Safety management method for financial data synchronization

Abstract

The invention relates to the technical field of data management, in particular to a safety management method for financial data synchronization, which comprises the following steps: acquiring an original financial data stream, and performing data fragmentation processing on the original financial data stream to obtain fragmented data packets and data fingerprints; performing storage node mapping on the redundant coded data packet according to the storage network to obtain a storage node address chain; and performing abnormal access behavior identification on the storage node address chain according to the security threshold model to obtain a security policy instruction, and deploying the financial scheduling model to a cloud edge collaboration platform to realize security management of financial data synchronization. According to the method, data fragmentation processing, encryption and identity binding are performed on the original financial data, so that the confidentiality of the data can be effectively protected, the data is prevented from being accessed or tampered by an unauthorized third party in a transmission process, and the encrypted data packet ensures secure transmission of the data.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and more specifically to a secure management method for synchronizing financial data. Background Technology

[0002] Traditional methods for transmitting financial data often lack sufficient encryption and authentication mechanisms, making them vulnerable to unauthorized access or tampering. Even with encryption, many traditional systems do not adequately guarantee security during encrypted data transmission, potentially falling victim to man-in-the-middle attacks. Furthermore, the lack of redundant coding mechanisms makes data recovery difficult in case of loss or corruption, impacting data integrity and reliability. Because traditional methods rely on static security rules and measures, they lack real-time dynamic monitoring and anomaly detection. Access to storage nodes is often not monitored promptly and accurately, leading to risks of malicious attacks, unauthorized access, or data leaks, thus failing to effectively guarantee data security. Additionally, traditional methods lack priority management mechanisms during data synchronization, making it impossible to rationally arrange the synchronization order based on the importance of different data and business needs. This can lead to delays in the transmission of critical financial data, impacting operational efficiency. Moreover, traditional methods often confine data synchronization and storage management to a single data center or traditional storage system, lacking support from cloud-edge collaborative platforms. Therefore, their real-time data management and dynamic adjustment capabilities are weak, making it difficult to flexibly address the synchronization needs of data from different sources and types. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a secure management method for financial data synchronization.

[0004] The technical solution adopted to solve the above-mentioned technical problems is: a secure management method for financial data synchronization, including: The original financial data stream is acquired, and the original financial data stream is processed by data fragmentation to obtain fragmented data packets and data fingerprints; the fragmented data packets are bound to the data source identity identifier to obtain encrypted financial data packets. Based on the encrypted financial data packets, a cross-node transmission path dynamic planning is performed on the storage network to obtain a transmission strategy; based on the transmission strategy, the encrypted financial data packets are redundantly encoded to obtain redundant encoded data packets. The redundant coded data packets are mapped to storage nodes according to the storage network to obtain a storage node address chain; the redundant coded data packets are associated with fragment check values ​​according to the storage node address chain to obtain a data check chain. A security threshold model is performed on the data verification chain to obtain a security threshold model; abnormal access behavior is identified on the storage node address chain based on the security threshold model to obtain security policy instructions. The security policy instructions are synchronized with priority weights to obtain a financial scheduling model; the financial scheduling model is then deployed to a cloud-edge collaboration platform to achieve secure management of financial data synchronization.

[0005] Preferably, the original financial data stream is subjected to data fragmentation processing to obtain fragmented data packets and data fingerprints, including: The original financial data stream is dynamically entropy monitored by the financial data acquisition terminal to obtain data content mutation points and logical fragmentation boundaries; The original financial data stream is fragmented according to the data content mutation points to obtain a fragmented data packet set, wherein the fragmented data packet set has a non-uniform granularity; Spatiotemporal coordinate encoding is embedded in each of the fragmented data packet sets to obtain an initial fragmentation unit; The fragment content within the initial fragment unit is hash-encoded to obtain a hash value; the fragment metadata within the initial fragment unit is fingerprint-encoded to obtain a data fingerprint.

[0006] Preferably, the fragmented data packets are bound to the source identity identifier to obtain encrypted financial data packets, including: The identity identifier of the data source is obtained based on the fragmented data packet, and the device fingerprint is determined based on the identity identifier of the data source. An asymmetric key pair is constructed based on the device fingerprint, wherein the asymmetric key pair includes a private key and a public key. The private key is embedded in the trusted execution environment of the data source, and the public key is bound to the original financial data stream to obtain an initial encryption carrier. The biometrics of the operator at the data source end are obtained, and the biometrics are associated and mapped with the initial encryption carrier to obtain the identity token corresponding to the original financial data stream; the identity token is cut into fragments matching the number of data packets to obtain the identity token fragments corresponding to the data packets. The identity token fragment is embedded into the fragmented data packet, and the fragmented data packet of the identity token fragment is encrypted using a chaotic sequence to obtain an encrypted fragmented data packet; The encrypted fragmented data packet is subjected to ciphertext field feature extraction according to the homomorphic encryption algorithm to obtain the encrypted metadata fingerprint; the encrypted metadata fingerprint and the identity token fragment are encoded to obtain the encrypted financial data packet.

[0007] Preferably, a transmission strategy is obtained by dynamically planning cross-node transmission paths in the storage network based on the encrypted financial data packets, including: The encrypted financial data packet is parsed to obtain the network topology; the link quality between nodes is evaluated based on the network topology to obtain the link quality evaluation set. The encrypted financial data packets are fragmented and prioritized according to the link quality assessment set to obtain fragmentation priority tags; node buffer queues are dynamically allocated according to the fragmentation priority tags to obtain node queue allocation status. The node queue allocation status is used to predict transmission delay trends to obtain path delay prediction data; path verification is performed based on the path delay prediction data to obtain a candidate path availability matrix. The encrypted financial data packet is subjected to path feature extraction based on the candidate path availability matrix to obtain a path feature vector; the transmission risk is quantitatively assessed based on the path feature vector to obtain a risk assessment coefficient table. The network topology is iteratively optimized using the risk assessment coefficient table to obtain a transmission strategy.

[0008] Preferably, the encrypted financial data packet is redundantly encoded according to the transmission strategy to obtain a redundant encoded data packet, including: The real-time bandwidth parameters and latency fluctuation characteristics of each transmission path are analyzed according to the transmission strategy to obtain a path feature matrix; the encrypted financial data packet is divided into time-series windows according to the path feature matrix to obtain a data block set. The data block set is evaluated for data importance level to obtain a data priority label sequence; the redundancy coding ratio of each data block is allocated according to the data priority label sequence to obtain a redundancy coding set; The data blocks in the data set are subjected to forward error correction coding based on the redundant coding set to obtain redundant coding packets; the redundant coding packets are subjected to temporal cross-validation to obtain temporal cross-validation related coding packets. Based on the path quality history record in the transmission strategy, the timing verification associated encoding packet is assigned a transmission risk weight to obtain a path adaptation encoding packet group; a dynamic redundancy backup strategy is generated based on the path adaptation encoding packet group to obtain a redundancy backup mapping table. The redundant backup mapping table and the timing verification associated encoding packet are bound and encapsulated to obtain a redundant encoded data packet.

