Risk control report analysis system and method based on AI intelligence

Abstract

The invention relates to the technical field of data analysis, and discloses a risk control report analysis system and method based on AI intelligence, which greatly reduces manual operation and remarkably improves processing speed and precision through automatic acquisition and analysis of financial, credit, market and other multi-source data in a risk control report. In a traditional risk control process, time and error rate depending on manual data collection and arrangement are effectively solved, through calculation of an anomaly judgment coefficient YCX and analysis of an anomaly magnitude value, the system can intelligently identify an abnormal condition in a risk control report, not only can judge an anomaly, but also can quantify the severity of the anomaly, and the risk control efficiency is improved. The analysis result is further refined through multi-dimensional abnormal coefficient calculation, an abnormal source is accurately positioned, an enterprise is helped to find the root of a problem in multiple risk dimensions, and after the system detects the abnormality, the result can be rapidly transmitted to a user through a feedback module, and real-time risk early warning and a detailed analysis report are provided.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and specifically to a risk control report analysis system and method based on AI intelligence. Background Art

[0002] In financial technology, risk management is a crucial sub-sector, especially in financial institutions such as banks, insurance, and investments. It is mainly responsible for identifying, assessing, and responding to potential financial, credit, market, and other risks. In risk management, risk control reports are one of the key tools. Through comprehensive analysis of financial data, credit data, and market data, they help companies understand risk status and optimize decision-making processes.

[0003] Traditional risk control report analysis has several obvious shortcomings and deficiencies. First, manual data extraction and processing is not only inefficient, but also prone to errors, especially when faced with a large amount of multi-source heterogeneous data, the error rate is even higher. Second, traditional analysis is highly dependent on the experience and judgment of analysts and is easily affected by subjective bias, resulting in some potential risks being ignored or misjudged. In addition, the generation cycle of traditional risk control reports is long, lacks real-time performance, cannot dynamically reflect market changes, and is difficult to provide early warning signals to companies in a timely manner. These problems make traditional risk control reports unable to cope with increasingly complex risks in the modern financial environment, and there is an urgent need to improve them through technical means.

[0004] Therefore, we proposed an AI-based risk control report analysis system to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a risk control report analysis system based on AI intelligence to solve the problem that manual data extraction and processing proposed in the above background technology is not only inefficient but also prone to errors, especially when faced with a large amount of multi-source heterogeneous data, the error rate is higher, and the system analysis is highly dependent on the experience and judgment of the analyst, which is easily affected by subjective bias, resulting in some potential risks being ignored or misjudged.

[0006] To achieve the above object, the present invention provides the following technical solution: an AI-based risk control report analysis system, comprising a first data extraction module, a second data extraction module, a first data processing module, a second data processing module, a first data calculation module, a second data calculation module, a first data analysis module and a second data analysis module;

[0007] The first data extraction module and the second data extraction module are used to extract financial data, credit data and market data in the risk control report, thereby forming a first multi-source data set and a second multi-source data set respectively;

[0008] The first data processing module is used to preprocess the first multi-source data set, and reorganize the preprocessed first multi-source data set into a first data set, a second data set, and a third data set;

[0009] The first data calculation module is used to integrate and calculate the first data set, the second data set and the third data set to generate an abnormality judgment coefficient YCX;

[0010] The first data analysis module is used to compare the abnormality judgment coefficient YCX with a preset first threshold value Y, so as to generate a first comparison result, and judge whether the risk control report is abnormal according to the first comparison result. If the first comparison result shows that the risk control report is abnormal, the abnormality judgment coefficient YCX is integrated and calculated with the first threshold value Y to generate a risk control abnormality magnitude value YLJ, and the abnormal magnitude value is compared with the first threshold value Y again, so as to generate an abnormal magnitude result of the risk control report;

[0011] The second data processing module preprocesses the second multi-source data set when the first comparison result of the first data analysis module shows that the risk control report is abnormal, and reorganizes the preprocessed second multi-source data set into a fourth data set, a fifth data set, and a sixth data set;

[0012] The second data calculation module is used to calculate the fourth data set, the fifth data set and the sixth data set respectively, so as to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC and the market data abnormality coefficient YSC respectively;

[0013] The second data analysis module is used to compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC and the market data anomaly coefficient YSC with the preset financial data anomaly threshold YC, the credit data anomaly threshold YX and the market data anomaly threshold YS, respectively, so as to generate a second comparison result, locate the abnormal position in the risk control report according to the second comparison result, and integrate and calculate the corresponding abnormal coefficient and the corresponding abnormal threshold according to the specific abnormal position located, so as to generate a corresponding abnormal magnitude value, and compare the calculated abnormal magnitude value with the corresponding abnormal threshold again, so as to generate an abnormal magnitude result;

[0014] The feedback module is used to send each comparison result and magnitude result to the terminal.

