Blockchain-based tamper-proofing method and system for financial instruments

Abstract

The application discloses a kind of financial voucher tamper-proofing method and system based on blockchain, it is related to blockchain technology field, the complete process of the method by voucher collection, data standardization processing, path consistency verification, time consensus detection to comprehensive reliability evaluation.Specifically set around financial voucher generation node, can real-time acquisition voucher data, and through TLS1.3 protocol encryption transmission to blockchain system, realizes the real-time on-chain of voucher data and structured delivery, significantly improves the data synchronization efficiency between voucher generation and on-chain.Through establishing distributed node database and association tracking calculation unit, so that voucher information has verifiability and anti-tamper property in the whole process of acquisition, transmission, verification, comparison, effectively avoids the data forgery, loss and node forgery problem that may occur in the process of voucher transmission in traditional financial system.

Description

Technical Field

[0001] This invention relates to the field of blockchain technology, specifically to a blockchain-based method and system for preventing the tampering of financial documents. Background Technology

[0002] With the rapid development of blockchain technology, digital financial management, and intelligent auditing systems, the data security, trustworthiness, and tamper-proof nature of financial vouchers have gradually become an important technical direction for enterprise internal control and compliance auditing. This is aimed at electronic financial vouchers generated, transmitted, and archived in the financial systems of enterprises or institutions, such as electronic invoices, contracts, payment vouchers, expense reports, and other electronic documents.

[0003] Currently, while some companies have adopted blockchain for on-chain document storage in their existing financial document management systems, most solutions only achieve hash-based tamper-proofing at the "single document level," lacking structured relationship management between documents and verifiable tracking mechanisms. This storage model fails to describe the upstream and downstream relationships of documents in the business process, resulting in a lack of correspondence between "source documents" and "sub-documents" in the financial document chain. When tracing fund flows or document sources, auditing departments need to rely on offline systems or manual comparisons, which is inefficient and prone to errors. Furthermore, traditional solutions do not establish a logical path matching mechanism between documents. When documents are tampered with, forged, or reused during transmission or splitting, the blockchain system cannot automatically identify broken paths or duplicate documents, thus failing to quantitatively assess the overall continuity and reliability of the document chain. These deficiencies significantly reduce the system's anti-counterfeiting and anti-tampering capabilities when facing complex business scenarios such as fund flows, multi-level reimbursements, and subcontracting settlements.

[0004] The main reason for the above problems is that existing blockchain evidence storage systems lack a hierarchical model of the credential association structure and a temporal verification mechanism. The data upload process is typically based on single-node writing, lacking a "family tree"-like parent-child relationship mapping between credentials, resulting in fragmented credential traceability paths. Furthermore, because blockchain systems only record transaction packaging time in the time dimension, without modeling the temporal characteristics of the entire process of credential generation, broadcasting, and confirmation, when multiple nodes concurrently submit credentials within a short period, the system struggles to distinguish between normal synchronization delays and malicious duplicate submissions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a blockchain-based method and system for preventing the tampering of financial documents, thus solving the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution, comprising the following steps:

[0007] S1. Set up voucher collection points around the financial voucher generation node to collect voucher data in real time and transmit the voucher data to the blockchain system. Then, preprocess the voucher data to obtain a standard voucher dataset.

[0008] S2. Based on the standard voucher dataset, perform voucher association consistency analysis, and based on the results of the voucher association consistency analysis, make a preliminary determination of the continuity reliability of the voucher association path;

[0009] S3. When it is initially determined that the continuity of the credibility of the credential association path is abnormal, the time consensus detection is automatically triggered, the time feature vector is extracted, and dynamic conflict consistency analysis is performed.

[0010] S4. Perform a comprehensive analysis of the dynamic conflict consistency analysis results and the voucher association consistency analysis results, and based on the comprehensive analysis results, conduct a comprehensive credibility assessment, and take corresponding actions based on the comprehensive credibility assessment results.

[0011] Preferably, S1 includes S11;

[0012] S11. Set up voucher collection points around the financial voucher generation node, and set up collection tools in the collection points to collect voucher data in real time; after collection, encapsulate the voucher data into a data block structure and transmit it to the blockchain system through the secure transmission protocol TLS1.3; after receiving the data block structure of the voucher data, the blockchain system stores it in the distributed node database.

[0013] The collection points include voucher recognition nodes, time nodes, signature nodes, and amount nodes;

[0014] The voucher data includes voucher number Vid, upstream voucher number Pid, ​​voucher amount Am, and generation time Ts.

[0015] Preferably, S1 further includes S12;

[0016] S12. Decapsulate the data block structure of the encapsulated voucher data to extract the voucher data, and preprocess the voucher data to obtain a standardized voucher dataset; the preprocessing includes verification processing, data cleaning, and dimensionless processing.

[0017] The verification process verifies the legitimacy of the node identity by verifying the credential generated based on the node signature NodeSig, and detects duplicates and conflicts between the credential number Vid and the upstream credential number Pid. When a duplicate credential number is detected, the credential correction logic is invoked to regenerate the credential with the duplicate number, forming a unique credential number Vid. When a conflict is detected, the conflicting credential data is automatically frozen, and an anomaly report is triggered to the audit system through a smart contract, prompting manual correction and re-uploading to the chain. After verification, a qualified credential set is output, and duplicate and conflicting credential records are removed.

[0018] The data cleaning process standardizes the voucher amount Am, including currency conversion, tax rate correction, and decimal precision adjustment, to form a standard amount field AmStd; and performs default completion operations on invalid fields and incomplete voucher information.

[0019] The dimensionless processing involves using the range normalization method to make the voucher data after data cleaning and verification dimensionless, thereby eliminating the unit dimension differences between voucher data.

[0020] After dimensionless processing, all data are aggregated to generate a standardized voucher dataset.

