as central inventory ledgers and cash flow records. How can we achieve consistency, reliability, and granular management of the aggregated management data in an unreliable network environment?

[0005] Therefore, this invention proposes an intelligent invoice management system that can operate offline. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent invoice management system that can operate offline. By using technologies such as edge computing, hash chain rights confirmation, and intelligent scheduling algorithms, it overcomes technical deficiencies in logistics operation risk management, resource allocation optimization, and global data consistency, thereby accelerating the intelligent operation and management of enterprises.

[0007] To achieve the above objectives, the present invention provides the following technical solution: An offline-operable intelligent invoice management system includes: Edge data synchronization module: It uses a hash chain structure to perform hash fingerprint verification on offline ticket operation data obtained from edge computing devices and stores it in the edge computing devices; it encapsulates the offline synchronization payload according to the edge network topology and business strategy. Risk perception module: Through the analysis unit built into the edge computing device, it performs cross-modal comparison and semantic verification on offline invoice operation data, identifies invoice structure anomalies and quantifies operational behavior deviations, and outputs edge business risk index stream; Flow decision module: Collects edge business risk index stream and set of pending bill transactions; integrates edge geographical location and preset bill business timeliness requirements, drives offline flow path simulation, and generates intelligent bill flow strategy; On-site execution module: The intelligent flow strategy for invoices drives the transaction processing of logistics entities through edge computing devices and records the results, forming an edge transaction confirmation flow, and pushes updates to the edge data synchronization module; Central Data Module: During network link reconstruction, the central scheduling and computing unit performs link activation, coordinates offline synchronization of payloads and edge transaction confirmation flows; aggregates to the central bill data storage, and calibrates and confirms the central bill data through business verification and global consistency arbitration mechanism.

[0008] Preferably, the process of performing hash fingerprint verification on offline ticket operation data obtained from the edge computing device using a hash chain structure and storing it on the edge computing device includes: obtaining offline ticket operation data; parsing the offline ticket operation data to identify the operation time, operation user identifier, and business operation description; obtaining the historical fingerprint of the previous offline ticket operation data already stored on the edge computing device; performing hash operation on the operation time, operation user identifier, business operation description, and historical fingerprint using a cryptographic hash algorithm to generate the current offline ticket operation data fingerprint; encapsulating the current offline ticket operation data fingerprint, the historical fingerprint, and the offline ticket operation data together into an offline data record; and storing the offline data record in the local storage unit of the edge computing device.

[0009] Preferably, the process of encapsulating the offline synchronization payload based on the edge network topology and service strategy includes: identifying the current edge network topology and obtaining available connection parameters and network status indicators; obtaining a data synchronization priority list and synchronization time window rules according to a preset service strategy; selecting target edge computing devices and gateways through a comprehensive evaluation model based on the available connection parameters, the network status indicators, the data synchronization priority list, and the synchronization time window rules; determining the range and priority of offline ticket operation data to be synchronized based on the data synchronization priority list and the synchronization time window rules; packaging the offline ticket operation data to be synchronized, the target edge computing device identifier, and metadata information for decryption into a data packet to be encrypted; and encrypting the data packet to be encrypted using a symmetric encryption algorithm to form the offline synchronization payload.

[0010] Preferably, the process by which the risk perception module performs cross-modal comparison and semantic verification of offline invoice operation data, identifies invoice structure anomalies and quantifies operational behavior deviations, and outputs an edge business risk index stream includes: receiving offline invoice operation data and decoupling the invoice operation data into structural field modal data and temporal behavior modal data; calling the structural field modal corresponding to the pre-stored business flow invoice template library and the temporal behavior modal corresponding to the historical operational behavior pattern library; performing a structural field modal comparison between the invoice operation data and the business flow invoice template library to identify invoice structure integrity and format deviations; performing a behavioral comparison between the temporal behavior modal of the invoice operation data and the historical operational behavior pattern library to identify operational behavior frequency and sequence deviations; calculating an invoice structure anomaly index based on the structural comparison results using a structural anomaly scoring model; calculating an operational behavior deviation index based on the behavioral comparison results using a temporal deviation scoring model; weighting and fusing the invoice structure anomaly index and the operational behavior deviation index according to a preset risk factor weight table to generate an edge business risk index; and attaching a timestamp to the edge business risk index and aggregating it into an edge business risk index stream for output.

[0011] Preferably, the process of collecting edge business risk index streams and sets of pending bill transactions; integrating edge geographic locations and preset bill transaction timeliness requirements to drive offline circulation path deduction and generate a bill intelligent flow strategy includes: collecting edge business risk index streams and sets of pending bill transactions; determining the geographic location information of the current edge computing device and the store where it is located; obtaining the preset bill transaction timeliness requirement set and associated resource availability information; initializing an offline circulation path model based on a directed graph model, inputting the edge business risk index stream, the set of pending bill transactions, the current edge geographic location, and the preset bill transaction timeliness requirement set; calculating the pending execution path and node allocation scheme of the offline circulation path based on a heuristic search algorithm, comprehensively considering the risk index, timeliness requirements, and geographic location; generating a structured bill intelligent flow strategy based on the pending execution path and node allocation scheme, the bill intelligent flow strategy including circulation path, node identifier, and link time limit data fields; and outputting the bill intelligent flow strategy.

[0012] Preferably, the process of the intelligent flow of invoices strategy driving logistics entity invoice transaction processing and recording results through edge computing devices to form an edge transaction confirmation stream, and pushing updates to the edge data synchronization module includes: receiving the intelligent flow of invoices strategy; parsing the intelligent flow of invoices strategy to determine a list of logistics entity invoice transactions to be executed; controlling the corresponding logistics entity devices to execute each invoice transaction in the list of logistics entity invoice transactions to be executed through the edge computing devices using a preset IoT communication protocol; obtaining the execution status and results of each invoice transaction by subscribing to the status feedback interface of the logistics entity devices; encapsulating the execution status and results, transaction identifiers and execution timestamps to form a structured edge transaction confirmation stream; and pushing the edge transaction confirmation stream to the edge data synchronization module through the edge computing devices to update the data.

[0013] Preferably, during network link reconstruction, the process of link activation performed by the central scheduling computing unit and coordinating offline synchronization payload and edge transaction confirmation stream includes: detecting the completion of network link reconstruction; initiating the link activation process through the central scheduling computing unit, including performing security authentication and establishing an encrypted transmission channel; synchronously receiving offline synchronization payload and edge transaction confirmation stream from the edge computing device; matching the offline synchronization payload and the edge transaction confirmation stream based on timestamps, transaction identifiers, and data fingerprints; performing collaborative processing of the offline synchronization payload and the edge transaction confirmation stream based on a preset collaborative processing algorithm to resolve business data conflicts, including data inconsistency, operation sequence differences, and transaction duplication; and generating collaboratively processed business data to be aggregated.