[0015] Preferably, the first data extraction module includes a first financial data extraction unit, a first credit data extraction unit and a first market data extraction unit, and the second data extraction module includes a second financial data extraction unit, a second credit data extraction unit and a second market data extraction unit;

[0016] The first financial data extraction unit and the second financial data extraction unit are respectively used to extract financial data in the risk control report, wherein the first financial data extraction unit is used to extract the cash flow ratio, net profit and return on assets in the risk control report; the second financial data extraction unit is used to extract the cash flow from operating activities, the cash flow from investment activities and the cash flow from financing activities in the risk control report;

[0017] The first credit data extraction unit and the second credit data extraction unit are respectively used to extract credit data from the risk control report, wherein the first credit data extraction unit is used to extract the credit rating change rate, overdue loan ratio and default rate from the risk control report, and the second credit data extraction unit is used to extract short-term debt, long-term debt and reserve coverage ratio from the risk control report;

[0018] The first market data extraction unit and the second market data extraction unit are respectively used to extract market data in the risk control report, wherein the first market data extraction unit is used to extract interest rate fluctuations, exchange rate fluctuations and industry average return change rate in the risk control report, and the second market data extraction unit is used to extract operating efficiency, market cost and supply chain risk;

[0019] The first multi-source data set includes cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation and industry average earnings change rate;

[0020] The second multi-source data set includes cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk.

[0021] Preferably, the first data processing module is used to preprocess the data contained in the first multi-source data set, and dimensionlessly transform the preprocessed data, and reorganize them into a first data set, a second data set, and a third data set;

[0022] The first data set includes cash flow ratio, net profit, and return on assets;

[0023] The cash flow ratios are recorded as Z1, Z2, Z3, ..., Zn according to the timestamps;

[0024] Net profit is recorded in months according to timestamps as J1, J2, J3, ..., Jn;

[0025] The asset returns are recorded as C1, C2, C3, ..., Cn according to the timestamp;

[0026] The second data set includes the credit rating change rate, the percentage of overdue loans, and the default rate;

[0027] The rating change rates are recorded as P1, P2, P3, ..., Pn according to the timestamps;

[0028] The overdue loan ratios are recorded as Y1, Y2, Y3, ..., Yn according to the timestamps;

[0029] The default rates are recorded as W1, W2, W3, ..., Wn according to the timestamps;

[0030] The third data set includes interest rate fluctuations, exchange rate fluctuations, and industry average earnings change rates;

[0031] The interest rate fluctuations are recorded as L1, L2, L3, ..., Ln according to the timestamps;

[0032] The exchange rate fluctuations are recorded as H1, H2, H3, ..., Hn according to the timestamps;

[0033] The industry average return change rate is recorded as S1, S2, S3, ..., Sn according to the timestamp;

[0034] The second data processing module is used to preprocess the data included in the second multi-source data set, and dimensionlessly transform the preprocessed data, and reorganize them into a fourth data set, a fifth data set, and a sixth data set;

[0035] The fourth data set includes cash flows from operating activities, investing activities, and financing activities;

[0036] The cash flows from operating activities are recorded as B1, B2, B3, ..., Bn according to the timestamps;

[0037] The cash flows from investment activities are recorded as T1, T2, T3, ..., Tn according to the timestamps;

[0038] The cash flows from financing activities are recorded as D1, D2, D3, ..., Dn according to the timestamps;

[0039] The fifth data set includes short-term debt, long-term debt, and reserve coverage ratios;

[0040] Short-term debts are recorded as E1, E2, E3, ..., En according to the timestamp;

[0041] Long-term debt is recorded as F1, F2, F3, ..., Fn according to the time stamp;

[0042] The reserve coverage ratios are recorded as G1, G2, G3, ..., Gn according to the timestamps;

[0043] The sixth data set includes operational efficiency, market costs, and supply chain risks;

[0044] The operational efficiency is recorded as I1, I2, I3, ..., In according to the timestamp;

[0045] The market costs are recorded as K1, K2, K3, ..., Kn according to the timestamp;

[0046] Supply chain risks are recorded as M1, M2, M3, ..., Mn according to timestamps.