[0009] Preferably, the redundant coded data packets are mapped to storage nodes according to the storage network to obtain a storage node address chain, including: The redundant encoded data packets are parsed to obtain fragmentation feature metadata; storage node load balancing analysis is performed based on the fragmentation feature metadata to obtain node evaluation information. Based on the node evaluation information, the redundant coded data packets are matched for storage requirements to obtain a storage strategy configuration parameter set; the storage strategy configuration parameter set is then subjected to distributed node screening to obtain candidate storage nodes. The candidate storage nodes are verified for trustworthiness to obtain node admission tokens; the redundant encoded data packets are fragmented and routed according to the node admission tokens to obtain storage path mapping records. The storage path mapping record is written into the storage ledger to obtain the initial storage node address chain; the initial storage node address chain is monitored in real time to obtain the availability tag; The initial storage node address chain is supplemented with redundant nodes based on the availability label to obtain the storage node address chain.

[0010] Preferably, the redundant coded data packets are fragmented and associated with checksums based on the storage node address chain to obtain a data checksum chain, including: The location of the storage node is determined based on the storage node address chain; the redundant encoded data packet is fragmented and verified based on the location of the storage node to obtain a hash fingerprint set; The hash fingerprint set and the storage node address chain corresponding to the storage ledger are associated and bound to obtain a mapping table between addresses and verification values; the mapping table between addresses and verification values ​​is organized in a hierarchical structure to obtain a verification chain structure. Verification rules are deployed on the verification chain structure to obtain a verification contract; the verification contract is cross-validated with external data sources to obtain verification logic. The redundant encoded data packet is subjected to integrity verification test according to the verification logic to obtain a verification pass mark; the verification pass mark and the storage node address chain corresponding to the storage network are timestamped to obtain a verification event record chain; The verification event record chain is broadcast to distributed storage nodes to obtain the data verification chain.

[0011] Preferably, a security threshold model is performed on the data verification chain to obtain a security threshold model, including: A verification chain feature set is determined based on the data verification chain; cross-domain access pattern clustering analysis is performed based on the verification chain feature set to obtain a behavioral pattern baseline; Anomaly detection is performed based on the behavioral pattern baseline to obtain an initial anomaly threshold boundary; the initial anomaly threshold boundary is dynamically weighted to obtain a threshold parameter set. Cross-domain data collaborative training is performed on the threshold parameter set to obtain a global security threshold model; the global security threshold model is deployed in a versioned manner to obtain a model update record chain; The global security threshold model is continuously learned and configured according to the model update record chain to obtain a threshold adjustment mechanism; the threshold adjustment mechanism is verified by attack and defense drills to obtain model evaluation information. The global security threshold model is iterated and optimized based on the model evaluation information to obtain the security threshold model.

[0012] Preferably, abnormal access behavior is identified in the storage node address chain based on the security threshold model to obtain security policy instructions, including: The storage node address chain is stream-processed according to the security threshold model to obtain a node access behavior feature stream; the node access behavior feature stream is then compared with a real-time threshold according to the security threshold model to obtain an abnormal alarm signal. Contextual behavior correlation analysis is performed on the abnormal alarm signals to obtain an abnormal event correlation graph; propagation path prediction is performed on the abnormal event correlation graph to obtain the attack entry point; Based on the attack entry point, the access path backtracking analysis of the storage node address chain is performed to obtain the attack evidence chain; the attacker profile is matched with the attack evidence chain to obtain the threat level label; Based on the threat level label, a pre-set response scheme is extracted from the data verification chain to obtain a candidate policy instruction set; the attack and defense benefits of the candidate policy instruction set are evaluated to obtain a policy execution sequence. The policy execution sequence is encapsulated into a security policy adjustment instruction set to obtain security policy instructions.

[0013] Preferably, the security policy instructions are synchronized with priority weights to obtain a financial scheduling model, including: The security policy adjustment instruction set is used to adjust the instruction types, execution sequence, and associated resource constraints in progress to obtain a policy instruction dataset; a state space is constructed based on the policy instruction dataset, wherein the state space includes the current network topology state, threat level distribution, and instruction execution queue; Define a reinforcement learning action space, wherein the action space includes instruction priority adjustment range, resource allocation ratio and synchronization timing offset; perform feasibility boundary labeling on the action space to obtain a set of safe and compliant action constraints; Design a composite reward function, which includes three dimensions: data synchronization timeliness, system load balancing, and security policy coverage. Perform dynamic weight calibration on the composite reward function to obtain an adaptive reward feedback mechanism. Initialize the parameters of the deep reinforcement learning network, and train the policy instruction dataset according to the parameters of the deep reinforcement learning network to obtain the financial scheduling model.

[0014] The beneficial effects of the present invention are as follows: (1) By performing data fragmentation, encryption and identity binding on the original financial data, the present invention can effectively protect the confidentiality of the data and prevent it from being accessed or tampered with by unauthorized third parties during transmission. The encrypted data packets ensure the secure transmission of the data. The redundancy coding technology provides a fault tolerance mechanism in the data transmission process, which can ensure that even if some data packets are lost or damaged during transmission, the data can still be recovered, thereby enhancing the reliability of data synchronization. (2) By establishing a security threshold model, this invention can monitor the access behavior of storage nodes in real time and identify potential abnormal behaviors. Timely identification of abnormal behaviors can help prevent malicious attacks, unauthorized access or data leakage, and ensure the integrity and security of financial data. Furthermore, by allocating synchronization priority weights, the synchronization priority of data can be reasonably arranged according to business needs and data importance, thereby ensuring the timely transmission and synchronization of key financial data, improving the efficiency of data synchronization and reducing unnecessary delays. (3) By deploying the financial scheduling model into the cloud-edge collaborative platform, the present invention can realize real-time management and dynamic adjustment of financial data. The cloud-edge collaborative platform can flexibly handle the synchronization needs of data from different sources and of different types, thereby improving the efficiency and flexibility of overall data management. Furthermore, through automated data sharding, encryption, path planning, and redundant coding, the manual management and operational complexity of financial data synchronization process is greatly reduced. The system's automatic security management and synchronization scheduling can reduce human error and improve the accuracy and efficiency of operations. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the overall method steps in one embodiment of the present invention. Detailed Implementation