[0014] Preferably, the process of aggregating data to the central bill data storage and calibrating and confirming the central bill data through a business verification and global consistency arbitration mechanism includes: receiving the coordinated business data to be aggregated; transmitting the business data to be aggregated to the central bill data storage unit; initiating a business verification and global consistency arbitration mechanism based on a multi-dimensional data verification model; performing multi-dimensional data verification on the data to be calibrated in the central bill database according to a preset consensus rule set; calculating a data deviation index based on the multi-dimensional data verification results and calibrating the data according to a deviation handling strategy; determining whether the calibrated data meets the global consistency threshold through the business verification and global consistency arbitration mechanism, and confirming the data that meets the threshold; and updating the confirmed business bill data in the central bill data storage unit.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention utilizes hash chains to ensure the sequential integrity and tamper-proof capability of offline transaction data; through heuristic scheduling of the flow decision module, it optimizes the average response time of edge computing nodes under complex multi-objective constraints; and after network reconstruction, through central coordination and business verification and global consistency arbitration mechanisms, it effectively solves the technical bottlenecks of data credibility and resource scheduling efficiency of distributed edge computing systems in offline environments.

[0016] 2. This invention utilizes an edge computing risk perception module to automate cross-modal comparison and semantic verification of offline invoice operation data. Compared to traditional manual comparison, this significantly improves comparison efficiency and process consistency, reduces comparison errors caused by non-standard manual operations, and can quickly identify invoice anomalies, providing quantified risk indicators for subsequent risk management and manual review. This mechanism, in conjunction with the edge computing flow decision module, integrates risk index flow, geographical location, and business timeliness requirements to intelligently generate invoice flow strategies. This enables the prediction and avoidance of potential risks for each invoice transaction before execution, effectively reducing operational chaos and profit losses caused by human error, and ensuring the accuracy and reliability of edge logistics operations.

[0017] 3. This invention uses an edge computing data synchronization module to retain and collaboratively update offline invoice operation data after fingerprint-based rights confirmation. A central data module coordinates the offline synchronization payload with the edge transaction confirmation flow during network link reconstruction. After calibration and confirmation through business verification and a global consistency arbitration mechanism, the data is aggregated to central storage. This edge-central collaborative and multi-verified data aggregation and calibration mechanism effectively overcomes the traditional information silos of edge data, providing enterprises with a business management data view that covers the entire chain and possesses high consistency and reliability. This significantly improves the scientific rigor and timeliness of key decisions such as logistics planning, resource scheduling, process optimization, and risk management, accelerating the intelligent transformation of logistics operations. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of an intelligent invoice management system that can operate offline, as proposed in this invention. Figure 2 This is a flowchart illustrating the offline invoice data fingerprinting and rights confirmation process proposed in an embodiment of the present invention; Figure 3 This is a flowchart illustrating the edge business risk perception process proposed in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 This invention provides a schematic diagram of the structure of an intelligent invoice management system that can operate offline. The technical solution is as follows: An offline-operable intelligent invoice management system includes: Edge data synchronization module: It uses a hash chain structure to perform hash fingerprint verification on offline ticket operation data obtained from edge computing devices and stores it in the edge computing devices; it encapsulates the offline synchronization payload according to the edge network topology and business strategy. Risk perception module: Through the analysis unit built into the edge computing device, it performs cross-modal comparison and semantic verification on offline invoice operation data, identifies invoice structure anomalies and quantifies operational behavior deviations, and outputs edge business risk index stream; Flow decision module: Collects edge business risk index stream and set of pending bill transactions; integrates edge geographical location and preset bill business timeliness requirements, drives offline flow path simulation, and generates intelligent bill flow strategy; On-site execution module: The intelligent flow strategy for invoices drives the transaction processing of logistics entities through edge computing devices and records the results, forming an edge transaction confirmation flow, and pushes updates to the edge data synchronization module; Central Data Module: During network link reconstruction, the central scheduling and computing unit performs link activation, coordinates offline synchronization of payloads and edge transaction confirmation flows; aggregates to the central bill data storage, and calibrates and confirms the central bill data through business verification and global consistency arbitration mechanism.

[0021] Example 1

[0022] This embodiment provides a specific application of an intelligent invoice management system that can operate offline. Its typical application scenario is offline inventory management. In this scenario, it ensures that invoice operation data can still be effectively recorded, risks can be detected, and intelligent circulation strategies can be generated when the network is interrupted, thus ensuring business continuity and data accuracy.

[0023] Furthermore, the process of performing hash fingerprint verification on offline ticket operation data obtained from the edge computing device using a hash chain structure and storing it on the edge computing device includes: obtaining offline ticket operation data; parsing the offline ticket operation data to identify the operation time, user identifier, and business operation description; obtaining the historical fingerprint of the previous offline ticket operation data already stored on the edge computing device; performing a hash operation on the operation time, user identifier, business operation description, and historical fingerprint using a cryptographic hash algorithm to generate the current offline ticket operation data fingerprint; encapsulating the current offline ticket operation data fingerprint, the historical fingerprint, and the offline ticket operation data together into an offline data record; and storing the offline data record in the local storage unit of the edge computing device. (Refer to...) Figure 2 This is a flowchart illustrating the offline invoice data fingerprinting and rights confirmation process according to an embodiment of the present invention.

[0024] Specifically, in offline inventory management scenarios, when the network connection is interrupted, store clerks use handheld edge computing devices to perform offline document operations such as "goods receipt," "goods issuance," and "inventory counting." The edge computing device first acquires this offline document operation data. For example, for a "goods receipt" operation, the system automatically extracts the operation time as "2023-10-27 10:30:15," the user identifier as "store employee A001," and the business operation description as "Inventory SKU: ABC123, Quantity: 50 pieces, received in shelf area A." The edge computing device first retrieves the historical fingerprint of the previous offline data record from local storage. Then, the device uses the SHA-256 cryptographic hash algorithm to hash the extracted operation time, user identifier, business operation description, and the historical fingerprint to generate a unique, fixed-length fingerprint for the current offline document operation data. Subsequently, the current fingerprint, the historical fingerprint, and the complete offline document operation data are encapsulated into an offline data record. Finally, the offline data record is stored in the local storage unit of the handheld edge computing device.