[0047] Preferably, the first data calculation module calculates and obtains the abnormality judgment coefficient YCX by the following formula:

[0048] YCX=a1×CK1+a2×CK2+a3×CK3;

[0049] In the formula: CK1 is the first reference coefficient, which is obtained by integrating and calculating the first data set; CK2 is the second reference coefficient, which is obtained by integrating and calculating the second data set; CK3 is the third reference coefficient, which is obtained by integrating and calculating the third data set; a1, a2 and a3 are weight values, and the values ​​of a1, a2 and a3 are adjusted and set by the user.

[0050] Preferably, the first data calculation module calculates and obtains the first reference coefficient CK1, the second reference coefficient CK2 and the third reference coefficient CK3 respectively through the following formulas;

[0051] CK1=b1×CZ+b1×CJ+b1×CC;

[0052] CK2=b1×CP+b1×CY+b1×CW;

[0053] CK3 = b1 × CL + b1 × CH + b1 × CS;

[0054]

[0055]

[0056] Where: CZ is the intervention value of cash flow ratio, CJ is the intervention value of net profit, CC is the intervention value of return on assets, CP is the intervention value of credit rating change rate, CY is the intervention value of overdue loan ratio, CW is the intervention value of default rate, CL is the intervention value of interest rate fluctuation, CH is the intervention value of exchange rate fluctuation, and CS is the intervention value of industry average return change rate, which are calculated and obtained by integrating the first data set, the second data set and the third data set respectively;

[0057] Zn, Jn, Cn, Pn, Yn, Wn, Ln, Hn and Sn are respectively the cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation and industry average earnings change rate at the nth time stamp;

[0058] Z1, J1, C1, P1, Y1, W1, L1, H1, and S1 are the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate at the first timestamp, respectively;

[0059] Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax, and Smax are the maximum values of the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate, respectively;

[0060] Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin, and Smin are the minimum values of the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate, respectively;

[0061] b1, b2, and b3 are weight values, and the values of b1, b2, and b3 are adjusted and set by the user.

[0062] Preferably, the first data analysis module alarms the first main analysis unit and the first secondary analysis unit. The first main analysis unit is used to generate a first comparison result, and the first secondary analysis unit is used to re-analyze the first comparison result;

[0063] The first comparison result is as follows;

[0064] When YCX ≤ Y, it means that the risk control report is normal. When YCX > Y, it means that the risk control report is abnormal;

[0065] The re-analysis result is as follows;

[0066] When YLJ ≤ Y × 60%, it means that the risk control has a first-level anomaly. When Y × 60% < YLJ ≤ Y × 90%, it means that the risk control has a second-level anomaly. When YLJ > Y × 90%, it means that the risk control has a third-level anomaly;

[0067] The risk control anomaly magnitude value YLJ is obtained by calculating through the following formula;

[0068]

[0069] In the formula: YCX is the anomaly judgment coefficient, and Y is the first threshold.

[0070] Preferably, the second data calculation module calculates and obtains the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC, and the market data anomaly coefficient YSC through the following formulas respectively;

[0071]

[0072] Where: Bn, Tn, Dn, En, Fn, Gn, In, Kn and Mn are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the nth time stamp respectively;

[0073] Bp, Tp, Dp, Ep, Fp, Gp, Ip, Kp and Mp are the average values ​​of cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk respectively;

[0074] d1, d2 and d3 are weight values, and the values ​​of d1, d2 and d3 are adjusted and set by the user.

[0075] Preferably, the second data calculation module calculates and obtains the average cash flow from operating activities Bp, the average cash flow from investing activities Tp, the average cash flow from financing activities Dp, the average short-term debt Ep, the average long-term debt Fp, the average reserve coverage ratio Gp, the average operating efficiency Ip, the average market cost Kp and the average supply chain risk Mp respectively through the following formulas;

[0076]

[0077] Where: Bn, Tn, Dn, En, Fn, Gn, In, Kn and Mn are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the nth time stamp respectively;

[0078] B1, T1, D1, E1, F1, G1, I1, K1 and M1 are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the first timestamp respectively.

[0079] Preferably, the second data analysis module includes a second main analysis unit and a secondary analysis unit, the second main analysis unit is used to generate a second comparison result, and the secondary analysis unit is used to reanalyze the second comparison result;

[0080] The second comparison results are as follows;

[0081] When YCC≤YC, it means that there is no abnormality in the financial data. When YCC>YC, it means that there is an abnormality in the financial data.

