Intelligent financial data management method and system

Abstract

The application provides a kind of financial data intelligent management method and system, the system includes multi-source heterogeneous data integration module, collects and standardizes fund flow data, bill information and operation index data;Intelligent analysis decision module, through time convolution network analysis data, generates fund dynamic prediction value, calculates risk coefficient and generates transaction analysis report;Risk visualization module, based on standardized data set, constructs fund correlation network, and generates interactive fund chain atlas, shows risk distribution;Audit tracking module, priority audit list is generated using path optimization algorithm, to improve audit efficiency;Data hub module, using blockchain technology generates a unique identification code for financial data, to ensure the security and traceability of data. The application provides an intelligent, real-time and secure solution for financial data management, effectively improving the transparency and decision support capability of financial management.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of financial data management, in particular to a financial data intelligent management method and system. BACKGROUND

[0002] Financial data management refers to the whole process of collecting, storing, processing, analyzing and reporting financial information of an enterprise or organization, involving multiple links, including financial data entry, cleaning, standardization, analysis and decision support, aiming to ensure the accuracy, integrity and timeliness of financial information. With the progress of information technology, financial data management systems gradually shift from traditional manual operation to automated and intelligent management mode, using big data, artificial intelligence, machine learning and other technologies to improve the efficiency and accuracy of financial data processing, and provide real-time capital flow monitoring and risk warning for enterprises.

[0003] Although the prior art has made certain progress in financial data management, there are still obvious deficiencies. First, the existing management system generally uses static rules to process invoices, bank statements and operation data, lacks dynamic cleaning and timestamp alignment mechanism, resulting in time sequence breakage problem in standardized data set. Second, the conventional audit method relies on fixed cycle full amount inspection, and does not establish a dynamic path evaluation model based on capital flow intensity and node correlation. In actual application, about 65% of audit resources are consumed in low-risk paths (risk coefficient <0.3), and the average identification delay of high-risk nodes (risk coefficient >2.5) is more than 36 hours. Finally, the existing blockchain solution only realizes single-point data storage, lacks cross-module identification code linking mechanism, and cannot form traceable verification chain between standardized data, risk analysis report and audit scheme. Statistics show that the operation history tracing failure rate of abnormal transactions is as high as 72%, which seriously restricts the credibility of audit conclusion. These problems make the existing financial data management system difficult to meet the needs of modern enterprises in the face of complex and dynamic financial environment. SUMMARY

[0004] In order to solve the technical problems mentioned in the background art, the purpose of the present application is to provide a financial data intelligent management method and system.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] A financial data intelligent management system, the system comprises:

[0007] M1, a multi-source heterogeneous data integration module, collects multi-source heterogeneous data and performs cleaning and standardization processing on the multi-source heterogeneous data, and outputs a standardized financial data set marked with a timestamp;

[0008] M2, the intelligent analysis and decision-making module, analyzes the capital flow patterns of the standardized financial dataset through a temporal convolutional network and generates dynamic capital forecast values; calculates the deviation rate between the dynamic capital forecast values ​​and the actual capital values ​​to obtain the capital risk coefficient; further determines whether the capital risk coefficient is abnormal through threshold comparison and outputs a transaction analysis and decision-making report;

[0009] M3, the risk visualization module, constructs a fund-related network topology model based on the standardized financial dataset, calculates the risk heat value through the fund-related network topology model, and finally generates an interactive fund chain map with the risk heat value through a 3D rendering engine.

[0010] M4, the audit tracking module, generates a priority audit list based on the risk heat value distribution in the aforementioned capital chain graph using a path optimization algorithm, and outputs a smart audit solution with time window constraints bound to the blockchain identifier code;

[0011] M5, the data hub module, establishes a blockchain distributed ledger to store data assets throughout the entire process, forming a traceable data relationship chain linked by blockchain identifier codes.

[0012] Furthermore, the multi-source heterogeneous data includes cash flow data, structured invoice information, and operational indicator data; the cash flow data is obtained through a bank transaction interface, the structured invoice information is obtained through a tax invoice parser, and the operational indicator data is obtained through an enterprise resource planning adapter.

[0013] Furthermore, the standardized financial dataset is modeled as a sequence of feature vectors in chronological order. X ,and ,in, Indicates the first Financial feature vectors at each point in time, T This represents the total time duration.