[0025] This invention uses a business verification and global consistency arbitration mechanism to perform final anti-tampering verification and data ownership confirmation, thereby achieving continuous, efficient, and secure management of offline services and ensuring the integrity of local data records and consistency with central data.

[0026] Furthermore, the process of encapsulating the offline synchronization payload based on the edge network topology and service strategy includes: identifying the current edge network topology and obtaining available connection parameters and network status indicators; obtaining a data synchronization priority list and synchronization time window rules according to a preset service strategy; selecting target edge computing devices and gateways through a comprehensive evaluation model based on the available connection parameters, the network status indicators, the data synchronization priority list, and the synchronization time window rules; determining the range and priority of offline ticket operation data to be synchronized based on the data synchronization priority list and the synchronization time window rules; packaging the offline ticket operation data to be synchronized, the target edge computing device identifier, and metadata information for decryption into a data packet to be encrypted; and encrypting the data packet to be encrypted using a symmetric encryption algorithm to form the offline synchronization payload.

[0027] Specifically, in offline inventory management scenarios, when edge computing devices, such as handheld terminals and smart shelf sensors in stores, detect network connectivity restoration and the achievement of preset synchronization conditions, they first identify the current edge network topology and obtain available connection parameters, such as available Wi-Fi networks, Bluetooth gateways, and other LAN connections, as well as network status indicators, such as signal strength, network latency, and gateway load. Simultaneously, the system obtains a structured data synchronization priority list based on preset business strategies. For example, priority 1: abnormal inventory adjustment data; priority 2: emergency outbound data; priority 3: regular inbound data; and synchronization time window rules, such as synchronizing inventory data hourly. The system utilizes an evaluation model based on multi-indicator weighted summation, and based on the identified edge network topology (e.g., highest signal strength, lowest load) and predetermined business strategies (e.g., minimizing synchronization data volume), selects the most suitable target edge computing device and gateway for data transmission. For example, it selects the store's local server with the highest signal strength and lowest load. The comprehensive evaluation model employs a weighted summation evaluation function. The evaluation value is used to select target edge computing devices and gateways. The input indicators of the evaluation function include available connection parameters and network status indicators. The business indicators include at least the amount of data to be synchronized, data synchronization priority, and the idle status of the target device. When selecting target devices, the system prioritizes devices with the highest evaluation value, or, provided that a preset minimum evaluation value threshold is met, prioritizes devices with the lowest load, to ensure load balancing under the premise of optimal data synchronization priority and connection quality. Subsequently, high-priority offline ticket operation data within a defined range, such as all outbound data and all abnormal adjustment data within the last 10 minutes, along with the unique identifier of the target device and decryption metadata information such as encryption key version and encryption algorithm type, are packaged to form a data packet to be encrypted. Finally, the data packet to be encrypted is encrypted using the AES-256 symmetric encryption algorithm to form a secure offline synchronization payload.

[0028] This invention enables intelligent and strategic filtering and encapsulation of offline ticket operation data in an edge network environment, effectively utilizing limited network resources, optimizing the efficiency and security of data synchronization, and ensuring timely and accurate data transmission after connection restoration.

[0029] Furthermore, the process by which the risk perception module performs cross-modal comparison and semantic verification of offline invoice operation data, identifies invoice structural anomalies, quantifies operational behavior deviations, and outputs an edge business risk index stream includes: receiving offline invoice operation data and decoupling the data into structural field modal data and temporal behavior modal data; calling the pre-stored structural field modal data corresponding to the business flow invoice template library and the temporal behavior modal data corresponding to the historical operational behavior pattern library; performing a structural field modal comparison between the invoice operation data and the business flow invoice template library to identify invoice structural integrity and format deviations; performing a behavioral comparison between the temporal behavior modal data and the historical operational behavior pattern library to identify operational behavior frequency and sequence deviations; calculating an invoice structural anomaly index based on the structural comparison results using a structural anomaly scoring model; calculating an operational behavior deviation index based on the behavioral comparison results using a temporal deviation scoring model; weighting and fusing the invoice structural anomaly index and the operational behavior deviation index according to a preset risk factor weight table to generate an edge business risk index; and appending a timestamp to the edge business risk index and aggregating it into an edge business risk index stream for output. (Refer to...) Figure 3 This is a flowchart of the edge business risk perception process proposed in an embodiment of the present invention.

[0030] Specifically, in offline inventory management scenarios, such as processing store return invoices, the risk perception module first parses the received offline synchronous payload to extract the original offline invoice operation data. The module decouples the data into structured field modal data (such as product SKU and quantity) and time-series behavioral modal data (such as operation timestamp and user ID). Subsequently, the module retrieves the corresponding business flow invoice template library (in this example, the "return business flow invoice template library") and historical operation behavior pattern library (in this example, the "store employee historical return behavior pattern library," which is constructed through statistical analysis of historical behavior data) from the pre-repository of the edge computing device.

[0031] Quantification of Structural Anomaly Scoring Model: The structural anomaly scoring model is responsible for calculating the document structural anomaly index. This index is quantified based on a preset set of structural integrity rules and severity levels. For example, the system will determine whether a field is a "necessary field" or a "minor field." If a necessary field is missing or formatted incorrectly, the anomaly index will accumulate significantly (e.g., increase by 0.4); if a minor field is incorrect, the index will accumulate less (e.g., increase by 0.1). The final structural anomaly index is the sum of the accumulated values ​​and is normalized to between 0 and 1 to reflect the degree of deviation from the document structure.

[0032] Quantification of the Time-Series Deviation Scoring Model: The time-series deviation scoring model is responsible for calculating the operational behavior deviation index. This model is pre-built based on a historical operational behavior pattern library, and the construction process includes: State Definition and Matrix Construction: The model's state is defined as a combination of the user identifier and the business operation description (e.g., "User A performed a warehouse location check"). This refined state definition enables the model to distinguish the risk differences brought about by different users and specific operation types. The model constructs a state transition probability matrix by statistically analyzing the transition frequency between consecutive states in historically confirmed, low-risk operation sequences. Sequence Anomaly Quantification: The time-series deviation scoring model decomposes the current operation sequence into consecutive state transition pairs, queries the pre-stored state transition probability matrix to obtain the corresponding transition probabilities, and calculates the joint probability L(Q) of the entire sequence. This calculation is based on the accepted first-order Markov chain assumption. If a zero-probability transition (i.e., a transition that has never occurred historically) occurs in the sequence, non-zero corrections are made using techniques such as Laplace smoothing. Deviation Index Calculation and Mapping: The index is calculated using the negative logarithmic probability normalized to L(Q) to reflect the degree of abnormality of the current operation sequence relative to historical normal behavior. The operation behavior deviation index R is inversely proportional to the sequence joint probability L(Q), i.e., the lower L(Q), the higher R.

