Intelligent financial management method and system based on multi-source data fusion and medium

Abstract

The invention discloses an intelligent financial management method and system based on multi-source data fusion and a medium, and relates to the technical field of financial management, and the method comprises the steps: carrying out the multi-source financial association data collection of a target enterprise through employing an ETL technology; performing data mapping and conversion on the multi-source financial associated data, and constructing an enterprise fund chain; performing fund chain fracture risk analysis on the enterprise fund chain to generate a fracture risk index; if the fracture risk index is greater than a preset risk index, generating a fund chain risk signal, performing fund chain health recovery analysis, and generating a fund chain recovery decision; and sending the fund chain risk signal and the fund chain recovery decision to a financial management terminal of the target enterprise for financial abnormity reminding management. The technical problems of data islands and insufficient risk prediction capability in financial management in the prior art are solved, and the technical effects of improving financial risk prediction precision and optimizing fund chain management are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of financial management, and particularly relates to an intelligent financial management method, system and medium for multi-source data fusion. Background Art

[0002] With the rapid development of the global economy and the continuous expansion of enterprise scale, the importance of financial management in enterprise operation has become increasingly prominent. However, with the diversification of business fields and the complexity of management processes, traditional financial management methods are facing increasingly severe challenges. Existing financial management systems often have problems such as data islands, lagging information processing, and insufficient financial risk prediction capabilities, making it difficult to meet the needs of modern enterprises for efficient, accurate and real-time decision-making. Summary of the Invention

[0003] This application provides an intelligent financial management method, system and medium for multi-source data fusion, which is used to solve the technical problems of data islands and insufficient risk prediction capabilities existing in the prior art in financial management.

[0004] In view of the above problems, this application provides an intelligent financial management method, system and medium for multi-source data fusion.

[0005] In the first aspect of this application, an intelligent financial management method for multi-source data fusion is provided, and the method includes: Collect multi-source financial correlation data of the target enterprise by adopting ETL technology; perform data mapping and transformation on the multi-source financial correlation data to construct an enterprise fund chain, where the enterprise fund chain includes multiple fund flow nodes and the fund flow relationships between the nodes, specifically including: correct the differences in cross-system mapping relationships of the multi-source financial correlation data, and then call a preset data template for unified processing to generate standard multi-source financial correlation data; identify the fund flow relationship, fund flow status, and fund flow type for any financial data in the standard multi-source financial correlation data to generate a fund flow feature set; based on the fund flow feature set, construct a fund flow chain based on the fund flow relationship, and mark the fund flow status and fund flow type to generate the enterprise fund chain; perform fund chain break risk analysis on the enterprise fund chain to generate a break risk indicator, specifically including: calculate the fund flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate based on the enterprise fund chain to generate a multi-index calculation result; based on the enterprise scale and business type of the target enterprise, collect the historical fund chain set of historical fund chain break enterprises within a preset time range before the break; perform abnormal radar chart distribution analysis of the fund flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate based on the historical fund chain set to construct an abnormal radar chart distribution set; after converting the multi-index calculation result into a radar chart, compare it with the abnormal radar chart distribution set to generate the break risk indicator; if the break risk indicator is greater than the preset risk indicator, generate a fund chain risk signal, and perform fund chain health recovery analysis to generate a fund chain recovery decision; send the fund chain risk signal and the fund chain recovery decision to the financial management terminal of the target enterprise for financial anomaly reminder management.

[0006] In the second aspect of this application, an intelligent financial management system for multi-source data fusion is provided. The system includes: A data collection module for collecting multi-source financial correlation data of the target enterprise by adopting ETL technology; a fund chain construction module for performing data mapping and transformation on the multi-source financial correlation data to construct an enterprise fund chain, where the enterprise fund chain includes multiple fund flow nodes and the fund flow relationships between the nodes; a risk analysis module for performing fund chain break risk analysis on the enterprise fund chain to generate a break risk indicator; a risk signal generation module for generating a fund chain risk signal if the break risk indicator is greater than the preset risk indicator, and performing fund chain health recovery analysis to generate a fund chain recovery decision; an anomaly reminder module for sending the fund chain risk signal and the fund chain recovery decision to the financial management terminal of the target enterprise for financial anomaly reminder management.

[0007] In the third aspect of the present application, a computer-readable storage medium is provided, storing a computer program, which when executed by a processor, implements an intelligent financial management method for multi-source data fusion provided by the present application.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: In the present application, ETL technology is adopted to collect multi-source financial associated data of the target enterprise; data mapping and conversion are performed on the multi-source financial associated data to construct an enterprise fund chain, where the enterprise fund chain includes multiple fund flow nodes and the fund flow relationships between the nodes; risk analysis of the enterprise fund chain is carried out to generate a break risk index; if the break risk index is greater than a preset risk index, a fund chain risk signal is generated, and fund chain health recovery analysis is performed to generate a fund chain recovery decision; the fund chain risk signal and the fund chain recovery decision are sent to the financial management terminal of the target enterprise for financial anomaly reminder management. The present invention solves the technical problems of data islands and insufficient risk prediction ability in the existing technology in financial management. By adopting ETL technology for multi-source data collection, constructing an enterprise fund chain, carrying out risk analysis and implementing a fund chain recovery decision, the technical effects of improving the accuracy of financial risk prediction and optimizing fund chain management are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0010] Figure 1 It is a schematic flowchart of an intelligent financial management method for multi-source data fusion provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of an intelligent financial management system for multi-source data fusion provided by an embodiment of the present application.