[0082] When YXC ≤ YX, it indicates that the credit data is normal. When YCC > YX, it indicates that the credit data is abnormal;

[0083] When YSC ≤ YS, it indicates that the market data is normal. When YCC > YS, it indicates that the market data report is abnormal;

[0084] The re - analysis results are as follows;

[0085] When YCJ ≤ YC × 60%, it indicates that the risk control has a first - level anomaly. When YC × 60% < YCJ ≤ YC × 90%, it indicates that the risk control has a second - level anomaly. When YCJ > YC × 90%, it indicates that the risk control has a third - level anomaly;

[0086] When YXJ ≤ YX × 60%, it indicates that the risk control has a first - level anomaly. When YX × 60% < YXJ ≤ YX × 90%, it indicates that the risk control has a second - level anomaly. When YXJ > Y × 90%, it indicates that the risk control has a third - level anomaly;

[0087] When YSJ ≤ YS × 60%, it indicates that the risk control has a first - level anomaly. When YS × 60% < YSJ ≤ YS × 90%, it indicates that the risk control has a second - level anomaly. When YSJ > YS × 90%, it indicates that the risk control has a third - level anomaly;

[0088] The abnormal magnitude values of financial data YCJ, credit data YXJ, and market data YSJ are obtained by calculating through the following formulas respectively;

[0089]

[0090] In the formula: YCC is the abnormal coefficient of financial data, YXC is the abnormal coefficient of credit data, YSC is the abnormal coefficient of market data, YC is the abnormal threshold of financial data, YX is the abnormal threshold of credit data, and YS is the abnormal threshold of market data.

[0091] This application also includes a risk control report analysis method based on AI intelligence. The specific steps are as follows:

[0092] S1. Perform multi - source extraction on the financial data, credit data, and market data in the risk control report to form a first multi - source data set, and extract the financial data, credit data, and market data in the risk control report to form a second multi - source data set;

[0093] S2. Perform pre - processing operations on the first multi - source data set, including data cleaning, denoising, missing value filling, and normalization. The processed data will be re - organized into: the first data set, the second data set, and the third data set. Calculate the abnormal judgment coefficient YCX for the pre - processed first data set, second data set, and third data set;

[0094] S3. Compare the generated abnormality judgment coefficient YCX with the preset first threshold value Y to determine whether there is an abnormality in the risk control report. If the first comparison result shows that there is an abnormality, the abnormality judgment coefficient YCX and the first threshold value Y are integrated and calculated again to generate an abnormality magnitude value. By comparing the abnormality magnitude value with the first threshold value Y, an abnormality magnitude result of the risk control report is generated to determine the severity of the abnormality.

[0095] S4. After the risk control report is determined to be abnormal, the fourth data set, the fifth data set, and the sixth data set are calculated respectively to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC, and the market data abnormality coefficient YSC, which are used to refine the location and analysis of the abnormal position;

[0096] S5. Compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC, and the market data anomaly coefficient YSC with the preset corresponding anomaly thresholds to generate a second comparison result. According to the comparison result, locate the location of the anomaly and the specific anomaly type, integrate and calculate the located anomaly coefficient with the corresponding anomaly threshold, generate a corresponding anomaly magnitude value, and compare it with the threshold again to output the anomaly magnitude result;

[0097] S6. Send each comparison result and abnormal magnitude result to the terminal for the terminal user to view, analyze and make decisions.

[0098] Compared with the prior art, the present invention has the following beneficial effects:

[0099] 1. This system greatly reduces manual operations and significantly improves processing speed and accuracy through the automated collection and analysis of financial, credit, market and other multi-source data in risk control reports. The time and error rate of relying on manual data collection and collation in traditional risk control processes have been effectively solved. Through the calculation of the abnormal judgment coefficient YCX and the analysis of the abnormal magnitude value, the system can intelligently identify abnormal situations in risk control reports, and can not only judge abnormalities, but also quantify the severity of abnormalities. The calculation of multi-dimensional abnormal coefficients further refines the analysis results and accurately locates the source of abnormalities, thereby helping enterprises find the root cause of problems in multiple risk dimensions. After detecting anomalies, the system can quickly convey the results to users through the feedback module, providing real-time risk warnings and detailed analysis reports, which enables managers to respond quickly when problems occur and reduce the potential damage of risks. The system adopts a quantitative analysis method and provides a more scientific and standardized risk control assessment by comparing with the preset threshold. This data-based quantitative analysis method avoids subjective bias in manual judgment, improves the accuracy of risk identification, and helps enterprises make accurate decisions at a more detailed level.