[0033] Edge Business Risk Index Generation and Weighting: Ultimately, the system weights and merges the document structure anomaly index and the operational behavior deviation index according to a preset risk factor weighting table to generate the edge business risk index. This weighting table is a configurable set of parameters, the values ​​of which are determined based on the current business type and the computing power status of the edge computing devices. For example, for "return" transactions involving cash or asset changes, the system allocates a higher weight to the structure anomaly index (e.g., 0.7) to prioritize data accuracy; while for high-frequency "picking" transactions, the weight of the time-series deviation index is increased to detect operational anomalies. The edge business risk index is the sum of the products of the two indices and their corresponding weights.

[0034] This invention enables real-time quantitative assessment of potential business risks by performing cross-modal comparison and semantic verification of offline invoice operation data. It effectively identifies invoice anomalies and behavioral deviations in an offline environment, thereby enhancing the internal control capabilities and risk warning level of inventory and transaction management.

[0035] Furthermore, the process of collecting edge business risk index streams and sets of pending bill transactions; integrating edge geographic locations and preset bill transaction timeliness requirements to drive offline circulation path deduction and generate intelligent bill flow strategy includes: collecting edge business risk index streams and sets of pending bill transactions; determining the geographic location information of the current edge computing device and its location in the store; obtaining the preset bill transaction timeliness requirement set and associated resource availability information; initializing an offline circulation path model based on a directed graph model, inputting the edge business risk index stream, the set of pending bill transactions, the current edge geographic location, and the preset bill transaction timeliness requirement set; calculating the pending execution path and node allocation scheme of the offline circulation path based on a heuristic search algorithm, comprehensively considering risk index, timeliness requirements, and geographic location; generating a structured intelligent bill flow strategy based on the pending execution path and node allocation scheme, the intelligent bill flow strategy including circulation path, node identifier, and link time limit data fields; and outputting the intelligent bill flow strategy.

[0036] Specifically, in offline inventory management scenarios, the flow decision module continuously collects edge business risk index streams output by the risk perception module, as well as sets of pending document transactions, such as "large-amount return orders awaiting review" and "online orders awaiting picking." Simultaneously, the module determines the spatial location (e.g., latitude and longitude coordinates) of the current edge computing device and the logistics entities (personnel / equipment) within the store, and obtains a preset set of document transaction timeliness requirements, such as "urgent order picking must be completed within 15 minutes" and "high-risk returns must be manually reviewed within 30 minutes," and obtains a list of available resources within the current store. This data is input into an offline flow path model built based on a directed graph structure. In this model, the nodes of the graph represent resource entities (such as human operators and picking robots) and key business links (such as storage locations and inspection areas) that execute document transactions; the edges of the graph represent feasible operational transfer paths from one entity or link to the next; and the weight of the edge is defined as the flow cost. This model utilizes A... *The heuristic path search algorithm comprehensively considers the priority, timeliness requirements, and resource geographic location matching of high-risk transactions to deduce and calculate the execution path and node allocation scheme for offline workflow. The algorithm's cost function g(n) (actual cost) is defined as the actual workflow cost from the starting node to the current node n, which is composed of the following three weighted factors: Resource cost (workload): Node consumption weight, reflecting the degree of resource node occupation by task processing. Physical cost (geographic distance): The time or energy required for personnel or equipment to move spatially. Risk penalty factor (risk index): The risk index acts as a penalty factor in the cost function. High-risk indices are not simply regarded as physical costs, but are significantly increased in terms of non-linearity or amplified weights, thereby prompting the search algorithm to prioritize low-risk business processing paths or resource nodes (e.g., avoiding processes marked as abnormal by the risk perception module). The final workflow cost is the sum of these three weighted factors. Its heuristic function h(n) (estimated cost) uses the shortest Euclidean geographic distance from the current node to the destination or the estimated minimum number of processing steps as a metric. For example, for a high-risk return transaction, the model might recommend a path of "store manager review to warehouse manager secondary verification" and assign the most recent and currently available "store manager" as the first processing node.

[0037] As a preferred implementation, the calculation of the offline processing path and node allocation scheme specifically includes: obtaining the current workload of each processing node in the path to be executed in real time, and using it as a constraint factor for the node allocation scheme, wherein the node allocation scheme is determined by minimizing the total workload of all nodes in the path to be executed while meeting the timeliness requirements of the ticket business.

[0038] Specifically, when calculating the execution path and node allocation scheme, the system queries the employee workload data in the store's local server in real time, such as the number of tickets to be processed by each employee and the estimated completion time. The employee's current total workload is converted into a "node consumption weight" for that node in the workflow model. When using a heuristic search algorithm for path deduction, the "ticket business timeliness requirement" is set as a hard time constraint, and "minimizing total workload" is set as the optimization objective. For example, for an urgent picking task that needs to be completed within 15 minutes, the model will prioritize selecting the idle employee node with the fewest pending tasks and the closest geographical location to the picking area, ensuring that the task meets the timeliness requirement while minimizing the impact on resource nodes on that path. This technology, by incorporating real-time workload into decision factors, achieves elastic allocation and local load balancing of workflow nodes, effectively avoiding task congestion and hotspot node phenomena, and ensuring the rapid and stable flow of high-risk and high-time-efficiency transactions. Based on this execution path and node allocation scheme, the system generates a structured intelligent ticket flow strategy, which clearly indicates the optimal processing flow, participating personnel, and time limits for each stage of ticket transactions in offline mode. Finally, the system outputs the intelligent flow strategy for the invoice.

[0039] This invention integrates multi-source information and drives offline circulation path deduction, realizing intelligent decision-making on the circulation path of bill transactions in an edge network environment. It balances business timeliness, risk control and resource allocation, and improves the efficiency, response speed and risk control level of business processing.