[0011] Description of the reference numerals: data collection module 11, fund chain construction module 12, risk analysis module 13, risk signal generation module 14, anomaly reminder module 15. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0012] The present application provides an intelligent financial management method, system and medium for multi-source data fusion, aiming to solve the technical problems of data islands and insufficient risk prediction ability in existing financial management. By adopting ETL technology for multi-source data collection, constructing an enterprise's capital chain, conducting risk analysis and implementing capital chain restoration decisions, the technical effects of improving the accuracy of financial risk prediction and optimizing capital chain management are achieved.

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the scope of protection of the present application.

[0014] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0015] Embodiment 1, as Figure 1 shown, the present application provides an intelligent financial management method for multi-source data fusion, and the method includes: Step S100: Collect multi-source financial related data of the target enterprise by adopting ETL technology.

[0016] In the embodiment of the present application, when collecting multi-source financial related data of the target enterprise by adopting ETL technology, relevant financial data is first extracted from multiple different data sources. These data sources include multiple internal systems of the enterprise, such as accounting systems, ERP systems, financial statements and bank transaction records, etc. These internal data record the daily operations and capital flows of the enterprise, covering detailed information of various financial activities. To ensure the smooth progress of the extraction process, interact with each internal system through interfaces, APIs or direct database connections, etc., to obtain and extract the required financial data in real time.

[0017] In addition, important market, supply chain and customer data, etc. are obtained from external data sources. These external data sources include market data, such as stock prices, interest rates, exchange rates, etc., supply chain data, such as supplier information, raw material prices, logistics conditions, etc., and customer data, such as sales orders, customer payment behaviors, etc. In order to comprehensively reflect the impact of the external economic environment and market changes on the financial situation of the enterprise, obtain relevant external data in a timely manner through open APIs or third-party data services, etc.

[0018] Through this series of data extraction steps, ETL technology extracts the required financial correlation data from data with different sources, formats, and structures, completing the collection of multi-source financial correlation data.

[0019] Step S200: Perform data mapping and transformation on the multi-source financial correlation data to construct an enterprise fund chain, where the enterprise fund chain includes multiple fund flow nodes and the fund flow relationships between the nodes.

[0020] In the embodiment of the present application, when performing data mapping and transformation on multi-source financial correlation data to construct an enterprise fund chain, first perform differential correction on the cross-system mapping relationships of the extracted multi-source financial correlation data. Since this data comes from different systems and sources, there may be problems such as inconsistent formats and units. Therefore, call a preset data template to uniformly process this data to ensure that all data can be compared and analyzed according to a unified standard. After this step, standardized multi-source financial correlation data is generated.

[0021] Next, analyze the standardized data to identify the fund flow relationships, fund flow statuses, and fund flow types in each piece of financial data. The goal of this process is to extract key features from the financial data, clarify the direction of fund flow, the fund flow status (such as income, expenditure, etc.), and the fund flow type (such as transfer, payment, receipt, etc.), generating a fund flow feature set. Finally, based on these fund flow feature sets, further construct an enterprise fund chain. The enterprise fund chain includes multiple fund flow nodes, and each node represents a fund activity point inside or outside the enterprise, such as a company account, a supplier account, or a customer account, etc. The fund flow relationships between the nodes describe how funds flow between these nodes. For example, funds may flow from the enterprise's account to the supplier account, or from the customer account to the enterprise account. By identifying and marking these fund flow relationships, clearly display the path of enterprise fund flow, and finally generate a complete enterprise fund chain.

[0022] Furthermore, in the method provided by the embodiment of the application, performing data mapping and transformation on the multi-source financial correlation data to construct an enterprise fund chain further includes: Perform differential correction on the cross-system mapping relationships of the multi-source financial correlation data, then call a preset data template for unified processing to generate standard multi-source financial correlation data; identify the fund flow relationships, fund flow statuses, and fund flow types in any piece of financial data in the standard multi-source financial correlation data to generate a fund flow feature set; based on the fund flow feature set, construct a fund flow chain based on the fund flow relationships, and mark the fund flow statuses and fund flow types to generate the enterprise fund chain.

[0023] Further, the method provided by the application embodiment further includes: Performing differential correction on the cross-system mapping relationship of the multi-source financial associated data, including cross-system numerical conversion, time format conversion, currency conversion, and classification mapping.

[0024] In the embodiment of the present application, when performing differential correction on the cross-system mapping relationship of the multi-source financial associated data, first, data mapping technology is used to identify and adjust the data format and unit differences between different systems. Specifically, currency conversion is performed to unify the currency units in the external market data into the currency units used in the enterprise internal system, and date format conversion is performed to ensure that all date fields conform to a unified format. This process also includes numerical conversion, which converts data from different sources into a unified numerical unit. For example, the amount is converted into the same currency unit. In addition, classification mapping technology is used to ensure that data fields and categories from different systems can be uniformly mapped to standardized financial classifications. After this stage, finally, standardized multi-source financial associated data is obtained, which has a unified format, unit, and classification, facilitating subsequent processing and analysis.

[0025] After completing the data differential correction, a preset data template is called to uniformly process the data. These data templates convert data from different sources into a unified format that meets the enterprise's financial management requirements by defining standard field names, data types, and related calculation rules. The use of the template ensures the consistency and accuracy of all data fields. For example, the "Accounts Receivable" field is standardized to "Accounts Receivable", and all data types are correctly identified as "currency type" or "numerical type". Through this process, standard multi-source financial associated data is generated.

[0026] Next, for any one of the financial data in the standard multi-source financial associated data, the fund flow relationship, fund flow status, and fund flow type are identified. The fund flow relationship describes how funds flow between different accounts or nodes, and the fund flow path is determined by analyzing the "source account" and "target account" in the transaction record. For example, funds may flow from the enterprise's bank account to the supplier account, or from the customer account to the enterprise account. The fund flow status identifies the current status of the funds, such as "paid", "unpaid", or "overdue". For receivables, it may be marked as "received" or "unreceived". The fund flow type indicates the purpose of the fund flow, such as "purchase payment", "sales revenue", or "loan repayment", etc. Through these identifications, a fund flow feature set is created.