[0016] Example 1, as Figure 1 As shown, the present invention proposes a secure management method for financial data synchronization, comprising: S1. Obtain the original financial data stream, perform data fragmentation on the original financial data stream to obtain fragmented data packets and data fingerprints; bind the data fragmented data packets to the identity identifier of the data source end to obtain encrypted financial data packets; S2. Perform cross-node transmission path dynamic planning on the storage network based on the encrypted financial data packets to obtain a transmission strategy; perform redundant encoding on the encrypted financial data packets based on the transmission strategy to obtain redundant encoded data packets. S3. Map the redundant coded data packets to storage nodes according to the storage network to obtain the storage node address chain; associate the fragmented check values ​​of the redundant coded data packets with tags according to the storage node address chain to obtain the data check chain. S4. Perform security threshold modeling on the data verification chain to obtain a security threshold model; identify abnormal access behavior on the storage node address chain based on the security threshold model to obtain security policy instructions. S5. Assign priority weights to security policy instructions to obtain a financial scheduling model; deploy the financial scheduling model to the cloud-edge collaboration platform to achieve secure management of financial data synchronization.

[0017] In this invention, the raw financial data stream refers to unprocessed data directly from the financial system or data source. In this method, this data requires subsequent encryption, fragmentation, and other processing to ensure data security and effective synchronization. The cloud-edge collaboration platform is an architecture combining cloud computing and edge computing, enabling distributed management of data processing and storage. In this platform, edge devices are responsible for real-time data processing, while the cloud handles large-scale data storage and efficient computation. The financial scheduling model refers to a scheduling scheme formulated based on different data synchronization needs and priorities. This model helps optimize the financial data synchronization process, ensuring that tasks are completed in a reasonable order and at the right time.

[0018] In an optional embodiment, the original financial data stream is fragmented to obtain fragmented data packets and data fingerprints, including: A1. Dynamically monitor the entropy value of the raw financial data stream using the financial data acquisition terminal to obtain the data content mutation points and logical segmentation boundaries; A2. The original financial data stream is fragmented according to the data content mutation points to obtain a fragmented data packet set, wherein the fragmented data packet set has a non-uniform granularity. A3. Embed spatiotemporal coordinate encoding for each fragmented data packet set to obtain the initial fragmented unit; A4. Hash the fragment content within the initial fragment unit to obtain a hash value; fingerprint the fragment metadata within the initial fragment unit to obtain a data fingerprint.

[0019] It should be noted that dynamic entropy monitoring utilizes financial data acquisition terminals to monitor the raw financial data stream in real time, analyzing the complexity and randomness of the data. Abrupt changes are points in the data stream where significant changes occur, potentially indicating anomalies or important turning points. Based on previously monitored abrupt changes, the raw financial data stream is segmented into multiple smaller data packets, forming a set of fragmented data packets. Non-uniform granularity means that the size of these data packets is not fixed; this ensures that data packets near abrupt changes are more granular, while larger data packets may exist in stable regions. Each data packet set is encoded with spatiotemporal coordinates; this identifier helps track the position and state of data packets within the data stream. Hash encoding performs a hash operation on the content of each fragment unit to obtain a fixed-length hash value. Data fingerprinting encodes the metadata of the fragments, generating a data fingerprint; the data fingerprint is used to ensure the security and integrity of data during transmission and storage.

[0020] In an optional embodiment, the fragmented data packets are bound to the source identity to obtain encrypted financial data packets, including: B1. Obtain the identity identifier of the data source end based on the fragmented data packets, and determine the device fingerprint based on the identity identifier of the data source end; construct an asymmetric key pair based on the device fingerprint, wherein the asymmetric key pair includes a private key and a public key, embed the private key into the trusted execution environment of the data source end, and bind the public key to the original financial data stream to obtain the initial encryption carrier; B2. Obtain the biometrics of the operators at the data source end, associate and map the biometrics with the initial encryption carrier to obtain the identity token corresponding to the original financial data stream; cut the identity token into fragments matching the number of data packets to obtain the identity token fragments corresponding to the data packets. B3. Embed the identity token fragments into the fragmented data packets, and perform chaotic sequence encryption on the fragmented data packets of the identity token fragments to obtain encrypted fragmented data packets; B4. Extract ciphertext field features from the encrypted fragmented data packets using the homomorphic encryption algorithm to obtain the encrypted metadata fingerprint; encode the encrypted metadata fingerprint and identity token fragments to obtain the encrypted financial data packets.

[0021] It should be noted that the source information of the data obtained from the fragmented data packets is mainly used to identify the sender of the data; the device fingerprint can be generated through hardware information, operating system characteristics, etc., and is a unique identifier for the device; based on the device fingerprint, a key pair is generated, including a private key and a public key; the operator's biometric features (such as fingerprints, facial recognition, etc.) are obtained, and these biometric features are used to verify identity; the biometric features are associated and mapped with the initial encryption carrier (the binding information of the public key and the financial data stream) to generate an identity token; the identity token is an authentication of the operator's identity and data access rights; the generated identity token is cut into multiple fragments, the number of which matches the number of fragmented data packets; the identity token fragments are embedded into the corresponding fragmented data packets to ensure that each data packet carries a portion of identity information; each fragmented data packet is encrypted using chaotic sequence encryption; the encrypted fragmented data packets are homomorphically encrypted; metadata is extracted from the homomorphically encrypted fragmented data packets, and encrypted metadata fingerprints are generated; these fingerprints are feature digests of the encrypted data and can be used to verify data integrity; the generated encrypted metadata fingerprints are encoded with the identity token fragments to generate the final encrypted financial data packet.

[0022] In an optional embodiment, dynamic cross-node transmission path planning is performed on the storage network based on encrypted financial data packets to obtain a transmission strategy, including: C1. Perform topology parsing on the encrypted financial data packets to obtain the network topology; perform link quality assessment between nodes based on the network topology to obtain the link quality assessment set. C2. Based on the link quality assessment set, fragment the encrypted financial data packets to mark the transmission priority, so as to obtain the fragment priority label; based on the fragment priority label, dynamically allocate the node buffer queue to obtain the node queue allocation status. C3. Perform transmission delay trend prediction on the node queue allocation status to obtain path delay prediction data; perform path verification based on the path delay prediction data to obtain the candidate path availability matrix. C4. Extract path features from encrypted financial data packets based on the candidate path availability matrix to obtain path feature vectors; quantify and assess transmission risks based on path feature vectors to obtain a risk assessment coefficient table. C5. Iteratively optimize the network topology using the risk assessment coefficient table to obtain the transmission strategy.