[0040] Furthermore, the process of the intelligent flow of invoices strategy driving logistics entity invoice transaction processing and recording results through edge computing devices to form an edge transaction confirmation stream, and pushing updates to the edge data synchronization module includes: receiving the intelligent flow of invoices strategy; parsing the intelligent flow of invoices strategy to determine a list of logistics entity invoice transactions to be executed; controlling the corresponding logistics entity devices to execute each invoice transaction in the list of logistics entity invoice transactions to be executed through the edge computing devices using a preset IoT communication protocol; obtaining the execution status and results of each invoice transaction by subscribing to the status feedback interface of the logistics entity devices; encapsulating the execution status and results, transaction identifiers and execution timestamps to form a structured edge transaction confirmation stream; and pushing the edge transaction confirmation stream to the edge data synchronization module through the edge computing devices to update the data.

[0041] Specifically, in offline inventory management scenarios, the on-site execution module receives the intelligent flow strategy generated by the flow decision module. For example, for an intelligent flow strategy of "product picking," the strategy might indicate: "SKU: XYZ456, quantity: 10 pieces, picked from shelf B, executed by the picking robot, completion time limit: 10 minutes." The on-site execution module parses this intelligent flow strategy to determine the list of logistics entity document transactions to be executed. Then, the system uses the MQTT protocol via edge computing devices (e.g., local servers in the store) to send JSON-formatted instructions to the corresponding logistics entity devices, such as picking robots or smart terminal trolleys, controlling them to execute the various document transactions in the list. For example, after receiving the instruction, the picking robot autonomously navigates to shelf B, identifies the product, and completes the picking. The edge computing device subscribes to the status feedback interface of the logistics entity devices in real time to obtain the execution status of these document transactions, such as "picking in progress," "completed," and the final result, such as "picking successful, actual quantity 10 pieces."

[0042] As a preferred implementation, obtaining the execution status and result of each ticket transaction specifically includes: if the obtained execution status indicates that the transaction execution has failed or exceeds a preset threshold, the edge computing device immediately triggers adaptive redirection calculation of the transaction based on the current offline flow path model, according to the remaining time limit and available resources, and sends a redirection instruction to the new processing node.

[0043] Specifically, when the edge computing device receives status feedback from the picking robot, displaying "Task failed: path blocked, estimated recovery time unknown," the system compares this failure status with the original time limit for the task (e.g., 8 minutes remaining). If it determines that recovery cannot be completed within the remaining time limit, the edge computing device invokes a simplified offline workflow path model, using "picking failure" as a new input condition, and combines it with currently available employee resources and geographical location to quickly calculate a new processing node, for example, assigning it to the nearest, idle "store employee B" for manual processing. Subsequently, the edge computing device immediately generates and pushes a redirection instruction to this new node, such as store employee B's handheld device. This solution provides immediate and adaptive processing capabilities for physical entity execution anomalies in an offline environment, greatly enhancing the robustness and continuity of business processes and effectively reducing the risk of task overdue. These execution statuses, results, transaction identifiers, execution timestamps, and device identifiers are encapsulated to form a structured edge transaction confirmation stream data packet. Subsequently, this edge transaction confirmation stream is pushed to the edge data synchronization module through the edge computing device to update relevant offline ticket operation data.

[0044] This invention enables the effective driving and execution recording of logistics entity transactions by the intelligent flow strategy of invoices in an offline environment, ensuring the automation of business operations and the verifiability of execution results, and providing real-time transaction confirmation for subsequent data synchronization.

[0045] Furthermore, during network link reconstruction, the process of link activation performed by the central scheduling computing unit and coordinating offline synchronization payloads and edge transaction confirmation streams includes: detecting the completion of network link reconstruction; initiating the link activation process through the central scheduling computing unit, including performing security authentication and establishing an encrypted transmission channel; synchronously receiving offline synchronization payloads and edge transaction confirmation streams from edge computing devices; matching the offline synchronization payloads and edge transaction confirmation streams based on timestamps, transaction identifiers, and data fingerprints; performing collaborative processing of the offline synchronization payloads and edge transaction confirmation streams based on a preset collaborative processing algorithm to resolve business data conflicts, including data inconsistency, operation sequence differences, and transaction duplication; and generating collaboratively processed business data to be aggregated.

[0046] Specifically, in offline inventory management scenarios, when the central data module detects that the network link reconstruction between it and the store edge computing devices is complete (e.g., when the WAN connection is restored), the central scheduling computing unit immediately initiates the link activation process. This process first performs security certificate authentication on the edge devices, and then establishes a stable encrypted data transmission channel between the central data module and each edge computing device. The link activation performed by the central scheduling computing unit specifically includes: before establishing the data transmission channel, the central scheduling computing unit queries the edge computing devices for the estimated total amount of offline data to be synchronized, and based on this estimated total amount and the current network status, dynamically calculates and allocates tiered bandwidth to ensure that edge computing devices with smaller total data volumes complete synchronization first. During the link activation process, the central scheduling computing unit first sends a data volume query message to the store edge computing devices. The edge computing devices reply with the total data volume of the offline synchronization payload and edge transaction confirmation stream stored locally; for example, store A replies with 20MB, and store B replies with 200KB. The central scheduling and computing unit dynamically allocates transmission quotas based on a pre-set bandwidth resource pool, such as a total bandwidth of 10Mbps and the amount of data from each store: stores with data volumes below 1MB (e.g., store B) are allocated the highest priority and 80% of peak bandwidth to ensure rapid synchronization; stores with large data volumes (e.g., store A) are allocated the remaining bandwidth to prevent them from unilaterally monopolizing resources. This hierarchical allocation is implemented in the data transmission channel through QoS policies. By using edge data volume as the decision-making basis for dynamic bandwidth allocation, this solution effectively alleviates the resource contention problem during network synchronization during link reconstruction, improves the fairness of synchronization tasks, and significantly enhances the overall efficiency of data aggregation.

[0047] Subsequently, the central data module synchronously receives the previously encapsulated offline synchronization payload from the edge computing device, containing the original offline ticket operation data fingerprint and data, and simultaneously receives the edge transaction confirmation stream generated by the field execution module, containing the actual execution status and results of the offline transactions. The central scheduling and computing unit precisely matches the received payload with the confirmation stream based on the embedded timestamp, unique transaction identifier, and hash fingerprint of the data. If data conflicts exist, including inconsistencies in quantity (e.g., the number of offline applications does not match the actual number of items entered into the database), incorrect operation order, or duplicate transaction reporting, the system will resolve them according to a preset collaborative processing algorithm and conflict type priority matrix. For example, for conflicts, the system will adjudicate according to the rule "based on the actual execution result"; for high-risk transaction conflicts, it will mark them according to the rule "requires manual intervention for review". Finally, a set of collaboratively processed, conflict-free business data to be aggregated is generated.