[0014] Will The financial feature vector at time point Input a temporal convolutional network to extract temporal dynamic features and calculate... Hidden state feature vector at time step The specific calculation formula is as follows:

[0015]

[0016] in, for The hidden state feature vector at time step; The input time series financial feature vector at time points The value at; The length of the convolution kernel; is the dilation coefficient of dilated convolution; For the first The convolution weight matrix corresponding to each time step; For bias terms; It is a non-linear activation function;

[0017] Based on the above Hidden state at time For the future The formula for predicting cash flow at any given time is as follows:

[0018]

[0019] in, Indicates from Time to Forecasted cash flow at any given time; For prediction functions; These are the parameters of the prediction function;

[0020] Will The predicted cash flow at time [time] and actual value of funds For comparison, the actual value of the funds was obtained through bank transaction records and calculated. deviation rate at time :

[0021]

[0022] Based on the above deviation rate at time Calculate the standard deviation accordingly :

[0023]

[0024] in, The sample size represents the total number of historical deviation rate data points.

[0025] Based on the above deviation rate at time and standard deviation calculate The financial risk coefficient at any time :

[0026]

[0027] Set threshold Standard deviation 2 times, that is ;like Exceeding the set threshold If so, it is determined to be an abnormal flow of funds; if Not exceeding the set threshold If so, the cash flow is considered normal;

[0028] Based on the aforementioned capital risk coefficient Different transaction analysis and decision reports are generated depending on whether an anomaly occurs. These reports are categorized into normal transaction analysis and decision reports and abnormal transaction analysis and decision reports.

[0029] Furthermore, the calculation steps for the risk heat value include calculating the amount of capital flow, calculating the frequency of capital flow, calculating the historical abnormal risk score, calculating the capital flow risk coefficient, and calculating the risk heat value.

[0030] The formulas for calculating the amount of funds flowing between the two nodes are as follows:

[0031]

[0032] in, For nodes To the node The amount of funds flowing between them; For the selected time interval; Indicates at a point in time τ Above, node To the node The amount of cash flow;

[0033] The formula for calculating the frequency of fund flows between two nodes is as follows:

[0034]

[0035] in, For nodes To the node The number of times funds flowed between them; For indicator functions;

[0036] The historical anomaly risk score is calculated based on the amount and frequency of fund flows. The calculation formula for the historical anomaly risk score of the two nodes is as follows:

[0037]

[0038] in, For nodes To the node Historical risk assessment scores between; The weighting factor for scoring the amount of cash flow is used to measure the proportion of the amount's influence in the overall risk score; The weighting factor for the frequency of fund flows is used to measure the proportion of the impact of frequency in the overall risk score; This represents the maximum total amount of funds flowing between all node pairs. This represents the maximum number of fund flows between all node pairs.

[0039] Furthermore, based on the aforementioned fund flow amount, fund flow frequency, and historical abnormal risk score, the edge is calculated. Capital flow risk coefficient That is, the cash flow risk on that side, calculated using the following formula:

[0040]

[0041] in, , and All of these are weighting factors, representing the degree of influence of the amount of cash flow, the frequency of cash flow, and the abnormal risk score on the cash flow risk coefficient, respectively.

[0042] Calculate the risk heat value for each node based on the aforementioned capital flow risk coefficient. To reflect the risk level of the node, the risk heat value is obtained by weighted summation of the risk coefficients of all adjacent edges of the node, and the calculation formula is as follows:

[0043]

[0044] in, For nodes The set of adjacent nodes, representing the set of nodes adjacent to each other. Other nodes with fund flows; The weight represents the intensity of capital flow;

[0045] The generated interactive capital chain graph is visualized based on the capital association network and the risk heat value. Each node and edge in the graph is displayed with different colors or thicknesses according to its corresponding risk heat value and risk coefficient. It also supports visualization and interaction functions for nodes and edges. The color of the node is used to represent its corresponding risk heat value, and the color or thickness of the edge reflects the risk intensity of capital flow. At the same time, users can view detailed capital flow data and risk coefficient information by clicking on nodes or edges, and can perform in-depth analysis of the overall network through zoom and rotation functions.

[0046] Furthermore, the steps for generating the priority audit list using the path optimization algorithm include node sorting, path evaluation, path selection and sorting, and generation of the priority audit list;

[0047] The node sorting is based on the risk heat value of the nodes in the capital chain diagram. The nodes are sorted from highest to lowest risk, with priority given to auditing nodes with higher risk heat values. The node sorting is represented as follows: ,in, Indicates the number is The node is denoted as the number node. Similarly, Indicates the number is The node is denoted as the number node. ; The total number of nodes that need to be audited;

[0048] The path assessment involves analyzing the fund flow paths between each high-risk node and other nodes. The risk value of the path is determined by the amount and frequency of fund flows between nodes. The specific assessment formula is as follows:

[0049]

[0050] in, Indicates numbered nodes To the numbered node Path risk value; Indicates numbered nodes To the intermediate numbered node The intensity of capital flow; intermediate numbered nodes To the numbered node The intensity of capital flow; This represents the total number of nodes involved in the path.