[0100] 2. The abnormal judgment coefficient YCX generated by integrated calculation can provide a clear risk picture and flexibly adjust the focus of analysis according to the actual needs of users to ensure that the risk control report can be consistent with the strategic needs of the enterprise. This provides a more reliable and accurate risk assessment basis for the decision makers of the enterprise, helping them to make more targeted decisions in an uncertain market environment. By calculating the abnormal coefficient YCC of financial data, the abnormal coefficient YXC of credit data and the abnormal coefficient YSC of market data respectively, the system can comprehensively detect abnormal situations from different data dimensions. The data of each dimension is calculated by a specific formula. This comprehensive analysis method improves the system's ability to detect risks and avoids misjudgments or omissions that may be caused by single-dimensional analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 It is a system flow chart of the present invention.

[0102] Figure 2 It is a diagram of the method steps of the present invention.

[0103] In the figure: 1. first data extraction module; 11. first financial data extraction unit; 12. first credit data extraction unit; 13. first market data extraction unit; 2. second data extraction module; 21. second financial data extraction unit; 22. second credit data extraction unit; 23. second market data extraction unit; 3. first data processing module; 4. second data processing module; 5. first data calculation module; 6. second data calculation module; 7. first data analysis module; 71. first main analysis unit; 72. first analysis unit; 8. second data analysis module; 81. second main analysis unit; 82. second analysis unit; 9. feedback module. DETAILED DESCRIPTION

[0104] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0105] Example 1: Please refer to Figure 1 , a risk control report analysis system based on AI intelligence, comprising a first data extraction module 1, a second data extraction module 2, a first data processing module 3, a second data processing module 4, a first data calculation module 5, a second data calculation module 6, a first data analysis module 7 and a second data analysis module 8;

[0106] The first data extraction module 1 and the second data extraction module 2 are used to extract financial data, credit data and market data in the risk control report, thereby forming a first multi-source data set and a second multi-source data set respectively;

[0107] The first data processing module 3 is used to preprocess the first multi-source data set, and reorganize the preprocessed first multi-source data set into a first data set, a second data set, and a third data set;

[0108] The first data calculation module 5 is used to integrate and calculate the first data set, the second data set and the third data set, so as to generate an abnormality judgment coefficient YCX;

[0109] The first data analysis module 7 is used to compare the abnormality judgment coefficient YCX with the preset first threshold value Y, so as to generate a first comparison result, and judge whether the risk control report is abnormal according to the first comparison result. If the first comparison result shows that the risk control report is abnormal, the abnormality judgment coefficient YCX is integrated and calculated with the first threshold value Y to generate a risk control abnormality magnitude value YLJ, and the abnormal magnitude value is compared with the first threshold value Y again, so as to generate a risk control report abnormal magnitude result;

[0110] When the first comparison result of the first data analysis module 7 shows that the risk control report is abnormal, the second data processing module 4 preprocesses the second multi-source data set and reorganizes the preprocessed second multi-source data set into a fourth data set, a fifth data set and a sixth data set;

[0111] The second data calculation module 6 is used to calculate the fourth data set, the fifth data set and the sixth data set respectively, so as to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC and the market data abnormality coefficient YSC respectively;

[0112] The second data analysis module 8 is used to compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC and the market data anomaly coefficient YSC with the preset financial data anomaly threshold YC, the credit data anomaly threshold YX and the market data anomaly threshold YS, respectively, so as to generate a second comparison result, locate the abnormal position in the risk control report according to the second comparison result, and integrate and calculate the corresponding abnormal coefficient and the corresponding abnormal threshold according to the specific abnormal position located, so as to generate a corresponding abnormal magnitude value, and compare the calculated abnormal magnitude value with the corresponding abnormal threshold again, so as to generate an abnormal magnitude result;

[0113] The feedback module 9 is used to send each comparison result and magnitude result to the terminal.

[0114] In this embodiment: the first data extraction module 1 realizes the integration and acquisition of multi-source data by automatically extracting financial data, credit data and market data in the risk control report. This process reduces the workload of manual data collection and improves the efficiency and accuracy of data collection. By forming the first multi-source data set, the system can quickly capture multi-dimensional risk information, thereby providing basic data support for subsequent analysis and processing.

[0115] When the risk control report is judged to be abnormal, the second data extraction module 2 extracts relevant data again to form a second multi-source data set. By linking with the first data extraction module 1, this module can ensure that more detailed data is further obtained when an abnormality occurs, thereby providing more accurate input for subsequent abnormality positioning. This process strengthens the intelligent response capability of the system and effectively avoids misjudgment and missed judgment due to insufficient data.

[0116] The first data processing module 3 preprocesses the extracted first multi-source data set, including operations such as denoising, missing value filling, and data normalization. By arranging the preprocessed data into the first data set, the second data set, and the third data set, the module ensures the quality and consistency of the data. High-quality preprocessing lays a solid foundation for subsequent data calculation and abnormality judgment, and improves the accuracy and reliability of the overall system.