[0048] This invention enables the central end to automatically receive and efficiently coordinate offline synchronous data and transaction execution confirmation data by performing link activation and data coordination during network link reconstruction. This effectively solves the data conflicts that occur during offline periods and ensures the accuracy of central data.

[0049] Furthermore, the process of aggregating data to the central bill data storage and calibrating and confirming the central bill data through a business verification and global consistency arbitration mechanism includes: receiving the coordinated business data to be aggregated; transmitting the business data to be aggregated to the central bill data storage unit; initiating a business verification and global consistency arbitration mechanism based on a multi-dimensional data verification model; performing multi-dimensional data verification on the data to be calibrated in the central bill database according to a preset consensus rule set; calculating a data deviation index based on the multi-dimensional data verification results and calibrating the data according to a deviation handling strategy; determining whether the calibrated data meets the global consistency threshold through the business verification and global consistency arbitration mechanism, and confirming the data that meets the threshold; and updating the confirmed business bill data in the central bill data storage unit.

[0050] Specifically, in offline inventory management scenarios, the central data module receives collaborative business data to be aggregated, generated by the central scheduling and computing unit. For example, this might be a collaborative "goods receipt" record, which resolves discrepancies between offline records and actual execution results. After conflict resolution, the collaborative business data to be aggregated is transmitted to the central invoice data storage unit, whereby a business verification and global consistency arbitration mechanism based on a multi-dimensional data verification model is initiated. The execution logic of the multi-dimensional data verification model is as follows: the "multi-dimensional" verification model covers at least the following core dimensions: Data Integrity Dimension Verification: Verifies whether the hash fingerprint chain of data blocks is continuous and tamper-free. The system verifies whether the fingerprint of the current document can be derived from the historical fingerprint of the previously confirmed data and the current data content. If the verification fails, the deviation value of this dimension is directly quantified to the highest value (1.0). Time Dimension Verification: Verifies the logical sequence between the timestamps of edge transactions and the timestamps of the central system. For example, it verifies whether the completion time of an edge transaction is earlier than its recording time in the central system. Business Status Dimension Verification: Verifies whether the inventory balance or cash flow before and after the operation is logically closed-loop. For example, for documents involving inventory changes, it verifies whether the central theoretical inventory quantity matches the actual execution quantity at the edge. If there is a difference, the inventory deviation percentage is used as the original deviation value for this dimension.

[0051] The quantification and fusion calculation of the data deviation index: A preset consensus rule set establishes a priority relationship: data integrity verification takes precedence over time dimension verification, and time dimension verification takes precedence over business status dimension verification. Based on the verification results, the system calculates the data deviation index. This index is a weighted fusion of verification results from different dimensions into a unified quantitative indicator, ensuring its value range is between 0 and 1. For qualitative verification (such as the success or failure of integrity verification), the deviation result is directly quantified as 0.0 or 1.0. For quantitative verification (such as inventory deviation percentage), the system uses a preset non-linear mapping function to normalize its value range to the range of 0 and 1. The final data deviation index is calculated through a priority weighted summation model, which is the sum of the normalized deviation values ​​of each dimension multiplied by their corresponding weight coefficients. Among them, the weight coefficients (for example, the integrity dimension has the highest weight) are set according to the priority relationship of the preset rule set, and the sum of all weight coefficients is 1. Only when the data deviation index of the calibrated central bill data is lower than the currently set global consistency threshold will the data be finally confirmed and marked as authoritative and accurate data.

[0052] This invention enables the final aggregation, calibration, and confirmation of offline business data at the central terminal. Through business verification and a global consistency arbitration mechanism, it ensures the global consistency and accuracy of central invoice data, thereby guaranteeing the reliability of the data center.

[0053] Example 2

[0054] This embodiment provides a specific application of an offline-operable intelligent document management system in a smart warehousing scenario, specifically offline picking and receiving management in a large smart warehouse. In a smart warehousing environment, when the warehouse is disconnected from the central management system due to network failures, signal blind spots, or maintenance, this intelligent document management system ensures the continuity of warehouse operations and the accuracy of data.

[0055] Edge Data Synchronization Module: When the warehouse network is interrupted, warehouse workers use handheld PDAs or other edge computing devices to perform offline picking or inbound operations. For example, when a worker scans a product barcode to complete an inbound operation, the edge computing device acquires this offline document operation data in real time. The system uses a hash chain structure to fingerprint and authenticate this data, associating the current operation data fingerprint with historical fingerprints and storing them along with the original operation data locally on the PDA. When a worker moves from a network-disconnected area to an area with local network coverage, or when the device detects a connection to the warehouse's local gateway, the edge data synchronization module encapsulates an offline synchronization payload based on preset business strategies, such as "prioritizing urgent outbound data," and the current edge network topology (e.g., using Wi-Fi or ZigBee connections). This payload contains packaged and encrypted offline document operation data, awaiting transmission to the nearest edge server.

[0056] Risk Awareness Module: After receiving the offline synchronized payload, the edge server's built-in analysis unit initiates risk awareness. For each piece of document operation data in the payload, the system performs cross-modal comparison and semantic verification. For example, for a "shortage" data entry, the module compares its structure to whether it conforms to the preset "shortage reporting template," and analyzes the operator's historical behavior patterns to determine whether the frequency of similar operations performed by this employee within a short period is abnormal, or whether the amount of shortage far exceeds the historical average. Based on these comparison results, the system calculates the document structure anomaly index and the operation behavior deviation index, and weights and fuses them to generate a real-time edge business risk index. For example, if the shortage amount is too large and the operator's behavior is abnormal, a high risk index stream will be output.

[0057] The workflow decision module continuously collects edge business risk index streams output by the risk perception module, as well as sets of pending document transactions, such as "large return orders awaiting review" and "online orders awaiting picking." Simultaneously, the module determines the geographical location of the current edge computing device or its location within the store, and obtains a preset set of document transaction timeliness requirements, such as "urgent order picking must be completed within 15 minutes" and "high-risk returns must be manually reviewed within 30 minutes," and retrieves a list of available resources within the current store. This data is input into an offline workflow path model built on a directed graph structure. The nodes in this model are defined as edge computing devices and assigned logistics entities (such as human operators and picking robots). Edges represent business workflow paths, and their weights are defined as workflow costs. The workflow cost is a weighted sum of the current node's workload, geographical distance (the movement cost of personnel or entities), and risk index.