[0051] The path selection and sorting are based on the risk value of each path. Select the path with the highest risk value, sort the paths according to their total risk value, and prioritize auditing the nodes with the highest risk on the path.

[0052] The priority audit list is generated based on the path optimization algorithm, which includes the order of high-risk node investigation and related transaction tracing rules.

[0053] The high-risk node screening order is to sort the nodes from high to low according to their risk heat value, and to audit high-risk nodes first; the related transaction tracing rule is to trace the related transaction paths and fund flows of each high-risk node in order to discover potential abnormal transactions.

[0054] Furthermore, the intelligent auditing scheme includes a high-risk node screening order sorted by risk heat value and related transaction tracing rules, while also considering time window constraints to ensure that auditing activities are executed within a specific time period. Each auditing scheme is bound to a unique blockchain identifier code.

[0055] A method for intelligent management of financial data, comprising the following steps:

[0056] S1. Collect financial data from multiple different sources, including cash flow data, bill information and operational indicator data, and clean and standardize this data to output a standardized financial dataset with timestamps.

[0057] S2. Analyze the standardized financial dataset through a temporal convolutional network, extract the cash flow pattern and generate a dynamic cash flow forecast value, calculate the deviation rate between the dynamic cash flow forecast value and the actual cash flow value, obtain the cash risk coefficient, determine whether the risk coefficient is abnormal based on the threshold, and generate a transaction analysis decision report.

[0058] S3. Construct a fund-related network topology model based on the standardized financial dataset, calculate the risk heat value, and use a 3D rendering engine to generate an interactive fund chain graph to display the risk distribution in fund flows and support user interaction with the graph.

[0059] S4. Based on the risk heat value in the aforementioned capital chain graph, generate an audit list using a path optimization algorithm, prioritize auditing high-risk nodes and related paths, and output an intelligent audit solution with time window constraints.

[0060] S5. Utilize blockchain technology to generate unique identifiers for all financial data, store standardized financial datasets, transaction analysis and decision reports, interactive capital chain diagrams, and audit plans on the blockchain, forming a traceable data relationship chain to ensure the traceability of financial data throughout its entire lifecycle.

[0061] Compared with the prior art, the advantages of the present invention are as follows:

[0062] 1. This invention breaks through the structural barriers of traditional heterogeneous data sources, realizes dynamic adaptation and cross-domain semantic alignment of tax invoices, bank statements and operational indicators. Through high-precision timestamp calibration and spatiotemporal dimension consistency modeling, it constructs a standardized financial data base across the entire domain, eliminates feature drift and information entropy decay caused by data silos, and provides unbiased input for high-level analysis.

[0063] 2. Based on the dynamic feature extraction of the topology of the capital flow network, this invention establishes a node risk heat value assessment system and a path risk intensity quantification model. Through a multi-dimensional risk propagation algorithm, it realizes the dynamic optimization configuration of audit paths, which significantly improves the early identification capability and response efficiency of high-risk related nodes in complex capital networks.

[0064] 3. This invention integrates blockchain distributed ledger technology with cryptographic hash algorithms to construct a full lifecycle traceability network for cross-module data assets. Through a smart contract binding mechanism with unique identifier codes, it forms a verifiable architecture across the entire chain, from original transaction data and risk analysis maps to audit decisions, thus establishing a mathematically credible foundation for financial data governance. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a schematic diagram of the system workflow of the present invention;

[0067] Figure 2 This is a schematic diagram of the intelligent analysis and decision-making module of the present invention;

[0068] Figure 3 This is a schematic diagram of the risk thermal value calculation process of the present invention. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0070] To achieve the above objectives, the present invention provides an intelligent financial data management system, such as... Figures 1-3 As shown, the system includes:

[0071] M1, Multi-source heterogeneous data integration module, collects multi-source heterogeneous data and cleans and standardizes the multi-source heterogeneous data, outputs a standardized financial dataset marked with a timestamp, and the standardized financial dataset is attached with a blockchain identifier code;

[0072] The multi-source heterogeneous data includes cash flow data, structured invoice information, and operational indicator data; the cash flow data is obtained through a bank transaction interface, the structured invoice information is obtained through a tax invoice parser, and the operational indicator data is obtained through an enterprise resource planning adapter.