[0027] Finally, these collections of capital flow characteristics are used to construct a capital flow chain, clarifying the flow relationships of funds between different nodes. Each node represents a point of capital activity, such as an enterprise account, a supplier account, a customer account, etc. The flow relationships between nodes describe how funds are transferred between these nodes. Each node and capital flow relationship are marked to identify the flow status and type of funds. For example, funds may flow from the "customer account" to the "enterprise account" and be marked as the "received payment" status and the "sales revenue" type. In this way, the capital flow path of the enterprise is clearly shown, and a complete enterprise capital chain is formed.

[0028] Step S300: Analyze the risk of capital chain breakage for the enterprise capital chain and generate breakage risk indicators.

[0029] In the embodiment of the present application, when analyzing the risk of capital chain breakage for the enterprise capital chain, first, key financial indicators such as the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate of the enterprise capital chain are calculated to generate calculation results of multiple financial indicators. Then, according to the scale and business type of the enterprise, relevant financial data of historical capital chain breakage enterprises before breakage are collected, especially the historical capital chain set within a preset time range. Then, based on these historical data, an abnormal radar chart distribution analysis is performed on indicators such as the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate to construct an abnormal radar chart distribution set, which is used to represent the financial characteristics of historical enterprise capital chains before breakage. Finally, the calculation results of multiple financial indicators of the current enterprise are converted into a radar chart and compared with the abnormal radar chart distribution set of historical capital chain breakage enterprises to generate breakage risk indicators.

[0030] Furthermore, in the method provided by the embodiment of the application, when analyzing the risk of capital chain breakage for the enterprise capital chain and generating breakage risk indicators, it further includes: Calculating the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate based on the enterprise capital chain to generate multi-indicator calculation results; collecting the historical capital chain set of historical capital chain breakage enterprises within a preset time range before breakage based on the scale and business type of the target enterprise; performing an abnormal radar chart distribution analysis of the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate based on the historical capital chain set to construct an abnormal radar chart distribution set; after converting the multi-indicator calculation results into a radar chart, comparing them with the abnormal radar chart distribution set to generate the breakage risk indicators.

[0031] In the embodiments of the present application, first, based on the enterprise's capital chain, the calculation of the current ratio, quick ratio, cash ratio, accounts receivable turnover ratio, and accounts payable turnover ratio is carried out. Among them, the calculation of the current ratio is based on the current assets and current liabilities of the enterprise, and its calculation formula is "current ratio = current assets ÷ current liabilities". Current assets include cash, accounts receivable, inventory, etc., and current liabilities include short-term borrowings, accounts payable, etc. The quick ratio is a further refinement of the current ratio. It excludes the influence of inventory and emphasizes the enterprise's solvency without relying on inventory. The calculation formula is "quick ratio = (current assets - inventory) ÷ current liabilities". The cash ratio evaluates the enterprise's cash payment ability by comparing the enterprise's cash and cash equivalents with current liabilities, and the calculation formula is "cash ratio = cash and cash equivalents ÷ current liabilities". The accounts receivable turnover ratio reflects the efficiency of the enterprise in converting sales revenue into cash, and the calculation formula is "accounts receivable turnover ratio = sales revenue ÷ average accounts receivable". The accounts payable turnover ratio measures the efficiency of the enterprise in paying accounts payable, and the calculation formula is "accounts payable turnover ratio = purchase expenditure ÷ average accounts payable". By calculating the results of multiple financial indicators, a multi-indicator calculation result is formed.

[0032] After calculating the above financial indicators, a comparative analysis with historical capital chain data is carried out. First, based on the scale and business type of the target enterprise, the data of historical enterprises with broken capital chains similar to the target enterprise are screened out. The enterprise scale is usually measured by indicators such as the number of employees, annual sales, total assets, etc.; while the business type is defined through dimensions such as industry classification, product types, and market distribution. According to these characteristics, historical enterprises that are most similar to the target enterprise in terms of market environment, industry background, and scale are identified. In this way, the relevance and applicability of the historical enterprise data are ensured. After screening out historical enterprises similar to the target enterprise, according to the data of these enterprises within a preset time range before the break, a historical capital chain set is constructed. The preset time range is selected within 3 months, 6 months, or 12 months before the capital chain break, with the aim of capturing the financial change trend of the capital chain before the break. The data in the historical capital chain set includes financial indicators such as the current ratio, quick ratio, cash ratio, accounts receivable turnover ratio, and accounts payable turnover ratio.

[0033] By analyzing the historical fund chain set, perform an analysis of the abnormal radar chart distribution. This analysis converts the financial indicators of historical enterprises with broken fund chains into radar charts to show the relative relationships and abnormal trends of the key financial indicators of enterprises before the fund chain breaks. By comparing the differences in the radar charts of multiple historical enterprises, identify the financial abnormal patterns of historical enterprises with broken fund chains before the break. Then, use cluster analysis to aggregate similar radar charts together, remove isolated and unrepresentative radar charts, and finally construct an abnormal radar chart distribution set. This set shows the typical financial patterns and abnormal financial performances of historical enterprises before the fund chain breaks, providing a basis for subsequent risk assessment.

[0034] Next, convert the calculation results of the target enterprise into a radar chart and compare it with each radar chart in the abnormal radar chart distribution set. By calculating the similarity between the radar chart of the current enterprise and the radar charts of historical enterprises, identify whether the current enterprise has similar financial risks to historical enterprises with broken fund chains, generate a set of radar chart similarities, and find the most similar historical radar chart through comparative calculations. Based on the similarity set, extract the maximum similarity value as the break risk indicator.