[0058] The flow decision module, based on a heuristic search algorithm, models the limited computing power and storage resources of edge computing devices, the geographical location of logistics entities, the business risk index of tasks, and preset timeliness requirements into a directed graph cost function for offline flow paths. This strategy aims to solve the optimization problem of task scheduling and computing resource allocation under multi-objective and hard constraints in the offline state of edge computing systems, thereby improving the accuracy of edge computing systems in responding to real-time business needs and task allocation.

[0059] The model utilizes a heuristic search algorithm for path deduction. Timeliness requirements are considered a hard constraint; the total time cost of any path must be less than a preset time limit, otherwise the path is excluded. Under the premise of satisfying the timeliness constraint, the algorithm aims to minimize the total flow cost of each path. The algorithm uses the shortest geographical distance from the current node to the destination or the estimated minimum number of processing steps as a heuristic function to guide the search direction. For example, for an inbound transaction marked as "high-risk" by the risk perception module, the system generates a "manager review to warehouse location verification" intelligent flow strategy and assigns it to the nearest warehouse manager and warehouse clerk with the lowest current workload. This strategy represents the path and node allocation scheme with the lowest total flow cost.

[0060] On-site execution module: Once the intelligent flow strategy is generated, the on-site execution module pushes it to the corresponding logistics entity via edge computing devices, such as designated warehouse robots, forklifts, or intelligent terminals of human operators. For example, for a "goods picking" task, the intelligent flow strategy instructs the picking robot to go to a specific shelf and pick a specified number of goods. The edge computing device sends control commands to the logistics entity through preset IoT communication protocols, such as MQTT or CoAP. To adapt to different types of devices, the edge device uses a common instruction set and converts it into instructions that can be recognized by the specific device through a protocol adaptation layer.

[0061] Edge computing devices control these logistics entities to execute specific document transactions and record their execution status and results in real time. The status feedback interface uses a publish / subscribe model to obtain real-time status information, with feedback data formatted uniformly as JSON. Feedback data includes at least a task identifier, current execution status (e.g., "Executing," "Success," "Failure"), and a timestamp. Edge devices monitor the heartbeat signals of logistics entities; if no status feedback is received within a preset time interval, the system determines that the device is offline or faulty and triggers adaptive redirection of the transaction, reassigning it to other robots or manual processing. These confirmation streams are then pushed back to the edge data synchronization module to update the locally stored data.

[0062] Central Data Module: When the warehouse network link is finally reconstructed and connected to the central system, the central scheduling and computing unit initiates the link activation process. It performs security authentication and establishes an encrypted transmission channel. The system intelligently and dynamically allocates bandwidth based on the amount of data to be synchronized on each edge computing device, prioritizing devices with smaller data volumes. The central system synchronously receives all accumulated offline synchronization loads and edge transaction confirmation streams from the edge computing devices. Collaborative Processing Algorithm Conflict Resolution: The system resolves business data conflicts, including data inconsistencies, differences in operation sequences, and duplicate transactions, based on a preset collaborative processing algorithm. This algorithm follows the following priorities and rules: Transaction duplicate conflict handling (highest priority): If two or more records are found to correspond to the same transaction identifier and have the same content, the system will retain the record that has passed the hash chain verification and has the earliest timestamp, and the remaining records will be marked as duplicates and deleted.

[0063] Data inconsistency conflict handling: If there is a discrepancy between the theoretical data in the central system and the results of the edge transaction confirmation stream (e.g., the theoretical inventory does not match the actual execution quantity), the system generally follows the principle of "subject to the actual execution result" because the confirmation stream is fingerprinted via a hash chain, which has higher credibility. However, if the edge business risk index of the transaction (output by the risk awareness module) exceeds the preset high-risk threshold (e.g., 0.8), the system will mark the conflict and automatically initiate a manual review process. This process is received and managed by the business verification and global consistency arbitration mechanism of the central data module.

[0064] Operation sequence difference conflict handling: If the system detects a logical conflict between the timestamp sequences of two related transactions, the cooperative algorithm will give priority to the operation sequence recorded by the record that has been confirmed by the hash chain and authenticated by multiple parties.

[0065] After conflict resolution, the coordinated business data to be aggregated is transmitted to the central bill data storage unit, and then a business verification and global consistency arbitration mechanism based on a multi-dimensional data verification model is initiated. The "multi-dimensional" verification model covers at least the following core dimensions: data integrity dimension (e.g., verifying whether the hash fingerprint chain of data blocks is continuous and tamper-free), time dimension (e.g., verifying the sequential logic between edge transaction timestamps and central system timestamps), and business status dimension (e.g., verifying the inventory balance or whether the cash flow is logically closed before and after the operation).

[0066] The preset consensus rule set is a configurable rule base used to guide the verification of different types of invoices (such as "inbound," "outbound," and "inventory count"). For example, for high-risk "inbound" invoices, the rule set will focus on performing inventory balance verification and supplier data traceability verification; while for "inventory shortage" invoices, it will focus on performing approval process integrity verification and inventory safety threshold verification. The priority relationship of this rule set is set as follows: data integrity verification takes precedence over time dimension verification, and time dimension verification takes precedence over business status dimension verification.

[0067] Based on the verification results, the system calculates the data deviation index. The global consistency threshold is a dynamically adjustable parameter. For highly sensitive business data (such as asset changes), the system uses a stricter threshold (e.g., below 0.001); while for low-sensitivity data (such as simple information query records), the system uses a more lenient threshold. Only when the calibrated central bill data deviation index is lower than the currently set global consistency threshold will the data be finally confirmed and marked as authoritative and accurate data.

[0068] The edge data synchronization module, through a hash chain structure, provides serialized proof of transactions generated under offline network conditions, solving the problem of data integrity assurance on edge nodes. The central data module's business verification and global consistency arbitration mechanism provides the ultimate technical guarantee for data synchronization and calibration of intermittently connected heterogeneous distributed nodes, effectively resolving the technical differences and transaction duplication issues in data consistency inherent in traditional edge cloud architectures, and improving the transaction processing reliability of the entire distributed system.