[0021] Preferably, S2 includes S21;

[0022] S21. Based on the standardized voucher dataset, an improved structure of the Merkle path matching algorithm is used to perform voucher association consistency analysis, and the voucher association consistency coefficient Cassoc is output; the voucher association consistency analysis includes the path construction stage, the node hash calculation stage, and the path integration stage;

[0023] The path construction phase establishes a hierarchical voucher association tree structure based on voucher number Vid and upstream voucher number Pid.

[0024] The node hash calculation stage generates a path matching function value by performing hash comparison on each pair of adjacent credential nodes;

[0025] The path integration stage outputs the credential association consistency coefficient Cassoc by counting the number of successfully matched nodes in the entire credential chain.

[0026] The document association consistency coefficient Cassoc is calculated and output using the following algorithm formula;

[0027] ;

[0028] In the formula, N represents the number of nodes in the credential chain, and Vid i Pid represents the document number of the i-th financial document in the blockchain system. iVid represents the upstream document number of the i-th financial document in the blockchain system. i+1 Pid represents the document number of the (i+1)th financial document in the blockchain system. i+1 represents the upstream document number of the (i+1)th financial document in the blockchain system, Hash represents the hash mapping function, match represents the hash matching function; the document association consistency coefficient Cassoc output result belongs to [0,1].

[0029] Preferably, S2 further includes S22;

[0030] S21. Based on the output of the document association consistency coefficient Cassoc and the preset consistency threshold Th1, a preliminary judgment is made to determine the continuity reliability of the document association path; the specific judgment content is as follows:

[0031] When the consistency coefficient Cassoc of the credential association is greater than or equal to the consistency threshold Th1, the credential association path is determined to be continuous and trustworthy, and the current credential chain is archived as a trustworthy chain.

[0032] When the consistency coefficient Cassoc of the credential association is less than the consistency threshold Th1, it is determined to be an abnormality in continuous credibility, and time consensus detection is automatically triggered.

[0033] Preferably, S3 includes S31;

[0034] S31. After initially determining that the time consensus detection has been triggered, a dynamic time consensus window △T is set in the blockchain system. The dynamic time consensus window △T is dynamically and adaptively generated based on the sum of the average consensus delay of the historical blockchain system and twice the standard deviation of the delay.

[0035] Within the dynamic time consensus window △T, the blockchain system synchronously obtains time parameters from two time collection points;

[0036] The time acquisition points include a first time acquisition point and a second time acquisition point;

[0037] The time parameters include the voucher generation time T1 and the block confirmation time T3.

[0038] Preferably, S3 further includes S32;

[0039] S31. Perform dynamic conflict consistency analysis based on time parameters and output the dynamic conflict consistency coefficient Cconf; the dynamic conflict consistency coefficient Cconf is calculated and output using the following algorithm formula.

[0040] ;

[0041] In the formula, M represents the number of detected credentials within the dynamic time consensus window ΔT, uT represents the average consensus latency, pT represents the latency standard deviation, and T3 represents the latency standard deviation. i T1 represents the block confirmation time of the i-th financial document in the blockchain system. i This indicates the time when the i-th financial document was generated in the blockchain system.

[0042] Preferably, S4 includes S41;

[0043] S41. The dynamic conflict consistency analysis and the document association consistency analysis output dynamic conflict consistency coefficient Cconf and document association consistency coefficient Cassoc are comprehensively analyzed. The comprehensive analysis is performed by calculating the average of the dynamic conflict consistency coefficient Cconf and the document association consistency coefficient Cassoc and outputting the comprehensive credibility evaluation index Sfinal.

[0044] Preferably, S4 further includes S42;

[0045] S42. Based on the comprehensive credibility assessment index Sfinal output by the comprehensive analysis, a comprehensive credibility assessment is conducted, and corresponding actions are taken based on the comprehensive credibility assessment results; the specific assessment content is as follows:

[0046] When the comprehensive credibility assessment index Sfinal≥0.9, the certificate genealogy chain is determined to be completely credible and on-chain confirmation is allowed;

[0047] When 0.7 ≤ comprehensive credibility assessment index Sfinal < 0.9, the credential chain is judged to have a moderate risk and is transferred to the manual review buffer for secondary verification;

[0048] When the comprehensive credibility assessment index Sfinal < 0.7, it is determined to be an abnormal credential chain, triggering the freezing and alarm mechanism.

[0049] A blockchain-based anti-tampering system for financial vouchers includes a voucher acquisition module, a consistency analysis module, a dynamic conflict analysis module, and a comprehensive analysis module.

[0050] The voucher collection module collects voucher data in real time by setting up voucher collection points around the financial voucher generation node, transmits the voucher data to the blockchain system, and then preprocesses the voucher data to obtain a standard voucher dataset.

[0051] The consistency analysis module performs document association consistency analysis based on a standard document dataset, and makes a preliminary determination of the continuity reliability of document association paths based on the results of the document association consistency analysis.

[0052] The dynamic conflict analysis module automatically triggers time consensus detection, extracts time feature vectors, and performs dynamic conflict consistency analysis when it initially determines that the continuity of the credential association path is abnormal.

[0053] The comprehensive analysis module integrates the dynamic conflict consistency analysis results with the voucher association consistency analysis results, performs a comprehensive credibility assessment based on the comprehensive analysis results, and executes corresponding actions based on the comprehensive credibility assessment results.