[0035] Furthermore, in the method provided by the application embodiment, based on the historical fund chain set, perform an analysis of the abnormal radar chart distribution of the current ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate, and construct an abnormal radar chart distribution set, which further includes: Calculate the current ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate and convert them into radar charts for each fund chain in the historical fund chain set to generate a set of historical indicator radar charts; compare the differences between every two radar charts in the set of historical indicator radar charts, perform clustering according to the difference comparison results, and delete isolated points to obtain multiple clusters of historical indicator radar charts; select one radar chart from each of the multiple clusters of historical indicator radar charts to generate the abnormal radar chart distribution set.

[0036] In the embodiment of the present application, first, by the same method as described above, the calculation of the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate is performed for each capital chain in the historical capital chain set. Then, the calculated data is converted into a radar chart. Since the radar chart can display the relationships between different financial indicators in multiple dimensions, it maps the various financial indicators of an enterprise to different axes, thus forming a multi-dimensional visualization graph. During the process of radar chart conversion, a normalization method is used to standardize different financial indicators to ensure the balanced contribution of each financial data in the radar chart. For example, by normalizing each financial indicator to the interval [0, 1], data deviation caused by the scale differences of different indicators is avoided. The financial data of each historical enterprise is converted into an independent radar chart, and these radar charts form the historical indicator radar chart set.

[0037] Then, a difference comparison is performed between every two radar charts in the historical indicator radar chart set, and the similarity between every two radar charts is calculated. By using methods such as Euclidean distance or cosine similarity, the differences between the two radar charts are quantified, and historical enterprises with similar financial performances are identified. The results of the difference comparison can reveal which historical enterprises have similar financial characteristics before the capital chain breaks, providing data support for subsequent cluster analysis.

[0038] After completing the difference comparison, a cluster analysis method is used to divide these radar charts into multiple clusters. Cluster analysis is based on the similarity between the radar charts, grouping similar financial patterns into one group, usually using methods such as K-means clustering or hierarchical clustering. These clustering algorithms group historical enterprises with similar financial characteristics into the same category by comparing the similarities of different radar charts. By cluster analysis, enterprise clusters with similar financial conditions are identified, and these enterprise groups usually represent enterprises that exhibit similar risks under similar economic environments and financial conditions. During the clustering process, those "isolated points" with too large differences in financial patterns from other radar charts are deleted to ensure the accuracy and representativeness of the clustering results. Through this process, multiple clusters of historical indicator radar charts are obtained.

[0039] After completing the cluster analysis, a representative radar chart is randomly selected from each cluster. These representative radar charts represent the typical financial patterns and capital chain risk performances of the enterprises within each cluster. By selecting these representative radar charts, an abnormal radar chart distribution set is finally constructed.

[0040] Furthermore, in the method provided by the application embodiment, after the radar chart conversion of the multi-index calculation results and the comparison with the abnormal radar chart distribution set to generate the break risk index, it further includes: After the multi-index calculation result is converted into a radar chart, it is compared with each abnormal radar chart distribution in the abnormal radar chart distribution set to calculate a radar chart similarity set; based on the radar chart similarity set, the maximum similarity is extracted as the fracture risk indicator.

[0041] In the embodiment of the present application, the multi-indicator calculation results are first converted into a radar chart. The radar chart uses each financial indicator as an axis of the graph to show the distribution status between indicators. The value of each indicator determines the position on the corresponding axis, and the points of all axes are connected to form a polygon, showing the overall status of the company in different financial dimensions.

[0042] Next, the radar chart of the target enterprise is compared with each abnormal radar chart in the abnormal radar chart distribution set. The abnormal radar chart distribution set contains radar charts of multiple historical companies with broken capital chains. These radar charts represent the typical financial characteristics and abnormal patterns of the companies before the capital chain broke. By comparing with the radar charts of these historical companies, the similarity of the radar charts is calculated to evaluate the similarity between the financial characteristics of the current company and the historical companies with broken capital chains. When making the comparison, similarity measurement methods such as Euclidean distance or cosine similarity are used to measure the similarity between each two radar charts through these algorithms. A higher similarity indicates that the financial characteristics of the target company are similar to those of the historical companies and may face similar risks of capital chain break.

[0043] After comparison, a radar chart similarity set is calculated, which contains all similarity values ​​between the target enterprise and the historical abnormal radar chart. Finally, the maximum similarity value is extracted from this similarity set as a fracture risk indicator.

[0044] Step S400: If the break risk index is greater than the preset risk index, a capital chain risk signal is generated, and a capital chain health recovery analysis is performed to generate a capital chain recovery decision.

[0045] In an embodiment of the present application, the rupture risk index is compared with a preset risk index, wherein the preset risk index is pre-set by a technical expert. If the rupture risk index is greater than the preset risk index, a capital chain risk signal is generated, indicating that the enterprise faces a potential risk of capital chain rupture.

[0046] Next, perform an analysis on the recovery of the capital chain health and generate a decision on the capital chain recovery. Specifically, first, collect the set of outstanding payment bills and the set of entities to be collected in the enterprise's capital chain. These data provide the uncollected funds faced by the enterprise and the corresponding customer information. Then, collect the historical collection behaviors of the entities to be collected. Based on these data, construct a set of collection difficulty indicators to reflect the collection difficulty of each customer. Next, according to the collection difficulty indicators, arrange the set of entities to be collected in ascending order to generate a sequence of entities to be collected, providing a basis for subsequent collection optimization. After that, conduct digital modeling on the enterprise's capital chain to generate a twin capital chain, which is used to simulate the collection process and predict the possibility of the capital chain recovery. Based on the twin capital chain and the set of outstanding payment bills, start the collection success simulation from the first entity in the sequence of entities to be collected, and simulate the results of the collection one by one. If the simulation results show that the break risk indicator is still greater than the preset risk indicator, continue to simulate the next entity until a collection-successful entity is found, such that the break risk indicator after the simulation is lower than the preset risk indicator, thereby generating the recommended collection entity information.