[0117] When the risk control report is judged as abnormal, the second data processing module 4 pre-processes the second multi-source data set and organizes it into the fourth data set, the fifth data set and the sixth data set. This pre-processing stage performs detailed processing on data of different dimensions, ensuring that the subsequent abnormal positioning in the three aspects of finance, credit and market is more accurate. By further processing the abnormal related data, the system can better conduct in-depth analysis on the source of the abnormality.

[0118] The first data calculation module 5 integrates and calculates the first data set, the second data set, and the third data set to generate an abnormality judgment coefficient YCX. This coefficient can quantitatively reflect the potential risks or abnormalities in the report, making risk judgment more scientific and standardized. This calculation step not only improves the efficiency of risk judgment, but also makes the analysis of risk control reports more intelligent and automated.

[0119] After the second data calculation module 6 finds an anomaly, it performs detailed calculations on the preprocessed fourth data set, fifth data set, and sixth data set to generate financial data anomaly coefficients YCC, credit data anomaly coefficients YXC, and market data anomaly coefficients YSC. These refined coefficients provide support for locating the specific source of the anomaly, so that the nature of the anomaly and the scope of possible impact can be more accurately determined. This module helps the system achieve multi-dimensional in-depth analysis and lays the foundation for accurate positioning of anomalies.

[0120] The first data analysis module 7 generates a first comparison result by comparing the abnormality judgment coefficient YCX with the preset first threshold value Y. When the comparison result shows that an abnormality exists, the abnormal magnitude value is further calculated and compared again. This process enables the system to not only determine whether an abnormality exists, but also quantitatively evaluate the severity of the abnormality, thereby providing managers with more valuable decision-making references. Through this analysis, the system can detect potential risks at an early stage and take corresponding measures before the problem becomes serious.

[0121] After discovering the anomaly, the second data analysis module 8 further analyzes the anomaly of each data set. The second comparison result is generated by comparing the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC and the market data anomaly coefficient YSC with their respective preset thresholds YC, YX and YS. This process achieves accurate positioning of the anomaly and can be refined to specific dimensions such as finance, credit or market, helping enterprises to take targeted management measures for different risk sources.

[0122] Feedback module 9 is used to send all comparison results and abnormal magnitude results to the user terminal. Through this feedback mechanism, the system can report potential risk situations to users in a timely and comprehensive manner, and help users make wise decisions through visual analysis results. This module improves the ease of use and information interaction efficiency of the system, making risk analysis results easier to understand and apply.

[0123] This system greatly reduces manual operations and significantly improves processing speed and accuracy through the automated collection and analysis of multi-source data such as finance, credit, and market in risk control reports. The time and error rate of the traditional risk control process that relies on manual collection and organization of data have been effectively solved. Through the calculation of the abnormal judgment coefficient YCX and the analysis of the abnormal magnitude value, the system can intelligently identify abnormal situations in risk control reports, and can not only judge abnormalities, but also quantify the severity of abnormalities. The calculation of multi-dimensional abnormal coefficients further refines the analysis results and accurately locates the source of abnormalities, thereby helping enterprises find the root cause of problems in multiple risk dimensions. After detecting an abnormality, the system can quickly convey the results to users through the feedback module 9, providing real-time risk warnings and detailed analysis reports, which enables managers to respond quickly when problems occur and reduce the potential damage of risks. The system adopts a quantitative analysis method and provides a more scientific and standardized risk control assessment by comparing with the preset threshold. This data-based quantitative analysis method avoids subjective bias in manual judgment, improves the accuracy of risk identification, and helps enterprises make accurate decisions at a more detailed level.

[0124] Example 2: Please refer to Figure 1, the first data extraction module 1 includes a first financial data extraction unit 11, a first credit data extraction unit 12 and a first market data extraction unit 13, and the second data extraction module 2 includes a second financial data extraction unit 21, a second credit data extraction unit 22 and a second market data extraction unit 23;

[0125] The first financial data extraction unit 11 and the second financial data extraction unit 21 are respectively used to extract financial data in the risk control report, wherein the first financial data extraction unit 11 is used to extract the cash flow ratio, net profit and return on assets in the risk control report; the second financial data extraction unit 21 is used to extract the cash flow from operating activities, the cash flow from investment activities and the cash flow from financing activities in the risk control report;