[0054] This invention provides a blockchain-based method and system for preventing the tampering of financial documents. It offers the following advantages:

[0055] (1) This method forms a complete process from voucher collection, data standardization processing, path consistency verification, time consensus detection to comprehensive credibility assessment through S1 to S4. In S1, by setting up a multi-point collection system around the financial voucher generation node, including voucher identification node, time node, signature node and amount node, key information such as voucher number Vid, upstream voucher number Pid, ​​voucher amount Am and generation time Ts can be collected in real time and transmitted to the blockchain system in encrypted form through the TLS1.3 protocol, realizing real-time on-chain and structured transmission of voucher data, which significantly improves the data synchronization efficiency between voucher generation and on-chain. By establishing a distributed node database and associated tracking computing unit, the voucher information has verifiability and tamper-proofness throughout the entire process of collection, transmission, verification and comparison, effectively avoiding data forgery, loss and node forgery problems that may occur during voucher transmission in traditional financial systems, thereby realizing complete tracking and reliable recording of voucher data throughout its entire lifecycle on the chain.

[0056] (2) This method constructs an algorithm system for joint verification of structural consistency and temporal consistency through the credential association consistency coefficient Cassoc and dynamic conflict consistency coefficient Cconf designed in S2 and S3. In S2, the improved Merkle path matching algorithm is used to construct the credential association tree structure. Through hash comparison and path integral calculation, the credential association consistency coefficient Cassoc is output, realizing the structural continuity verification of the credential chain. When the credential association consistency coefficient Cassoc is lower than the preset consistency threshold Th1, the time consensus detection module of S3 is automatically triggered, thus entering the dynamic conflict analysis stage based on time parameters. In this stage, the dynamic time consensus window △T is constructed, and the dynamic conflict consistency coefficient Cconf is calculated by combining the statistical model of average consensus delay and standard deviation. This coefficient can effectively identify forged credential behaviors such as abnormal concurrency, repeated on-chain, and time anomalies. Through the joint analysis of the credential association consistency coefficient Cassoc and the dynamic conflict consistency coefficient Cconf, the system provides dual protection for credential credibility at both the structural and temporal levels, upgrading the anti-tampering detection from static data verification to dynamic behavior analysis, realizing the transformation from "passive anti-counterfeiting" to "active identification" of a trusted anti-tampering mechanism.

[0057] (3) This method calculates the average of the certificate association consistency coefficient Cassoc and the dynamic conflict consistency coefficient Cconf, and outputs the comprehensive credibility assessment index Sfinal, thereby achieving a quantitative assessment of the overall credibility level of the certificate family chain. When the comprehensive credibility assessment index Sfinal ≥ 0.9, the chain is automatically determined to be a fully trustworthy chain and allowed to be confirmed on the chain; when 0.7 ≤ comprehensive credibility assessment index Sfinal < 0.9, the system marks the certificate chain as medium risk and enters the manual review buffer; when the comprehensive credibility assessment index Sfinal < 0.7, the smart contract module is automatically triggered to perform freezing and anomaly reporting operations. Through this comprehensive assessment and execution strategy, the system realizes automatic hierarchical management and dynamic handling of certificate status, avoids misjudgment and delay caused by manual intervention, and improves the security and autonomy of the blockchain system in a multi-node collaborative environment. In addition, the comprehensive credibility judgment mechanism also supports the audit system to call the results in real time, realize the visual supervision and dynamic risk warning of the certificate family chain, and significantly improve the anti-counterfeiting accuracy and intelligence level of the financial certificate system. Attached Figure Description

[0058] Figure 1 This is a schematic diagram illustrating the steps of a blockchain-based method for preventing tampering with financial documents according to the present invention.

[0059] Figure 2 This is a schematic diagram of a blockchain-based anti-tampering system for financial documents according to the present invention.

[0060] Figure 3This is a graph showing the change in the dynamic conflict consistency coefficient Cconf. Detailed Implementation

[0061] 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.

[0062] Example 1

[0063] Please see Figure 1 and Figure 3 This invention provides a blockchain-based method for preventing the tampering of financial documents. To achieve the above objectives, this invention employs the following technical solution, comprising the following steps:

[0064] S1. Set up voucher collection points around the financial voucher generation node to collect voucher data in real time and transmit the voucher data to the blockchain system. Then, preprocess the voucher data to obtain a standard voucher dataset.

[0065] S2. Based on the standard voucher dataset, perform voucher association consistency analysis, and based on the results of the voucher association consistency analysis, make a preliminary determination of the continuity reliability of the voucher association path;

[0066] S3. When it is initially determined that the continuity of the credibility of the credential association path is abnormal, the time consensus detection is automatically triggered, the time feature vector is extracted, and dynamic conflict consistency analysis is performed.

[0067] S4. Perform a comprehensive analysis of the dynamic conflict consistency analysis results and the voucher association consistency analysis results, and based on the comprehensive analysis results, conduct a comprehensive credibility assessment, and take corresponding actions based on the comprehensive credibility assessment results.