[0047] In addition, mark the entities in the sequence of entities to be collected whose collection difficulty indicators are greater than the preset difficulty threshold as bad debt entities. Finally, combine the recommended collection entity information and the bad debt entity marks to generate a decision on the capital chain recovery.

[0048] Furthermore, in the method provided by the application embodiment, if the break risk indicator is greater than the preset risk indicator, generate a capital chain risk signal, perform an analysis on the recovery of the capital chain health, generate a decision on the capital chain recovery, and further include: Collect the set of outstanding payment bills and the set of entities to be collected in the enterprise's capital chain; collect the historical collection behaviors of the set of entities to be collected and construct a set of collection difficulty indicators; arrange the set of entities to be collected in ascending order based on the set of collection difficulty indicators to generate a sequence of entities to be collected; conduct digital modeling on the enterprise's capital chain to generate a twin capital chain; based on the twin capital chain and the set of outstanding payment bills, start the collection success simulation from the first entity in the sequence of entities to be collected. If the break risk indicator of the simulation is still greater than the preset risk indicator, then conduct the collection success simulation on the second entity until the break risk indicator of the simulation is still less than the preset risk indicator, generating the recommended collection entity information; mark the entities in the sequence of entities to be collected whose collection difficulty indicators are greater than the preset difficulty threshold as bad debt entities; generate the decision on the capital chain recovery with the recommended collection entity information and the bad debt entity marks.

[0049] In the embodiment of the present application, first, a collection of bills receivable and a collection of entities receivable are collected from the enterprise's financial management system. The collection of bills receivable contains all records of uncollected amounts, while the collection of entities receivable includes the list of customers or debtors behind these bills. These data are sourced from the enterprise's ERP system or financial system, and all outstanding bills and relevant customer information are automatically collected through database queries to ensure the accuracy and comprehensiveness of the data.

[0050] Next, historical collection behaviors are collected for the collection of entities receivable. The machine learning models used in this process, such as decision trees or random forests, are used to analyze historical collection data and generate a collection difficulty index for each entity receivable. Specifically, first, the historical payment behaviors of each customer are collected, including the collection cycle, whether there are late payments, default history, volatility of payment records, etc. The collection difficulty of each customer is determined by multiple factors, including the timeliness of payment, the length of the payment cycle, and whether there are records of overdue payments or defaults.

[0051] When constructing the machine learning model, historical data is first collected, including the collection history of each customer, as the training dataset. The features in this training set may include the historical payment behaviors of the customers (such as the average collection cycle, whether there are defaults, timeliness of payment, etc.), and the target variable is the collection difficulty score of the customer, which is a quantified value pre-labeled by technical experts and reflects the collection difficulty of the customer. To train the model, decision trees or random forests are used to automatically discover the key factors affecting the collection difficulty through historical data and predict the collection difficulty of new customers based on these factors. For example, if a customer's past payment behavior shows long-term late payments and irregular payment cycles, the model assigns a higher collection difficulty score (such as 80 points) to this customer. On the contrary, if a customer has made timely payments and has no defaults historically, a lower collection difficulty score (such as 30 points) is assigned to this customer.

[0052] Through this process, a collection difficulty index set is generated. This set is a specific set of numerical values, and each value represents the collection difficulty score of a customer. This set is used to sort the entities receivable, and a sequence of entities receivable is generated in ascending order of the collection difficulty index. This sorting helps to determine the collection priority. Customers with low collection difficulty are ranked first, which means that the collection of these customers may be smoother, and the enterprise can give priority to collecting payments from these customers to ensure the stability of the cash flow.

[0053] After sorting is completed, digital modeling is carried out to generate a twin fund chain. The twin fund chain is a virtual fund chain model used to simulate the actual fund flow of an enterprise and predict the stability of the fund chain under different collection strategies. The twin fund chain is constructed through a system dynamics model or a discrete event simulation model. In this model, the input data includes the financial data of the enterprise (such as existing cash flow, accounts receivable, accounts payable, etc.) and the set of bills to be collected. Through simulation, the state and potential risks of the fund chain are simulated under different collection orders, and then the impact of different collection strategies on the enterprise's fund chain is evaluated.

[0054] Based on the twin fund chain and the set of bills to be collected, the collection success simulation starts. Starting from the first entity in the sequence of entities to be collected, it is simulated whether the collection from this customer is successful. If the collection is successful and the break risk indicator is still greater than the preset risk threshold, the collection simulation for the second entity will continue. This process uses Monte Carlo simulation or Markov decision process (MDP). By simulating different collection paths and collection results multiple times, the possibility of the fund chain recovery is predicted. During the simulation process, the break risk indicator is calculated in real time and compared with the preset risk threshold. If the simulation results show that the collection is successful and the risk indicator drops below the threshold, the simulation will stop and the recommended collection entity information will be generated, that is, the list of customers recommended for priority collection.

[0055] Meanwhile, customers with a collection difficulty indicator greater than the preset difficulty threshold in the sequence of entities to be collected are marked as bad debt entities. This marking process is based on the threshold method. If the collection difficulty score of a certain customer exceeds the preset threshold (for example, the difficulty score is greater than 70 points), then this customer is considered a customer with a relatively high bad debt risk. After being marked as a bad debt entity, different collection strategies may be adopted, including reducing the credit limit for this customer, taking legal actions, or transferring it to a collection agency, etc.

[0056] Finally, combining the generated recommended collection entity information and the bad debt entity marking, the final decision on the fund chain recovery is generated. This decision recommends that the enterprise give priority to collecting the payments from customers with lower collection difficulties, that is, collect payments in the order of the recommended collection entities. At the same time, for customers marked as bad debt entities, it is recommended that the enterprise stop extending credit and instead adopt other risk control measures, such as through prepayment, guarantee, or direct collection, etc., to ensure that the enterprise's fund chain is no longer affected by these high-risk customers.