[0073] In this embodiment, the tax invoice parser combines optical character recognition technology to automatically identify and extract key fields from paper and electronic invoices, such as invoice number, amount, tax rate, supplier information, etc., and converts the information into structured data through the tax invoice parser.

[0074] In this embodiment, the enterprise resource planning adapter is used to automatically extract key operational data from the enterprise resource planning system of the enterprise, such as revenue, cost, gross profit margin, inventory status, order quantity, etc. The data comes from various departments and business modules of the enterprise. It connects to the enterprise resource planning system through the adapter's interface to obtain relevant data in real time. The extracted operational indicator data is organized into a structured format.

[0075] During the data cleaning process, data quality is improved by deleting duplicate records, filling in missing data, and correcting outliers. Standardization processing includes unifying the format of fields such as date, time, and amount, ensuring that all data items use consistent naming rules and units, and ensuring data type consistency. Through these steps, heterogeneous data from different data sources can be compatible, compared, and used in subsequent analysis.

[0076] The standardized financial dataset is attached with a blockchain identifier, which is a unique identity generated for each data record based on blockchain technology. It is a string of characters obtained by encrypting the data content using a hash algorithm, and has uniqueness and immutability. Specifically, after the standardized financial dataset is generated, the system will perform encrypted hashing on each data content (such as time, amount, and transaction method) to generate a unique identifier, and write the identifier along with the data into the blockchain. Once the record is on the chain, it cannot be modified or deleted.

[0077] M2, the intelligent analysis and decision-making module, analyzes the capital flow patterns of the standardized financial dataset through a temporal convolutional network and generates dynamic capital forecast values; calculates the deviation rate between the dynamic capital forecast values ​​and the actual capital values ​​to obtain the capital risk coefficient; further determines whether the capital risk coefficient is abnormal through threshold comparison and outputs a transaction analysis and decision-making report;

[0078] 1) The standardized financial dataset output from module M1 will be modeled into a feature vector sequence in chronological order. X ,and ,in, Indicates the first Financial feature vectors at each point in time, T This represents the total time duration.

[0079] 2) The financial feature vector at time point Input a temporal convolutional network to extract temporal dynamic features and calculate... Hidden state feature vector at time step The specific calculation formula is as follows:

[0080]

[0081] in, for The hidden state feature vector at time step; The input time series financial feature vector at time points The value at; The length of the convolution kernel; is the dilation coefficient of dilated convolution; For the first The convolution weight matrix corresponding to each time step; For bias terms; It is a non-linear activation function;

[0082] In this embodiment, the expansion coefficient The value is 2; the bias term The value is 0.1.

[0083] 3) Based on the above Hidden state at time For the future The formula for predicting cash flow at any given time is as follows:

[0084]

[0085] in, Indicates from Time to Forecasted cash flow at any given time; For prediction functions; These are the parameters of the prediction function;

[0086] 4) The predicted cash flow at time [time] and actual value of funds For comparison, the actual value of the funds was obtained through bank transaction records and calculated. deviation rate at time :

[0087]

[0088] Based on the above deviation rate at time Calculate the standard deviation accordingly :

[0089]

[0090] in, The sample size represents the total number of historical deviation rate data points.

[0091] 5) Based on the above deviation rate at time and standard deviation calculate The financial risk coefficient at any time :

[0092]

[0093] Set threshold Standard deviation 2 times, that is ;like Exceeding the set threshold If so, it is determined to be an abnormal flow of funds; if Not exceeding the set threshold If so, the cash flow is considered normal;

[0094] 6) Based on the aforementioned capital risk coefficient Different transaction analysis and decision reports are generated depending on whether an anomaly is detected.

[0095] When the aforementioned capital risk coefficient Under normal circumstances, a normal transaction analysis and decision report is generated at this time. The report includes: the predicted and actual values ​​of cash flow, the deviation rate, the risk coefficient, and the normal risk assessment (stable cash flow, in line with expectations, and low risk).

[0096] When the aforementioned capital risk coefficient In the event of an anomaly, an anomaly transaction analysis and decision report is generated. The report includes: the predicted and actual cash flow values, the deviation rate, the risk coefficient, and an anomaly risk warning (abnormal cash flow indicates significant risk, and it is recommended to take measures to adjust). This embodiment provides the following recommendations: ① Immediately review the relevant cash flow data to identify potential errors or anomalies; ② Strengthen monitoring of cash outflows to prevent potential financial risks; ③ Take cash adjustment measures based on the risk situation to ensure the health and stability of cash flow.