[0126] The first credit data extraction unit 12 and the second credit data extraction unit 22 are respectively used to extract credit data from the risk control report, wherein the first credit data extraction unit 12 is used to extract the credit rating change rate, overdue loan ratio and default rate from the risk control report, and the second credit data extraction unit 22 is used to extract short-term debt, long-term debt and reserve coverage ratio from the risk control report;

[0127] The first market data extraction unit 13 and the second market data extraction unit 23 are respectively used to extract market data in the risk control report, wherein the first market data extraction unit 13 is used to extract interest rate fluctuations, exchange rate fluctuations and industry average return change rate in the risk control report, and the second market data extraction unit 23 is used to extract operational efficiency, market cost and supply chain risk;

[0128] The first multi-source data set includes cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate;

[0129] The second multi-source data set includes cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage, operating efficiency, market costs, and supply chain risks.

[0130] In this embodiment: This AI-based risk control report analysis system has been significantly optimized in the data extraction module. Through hierarchical financial, credit, and market data extraction units, the system can automatically and finely extract key information of different categories. Compared with the traditional manual data collection method, the system greatly improves the efficiency and accuracy of data extraction. First, the first financial data extraction unit 11 and the second financial data extraction unit 21 respectively extract financial data of different dimensions in the risk control report. The first financial data extraction unit 11 focuses on cash flow ratio, net profit and return on assets, while the second financial data extraction unit 21 focuses on cash flow from operating activities, cash flow from investment activities and cash flow from financing activities. Such a design helps the system obtain key information on financial health from multiple dimensions to ensure a more solid foundation for risk assessment.

[0131] The first credit data extraction unit 12 and the second credit data extraction unit 22 are used to extract the credit data of the enterprise. The first credit data extraction unit 12 focuses on the credit rating change rate, overdue loan ratio and default rate, and can evaluate the credit performance of the enterprise, while the second credit data extraction unit 22 extracts short-term debt, long-term debt and reserve coverage ratio to help the system analyze the debt situation of the enterprise and its ability to deal with debt risks. These data provide a solid foundation for a more comprehensive credit risk analysis. The first market data extraction unit 13 and the second market data extraction unit 23 respectively extract data for the market risk of the enterprise. The first market data extraction unit 13 focuses on interest rate fluctuations, exchange rate fluctuations and the average industry earnings change rate to ensure that the system can track changes in the external market environment; the second market data extraction unit 23 focuses on extracting operational efficiency, market costs and supply chain risks, thereby providing an in-depth analysis of the operating environment and market competitiveness of the enterprise.

[0132] Example 3: Please refer to Figure 1 The first data processing module 3 is used to preprocess the data contained in the first multi-source data set, and non-dimensionalize the pre-processed data, and reorganize it into a first data set, a second data set, and a third data set;

[0133] The first data set includes cash flow ratio, net profit, and return on assets;

[0134] The cash flow ratios are recorded as Z1, Z2, Z3, ..., Zn according to the timestamps;

[0135] Net profit is recorded in months according to timestamps as J1, J2, J3, ..., Jn;

[0136] The asset returns are recorded as C1, C2, C3, ..., Cn according to the timestamp;

[0137] The second data set includes the credit rating change rate, the percentage of overdue loans, and the default rate;

[0138] The rating change rates are recorded as P1, P2, P3, ..., Pn according to the timestamps;

[0139] The overdue loan ratios are recorded as Y1, Y2, Y3, ..., Yn according to the timestamps;

[0140] The default rates are recorded as W1, W2, W3, ..., Wn according to the timestamps;

[0141] The third data set includes interest rate fluctuations, exchange rate fluctuations, and industry average earnings change rates;

[0142] The interest rate fluctuations are recorded as L1, L2, L3, ..., Ln according to the timestamps;

[0143] The exchange rate fluctuations are recorded as H1, H2, H3, ..., Hn according to the timestamps;

[0144] The industry average return change rate is recorded as S1, S2, S3, ..., Sn according to the timestamp;

[0145] The second data processing module 4 is used to preprocess the data included in the second multi-source data set, and dimensionlessly transform the preprocessed data, and reorganize them into a fourth data set, a fifth data set, and a sixth data set;

[0146] The fourth data set includes cash flows from operating activities, investing activities, and financing activities;

[0147] The cash flows from operating activities are recorded as B1, B2, B3, ..., Bn according to the timestamps;

[0148] The cash flows from investment activities are recorded as T1, T2, T3, ..., Tn according to the timestamps;

[0149] The cash flows from financing activities are recorded as D1, D2, D3, ..., Dn according to the timestamps;

[0150] The fifth data set includes short-term debt, long-term debt, and reserve coverage ratios;

[0151] Short-term debts are recorded as E1, E2, E3, ..., En according to the timestamp;

[0152] Long-term debt is recorded as F1, F2, F3, ..., Fn according to the time stamp;

[0153] The reserve coverage ratios are recorded as G1, G2, G3, ..., Gn according to the timestamps;

[0154] The sixth data set includes operational efficiency, market costs, and supply chain risks;

[0155] The operational efficiency is recorded as I1, I2, I3, ..., In according to the timestamp;

[0156] The market costs are recorded as K1, K2, K3, ..., Kn according to the timestamp;

[0157] Supply chain risks are recorded as M1, M2, M3, ..., Mn according to timestamps.