[0057] Step S500: Send the fund chain risk signal and the fund chain recovery decision to the financial management terminal of the target enterprise for financial anomaly reminder management.

[0058] In the embodiments of the present application, after generating the risk signals of the capital chain and the decision on the restoration of the capital chain, these information are sent to the financial management terminal of the target enterprise through an automated data interface. This terminal is the financial management system within the enterprise (such as an ERP system or a dedicated financial information platform). The sending process uses an API (Application Programming Interface) or message queue technology for data transmission to ensure that the information can be transmitted to the financial management system of the target enterprise in real time and without error.

[0059] Once receiving the risk signals of the capital chain and the decision on the restoration of the capital chain, the financial management terminal triggers the financial anomaly reminder management function to present the risk warning to the enterprise's financial personnel in a prominent manner. Specifically, the financial management terminal notifies the financial personnel through pop-up notifications, dashboard warnings, or email / sms reminders, etc., to ensure that they can understand the risks faced by the capital chain and the restoration measures to be taken in the shortest possible time.

[0060] Furthermore, in the method provided by the application embodiments, for the financial anomaly reminder management, it further includes: Collect historical bank credit approval records, conduct analysis on the associated indicators of credit approvals, and construct a credit approval analysis model based on the associated indicators; collect the associated indicator values of the target enterprise based on the associated indicators, input them into the credit approval analysis model for analysis, and output the credit approval analysis results; conduct financial anomaly reminder management based on the credit approval analysis results.

[0061] In the embodiments of the present application, first, historical bank credit approval records are collected. By obtaining the historical loan data of the target enterprise from a preset database. The historical credit approval records include information such as loan amount, approval date, loan term, interest rate, loan purpose, repayment method, and the credit rating of the borrowing enterprise, etc. These data are extracted through database query methods to form the basic data for subsequent analysis. For example, assume that the loan data extracted from the historical credit records includes a loan amount of 5 million yuan, a loan approval date of May 2019, a loan term of 5 years, an annual interest rate of 5.5%, and the credit rating of the borrowing enterprise is AA. This data provides key background information for subsequent analysis of whether the target enterprise can obtain loan approval.

[0062] Next, perform an analysis of the correlation indicators for credit approvals on these historical credit approval records. This step uses regression analysis methods. The purpose is to identify key correlation indicators by analyzing the relationships between multiple factors in the historical credit approval records (such as loan amount, interest rate, loan term, etc.) and the financial conditions of the enterprise (such as cash flow, debt ratio, asset - liability ratio, etc.), that is, which financial characteristics will affect the likelihood of an enterprise obtaining credit approval. For example, analyze the impact of the debt ratio on loan approval. Are enterprises with a higher debt ratio more likely to be rejected for loans, or can cash flow significantly increase the probability of loan approval? The regression analysis will output a series of correlation indicators (such as debt ratio, cash flow, asset - liability ratio, etc.). Based on these extracted correlation indicators, a credit approval analysis model is then constructed. In this process, a logistic regression model is used for modeling. The logistic regression model is an algorithm commonly used for binary classification problems, especially suitable for scenarios such as credit approval. When constructing the model, the historical credit approval data is used as the training set. The training set includes the financial data of the enterprise (such as debt ratio, cash flow, credit score, etc.) and the result of whether the loan is approved (label 1 indicates loan approval, 0 indicates non - approval). Through supervised learning, the logistic regression model fits according to the correlation indicators in the training data and finally outputs a formula to calculate the probability of an enterprise obtaining loan approval under given financial indicators. For example, if the debt ratio of the target enterprise is 60%, the cash flow is 5 million yuan, and the credit score is B, the logistic regression model may calculate that the probability of this enterprise obtaining loan approval is 70%. Such a model can help banks or financial managers predict whether a target enterprise can obtain credit based on its financial conditions.

[0063] Then, based on the constructed credit approval analysis model, collect the values of the correlation indicators of the target enterprise based on the correlation indicators. Collect the latest financial data of the target enterprise through financial statements (such as balance sheet, cash flow statement, etc.), including correlation indicators such as debt ratio, cash flow, and credit score. For example, assume that the financial data of the target enterprise includes a debt ratio of 65%, a cash flow of 4 million yuan, and a credit score of C. Take these correlation indicators as input data and pass them to the credit approval analysis model. Through the calculation of the model, a probability of loan approval is obtained. For example, the model may output that the probability of this enterprise obtaining loan approval is 60%, indicating that the likelihood of it obtaining credit approval is relatively low.

[0064] Finally, based on the credit approval analysis results, financial anomaly reminder management is carried out. If the output result of the model indicates that the loan approval probability of the target enterprise is relatively low (such as lower than the preset risk threshold, which may be 50%), or there are significant risks in the financial condition of the enterprise (such as too high debt ratio, insufficient cash flow, low credit score, etc.), a financial anomaly reminder is generated. This reminder will be sent to the financial staff through the financial management terminal, using an automated notification method (such as pop-up reminder, email notification, SMS, etc.). After receiving the reminder, the financial staff can immediately adjust their financial decisions, such as strengthening the risk control of the enterprise, requiring additional guarantees, or adjusting the loan conditions.