[0069] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent invoice management system capable of offline operation, characterized in that, include: Edge data synchronization module: It uses a hash chain structure to perform hash fingerprint verification on offline ticket operation data obtained from edge computing devices and stores it in the edge computing devices; Based on the edge network topology and service strategy, encapsulate the offline synchronization payload; Risk perception module: Through the analysis unit built into the edge computing device, it performs cross-modal comparison and semantic verification on offline invoice operation data, identifies invoice structure anomalies and quantifies operational behavior deviations, and outputs edge business risk index stream; Flow decision module: Collects edge business risk index stream and set of pending bill transactions; integrates edge geographical location and preset bill business timeliness requirements, drives offline flow path simulation, and generates intelligent bill flow strategy; On-site execution module: The intelligent flow strategy for invoices drives the transaction processing of logistics entities through edge computing devices and records the results, forming an edge transaction confirmation flow, and pushes updates to the edge data synchronization module; Central Data Module: During network link reconstruction, the central scheduling and computing unit performs link activation, coordinates offline synchronization of payloads and edge transaction confirmation flows; aggregates to the central bill data storage, and calibrates and confirms the central bill data through business verification and global consistency arbitration mechanism.

2. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The process of performing hash fingerprint verification on offline ticket operation data obtained from the edge computing device using a hash chain structure and storing it on the edge computing device includes: obtaining offline ticket operation data; parsing the offline ticket operation data to identify the operation time, operation user identifier, and business operation description; obtaining the historical fingerprint of the previous offline ticket operation data already stored on the edge computing device; using a cryptographic hash algorithm to perform hash operations on the operation time, operation user identifier, business operation description, and historical fingerprint to generate the current offline ticket operation data fingerprint; encapsulating the current offline ticket operation data fingerprint, the historical fingerprint, and the offline ticket operation data together into an offline data record; and storing the offline data record in the local storage unit of the edge computing device.

3. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The process of encapsulating the offline synchronization payload based on the edge network topology and service strategy includes: identifying the current edge network topology and obtaining available connection parameters and network status indicators; obtaining a data synchronization priority list and synchronization time window rules according to a preset service strategy; selecting target edge computing devices and gateways through a comprehensive evaluation model based on the available connection parameters, network status indicators, data synchronization priority list, and synchronization time window rules; determining the range and priority of offline ticket operation data to be synchronized based on the data synchronization priority list and synchronization time window rules; packaging the offline ticket operation data to be synchronized, the target edge computing device identifier, and metadata information for decryption into a data packet to be encrypted; and encrypting the data packet to be encrypted using a symmetric encryption algorithm to form the offline synchronization payload.

4. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The process by which the risk perception module performs cross-modal comparison and semantic verification of offline invoice operation data, identifies invoice structure anomalies, quantifies operational behavior deviations, and outputs an edge business risk index stream includes: receiving offline invoice operation data and decoupling the data into structural field modal data and temporal behavior modal data; calling the pre-stored structural field modal data corresponding to the business flow invoice template library and the temporal behavior modal data corresponding to the historical operational behavior pattern library; performing a structural field modal comparison between the invoice operation data and the business flow invoice template library to identify invoice structure integrity and format deviations; performing a behavioral comparison between the temporal behavior modal data and the historical operational behavior pattern library to identify operational behavior frequency and sequence deviations; calculating an invoice structure anomaly index based on the structural comparison results using a structural anomaly scoring model; calculating an operational behavior deviation index based on the behavioral comparison results using a temporal deviation scoring model; weighting and fusing the invoice structure anomaly index and the operational behavior deviation index according to a preset risk factor weight table to generate an edge business risk index; and appending a timestamp to the edge business risk index and aggregating it into an edge business risk index stream for output.

5. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The data collection includes the edge business risk index stream and the set of pending bill transactions. The process of integrating edge geographic location and preset timeliness requirements for invoice transactions to drive offline circulation path deduction and generate an intelligent invoice flow strategy includes: collecting edge business risk index streams and a set of invoice transactions to be executed; determining the geographic location information of the current edge computing device and its location in the store; obtaining a set of preset timeliness requirements for invoice transactions and associated resource availability information; initializing an offline circulation path model based on a directed graph model, inputting the edge business risk index stream, the set of invoice transactions to be executed, the current edge geographic location, and the set of preset timeliness requirements for invoice transactions; calculating the execution path and node allocation scheme of the offline circulation path based on a heuristic search algorithm, comprehensively considering the risk index, timeliness requirements, and geographic location; generating a structured intelligent invoice flow strategy based on the execution path and node allocation scheme, the intelligent invoice flow strategy including circulation path, node identifier, and time limit data fields; and outputting the intelligent invoice flow strategy.

6. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The process of the intelligent flow of invoices strategy driving logistics entity invoice transaction processing and recording results through edge computing devices to form an edge transaction confirmation stream and pushing updates to the edge data synchronization module includes: receiving the intelligent flow of invoices strategy; parsing the intelligent flow of invoices strategy to determine a list of logistics entity invoice transactions to be executed; controlling the corresponding logistics entity devices to execute each invoice transaction in the list of logistics entity invoice transactions to be executed through the edge computing devices using a preset IoT communication protocol; obtaining the execution status and results of each invoice transaction by subscribing to the status feedback interface of the logistics entity devices; encapsulating the execution status and results, transaction identifiers and execution timestamps to form a structured edge transaction confirmation stream; and pushing the edge transaction confirmation stream to the edge data synchronization module through the edge computing devices to update the data.

7. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, During network link reconstruction, the process of link activation performed by the central scheduling computing unit and the coordination of offline synchronization payload and edge transaction confirmation stream includes: detecting the completion of network link reconstruction; initiating the link activation process through the central scheduling computing unit, including performing security authentication and establishing an encrypted transmission channel; synchronously receiving offline synchronization payload and edge transaction confirmation stream from the edge computing device; matching the offline synchronization payload and the edge transaction confirmation stream based on timestamps, transaction identifiers, and data fingerprints; performing collaborative processing of the offline synchronization payload and the edge transaction confirmation stream based on a preset collaborative processing algorithm to resolve business data conflicts, including data inconsistency, operation sequence differences, and transaction duplication; and generating collaboratively processed business data to be aggregated.

8. The intelligent invoice management system capable of offline operation according to claim 1, characterized in that, The process of aggregating data to the central bill data storage and calibrating and confirming the central bill data through a business verification and global consistency arbitration mechanism includes: receiving the coordinated business data to be aggregated; transmitting the business data to be aggregated to the central bill data storage unit; initiating a business verification and global consistency arbitration mechanism based on a multi-dimensional data verification model; performing multi-dimensional data verification on the data to be calibrated in the central bill database according to a preset consensus rule set; calculating a data deviation index based on the multi-dimensional data verification results and calibrating the data according to a deviation handling strategy; determining whether the calibrated data meets the global consistency threshold through the business verification and global consistency arbitration mechanism, and confirming the data that meets the threshold; and updating the confirmed business bill data in the central bill data storage unit.

Images