[0023] It should be noted that network topology refers to the nodes in a network and their interconnections. Transmission nodes refer to the devices involved in data transmission (such as routers and switches). Topology resolution analyzes these devices and connections to determine how data packets flow through the network. Link quality assessment evaluates the quality of each transmission path in the network. The purpose of the assessment is to understand the performance of different links and select the most suitable data path for transmission. The link quality assessment set refers to the quality assessment of each link in the network. This data provides the basis for subsequent priority marking and path selection. Data packets are often fragmented to improve transmission efficiency and fault tolerance. In the fragmentation transmission priority marking step, each fragment is prioritized based on the link quality assessment results. By marking the priority of different fragments, the system can rationally arrange data transmission according to the urgency and importance of different fragments. Each network node (e.g., a router) has a buffer queue to store data packets to be forwarded. Based on the priority label of the fragment, the node dynamically adjusts the allocation of the buffer queue to ensure that high-priority data packets are processed first, reducing transmission latency. By analyzing past data transmission latency data... The system performs analysis to predict future path transmission delays. The purpose of prediction is to identify potential delay issues in advance and optimize data packet transmission strategies. Path verification verifies each candidate path based on the predicted path delay data to determine whether it can effectively transmit data and meet security and delay requirements. After verification, a matrix is ​​generated, recording the availability of each candidate path, including path stability, delay, and bandwidth information. This matrix helps select the most suitable transmission path. Through feature extraction, the system can comprehensively understand the advantages and disadvantages of each path, providing a basis for subsequent decisions. These extracted feature information are converted into a vector. The system then quantitatively assesses the risks of path transmission. The goal of the assessment is to identify potential risks and quantify their potential impact. A risk assessment coefficient table calculates the risk coefficients for different paths based on the characteristics of the transmission path and environmental factors, and records them in the table. Based on the preceding risk assessment, the network topology is repeatedly optimized. By adjusting transmission paths, node allocation, priority strategies, etc., optimal network performance and security are achieved. The optimized transmission strategy is the final scheme used to guide data transmission, supporting the efficient and low-risk transmission of encrypted financial data packets.

[0024] In an optional embodiment, the encrypted financial data packets are redundantly encoded according to a transmission strategy to obtain redundantly encoded data packets, including: D1. Analyze the real-time bandwidth parameters and delay fluctuation characteristics of each transmission path according to the transmission strategy to obtain the path feature matrix; divide the encrypted financial data packets into time-series windows according to the path feature matrix to obtain the data block set. D2. Evaluate the importance level of the data block set to obtain a data priority label sequence; allocate the redundancy coding ratio of each data block according to the data priority label sequence to obtain a redundancy coding set; D3. Perform forward error correction coding on the data blocks in the dataset according to the redundant coding set to obtain the redundant coding packet; perform time-series cross-validation on the redundant coding packet to obtain the time-series cross-validation associated coding packet. D4. Assign transmission risk weights to the timing verification associated coding packets based on the path quality history records in the transmission strategy to obtain the path adaptation coding packet group; generate a dynamic redundancy backup strategy based on the path adaptation coding packet group to obtain the redundancy backup mapping table. D5. Bind and encapsulate the redundant backup mapping table and the timing verification associated encoding packet to obtain the redundant encoded data packet.

[0025] It should be noted that the system needs to collect real-time bandwidth information and latency fluctuations for each transmission path according to the transmission strategy. This is to understand the stability and performance of each transmission path and help optimize data transmission. By collecting this real-time information, the system generates a path feature matrix. Based on the path feature matrix, the data stream (encrypted financial data packets) is divided into multiple time-series windows. These windows will block the data according to time order, so that the data can be better matched with network conditions. Each time-series window is divided into multiple data blocks, which are transmitted and processed separately. The priority of each data block may depend on its content, purpose, or urgency related to the financial data. Importance assessment involves assigning a priority label (e.g., high, medium, low priority) to each data block, forming a data priority label sequence. To improve the reliability of data transmission in the network, data blocks are redundantly encoded. Redundancy encoding increases redundancy information so that even if some data packets are lost or corrupted, the data can be recovered using the redundancy information. Based on the data priority label, the system allocates different redundancy encoding ratios to each data block; higher-priority data blocks are allocated more redundancy information to improve transmission reliability. Forward error correction (FEC) coding is a key technology for redundancy encoding of data blocks. By adding redundancy information to data blocks, errors can be corrected at the receiving end. The system corrects lost data blocks without retransmitting them; through forward error correction coding, it generates coded packets containing redundant information, which can be recovered in the event of data loss; this ensures the correctness of redundant coded data in different time windows at the receiving end, and cross-checking verifies data integrity and avoids data loss or mistransmission; the system analyzes the quality of each transmission path based on historical records in the transmission strategy; it assigns transmission risk weights to each time-series check-associated coded packet, with the weight value reflecting the reliability of the path; if a path has poor historical quality, its risk weight is higher; after assigning transmission risk weights, a path-adaptive coded packet group is generated. This packet group is based on the path quality assessment results. Based on the path adaptation coding packet group, the system dynamically generates a redundancy backup strategy. This strategy determines the number and distribution of redundant backups based on the real-time quality changes of the path, ensuring that other paths can quickly take over transmission when a problem occurs on a certain path. The redundancy backup mapping table contains all redundant backup paths and their priorities, ensuring efficient redundancy backup management during transmission. The redundancy backup mapping table and the timing check associated coding packet are bound together and encapsulated into a new redundant coding data packet. This redundant coding data packet will be transmitted to the receiving end, ensuring that data can be recovered through redundancy backup and error correction coding even if it is lost or damaged during transmission.

[0026] In an optional embodiment, redundant coded data packets are mapped to storage nodes according to the storage network to obtain a storage node address chain, including: E1. Perform data feature parsing on redundant coded data packets to obtain fragmentation feature metadata; perform storage node load balancing analysis based on fragmentation feature metadata to obtain node evaluation information; E2. Based on the node evaluation information, perform storage requirement matching on redundant coded data packets to obtain a storage policy configuration parameter set; perform distributed node screening on the storage policy configuration parameter set to obtain candidate storage nodes; E3. Verify the trustworthiness of candidate storage nodes to obtain node admission tokens; perform fragmented routing allocation of redundant coded data packets based on node admission tokens to obtain storage path mapping records. E4. Write the storage path mapping record into the storage ledger to obtain the initial storage node address chain; perform real-time status monitoring on the initial storage node address chain to obtain the availability tag; E5. Based on the availability tags, the initial storage node address chain is supplemented with redundant nodes to obtain the storage node address chain.