[0097] M3, the risk visualization module, constructs a fund-related network topology model based on the standardized financial dataset, calculates risk heat values ​​through the fund-related network topology model, and finally generates an interactive fund chain map with the risk heat values ​​through a 3D rendering engine.

[0098] The steps for calculating the risk heat value include calculating the amount of capital flow, calculating the frequency of capital flow, calculating the historical abnormal risk score, calculating the capital flow risk coefficient, and calculating the risk heat value.

[0099] 1) Calculate the amount of cash flow

[0100] The formulas for calculating the amount of funds flowing between the two nodes are as follows:

[0101]

[0102] in, For nodes To the node The amount of funds flowing between them; For the selected time interval; Indicates at a point in time τ Above, node To the node The amount of cash flow.

[0103] 2) Calculate the frequency of fund flows

[0104] The formula for calculating the frequency of fund flows between two nodes is as follows:

[0105]

[0106] in, For nodes To the node The number of times funds flowed between them; For indicator functions;

[0107] If at the point of time τ Above, node To the node There was indeed a flow of funds between them, then The value is 1; if there is no flow of funds, then The value is 0.

[0108] 3) Calculate historical anomaly risk scores

[0109] The historical anomaly risk score is calculated based on the amount and frequency of fund flows. The calculation formula for the historical anomaly risk score of the two nodes is as follows:

[0110]

[0111] in, For nodes To the node Historical risk assessment scores between; The weighting factor for scoring the amount of cash flow is used to measure the proportion of the amount's influence in the overall risk score; The weighting factor for the frequency of fund flows is used to measure the proportion of the impact of frequency in the overall risk score; This represents the maximum total amount of funds flowing between all node pairs. This represents the maximum number of fund flows between all node pairs.

[0112] In this embodiment, the weighting factor The value is 0.7; the weighting factor The value is 0.3.

[0113] 4) Calculate the cash flow risk coefficient

[0114] The edge is calculated based on the amount of fund flow, the frequency of fund flow, and the historical abnormal risk score. Capital flow risk coefficient That is, the cash flow risk on that side, calculated using the following formula:

[0115]

[0116] in, , and All of these are weighting factors, representing the degree of influence of the amount of cash flow, the frequency of cash flow, and the abnormal risk score on the cash flow risk coefficient, respectively.

[0117] In this embodiment, the weighting factor The value is 0.5; the weighting factor The value is 0.2; the weighting factor The value is 0.3.

[0118] 5) Calculate the risk heat value

[0119] Calculate the risk heat value for each node based on the aforementioned capital flow risk coefficient. To reflect the risk level of the node, the risk heat value is obtained by weighted summation of the risk coefficients of all adjacent edges of the node, and the calculation formula is as follows:

[0120]

[0121] in, For nodes The set of adjacent nodes, representing the set of nodes adjacent to each other. Other nodes with fund flows; The weight represents the intensity of capital flow;

[0122] In this embodiment, the weight The value is 0.5.

[0123] The generated interactive capital chain graph is visualized based on the capital association network and the risk heat value. Each node and edge in the graph is displayed with different colors or thicknesses according to its corresponding risk heat value and risk coefficient. It also supports visualization and interaction functions for nodes and edges. The color of the node is used to represent its corresponding risk heat value, and the color or thickness of the edge reflects the risk intensity of capital flow. At the same time, users can view detailed capital flow data and risk coefficient information by clicking on nodes or edges, and can perform in-depth analysis of the overall network through zoom and rotation functions.

[0124] This embodiment provides an example: the color of the node gradually changes from green (low risk) to red (high risk), and the thickness of the edge is adjusted according to the risk intensity;

[0125] Ultimately, the generated interactive cash flow graph is embedded into the visualization interface of the financial management system for use by decision-makers and auditors. Through this graph, users can intuitively identify potential high-risk cash flows and anomalies.

[0126] M4, the audit tracking module, generates a priority audit list based on the risk heat value distribution in the aforementioned capital chain graph using a path optimization algorithm, and outputs a smart audit solution with time window constraints bound to the blockchain identifier code;

[0127] The steps for generating a priority audit list using the path optimization algorithm include node sorting, path evaluation, path selection and sorting, and generation of the priority audit list.

[0128] 1) Node sorting

[0129] The node sorting is based on the risk heat value of the nodes in the capital chain diagram. The nodes are sorted from highest to lowest risk, with priority given to auditing nodes with higher risk heat values. The node sorting is represented as follows: ,in, Indicates the number is The node is denoted as the number node. Similarly, Indicates the number is The node is denoted as the number node. ; This represents the total number of nodes that need to be audited.