[0068] In this embodiment, in stage S1, the method sets up multiple data acquisition devices around the financial voucher generation node to achieve real-time data acquisition and on-chaining of voucher data. This setup is because the generation of financial vouchers is often distributed across different departments and nodes. Delays in data acquisition or manual entry can lead to incorrect voucher numbers or timestamp offsets, introducing uncertainty into subsequent verification. By deploying acquisition points at the source and using TLS 1.3 encrypted transmission, data can be on-chained instantly upon voucher generation, ensuring time synchronization and data uniqueness, effectively preventing voucher forgery or subsequent tampering. Secondly, in stage S2, the system uses a standardized voucher dataset to establish a voucher association tree structure and calculates the voucher association consistency coefficient Cassoc using an improved Merkle path matching algorithm. This algorithm structurally reflects the continuity of the inheritance relationship between vouchers. If a node's hash value is abnormal or the match fails, it indicates a break in the voucher's transmission chain or a risk of replacement. This structured analysis method essentially ensures the authenticity of the voucher path through the irreversibility of mathematical hashing, making each voucher transmission traceable to its upstream source. When an anomaly is detected in the credential path, the S3 phase initiates the time consensus detection module, automatically constructing a dynamic time consensus window △T to compare the credential generation time T1, broadcast time T2, and block confirmation time T3. This design is because in a blockchain network environment, the latency between different nodes is not fixed. Without considering the time dimension, confirmation delays caused by network fluctuations can easily be misjudged as forgery. By calculating the dynamic conflict consistency coefficient Cconf, the system can quantitatively reflect the normality of credential time behavior. For example, when a node submits the same credential number multiple times within a short period, the deviation between its generation time and block confirmation time increases significantly, and the dynamic conflict consistency coefficient Cconf value decreases accordingly, thus accurately identifying abnormal concurrency or repeated on-chain behavior. This mechanism is equivalent to adding a dynamic layer of protection to structural verification in the time dimension, ensuring that even if forgery bypasses structural matching, it will still be identified in the time feature detection. Finally, in the S4 phase, the credential association consistency coefficient Cassoc and the dynamic conflict consistency coefficient Cconf are comprehensively analyzed to output the comprehensive credibility assessment index Sfinal, achieving a quantitative determination of the credential chain's credibility. By setting tiered thresholds, automated risk stratification and response strategies are implemented. This design takes into account uncontrollable factors such as network jitter and node load differences in actual business operations, avoiding "one-vote veto" type misjudgments, enabling trusted credentials to be quickly confirmed, while risky credentials are intervened in a timely manner. In a physical sense, this mechanism achieves comprehensive constraints on credential behavior through two-dimensional indicators (structure + time), improving data credibility from static evidence storage to dynamic self-verification.In summary, this implementation process achieves fully automated anti-tampering and adaptive trust assessment of financial vouchers from generation to on-chain verification. It not only improves the anti-counterfeiting accuracy and traceability efficiency of voucher management, but also significantly enhances the system's autonomy and audit transparency, providing stable and quantifiable technical support for the digital supervision of corporate finance.

[0069] Example 2

[0070] Please see Figure 1 Specifically: S1 includes S11;

[0071] S11. Set up voucher collection points around the financial voucher generation node, and set up collection tools in the collection points to collect voucher data in real time; after collection, encapsulate the voucher data into a data block structure and transmit it to the blockchain system through the secure transmission protocol TLS1.3; after receiving the data block structure of the voucher data, the blockchain system stores it in the distributed node database for subsequent correlation tracking calculation units to use for matching calculation, thereby realizing the real-time on-chain and structured transmission of voucher information;

[0072] The data collection points include voucher recognition nodes, time nodes, signature nodes, and amount nodes;

[0073] Voucher data includes voucher number Vid, upstream voucher number Pid, ​​voucher amount Am, and generation time Ts;

[0074] Among them: the voucher recognition node automatically reads the unique voucher number Vid and the upstream voucher number Pid for each newly generated financial voucher, and binds the voucher number with related contract, invoice or report information through optical character recognition (OCR);

[0075] The time node synchronizes with the blockchain time node through the Network Time Protocol (NTP) to generate a standardized timestamp, and writes the timestamp in hexadecimal sequence form into the header of the credential identifier to obtain the generation time Ts;

[0076] The signing node generates a digital signature NodeSig based on the node's public and private key system, which binds and verifies the identity of the credential generation node.

[0077] The amount node extracts the amount field from the voucher data structure and performs a unified unit conversion for amounts in different currencies, outputting the voucher amount Am.

[0078] S1 also includes S12;

[0079] S12. Decapsulate the data block structure of the encapsulated voucher data to extract the voucher data, and preprocess the voucher data to obtain a standardized voucher dataset; the preprocessing includes verification processing, data cleaning and dimensionless processing.

[0080] The verification process verifies the legitimacy of the node identity by verifying the credential generated based on the node signature NodeSig, and detects duplicates and conflicts between the credential number Vid and the upstream credential number Pid. When a duplicate credential number is detected, the credential correction logic is invoked to regenerate the credential with the duplicate number, forming a unique credential number Vid. When a conflict is detected, i.e., two different upstream numbers Pid point to the same Vid, the conflicting credential data is automatically frozen, and an anomaly report is triggered to the audit system via a smart contract, prompting manual correction and re-uploading to the chain. After verification, a qualified credential set is output, and duplicate and conflicting credential records are removed.

[0081] Data cleaning involves standardizing the voucher amount Am, including currency conversion, tax rate correction, and decimal precision adjustment, to form a standard amount field AmStd; and performing default padding for invalid fields and incomplete voucher information to ensure data integrity.

[0082] Dimensionless processing eliminates the differences in unit dimensions between voucher data by using the range normalization method to make the voucher data dimensionless after data cleaning and verification.

[0083] After dimensionless processing, all data are aggregated to generate a standardized voucher dataset.

[0084] In this embodiment, step S1 of the method deploys multi-point data collection at the source of the voucher, including voucher identification nodes, time nodes, signature nodes, and amount nodes. Voucher data is captured in real-time via OCR / interface and encrypted and packaged for blockchain uploading using TLS 1.3. Subsequently, the received data blocks are desealed, node signatures are verified, duplicate / conflict detection is performed, amount and format are cleaned, and range normalization is applied. Finally, a standardized dimensionless voucher dataset is output for subsequent comparison. The direct purpose of this is to capture and fix the "original evidence of the event" at the moment of generation: the time node (NTP synchronization and writing of a hexadecimal timestamp) physically ensures the immutability of the time sequence; the signature node (NodeSig) cryptographically binds the voucher source; the voucher identification node ensures the traceability of the number and upstream references; and the amount node and dimensionless processing eliminate dimensional differences caused by currency, tax rate, and unit, ensuring the comparability of subsequent hash comparisons and statistical analyses. For example, in cross-border subcontracting scenarios, discrepancies in the numerical values ​​of the same invoice due to differences in currency or decimal places are masked by dimensionless encoding, thus avoiding false alarms. If a node submits duplicate entries or two upstream nodes point to the same voucher number Vid, the verification phase immediately triggers the regeneration or freezing of the number and reports it via a smart contract, physically blocking the path of subsequent tampering and duplicate accounting. In summary, the implementation of S1 achieves simultaneous fixation and standardization of the four dimensions of "time, identity, content, and value" at the source, providing a high-quality, comparable, and unforgeable data foundation for subsequent verification of related paths and detection of temporal conflicts. This significantly reduces the costs of forgery, duplicate on-chain entries, and manual review, and improves the accuracy and efficiency of voucher tamper-proofing.