[0027] It should be noted that in the data feature parsing step, the system analyzes the data packets and extracts feature information related to data blocks. By parsing the features of redundant coded data packets, the system can obtain "fragmentation feature metadata," which involves dividing the data packets into multiple small fragments and labeling their features. This information will help to efficiently process data during storage and transmission. Based on the "fragmentation feature metadata," the system will analyze the load of the current storage nodes. The purpose of load balancing analysis is to ensure that data is not concentrated on a single storage node, avoiding node overload. The results of load balancing will help evaluate the processing capacity and load status of each storage node. Based on the load balancing analysis results, the system... The system generates "node evaluation information" for each storage node. This information includes metrics such as the node's storage capacity, current load, and response time, which guide storage decisions. Based on the "node evaluation information" and the characteristics of redundant coded data packets, the system determines which storage nodes can most effectively store the data packet. This step matches storage requirements (such as storage capacity and speed) with the capabilities of each node. The matching results generate a storage policy configuration parameter set, which includes the storage policy for the data packet. Based on the "storage policy configuration parameter set," the system filters out qualified distributed storage nodes. These nodes should possess certain storage capacity, network connectivity, and reliability. Features ensure efficient and secure data storage; when selecting candidate storage nodes, their reliability and security must be guaranteed; after trust verification, qualified nodes are granted a "node admission token," meaning the node is approved to participate in data storage; based on the "node admission token" and the "candidate storage nodes," the system will route redundant coded data packets; the storage path of each data fragment (i.e., the address information of the storage node) will be recorded in a storage path mapping table, which is used to track the location of data storage; the storage path mapping record will be written to the "storage ledger," forming a global record of data storage; initial storage node address. This step generates an initial storage node address chain, which is generated based on storage path mapping records and contains the addresses of all storage nodes and data shard allocation information. The system monitors the real-time status of storage nodes to ensure their health. Based on the monitoring results, an "availability tag" is assigned to each storage node. The availability tag reflects the current availability status of each node, such as "normal," "overloaded," or "faulty." If a storage node fails or its availability decreases, the system automatically supplements it with redundant nodes based on the availability tag. After the redundant nodes are supplemented, the storage node address chain is updated to ensure that data can always be accessed through a valid path.

[0028] In an optional embodiment, redundant coded data packets are fragmented and associated with checksums based on the storage node address chain to obtain a data checksum chain, including: F1. Determine the location of the storage node based on the storage node address chain; perform fragmentation verification on the redundant encoded data packet based on the location of the storage node to obtain the hash fingerprint set; F2. Associate and bind the hash fingerprint set with the storage node address chain corresponding to the storage ledger to obtain a mapping table between addresses and check values; organize the mapping table between addresses and check values ​​into a hierarchical structure to obtain a check chain structure. F3. Deploy verification rules on the verification chain structure to obtain the verification contract; perform cross-validation of the verification contract with external data sources to obtain the verification logic. F4. Perform integrity verification tests on redundant encoded data packets according to the verification logic to obtain a verification pass mark; timestamp the verification pass mark and the storage node address chain corresponding to the storage network to obtain the verification event record chain; F5. Broadcast the verification event record chain to distributed storage nodes to obtain the data verification chain.

[0029] It should be noted that the storage node address chain is a chain that records the addresses of storage nodes; the address and status information of each storage node are reflected in this chain. Based on this address chain, the system can determine the specific location of the storage node (e.g., physical location, network location, etc.). In redundant coded data packets, data is divided into multiple fragments, each of which may be stored independently on different nodes. When verifying each fragment, the system calculates the hash value of that fragment. By performing hash calculations on each fragment, a hash fingerprint set can be obtained. This is crucial for data integrity verification because the hash value changes whenever the data changes, facilitating the detection of data integrity. By comparing the hash fingerprint set with the storage node addresses in the storage ledger... By associating and binding addresses through a chain, the system can map the hash value of each shard to its specific storage location, forming a "mapping table between addresses and checksums." This mapping table records the storage address of each shard and its hash fingerprint, allowing the system to track the specific location and integrity status of each shard. To optimize data storage and retrieval efficiency, the system organizes the mapping table into a hierarchical structure. This hierarchical structure facilitates rapid data location and management. The system generates a "checksum chain structure," which records the relationship between each shard, its hash value, and storage nodes. This structure facilitates verification and validation operations. Based on the checksum chain structure, the system deploys a set of verification rules that define how to verify the integrity of data packets. After deploying these rules, the system can automatically perform integrity checks on the stored data. Based on the verification rules, the system generates a "verification contract," which is a specific protocol or rule set for data verification, specifying how to perform verification and how to verify the integrity of data packets. To enhance the reliability and security of verification, the system also performs cross-validation of the verification contract with external data sources. This may mean comparing data from different sources to ensure that the stored data has not been tampered with. Through cross-validation, the system obtains the final "verification logic," which consists of the rules and methods required to ensure data integrity. The verification logic is the core of the final data integrity verification. After obtaining the verification logic, the system uses it to process redundant encoded data packets. Integrity verification: If the integrity of a data packet passes verification, the system marks the packet as "verification passed," indicating that the data integrity has been verified. To ensure the traceability and immutability of the verification process, the system timestamps the "verification passed marker" and the "storage node address chain." Each verification event is assigned a timestamp, ensuring that the historical record of data verification can be audited and tracked. After timestamp anchoring, the system generates a "verification event record chain," which records each successful verification event and its time. These records can provide a basis for subsequent auditing or analysis. To ensure that the verification results can be widely stored and accessed, the system broadcasts the "verification event record chain" to distributed storage nodes.The purpose of this is to ensure that verification results can be stored in multiple locations, preventing data loss or tampering; the chain of verification event records after broadcasting forms a "data verification chain".

[0030] In an optional embodiment, a security threshold model is performed on the data verification chain to obtain a security threshold model, including: G1. Determine the verification chain feature set based on the data verification chain; perform cross-domain access pattern clustering analysis based on the verification chain feature set to obtain the behavioral pattern baseline; G2. Detect outliers based on the behavioral pattern baseline to obtain the initial outlier threshold boundary; dynamically adjust the weights of the initial outlier threshold boundary to obtain the threshold parameter set. G3. Perform cross-domain data collaborative training on the threshold parameter set to obtain a global security threshold model; deploy the global security threshold model in a versioned manner to obtain a model update record chain. G4. Continuously learn and configure the global security threshold model based on the model update record chain to obtain the threshold adjustment mechanism; conduct attack and defense drills to verify the threshold adjustment mechanism to obtain model evaluation information. G5. Based on the model evaluation information, perform parameter optimization iteration on the global security threshold model to obtain the security threshold model.