[0130] 2) Path evaluation

[0131] The path assessment involves analyzing the fund flow paths between each high-risk node and other nodes. The risk value of the path is determined by the amount and frequency of fund flows between nodes. The specific assessment formula is as follows:

[0132]

[0133] in, Indicates numbered nodes To the numbered node Path risk value; Indicates numbered nodes To the intermediate numbered node The intensity of capital flow; intermediate numbered nodes To the numbered node The intensity of capital flow; This represents the total number of nodes involved in the path.

[0134] 3) Path selection and sorting

[0135] The path selection and sorting are based on the risk value of each path. Select the path with the highest risk value, sort the paths according to their total risk value, and prioritize auditing the nodes with the highest risk on the path.

[0136] 4) Generation of the priority audit list

[0137] The priority audit list is generated based on the path optimization algorithm, which includes the order of high-risk node investigation and related transaction tracing rules.

[0138] The high-risk node screening order is to sort the nodes from high to low according to their risk heat value, and to audit high-risk nodes first; the related transaction tracing rule is to trace the related transaction paths and fund flows of each high-risk node in order to discover potential abnormal transactions.

[0139] The output intelligent audit solution includes a high-risk node investigation order sorted by risk heat value and related transaction tracing rules. It also considers time window constraints to ensure that audit activities are executed within a specific time. Each audit solution is bound to a unique blockchain identifier code to ensure that its data is tamper-proof and traceable, thereby enhancing the transparency and credibility of the audit process.

[0140] M5, the data hub module, establishes a blockchain distributed ledger to store data assets throughout the entire process, forming a traceable data relationship chain linked by blockchain identifier codes.

[0141] In this embodiment, the blockchain distributed ledger is used for data storage and management. The end-to-end data assets include the standardized financial dataset, transaction analysis and decision report, interactive capital chain diagram and audit plan. At the same time, a unique identity is generated for each piece of data using a blockchain identifier code, and different data elements are connected through the identifier code to form a traceable data relationship chain.

[0142] The full-process data assets support data traceability, accessibility, and security, as well as the establishment of data relationship chains;

[0143] The traceability of the data means that once all financial data and related documents (such as reports, graphs, etc.) are recorded in the blockchain, each data read can trace its source and modification history through the blockchain identifier code to ensure that the data can be verified at any time, prevent data tampering, and improve the transparency of the system.

[0144] The accessibility and security mentioned above are achieved by using blockchain technology to ensure the integrity and immutability of financial data. At the same time, through the permission management mechanism, it is ensured that different roles (such as auditors, decision-makers, etc.) can access the corresponding data according to authorization. The decentralized nature of blockchain avoids the dependence on a single point of control, making data storage and management more decentralized, thereby improving the system's security and anti-attack capabilities. This structure not only ensures the transparency and traceability of data, but also prevents data tampering or loss, ensuring the security and reliability of financial data.

[0145] The establishment of the data relationship chain is achieved by linking different data elements (such as financial data, analysis reports, and capital chain diagrams) through a data relationship chain using blockchain identifiers. The establishment and maintenance of this relationship chain ensures close correlation between data, enabling the output results of each module to verify and support each other.

[0146] A method for intelligent management of financial data, comprising the following steps:

[0147] S1. Collect financial data from multiple different sources, including cash flow data, bill information and operational indicator data, and clean and standardize this data to output a standardized financial dataset with timestamps.

[0148] S2. Analyze the standardized financial dataset through a temporal convolutional network, extract the cash flow pattern and generate a dynamic cash flow forecast value, calculate the deviation rate between the dynamic cash flow forecast value and the actual cash flow value, obtain the cash risk coefficient, determine whether the risk coefficient is abnormal based on the threshold, and generate a transaction analysis decision report.

[0149] S3. Construct a fund-related network topology model based on the standardized financial dataset, calculate the risk heat value, and use a 3D rendering engine to generate an interactive fund chain graph to display the risk distribution in fund flows and support user interaction with the graph.

[0150] S4. Based on the risk heat value in the aforementioned capital chain graph, generate an audit list using a path optimization algorithm, prioritize auditing high-risk nodes and related paths, and output an intelligent audit solution with time window constraints.

[0151] S5. Utilize blockchain technology to generate unique identifiers for all financial data, store standardized financial datasets, transaction analysis and decision reports, interactive capital chain diagrams, and audit plans on the blockchain, forming a traceable data relationship chain to ensure the traceability of financial data throughout its entire lifecycle.