[0085] Example 3

[0086] Please see Figure 1 Specifically: S2 includes S21;

[0087] S21. Based on the standardized voucher dataset, an improved structure of the Merkle path matching algorithm is used to perform voucher association consistency analysis and output the voucher association consistency coefficient Cassoc. The voucher association consistency analysis includes the path construction stage, the node hash calculation stage, and the path integration stage.

[0088] In the path construction phase, a hierarchical voucher association tree structure is established based on the voucher number Vid and the upstream voucher number Pid. The construction of the voucher association tree structure follows a tree topology, with the root node corresponding to the initial contract or fund disbursement voucher and the leaf nodes corresponding to the final reimbursement or settlement voucher. The formation of the voucher association tree structure ensures that there is a unique path between any two nodes, which facilitates subsequent path continuity matching.

[0089] The node hash calculation stage generates a path matching function value by performing hash comparison on each pair of adjacent voucher nodes;

[0090] The path integration stage outputs the credential association consistency coefficient Cassoc by counting the number of successfully matched nodes in the entire credential chain.

[0091] The document association consistency coefficient Cassoc is calculated and output using the following algorithm formula;

[0092] ;

[0093] In the formula, N represents the number of nodes in the credential chain, and Vid i Pid represents the document number of the i-th financial document in the blockchain system. i Vid represents the upstream document number of the i-th financial document in the blockchain system. i+1 Pid represents the document number of the (i+1)th financial document in the blockchain system. i+1 This represents the upstream document number of the (i+1)th financial document in the blockchain system. Hash represents the hash mapping function, and match represents the hash matching function. The document association consistency coefficient Cassoc output result belongs to [0,1]; match=1 indicates consistency, and match=0 indicates inconsistency.

[0094] The derivation of the formula: This formula is derived from the classic Merkle tree path verification algorithm. The core idea of ​​the Merkle path algorithm is to verify the legality and consistency of the position of a data block in the tree structure by comparing the hashes of data nodes layer by layer. It is used to verify whether a single node exists in the Merkle tree. Based on the classic Merkle path verification, this formula extends it to a multi-level continuous node path consistency calculation model. By defining the matching function match(), it realizes the comparison of the local hash structure of adjacent nodes on the certificate chain, and introduces the idea of ​​path integral to calculate the overall continuity of the entire certificate chain.

[0095] Dimensional consistency explanation: The standardized voucher dataset consists of dimensionless values, while the Match function is a dimensionless Boolean function whose output only represents the path continuity state and does not carry physical dimensions.

[0096] S2 also includes S22;

[0097] S21. Based on the output of the document association consistency coefficient Cassoc and the preset consistency threshold Th1, a preliminary judgment is made to determine the continuity reliability of the document association path; the specific judgment content is as follows:

[0098] When the consistency coefficient Cassoc of the credential association is greater than or equal to the consistency threshold Th1, the credential association path is determined to be continuous and trustworthy, and the current credential chain is archived as a trustworthy chain.

[0099] When the consistency coefficient Cassoc of the credential association is less than the consistency threshold Th1, it is judged as an abnormality in continuous credibility, indicating that there is a broken chain or forgery in the path, and the time consensus detection is automatically triggered.

[0100] Among them, when the success rate of matching adjacent nodes in the credential chain reaches 90% or more, it is considered trustworthy. Therefore, the consistency threshold Th1 is directly set to 0.9.

[0101] In this embodiment, step S2 of the method organizes the standardized credential dataset into a hierarchical credential association tree and performs hash comparison and path integration on adjacent nodes to form a credential association consistency coefficient Cassoc, thereby quantifying the integrity of the credential chain at the structural level. The real-time purpose of this is to use both the "credential content fingerprint (an irreversible digest obtained through hashing)" and the "parent-child reference relationship (Vid-Pid)" as verification criteria: when the hashes of adjacent nodes do not match, Match=0 indicates that the node has been tampered with or replaced in terms of content or source; when most nodes fail to match, the credential association consistency coefficient Cassoc decreases and triggers subsequent time dimension detection, avoiding misjudging normal delays or format differences based solely on a single point hash. Physically, hash comparison is equivalent to "fingerprint comparison" of the voucher content, while path integration is like "integrity scoring" of the entire transmission chain. The setting of Th1=0.9 (i.e., 90% successful matching is considered trustworthy) is an engineered trade-off between fault tolerance and security: it allows for a small number of non-fraudulent differences caused by formatting / rounding, while also providing rapid alerts for chain breaks or replacements with more than 10% of abnormal nodes. For example, in a multi-level subcontracting scenario, if a subcontractor maliciously replaces an invoice or duplicates it on the chain, it will cause a hash mismatch in its corresponding node, thereby reducing the voucher association consistency coefficient Cassoc and automatically entering further time consensus and conflict analysis; conversely, normal single-point differences caused by currency decimals or OCR misidentification usually will not cause the voucher association consistency coefficient Cassoc to fall below 0.9. This enables real-time structured verification of the voucher source and transmission path, allowing for early identification and isolation of forgery / replacement risks, significantly reducing manual traceability costs, improving audit efficiency, and enhancing the tamper-proof reliability of the entire voucher chain.