[0031] It's important to note that the data verification chain refers to tracing the source, integrity, and legality of data through a data verification and auditing chain. By analyzing each data node in the data verification chain, its characteristic set is determined. These characteristics may include data traffic, transmission patterns, timestamps, user behavior, etc., thus providing data support for subsequent analysis. Cross-domain access patterns refer to access behavior across multiple networks or platforms, potentially involving different systems, applications, or geographical locations. Analyzing cross-domain access patterns can reveal the interaction methods between different domains. Using clustering algorithms to analyze cross-domain access patterns can identify user or data behavior patterns. The clustering results can help the system understand normal behavior patterns, thereby aiding in detection... Anomalies are provided as a reference; the behavioral pattern baseline refers to the baseline of normal behavior obtained through cluster analysis; it provides a standard for subsequent anomaly detection, and any behavior deviating from this baseline may be abnormal; anomalies may include abnormal access frequency, sudden changes in request timing or content, etc.; after detecting an anomaly, the system sets an initial anomaly threshold based on the statistical characteristics of the data or the results of the model; to adapt to changes in the network or environment, the initial anomaly threshold boundary needs to be dynamically adjusted; weight adjustments may be made based on newly acquired behavioral data, environmental changes, or changes in attack patterns to ensure that the threshold can adapt to constantly changing security threats; the adjusted threshold forms a set of parameters, which will be based on... Continuous optimization based on real-time data changes maintains system sensitivity and accuracy; data sharing and collaborative training across multiple domains or systems enhance model generalization and accuracy; the global security threshold model, obtained through cross-domain data collaborative training, comprehensively considers the security characteristics of different domains to formulate security threshold strategies suitable for the entire system; versioned deployment refers to version management of the trained and optimized global security threshold model and its deployment to the actual network environment; each model update generates a new version, facilitating retrospective review, inspection, and management of model changes; each update and deployment is recorded in a chain, forming the model's historical versions; this... The record chain helps analyze the model's evolution and the performance of different versions, ensuring that it can be restored to a previous stable version when necessary. This means that the system continuously adjusts and optimizes the model based on new data and environmental changes, maintaining its adaptability to security threats. Continuous learning configuration aims to automatically adjust threshold parameters when new security threats emerge. This mechanism ensures that the system can respond promptly to changing attack methods or user behaviors, improving detection accuracy. The effectiveness of the model is verified through simulated attack and defense strategies. The results of attack and defense drills provide an evaluation basis for the model's performance. This evaluation information can be used to improve the model. Through multiple iterations of optimization, the model can more accurately adapt to the network environment and security threats.Ultimately, the optimized model becomes the security threshold model, used to guide security decisions and network behavior analysis.

[0032] In an optional embodiment, abnormal access behavior is identified in the storage node address chain based on a security threshold model to obtain security policy instructions, including: H1. Perform streaming processing on the storage node address chain according to the security threshold model to obtain the node access behavior feature stream; perform real-time threshold comparison on the node access behavior feature stream according to the security threshold model to obtain abnormal alarm signals. H2. Perform contextual behavior correlation analysis on abnormal alarm signals to obtain an abnormal event correlation graph; predict the propagation path of the abnormal event correlation graph to obtain the attack entry point. H3. Perform access path backtracking analysis on the storage node address chain based on the attack entry point to obtain the attack evidence chain; perform attacker profile matching on the attack evidence chain to obtain threat level labels; H4. Extract pre-defined response schemes from the data verification chain based on threat level labels to obtain candidate policy instruction sets; evaluate the offensive and defensive benefits of the candidate policy instruction sets to obtain policy execution sequences. H5. Encapsulate the policy execution sequence into a security policy adjustment instruction set to obtain security policy instructions.

[0033] It should be noted that in a network, the access path of each storage node is represented by an address chain, which may include IP address, port number, etc.; the storage node address chain tracks the path of data storage; streaming processing refers to processing continuously flowing data in real time, and the processed data can be immediately used for subsequent analysis; here, the address chain data stream of storage nodes is analyzed in real time to extract the access behavior characteristics of each node, such as access frequency, access time period, request type, etc.; after streaming processing, the extracted behavioral characteristics of each node are used as the basis for anomaly detection, helping to identify whether there are abnormal behavioral patterns; based on a pre-set security threshold model (such as a behavioral pattern baseline), the node access behavior is compared with thresholds in real time; Changes in the characteristics are compared with preset normal values ​​or ranges. Once the value exceeds the threshold, it is considered abnormal behavior. When a node's access behavior does not conform to the normal range, an abnormal alarm signal is triggered. This signal indicates that the system has detected a potential security threat, which may require further analysis and response. Abnormal alarm signals are often isolated and need to be analyzed in conjunction with contextual information (such as the behavior of other related nodes, time, location, etc.). Through behavioral correlation analysis, it is possible to more accurately determine whether these abnormal alarms belong to the same event or whether they are related to other abnormal behaviors. Based on behavioral correlation analysis, a graph is constructed to represent the relationships between abnormal events. This graph helps to understand the propagation path and interrelationships of abnormal events. This provides a clear view of security incidents; abnormal events may spread across the network, or attackers may employ various methods to expand the attack scope; by analyzing the correlation and propagation methods between abnormal events, possible attack propagation paths can be predicted; the attack entry point refers to the point where the attacker first intrudes into the system; by predicting the propagation path, the attacker's entry point can be located, providing important clues for subsequent attack analysis and defense preparation; once the attack entry point is located, the next step is to trace how the attacker accessed the network and entered the system through different storage nodes; this process obtains a complete view of the attack chain by tracing the access path; through backtracking analysis, the path from the entry point to the attacker's final goal is obtained, forming a complete [system / mechanism]. Attack evidence chain; this chain serves as the basis for post-attack analysis and proof of attack behavior; by analyzing the attack evidence chain; based on the attacker's profile and factors such as the severity and scope of the attack, each attack behavior is labeled with a threat level; the threat level helps decision-makers quickly understand the risk level of the attack and decide what response measures to take; the data verification chain includes all processes used to ensure data security and legitimacy; based on the threat level label, the system can extract pre-set response plans from the verification chain; from the pre-set response plans, strategies suitable for the current threat level are selected, generating a set of candidate strategy instructions; these strategy instructions will be used for subsequent defense or remediation measures; the candidate strategies are evaluated, and the cost and benefits of each strategy are analyzed;The goal of the assessment is to find the most suitable response strategy that can maximally prevent attacks without incurring excessive resource consumption or business impact. After evaluating the cost-benefit of attack and defense, an execution sequence for the strategy is determined. Based on the strategy execution sequence, all defensive measures are translated into specific security policy instructions to strengthen protection. Finally, the instructions are sent to the system for execution, changing the system's behavior to address the current security threat and ultimately protecting the system from attacks.