[0065] In the embodiments of the present application, in summary, the embodiments of the present application at least have the following technical effects: In the present application, ETL technology is adopted to collect multi-source financial correlation data of the target enterprise; data mapping and conversion are performed on the multi-source financial correlation data to construct an enterprise capital chain, where the enterprise capital chain includes multiple capital flow nodes and the capital flow relationships between the nodes; a risk analysis of capital chain breakage is performed on the enterprise capital chain to generate a breakage risk index; if the breakage risk index is greater than the preset risk index, a capital chain risk signal is generated, and a capital chain health recovery analysis is performed to generate a capital chain recovery decision; the capital chain risk signal and the capital chain recovery decision are sent to the financial management terminal of the target enterprise for financial anomaly reminder management. The present invention solves the technical problems of data islands and insufficient risk prediction ability existing in the prior art in financial management. By adopting ETL technology for multi-source data collection, constructing an enterprise capital chain, performing risk analysis and implementing a capital chain recovery decision, the technical effects of improving the accuracy of financial risk prediction and optimizing capital chain management are achieved.

[0066] Embodiment 2, based on the same inventive concept as the intelligent financial management method of multi-source data fusion in the foregoing embodiment, as Figure 2 shown, the present application provides an intelligent financial management system of multi-source data fusion. The system in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the system includes: A data collection module 11, which is used to collect multi-source financial correlation data of the target enterprise by adopting ETL technology; a capital chain construction module 12, which is used to perform data mapping and conversion on the multi-source financial correlation data to construct an enterprise capital chain, wherein the enterprise capital chain includes multiple capital flow nodes and the capital flow relationships between the nodes; a risk analysis module 13, which is used to analyze the risk of capital chain breakage of the enterprise capital chain and generate a breakage risk index; a risk signal generation module 14, which is used to generate a capital chain risk signal if the breakage risk index is greater than a preset risk index, and perform an analysis on the restoration of the health of the capital chain to generate a capital chain restoration decision; an anomaly reminder module 15, which is used to send the capital chain risk signal and the capital chain restoration decision to the financial management terminal of the target enterprise for financial anomaly reminder management.

[0067] Furthermore, the system is also used to implement the following functions: Perform differential correction on the cross-system mapping relationships of the multi-source financial correlation data, and then call a preset data template for unified processing to generate standard multi-source financial correlation data; identify the capital flow relationship, capital flow status, and capital flow type for any piece of financial data in the standard multi-source financial correlation data to generate a capital flow feature set; based on the capital flow feature set, construct a capital flow chain based on the capital flow relationship, and mark the capital flow status and capital flow type to generate the enterprise capital chain.

[0068] Furthermore, the system is also used to implement the following functions: Perform differential correction on the cross-system mapping relationships of the multi-source financial correlation data, including cross-system numerical conversion, time format conversion, currency conversion, and classification mapping.

[0069] Furthermore, the system is also used to implement the following functions: Calculate the current ratio, quick ratio, cash ratio, accounts receivable turnover ratio, and accounts payable turnover ratio based on the enterprise capital chain to generate a multi-index calculation result; collect the historical capital chain set of historical capital chain breakage enterprises within a preset time range before the breakage based on the enterprise scale and business type of the target enterprise; perform an abnormal radar chart distribution analysis on the current ratio, quick ratio, cash ratio, accounts receivable turnover ratio, and accounts payable turnover ratio based on the historical capital chain set to construct an abnormal radar chart distribution set; after converting the multi-index calculation result into a radar chart, compare it with the abnormal radar chart distribution set to generate the breakage risk index.

[0070] Furthermore, the system is also used to implement the following functions: For each capital chain in the historical capital chain set, calculate the capital flow ratio, quick ratio, cash ratio, collection turnover rate, and payment turnover rate, and perform radar chart conversion to generate a set of historical index radar charts; compare every two radar charts in the set of historical index radar charts, perform clustering based on the result of the difference comparison, and delete isolated points to obtain multiple clusters of historical index radar charts; select one radar chart from each of the multiple clusters of historical index radar charts to generate the set of abnormal radar chart distributions.

[0071] Further, the system is also used to implement the following functions: After performing radar chart conversion on the multi-index calculation results, compare them with each abnormal radar chart distribution in the set of abnormal radar chart distributions, and calculate the set of radar chart similarities; based on the set of radar chart similarities, extract the maximum similarity as the break risk indicator.

[0072] Further, the system is also used to implement the following functions: Collect the set of bills receivable and the set of bill receivable entities in the enterprise capital chain; collect the historical collection behaviors of the set of bill receivable entities to construct a set of collection difficulty indicators; arrange the set of bill receivable entities in ascending order of the indicators based on the set of collection difficulty indicators to generate a sequence of bill receivable entities; perform digital modeling on the enterprise capital chain to generate a twin capital chain; based on the twin capital chain and the set of bills receivable, start the collection success simulation from the first entity in the sequence of bill receivable entities. If the simulated break risk indicator is still greater than the preset risk indicator, then perform the collection success simulation on the second entity until the simulated break risk indicator is less than the preset risk indicator to generate recommended bill receivable entity information; mark the entities with collection difficulty indicators greater than the preset difficulty threshold in the sequence of bill receivable entities as bad debt entities; generate the capital chain recovery decision with the recommended bill receivable entity information and the bad debt entity mark.

[0073] Further, the system is also used to implement the following functions: Collect historical bank credit approval records, perform analysis of associated indicators of credit approval, and construct a credit approval analysis model with the associated indicators; collect the associated indicator values of the target enterprise based on the associated indicators, input them into the credit approval analysis model for analysis, and output the credit approval analysis result; perform financial anomaly reminder management based on the credit approval analysis result.

[0074] Embodiment 3. Based on a multi-source data fusion intelligent financial management method in the foregoing embodiments, with the same inventive concept, the present application also provides a computer-readable storage medium, on which a computer program is stored, and the computer program implements the steps of any one of the methods in Embodiment 1 above when executed.