[0102] Example 4

[0103] Please see Figure 1 and Figure 3 Specifically: S3 includes S31;

[0104] S31. After initially determining that the time consensus detection has been triggered, a dynamic time consensus window △T is set in the blockchain system. The dynamic time consensus window △T is dynamically and adaptively generated based on the sum of the average consensus delay of the historical blockchain system and twice the standard deviation of the delay. This setting method can ensure that the time window covers the normal block confirmation range and eliminate the impact of occasional communication jitter.

[0105] Within the dynamic time consensus window △T, the blockchain system synchronously obtains time parameters from two time collection points;

[0106] The time collection points include the first time collection point and the second time collection point;

[0107] The time parameters include the voucher generation time T1 and the block confirmation time T3;

[0108] The first data collection point is set at the voucher generation node, and the voucher generation time T1 is recorded through the time synchronization unit.

[0109] The second time point is set up on the consensus node to collect the block confirmation time T3 after the block is officially packaged.

[0110] S3 also includes S32;

[0111] S31. Perform dynamic conflict consistency analysis based on time parameters and output the dynamic conflict consistency coefficient Cconf. The dynamic conflict consistency coefficient Cconf is calculated and output using the following algorithm formula.

[0112] ;

[0113] In the formula, M represents the number of detected credentials within the dynamic time consensus window ΔT, uT represents the average consensus latency, pT represents the latency standard deviation, and T3 represents the latency standard deviation. i T1 represents the block confirmation time of the i-th financial document in the blockchain system. i This indicates the time when the i-th financial document was generated in the blockchain system;

[0114] Derivation of the formula: This formula is an improved form of the time delay difference model. Its original idea originates from the standardized deviation formula in mathematical statistics, and it also draws on the principle of distributed delay models in computer network delay detection. The classic delay difference model is used to measure the deviation of each event's delay from the average delay in a system. However, in the blockchain environment, conflict detection in this formula needs to consider not only the average delay deviation but also the stability distribution characteristics of the system. Therefore, the delay standard deviation pT is introduced for normalization. Furthermore, to ensure the comparability of the calculation results with an upper limit of 1 (i.e., no conflict) and a lower limit of 0 (severe conflict), a "1" is introduced. The inverse indicator form of "standardized delay bias";

[0115] Dimensional consistency analysis: T3 and T1 are time units, and the overall fractional ratio is dimensionless. Therefore, the dynamic conflict consistency coefficient Cconf output is dimensionless.

[0116] In this embodiment, step S3 of the method introduces a dynamic time consensus detection mechanism after the abnormal certificate chain is identified, enabling the system to re-verify the authenticity of the certificate from a time dimension. Specifically, the dynamic time consensus window △T is not a fixed value, but is adaptively generated based on the average consensus latency of the blockchain system and twice the standard deviation. The reason for this design is that the consensus confirmation speed of the actual blockchain network is affected by factors such as node load, network congestion, and block size. If a fixed time threshold is used, normal delays are easily misjudged as forgery or potential concurrent conflicts are ignored. By introducing a dynamic window △T, the normal confirmation range of the system can be effectively covered, while eliminating occasional jitter and ensuring a balance between the stability and sensitivity of the detection. Within the dynamic window △T, three key time-series parameters—certificate generation time T1, broadcast time T2, and block confirmation time T3—are collected synchronously. Their physical meaning is to reconstruct the time behavior curve of the certificate from "generation to broadcast to on-chain". For example, if a node attempts to tamper with the certificate content afterward and re-upload it to the chain, T1 will lag significantly behind the time distribution of similar historical certificates, resulting in a confirmation delay (T3). T1) is much larger than the average consensus latency uT. At this point, by calculating the dynamic conflict consistency coefficient Cconf, the system can quantitatively reflect the degree of anomaly in time behavior. When the dynamic conflict consistency coefficient Cconf approaches 0, it indicates the presence of obvious forgery or replay behavior; when Cconf approaches 1, it represents the stable consistency of the credential within the consensus time. The essence of this implementation is to introduce "time" into anti-tampering, supplementing the static shortcomings of relying solely on hash structure verification. Through the normalization of the latency standard deviation pT, the dynamic conflict consistency coefficient Cconf achieves automatic suppression and quantification of abnormal latency, enabling the system to identify forgery and concurrent conflict behavior at millisecond-level time differences. The beneficial effects of this mechanism are: on the one hand, it significantly improves the detection accuracy of the blockchain financial credential system for covert behaviors such as "repeated on-chain," "delayed replacement," and "forged synchronization"; on the other hand, it avoids the risk of misjudgment caused by fixed thresholds, making the anti-tampering mechanism adaptive and environmentally robust, thereby achieving dynamic and reliable evaluation of credential behavior and ensuring time consistency.

[0117] Example 5

[0118] Please see Figure 1 Specifically: S4 includes S41;

[0119] S41. Combine the dynamic conflict consistency analysis and the voucher association consistency analysis to perform a comprehensive analysis on the output dynamic conflict consistency coefficient Cconf and voucher association consistency coefficient Cassoc. The comprehensive analysis is then used to calculate and output the comprehensive credibility evaluation index Sfinal by the average of the dynamic conflict consistency coefficient Cconf and the voucher association consistency coefficient Cassoc.