[0152] This invention provides an intelligent financial data management method and system, aiming to address the shortcomings of existing financial management systems in data integration, cash flow forecasting, risk analysis, and auditing. It standardizes and cleans data from different sources through a multi-source heterogeneous data integration module, ensuring the consistency and comparability of financial data. It utilizes temporal convolutional networks for intelligent analysis of cash flow patterns and identifies anomalies in cash flow through risk coefficients, generating timely transaction analysis reports. Through the calculation of cash flow correlation networks and risk heat values, combined with 3D rendering technology, it provides an intuitive tool for visualizing cash flow risks, enhancing the efficiency and accuracy of cash flow monitoring. In the audit tracking module, path optimization algorithms optimize the audit process, ensuring priority auditing of high-risk nodes and improving the intelligence and timeliness of audit decisions. Finally, it uses blockchain technology to achieve tamper-proof storage of financial data and constructs a traceable data relationship chain through blockchain identifiers, ensuring data security and integrity. Through these technical means, this invention significantly improves the intelligence, real-time performance, and security of financial data management systems, providing strong support for enterprise financial risk control and decision-making.

[0153] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0154] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A financial data intelligent management system, characterized in that, The system includes: M1, Multi-source heterogeneous data integration module, collects multi-source heterogeneous data and cleans and standardizes the multi-source heterogeneous data, and outputs a standardized financial dataset marked with timestamps; M2, the intelligent analysis and decision-making module, analyzes the capital flow patterns of the standardized financial dataset through a temporal convolutional network and generates dynamic capital forecast values; calculates the deviation rate between the dynamic capital forecast values ​​and the actual capital values ​​to obtain the capital risk coefficient; further determines whether the capital risk coefficient is abnormal through threshold comparison and outputs a transaction analysis and decision-making report; M3, the risk visualization module, constructs a fund-related network topology model based on the standardized financial dataset, calculates the risk heat value through the fund-related network topology model, and finally generates an interactive fund chain map with the risk heat value through a 3D rendering engine. M4, the audit tracking module, generates a priority audit list based on the risk heat value distribution in the aforementioned capital chain graph using a path optimization algorithm, and outputs a smart audit solution with time window constraints bound to the blockchain identifier code; M5, the data hub module, establishes a blockchain distributed ledger to store data assets throughout the entire process, forming a traceable data relationship chain linked by blockchain identification codes; The standardized financial dataset is modeled as a sequence of feature vectors in chronological order. , ,in, Indicates the first The feature vectors at each time point, where T is the total time length; Will The financial feature vector at time point Input a temporal convolutional network to extract temporal dynamic features and calculate... Hidden state at time The specific calculation formula is as follows: in, for The hidden state features at time twentieth moment contain The temporal dynamic characteristics at any given moment; This is the processing function for the temporal convolutional network; These are the parameters of the temporal convolutional network processing function; Based on the above Hidden state at time For the future The formula for predicting cash flow at any given time is as follows: in, Indicates from Time to Forecasted cash flow at any given time; For prediction functions; These are the parameters of the prediction function; Will The predicted cash flow at time [time] and actual value of funds For comparison, the actual value of the funds was obtained through bank transaction records and calculated. deviation rate at time : Based on the above deviation rate at time Calculate the standard deviation accordingly : in, The sample size represents the total number of historical deviation rate data points. Based on the above deviation rate at time and standard deviation calculate The financial risk coefficient at any time : Set threshold Standard deviation 2 times, that is ;like Exceeding the set threshold If so, it is determined to be an abnormal flow of funds; if Not exceeding the set threshold If so, the cash flow is considered normal; Based on the aforementioned capital risk coefficient Different transaction analysis and decision reports are generated depending on whether an anomaly occurs. These reports are categorized into normal transaction analysis and decision reports and abnormal transaction analysis and decision reports. The steps for generating a priority audit list using the path optimization algorithm include node sorting, path evaluation, path selection and sorting, and generation of the priority audit list.

2. The intelligent financial data management system according to claim 1, characterized in that, The multi-source heterogeneous data includes cash flow data, structured invoice information, and operational indicator data; the cash flow data is obtained through a bank transaction interface, the structured invoice information is obtained through a tax invoice parser, and the operational indicator data is obtained through an enterprise resource planning adapter.

3. The intelligent management system for financial data according to claim 1, characterized in that, The steps for calculating the risk heat value include calculating the amount of capital flow, calculating the frequency of capital flow, calculating the historical abnormal risk score, calculating the capital flow risk coefficient, and calculating the risk heat value. The formula for calculating the amount of funds flowing between two nodes in a fund-related network topology model is as follows: in, For nodes To the node The amount of funds flowing between them; For the selected time interval; This indicates that at time point τ, the node To the node The amount of cash flow; The formula for calculating the frequency of fund flows between two nodes in a fund-related network topology model is as follows: in, For nodes To the node The number of times funds flowed between them; For indicator functions; The historical anomaly risk score is calculated based on the amount and frequency of fund flows. The calculation formula for the historical anomaly risk score of the two nodes is as follows: in, For nodes To the node Historical risk assessment scores between; This is a function for calculating the anomaly risk score.