[0075] It should be noted that the above order of the embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above specific embodiments of this specification have been described. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0076] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0077] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An intelligent financial management method based on multi-source data fusion, characterized in that: include: Collect multi-source financial data of target enterprises by using ETL technology; Data mapping and conversion are performed on the multi-source financial related data to construct an enterprise capital chain, wherein the enterprise capital chain includes multiple capital flow nodes and capital flow relationships between the nodes, specifically including: Correcting the differences in the cross-system mapping relationship of the multi-source financial related data, and then calling the preset data template for unified processing to generate standard multi-source financial related data; Identify the capital flow relationship, capital flow status and capital flow type of any piece of financial data in the standard multi-source financial association data to generate a capital flow feature set; Based on the capital flow feature set and the capital flow relationship, a capital flow chain is constructed, and the capital flow status and capital flow type are marked to generate the enterprise capital chain; Conduct a capital chain rupture risk analysis on the capital chain of the enterprise and generate rupture risk indicators, including: Calculate the capital flow ratio, quick ratio, cash ratio, collection turnover rate and payment turnover rate based on the enterprise's capital chain to generate multi-indicator calculation results; Based on the enterprise scale and business type of the target enterprise, a collection of historical capital chains of the enterprise with a broken capital chain within a preset time range before the break is collected; Based on the historical capital chain set, an abnormal radar chart distribution analysis of the capital flow ratio, quick ratio, cash ratio, collection turnover rate and payment turnover rate is performed to construct an abnormal radar chart distribution set; After converting the multi-index calculation results into radar charts, the results are compared with the abnormal radar chart distribution set to generate the fracture risk index; If the break risk index is greater than the preset risk index, a capital chain risk signal is generated, and a capital chain health recovery analysis is performed to generate a capital chain recovery decision; The capital chain risk signal and the capital chain recovery decision are sent to the financial management terminal of the target enterprise for financial abnormality reminder management.

2. The intelligent financial management method of multi-source data fusion according to claim 1, characterized in that: The multi-source financial related data is corrected for differences in cross-system mapping relationships, including cross-system numerical conversion, time format conversion, currency conversion, and category mapping.

3. The intelligent financial management method of multi-source data fusion according to claim 1, characterized in that: Based on the historical capital chain set, abnormal radar chart distribution analysis of capital flow ratio, quick ratio, cash ratio, collection turnover rate and payment turnover rate is performed to construct an abnormal radar chart distribution set, including: Calculate and convert the capital flow ratio, quick ratio, cash ratio, collection turnover rate and payment turnover rate of each capital chain in the historical capital chain set into a radar chart to generate a historical indicator radar chart set; Performing a difference comparison on every two radar charts in the historical indicator radar chart set, clustering according to the difference comparison results and deleting isolated points to obtain a multi-cluster historical indicator radar chart; A radar chart is selected from each of the multiple clusters of historical indicator radar charts to generate the abnormal radar chart distribution set.

4. The intelligent financial management method of multi-source data fusion according to claim 1, characterized in that: After converting the multi-index calculation results into radar charts, the results are compared with the abnormal radar chart distribution set to generate the fracture risk index, including: After performing radar chart conversion on the multi-index calculation result, the result is compared with each abnormal radar chart distribution in the abnormal radar chart distribution set to calculate a radar chart similarity set; Based on the radar chart similarity set, the maximum similarity is extracted as the fracture risk indicator.

5. The intelligent financial management method of multi-source data fusion according to claim 1, characterized in that: If the rupture risk index is greater than the preset risk index, a capital chain risk signal is generated, and a capital chain health recovery analysis is performed to generate a capital chain recovery decision, including: Collecting a collection of bills to be collected and a collection of entities to be collected in the enterprise's capital chain; Collecting historical payment collection behaviors of the set of entities to be paid, and constructing a set of payment collection difficulty indicators; Based on the payment collection difficulty index set, the set of entities to be collected are arranged in order of the index from small to large, to generate a sequence of entities to be collected; Conduct digital modeling for the enterprise's capital chain to generate a twin capital chain; Based on the twin capital chain and the set of bills to be collected, a successful collection simulation is performed starting from the first subject in the sequence of subjects to be collected. If the simulated breakage risk index is still greater than the preset risk index, a successful collection simulation is performed on the second subject until the simulated breakage risk index is still less than the preset risk index, and recommended collection subject information is generated; Marking the subject whose collection difficulty index in the subject sequence to be collected is greater than a preset difficulty threshold as a bad debt subject; The capital chain recovery decision is generated based on the recommended payment entity information and the bad debt entity mark.

6. The intelligent financial management method of multi-source data fusion according to claim 1, characterized in that: Financial abnormality reminder management also includes: Collect historical bank credit approval records, analyze the associated indicators of credit approval, and build a credit approval analysis model based on the associated indicators; Based on the correlation index, the correlation index value of the target enterprise is collected, input into the credit approval analysis model for analysis, and the credit approval analysis result is output; Financial anomaly reminder management is performed based on the credit approval analysis results.

7. An intelligent financial management system for multi-source data fusion, characterized in that: The system is used to implement the intelligent financial management method of multi-source data fusion as described in any one of claims 1 to 6, comprising: Data collection module, used to collect multi-source financial related data of target enterprises by using ETL technology; A capital chain construction module, used to perform data mapping and conversion on the multi-source financial related data to construct an enterprise capital chain, wherein the enterprise capital chain includes a plurality of capital flow nodes and capital flow relationships between the nodes; A risk analysis module, used to analyze the capital chain rupture risk of the enterprise capital chain and generate a rupture risk index; A risk signal generation module is used to generate a capital chain risk signal if the break risk index is greater than a preset risk index, and to perform a capital chain health recovery analysis to generate a capital chain recovery decision; The abnormality reminder module is used to send the capital chain risk signal and the capital chain recovery decision to the financial management terminal of the target enterprise for financial abnormality reminder management.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, an intelligent financial management method for multi-source data fusion as described in any one of claims 1 to 6 is implemented.

Images