[0120] S4 also includes S42;

[0121] S42. Based on the comprehensive credibility assessment index Sfinal output by the comprehensive analysis, a comprehensive credibility assessment is conducted, and corresponding actions are taken based on the comprehensive credibility assessment results; the specific assessment content is as follows:

[0122] When the comprehensive credibility assessment index Sfinal≥0.9, the certificate genealogy chain is determined to be completely credible and on-chain confirmation is allowed;

[0123] When 0.7 ≤ comprehensive credibility assessment index Sfinal < 0.9, the credential chain is judged to have a moderate risk and is transferred to the manual review buffer for secondary verification;

[0124] When the comprehensive credibility assessment index Sfinal < 0.7, it is determined to be an abnormal credential chain, triggering the freezing and alarm mechanism.

[0125] In this embodiment, step S4 of the method merges the structural dimension's credential association consistency Cassoc and the temporal dimension's dynamic conflict consistency Cconf into a single comprehensive credibility index Sfinal = (Cassoc + Cconf) / 2 to drive the automated processing flow. The purpose of this implementation is to make executable judgments using dual evidence of "structure + time": credentials with high structural consistency but temporal anomalies (e.g., replayed afterward or concurrently submitted repeatedly) will not be misjudged as credible, and conversely, credentials with normal time but structural breaks will not be simply allowed. Physically, the credential association consistency coefficient Cassoc is the integrity score of the credential's "content fingerprint and reference chain," and the credential association consistency coefficient Cassoc is the stability score of the credential's "generation, broadcasting, and confirmation" temporal behavior. The average of these two values ​​directly reflects the true credibility of the credential in both business flow and network consensus. The threshold grading is an engineering compromise: 0.9 ensures high-confidence decisions for automatic confirmation, while 0.7 serves as a fault tolerance lower limit to avoid large-scale false freezes caused by network jitter or minor identification errors. For example, when a subcontractor repeatedly submits the same credential within a short period, the dynamic conflict consistency (Cconf) will significantly decrease, causing the overall credibility assessment index (Sfinal) to fall into the freeze zone, triggering a smart contract freeze and accountability. Conversely, in cases where a single field mismatch is caused by OCR misidentification, the credential association consistency (Cassoc) may decrease slightly but typically will not cause the overall credibility assessment index (Sfinal) to fall below 0.7, thus avoiding unnecessary manual intervention. The overall beneficial effects are: automated and quantifiable risk stratification and handling, which improves the detection rate of forgery and concurrent attacks, significantly reduces false positives and the burden of manual review, and enhances the speed of credential on-chain processing, audit efficiency, and system autonomy.

[0126] Example 6

[0127] Please see Figure 1 and Figure 2 A blockchain-based anti-tampering system for financial vouchers includes a voucher acquisition module, a consistency analysis module, a dynamic conflict analysis module, and a comprehensive analysis module.

[0128] The voucher collection module collects voucher data in real time by setting up voucher collection points around the financial voucher generation node, and transmits the voucher data to the blockchain system. The voucher data is then preprocessed to obtain a standard voucher dataset.

[0129] The consistency analysis module performs document association consistency analysis based on a standard document dataset, and makes a preliminary determination of the continuity reliability of document association paths based on the results of the document association consistency analysis.

[0130] The dynamic conflict analysis module automatically triggers time consensus detection, extracts time feature vectors, and performs dynamic conflict consistency analysis when it initially determines that the continuity of the credential association path is abnormal.

[0131] The comprehensive analysis module integrates the results of dynamic conflict consistency analysis with the results of document association consistency analysis, performs a comprehensive credibility assessment based on the comprehensive analysis results, and takes corresponding actions based on the comprehensive credibility assessment results.

[0132] 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.

Claims

1. A method for preventing tampering with financial documents based on blockchain, characterized in that: Includes the following steps: S1. Set up voucher collection points around the financial voucher generation node to collect voucher data in real time and transmit the voucher data to the blockchain system. Then, preprocess the voucher data to obtain a standard voucher dataset. S2. Based on the standard voucher dataset, perform voucher association consistency analysis, and based on the results of the voucher association consistency analysis, make a preliminary determination of the continuity reliability of the voucher association path; S2 includes S21; S21. Based on the standardized voucher dataset, an improved structure of the Merkle path matching algorithm is used to perform voucher association consistency analysis, and the voucher association consistency coefficient Cassoc is output; the voucher association consistency analysis includes the path construction stage, the node hash calculation stage, and the path integration stage; The path construction phase establishes a hierarchical voucher association tree structure based on voucher number Vid and upstream voucher number Pid. The node hash calculation stage generates a path matching function value by performing hash comparison on each pair of adjacent credential nodes; The path integration stage outputs the credential association consistency coefficient Cassoc by counting the number of successfully matched nodes in the entire credential chain. The document association consistency coefficient Cassoc is calculated and output using the following algorithm formula; ； In the formula, N represents the number of nodes in the credential chain, and Vid i Pid represents the document number of the i-th financial document in the blockchain system. i Vid represents the upstream document number of the i-th financial document in the blockchain system. i+1 Pid represents the document number of the (i+1)th financial document in the blockchain system. i+1 This represents the upstream document number of the (i+1)th financial document in the blockchain system, where Hash represents the hash mapping function and match represents the hash matching function; the document association consistency coefficient Cassoc output result belongs to [0,1]. S3. When it is initially determined that the continuity of the credibility of the credential association path is abnormal, the time consensus detection is automatically triggered, the time feature vector is extracted, and dynamic conflict consistency analysis is performed. S3 includes S31; S31. After initially determining that the time consensus detection has been triggered, a dynamic time consensus window △T is set in the blockchain system. The dynamic time consensus window △T is dynamically and adaptively generated based on the sum of the average consensus delay of the historical blockchain system and twice the standard deviation of the delay. Within the dynamic time consensus window △T, the blockchain system synchronously obtains time parameters from two time collection points; The time acquisition points include a first time acquisition point and a second time acquisition point; The time parameters include the voucher generation time T1 and the block confirmation time T3; S3 further includes S32; S31. Perform dynamic conflict consistency analysis based on time parameters and output the dynamic conflict consistency coefficient Cconf; the dynamic conflict consistency coefficient Cconf is calculated and output using the following algorithm formula. ； In the formula, M represents the number of detected credentials within the dynamic time consensus window ΔT, uT represents the average consensus latency, pT represents the latency standard deviation, and T3 represents the latency standard deviation. i T1 represents the block confirmation time of the i-th financial document in the blockchain system. i This indicates the time when the i-th financial document was generated in the blockchain system; S4. Perform a comprehensive analysis of the dynamic conflict consistency analysis results and the voucher association consistency analysis results, and based on the comprehensive analysis results, conduct a comprehensive credibility assessment, and take corresponding actions based on the comprehensive credibility assessment results.