[0034] In an optional embodiment, the security policy instructions are synchronized with priority weights to obtain a financial scheduling model, including: I1. Adjust the instruction types, execution sequence, and associated resource constraints of the security policy instruction set to obtain the policy instruction dataset; construct a state space based on the policy instruction dataset, where the state space includes the current network topology state, threat level distribution, and instruction execution queue; I2. Define the reinforcement learning action space, which includes the instruction priority adjustment range, resource allocation ratio, and synchronization timing offset; perform feasibility boundary labeling on the action space to obtain a set of safe and compliant action constraints; I3. Design a composite reward function, which includes three dimensions: data synchronization timeliness, system load balancing, and security policy coverage. Perform dynamic weight calibration on the composite reward function to obtain an adaptive reward feedback mechanism. I4. Initialize the deep reinforcement learning network parameters, and train the policy instruction dataset based on the deep reinforcement learning network parameters to obtain the financial scheduling model.

[0035] It's important to note that in reinforcement learning, the state space represents all possible states that an agent (system) can be in; the distribution of security threats in the network according to severity levels, such as which nodes are likely to be attacked and which nodes are at higher risk; the queue of security policy adjustment instructions currently pending execution, the execution order of the instructions, and their priorities; this state information helps the system understand the current environment and security situation so that it can take appropriate actions to optimize. In reinforcement learning, the action space is the set of all possible operations that an agent can take; according to the above description, the action space includes: the instruction priority adjustment range, which indicates the range within which the priority of instructions can be adjusted when adjusting the policy to optimize response speed and security effectiveness; the resource allocation ratio, which refers to the proportion of resources allocated to different tasks or network nodes, such as the optimization of CPU, bandwidth, or storage resource allocation; and the synchronization timing offset, which involves adjusting the timing order of instruction execution, such as how to optimize the coordination and synchronization between multiple tasks when they are running simultaneously. These actions define the policies that an agent can take in the environment, optimizing security policies and resource usage by selecting appropriate actions; not all actions are feasible, for example... For example, adjusting instruction priority or resource allocation may be limited by network topology, bandwidth, computing power, and other resources. By marking the feasibility boundaries of the action space, the system can eliminate infeasible actions and avoid non-compliant operations. To ensure safety and compliance, the system marks the action space, forming a set of constraints. This set of constraints ensures that all selected actions meet system safety requirements and resource usage limits. In reinforcement learning, the reward function is used to measure the feedback effect after the agent takes an action. The adaptive reward feedback mechanism can dynamically adjust the reward mechanism according to the actual operation effect, so that the system can adjust the optimization goal according to environmental changes and needs in continuous iteration. Deep reinforcement learning networks are usually based on deep neural networks to estimate and optimize policies. Initializing parameters is the first step in the training process, which usually includes the selection of network structure and the setting of initial weights. After initializing parameters, the system begins policy training by interacting with the environment. Through multiple iterations, the system learns how to adjust safety policies and resource allocation to optimize safety and efficiency in different network environments. The goal of training is to continuously improve the total score of the composite reward function, thereby deriving an optimal scheduling policy.

[0036] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A secure management method for synchronizing financial data. Includes, characterized in that: The original financial data stream is acquired, and the original financial data stream is processed by data fragmentation to obtain fragmented data packets and data fingerprints; the fragmented data packets are bound to the data source identity identifier to obtain encrypted financial data packets. Based on the encrypted financial data packets, a cross-node transmission path dynamic planning is performed on the storage network to obtain a transmission strategy; based on the transmission strategy, the encrypted financial data packets are redundantly encoded to obtain redundant encoded data packets. The redundant coded data packets are mapped to storage nodes according to the storage network to obtain a storage node address chain; the redundant coded data packets are associated with fragment check values ​​according to the storage node address chain to obtain a data check chain. A security threshold model is performed on the data verification chain to obtain a security threshold model; abnormal access behavior is identified on the storage node address chain based on the security threshold model to obtain security policy instructions. The security policy instructions are synchronized with priority weights to obtain a financial scheduling model; the financial scheduling model is then deployed to a cloud-edge collaboration platform to achieve secure management of financial data synchronization.

2. The security management method for financial data synchronization according to claim 1, characterized in that, The original financial data stream is fragmented to obtain fragmented data packets and data fingerprints, including: The original financial data stream is dynamically entropy monitored by the financial data acquisition terminal to obtain data content mutation points and logical fragmentation boundaries; The original financial data stream is fragmented according to the data content mutation points to obtain a fragmented data packet set, wherein the fragmented data packet set has a non-uniform granularity; Spatiotemporal coordinate encoding is embedded in each of the fragmented data packet sets to obtain an initial fragmentation unit; The fragment content within the initial fragment unit is hash-encoded to obtain a hash value; the fragment metadata within the initial fragment unit is fingerprint-encoded to obtain a data fingerprint.

3. The security management method for financial data synchronization according to claim 2, characterized in that, Binding the fragmented data packets to the source identity identifier to obtain encrypted financial data packets, including: The identity identifier of the data source is obtained based on the fragmented data packet, and the device fingerprint is determined based on the identity identifier of the data source. An asymmetric key pair is constructed based on the device fingerprint, wherein the asymmetric key pair includes a private key and a public key. The private key is embedded in the trusted execution environment of the data source, and the public key is bound to the original financial data stream to obtain an initial encryption carrier. The biometrics of the operator at the data source end are obtained, and the biometrics are associated and mapped with the initial encryption carrier to obtain the identity token corresponding to the original financial data stream; the identity token is cut into fragments matching the number of data packets to obtain the identity token fragments corresponding to the data packets. The identity token fragment is embedded into the fragmented data packet, and the fragmented data packet of the identity token fragment is encrypted using a chaotic sequence to obtain an encrypted fragmented data packet; The encrypted fragmented data packet is subjected to ciphertext field feature extraction according to the homomorphic encryption algorithm to obtain the encrypted metadata fingerprint; the encrypted metadata fingerprint and the identity token fragment are encoded to obtain the encrypted financial data packet.