4. The intelligent management system for financial data according to claim 3, characterized in that, The edge is calculated based on the amount of fund flow, the frequency of fund flow, and the historical abnormal risk score. Capital flow risk coefficient That is, the risk of capital flow on that side, calculated using the following formula: in, , and All of these are weighting factors, representing the degree of influence of the amount of cash flow, the frequency of cash flow, and the abnormal risk score on the cash flow risk coefficient, respectively. Calculate the risk heat value for each node based on the aforementioned capital flow risk coefficient. To reflect the risk level of the node, the risk heat value is obtained by weighted summation of the risk coefficients of all adjacent edges of the node, and the calculation formula is as follows: in, For nodes The set of adjacent nodes, representing the set of nodes connected to the node. Other nodes with fund flows; The weight represents the intensity of capital flow; The generated interactive capital chain graph is visualized based on the capital association network and the risk heat value. Each node and edge in the graph is displayed with different colors or thicknesses according to its corresponding risk heat value and risk coefficient. It also supports visualization and interactive functions for nodes and edges. In terms of visualization, the color of a node represents the risk heat value of that node, and the thickness or color of an edge represents the risk intensity of capital flow. The interactive function allows users to click on a node or edge to view detailed capital flow data and risk coefficient information, and perform zoom and rotation operations to conduct in-depth analysis of the risk of capital flow.

5. The intelligent management system for financial data according to claim 4, characterized in that, The node sorting is based on the risk heat value of the nodes in the capital chain diagram. The nodes are sorted from highest to lowest risk, with priority given to auditing nodes with higher risk heat values. The node sorting is represented as follows: ,in, Representing the 1 node The total number of nodes that need to be audited; The path assessment involves analyzing the fund flow paths between each high-risk node and other nodes. The risk value of the path is determined by the amount and frequency of fund flows between nodes. The specific assessment formula is as follows: in: Indicates from node arrive Path risk value; and These are nodes To the node and nodes To the node The intensity of capital flow between them; This represents the number of nodes involved in the path. The path selection and sorting are based on the risk value of each path. Select the path with the highest risk value, sort the paths according to their total risk value, and prioritize auditing the nodes with the highest risk on the path. The priority audit list is generated based on the path optimization algorithm, which includes the order of high-risk node investigation and related transaction tracing rules. The high-risk node screening order is to sort the nodes from high to low according to their risk heat value, and to audit high-risk nodes first; the related transaction tracing rule is to trace the related transaction paths and fund flows of each high-risk node in order to discover potential abnormal transactions.

6. The intelligent management system for financial data according to claim 1, characterized in that, The intelligent auditing scheme includes a high-risk node screening order sorted by risk heat value and rules for tracing related transactions. It also considers time window constraints to ensure that auditing activities are executed within a specific time. Each auditing scheme is bound to a unique blockchain identifier.

7. A method for intelligent management of financial data, wherein the method is applied to the intelligent management system for financial data as described in claim 1, characterized in that, The method includes the following steps: S1. Collect financial data from multiple different sources, including cash flow data, bill information and operational indicator data, and clean and standardize this data to output a standardized financial dataset with timestamps. S2. Analyze the standardized financial dataset through a temporal convolutional network, extract the cash flow pattern and generate a dynamic cash flow forecast value, calculate the deviation rate between the dynamic cash flow forecast value and the actual cash flow value, obtain the cash risk coefficient, determine whether the risk coefficient is abnormal based on the threshold, and generate a transaction analysis decision report. S3. Construct a fund-related network topology model based on the standardized financial dataset, calculate the risk heat value, and use a 3D rendering engine to generate an interactive fund chain graph to display the risk distribution in fund flows and support user interaction with the graph. S4. Based on the risk heat value in the aforementioned capital chain graph, generate an audit list using a path optimization algorithm, prioritize auditing high-risk nodes and related paths, and output an intelligent audit solution with time window constraints. S5. Utilize blockchain technology to generate unique identifiers for all financial data, store standardized financial datasets, transaction analysis and decision reports, interactive capital chain diagrams, and audit plans on the blockchain, forming a traceable data relationship chain to ensure the traceability of financial data throughout its entire lifecycle.

Images