2. The method for preventing tampering with financial documents based on blockchain according to claim 1, characterized in that: S1 includes S11; S11. Set up voucher collection points around the financial voucher generation node, and set up collection tools in the collection points to collect voucher data in real time; after collection, encapsulate the voucher data into a data block structure and transmit it to the blockchain system through the secure transmission protocol TLS1.3; after receiving the data block structure of the voucher data, the blockchain system stores it in the distributed node database. The collection points include voucher recognition nodes, time nodes, signature nodes, and amount nodes; The voucher data includes voucher number Vid, upstream voucher number Pid, ​​voucher amount Am, and generation time Ts.

3. The method for preventing tampering with financial documents based on blockchain according to claim 2, characterized in that: S1 further includes S12; S12. Decapsulate the data block structure of the encapsulated voucher data to extract the voucher data, and preprocess the voucher data to obtain a standardized voucher dataset; the preprocessing includes verification processing, data cleaning, and dimensionless processing. The verification process verifies the legitimacy of the node identity by verifying the credential generated based on the node signature NodeSig, and detects duplicates and conflicts between the credential number Vid and the upstream credential number Pid. When a duplicate credential number is detected, the credential correction logic is invoked to regenerate the credential with the duplicate number, forming a unique credential number Vid. When a conflict is detected, the conflicting credential data is automatically frozen, and an anomaly report is triggered to the audit system through a smart contract, prompting manual correction and re-uploading to the chain. After verification, a qualified credential set is output, and duplicate and conflicting credential records are removed. The data cleaning process standardizes the voucher amount Am, including currency conversion, tax rate correction, and decimal precision adjustment, to form a standard amount field AmStd; and performs default completion operations on invalid fields and incomplete voucher information. The dimensionless processing involves using the range normalization method to make the voucher data after data cleaning and verification dimensionless, thereby eliminating the unit dimension differences between voucher data. After dimensionless processing, all data are aggregated to generate a standardized voucher dataset.

4. The method for preventing tampering with financial documents based on blockchain according to claim 1, characterized in that: S2 further includes S22; S21. Based on the output of the document association consistency coefficient Cassoc and the preset consistency threshold Th1, a preliminary judgment is made to determine the continuity reliability of the document association path; the specific judgment content is as follows: When the consistency coefficient Cassoc of the credential association is greater than or equal to the consistency threshold Th1, the credential association path is determined to be continuous and trustworthy, and the current credential chain is archived as a trustworthy chain. When the consistency coefficient Cassoc of the credential association is less than the consistency threshold Th1, it is determined to be an abnormality in continuous credibility, and time consensus detection is automatically triggered.

5. The method for preventing tampering with financial documents based on blockchain according to claim 1, characterized in that: S4 includes S41; S41. The dynamic conflict consistency analysis and the document association consistency analysis output dynamic conflict consistency coefficient Cconf and document association consistency coefficient Cassoc are comprehensively analyzed. The comprehensive analysis is performed by calculating the average of the dynamic conflict consistency coefficient Cconf and the document association consistency coefficient Cassoc and outputting the comprehensive credibility evaluation index Sfinal.

6. The method for preventing tampering with financial documents based on blockchain according to claim 5, characterized in that: S4 also includes S42; S42. Based on the comprehensive credibility assessment index Sfinal output by the comprehensive analysis, a comprehensive credibility assessment is conducted, and corresponding actions are taken based on the comprehensive credibility assessment results; the specific assessment content is as follows: When the comprehensive credibility assessment index Sfinal≥0.9, the certificate genealogy chain is determined to be completely credible and on-chain confirmation is allowed; When 0.7 ≤ comprehensive credibility assessment index Sfinal < 0.9, the credential chain is judged to have a moderate risk and is transferred to the manual review buffer for secondary verification; When the comprehensive credibility assessment index Sfinal < 0.7, it is determined to be an abnormal credential chain, triggering the freezing and alarm mechanism.

7. A blockchain-based anti-tampering system for financial documents, applied to the blockchain-based anti-tampering method for financial documents as described in any one of claims 1-6, characterized in that: It includes a voucher acquisition module, a consistency analysis module, a dynamic conflict analysis module, and a comprehensive analysis module; The voucher collection module collects voucher data in real time by setting up voucher collection points around the financial voucher generation node, transmits the voucher data to the blockchain system, and then preprocesses the voucher data to obtain a standard voucher dataset. The consistency analysis module performs document association consistency analysis based on a standard document dataset, and makes a preliminary determination of the continuity reliability of document association paths based on the results of the document association consistency analysis. The dynamic conflict analysis module automatically triggers time consensus detection, extracts time feature vectors, and performs dynamic conflict consistency analysis when it initially determines that the continuity of the credential association path is abnormal. The comprehensive analysis module integrates the dynamic conflict consistency analysis results with the voucher association consistency analysis results, performs a comprehensive credibility assessment based on the comprehensive analysis results, and executes corresponding actions based on the comprehensive credibility assessment results.

Images