4. A security management method for synchronizing financial data according to claim 3, characterized in that, Based on the encrypted financial data packets, dynamic cross-node transmission path planning is performed on the storage network to obtain a transmission strategy, including: The encrypted financial data packet is parsed to obtain the network topology; the link quality between nodes is evaluated based on the network topology to obtain the link quality evaluation set. The encrypted financial data packets are fragmented and prioritized according to the link quality assessment set to obtain fragmentation priority tags; node buffer queues are dynamically allocated according to the fragmentation priority tags to obtain node queue allocation status. The node queue allocation status is used to predict transmission delay trends to obtain path delay prediction data; path verification is performed based on the path delay prediction data to obtain a candidate path availability matrix. The encrypted financial data packet is subjected to path feature extraction based on the candidate path availability matrix to obtain a path feature vector; the transmission risk is quantitatively assessed based on the path feature vector to obtain a risk assessment coefficient table. The network topology is iteratively optimized using the risk assessment coefficient table to obtain a transmission strategy.

5. A security management method for financial data synchronization according to claim 4, characterized in that, The encrypted financial data packet is redundantly encoded according to the transmission strategy to obtain a redundant encoded data packet, including: The real-time bandwidth parameters and latency fluctuation characteristics of each transmission path are analyzed according to the transmission strategy to obtain a path feature matrix; the encrypted financial data packet is divided into time-series windows according to the path feature matrix to obtain a data block set. The data block set is evaluated for data importance level to obtain a data priority label sequence; the redundancy coding ratio of each data block is allocated according to the data priority label sequence to obtain a redundancy coding set; The data blocks in the data set are subjected to forward error correction coding based on the redundant coding set to obtain redundant coding packets; the redundant coding packets are subjected to temporal cross-validation to obtain temporal cross-validation related coding packets. Based on the path quality history record in the transmission strategy, the timing verification associated encoding packet is assigned a transmission risk weight to obtain a path adaptation encoding packet group; a dynamic redundancy backup strategy is generated based on the path adaptation encoding packet group to obtain a redundancy backup mapping table. The redundant backup mapping table and the timing verification associated encoding packet are bound and encapsulated to obtain a redundant encoded data packet.

6. A security management method for synchronizing financial data according to claim 5, characterized in that, Based on the storage network, the redundant coded data packets are mapped to storage nodes to obtain a storage node address chain, including: The redundant encoded data packets are parsed to obtain fragmentation feature metadata; storage node load balancing analysis is performed based on the fragmentation feature metadata to obtain node evaluation information. Based on the node evaluation information, the redundant coded data packets are matched for storage requirements to obtain a storage strategy configuration parameter set; the storage strategy configuration parameter set is then subjected to distributed node screening to obtain candidate storage nodes. The candidate storage nodes are verified for trustworthiness to obtain node admission tokens; the redundant encoded data packets are fragmented and routed according to the node admission tokens to obtain storage path mapping records. The storage path mapping record is written into the storage ledger to obtain the initial storage node address chain; the initial storage node address chain is monitored in real time to obtain the availability tag; The initial storage node address chain is supplemented with redundant nodes based on the availability label to obtain the storage node address chain.

7. A security management method for synchronizing financial data according to claim 6, characterized in that, The redundant coded data packets are fragmented and associated with checksums based on the storage node address chain to obtain a data checksum chain, including: The location of the storage node is determined based on the storage node address chain; the redundant encoded data packet is fragmented and verified based on the location of the storage node to obtain a hash fingerprint set; The hash fingerprint set and the storage node address chain corresponding to the storage ledger are associated and bound to obtain a mapping table between addresses and verification values; the mapping table between addresses and verification values ​​is organized in a hierarchical structure to obtain a verification chain structure. Verification rules are deployed on the verification chain structure to obtain a verification contract; the verification contract is cross-validated with external data sources to obtain verification logic. The redundant encoded data packet is subjected to integrity verification test according to the verification logic to obtain a verification pass mark; the verification pass mark and the storage node address chain corresponding to the storage network are timestamped to obtain a verification event record chain; The verification event record chain is broadcast to distributed storage nodes to obtain the data verification chain.

8. A security management method for synchronizing financial data according to claim 7, characterized in that, A security threshold model is performed on the data verification chain to obtain a security threshold model, including: A verification chain feature set is determined based on the data verification chain; cross-domain access pattern clustering analysis is performed based on the verification chain feature set to obtain a behavioral pattern baseline; Anomaly detection is performed based on the behavioral pattern baseline to obtain an initial anomaly threshold boundary; the initial anomaly threshold boundary is dynamically weighted to obtain a threshold parameter set. Cross-domain data collaborative training is performed on the threshold parameter set to obtain a global security threshold model; the global security threshold model is deployed in a versioned manner to obtain a model update record chain; The global security threshold model is continuously learned and configured according to the model update record chain to obtain a threshold adjustment mechanism; the threshold adjustment mechanism is verified by attack and defense drills to obtain model evaluation information. The global security threshold model is iterated and optimized based on the model evaluation information to obtain the security threshold model.

9. A security management method for synchronizing financial data according to claim 8, characterized in that, Based on the security threshold model, abnormal access behavior is identified in the storage node address chain to obtain security policy instructions, including: The storage node address chain is stream-processed according to the security threshold model to obtain a node access behavior feature stream; the node access behavior feature stream is then compared with a real-time threshold according to the security threshold model to obtain an abnormal alarm signal. Contextual behavior correlation analysis is performed on the abnormal alarm signals to obtain an abnormal event correlation graph; propagation path prediction is performed on the abnormal event correlation graph to obtain the attack entry point; Based on the attack entry point, the access path backtracking analysis of the storage node address chain is performed to obtain the attack evidence chain; the attacker profile is matched with the attack evidence chain to obtain the threat level label; Based on the threat level label, a pre-set response scheme is extracted from the data verification chain to obtain a candidate policy instruction set; the attack and defense benefits of the candidate policy instruction set are evaluated to obtain a policy execution sequence. The policy execution sequence is encapsulated into a security policy adjustment instruction set to obtain security policy instructions.

10. A security management method for financial data synchronization according to claim 9, characterized in that, The security policy instructions are synchronized with priority weights to obtain a financial scheduling model, including: The security policy adjustment instruction set is used to adjust the instruction types, execution sequence, and associated resource constraints in progress to obtain a policy instruction dataset; a state space is constructed based on the policy instruction dataset, wherein the state space includes the current network topology state, threat level distribution, and instruction execution queue; Define a reinforcement learning action space, wherein the action space includes instruction priority adjustment range, resource allocation ratio and synchronization timing offset; perform feasibility boundary labeling on the action space to obtain a set of safe and compliant action constraints; Design a composite reward function, which includes three dimensions: data synchronization timeliness, system load balancing, and security policy coverage. Perform dynamic weight calibration on the composite reward function to obtain an adaptive reward feedback mechanism. Initialize the parameters of the deep reinforcement learning network, and train the policy instruction dataset according to the parameters of the deep reinforcement learning network to obtain the financial scheduling model.

Images