[0158] In this embodiment, by preprocessing and dimensionlessizing multi-source data sets, the influence of different dimensions and scales on data analysis can be eliminated, so that data can be compared on the same level. This process not only improves the comparability of data, but also lays a solid foundation for subsequent anomaly detection and risk analysis.

[0159] The first data processing module 3 and the second data processing module 4 systematically organize the extracted financial, credit and market data to form the first, second, third, fourth, fifth and sixth data sets. This multi-dimensional data integration enables the system to comprehensively analyze the financial status, credit risk and market environment of the enterprise from multiple angles, providing the enterprise with rich analytical information.

[0160] Automated data preprocessing and organization significantly reduces the need for manual operations, saving time and labor costs. Compared with traditional methods, the system can respond to data needs more quickly, improve the timeliness of risk analysis, and provide enterprises with faster decision-making support. The segmentation of the first, second and third data sets, and the construction of the fourth, fifth and sixth data sets ensure that the system can deeply analyze the relationship between each data point, improving the depth and breadth of data analysis. Through meticulous data analysis, enterprises can more accurately identify potential risks and formulate targeted response measures. The integrated data set provides a comprehensive information foundation, which can help enterprises to be more scientific when formulating strategies. Management can use this data for risk assessment and control to ensure that the company remains competitive in a changing market environment. Organizing and analyzing data with different timestamps can support enterprises to establish real-time monitoring and early warning mechanisms. Once certain indicators fluctuate abnormally, the system can issue an alarm in time to help enterprises quickly respond to potential risks and reduce losses.

[0161] Example 4: Please refer to Figure 1 , the first data calculation module 5 calculates and obtains the abnormality judgment coefficient YCX through the following formula;

[0162] YCX=a1×CK1+a2×CK2+a3×CK3;

[0163] In the formula: CK1 is the first reference coefficient, which is obtained by integrating and calculating the first data set; CK2 is the second reference coefficient, which is obtained by integrating and calculating the second data set; CK3 is the third reference coefficient, which is obtained by integrating and calculating the third data set; a1, a2 and a3 are weight values, and the values ​​of a1, a2 and a3 are adjusted and set by the user.

[0164] The first data calculation module 5 calculates and obtains the first reference coefficient CK1, the second reference coefficient CK2 and the third reference coefficient CK3 through the following formulas respectively;

[0165] CK1=b1×CZ+b1×CJ+b1×CC;

[0166] CK2=b1×CP+b1×CY+b1×CW;

[0167] CK3 = b1 × CL + b1 × CH + b1 × CS;

[0168]

[0169] Where: CZ is the intervention value of cash flow ratio, CJ is the intervention value of net profit, CC is the intervention value of return on assets, CP is the intervention value of credit rating change rate, CY is the intervention value of overdue loan ratio, CW is the intervention value of default rate, CL is the intervention value of interest rate fluctuation, CH is the intervention value of exchange rate fluctuation, and CS is the intervention value of industry average return change rate, which are calculated and obtained by integrating the first data set, the second data set and the third data set respectively;

[0170] Zn, Jn, Cn, Pn, Yn, Wn, Ln, Hn and Sn are respectively the cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation and industry average earnings change rate at the nth time stamp;

[0171] Z1, J1, C1, P1, Y1, W1, L1, H1 and S1 are respectively the cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation and industry average earnings change rate at the first timestamp;

[0172] Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax, and Smax are the maximum values ​​of cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation, and industry average earnings change rate, respectively;

[0173] Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin, and Smin are the minimum values ​​of cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation, and industry average earnings change rate, respectively;

[0174] b1, b2 and b3 are weight values, and the values ​​of b1, b2 and b3 are adjusted and set by the user.

[0175] The second data calculation module 6 calculates and obtains the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC and the market data abnormality coefficient YSC respectively through the following formulas;

[0176]

[0177] Where: Bn, Tn, Dn, En, Fn, Gn, In, Kn and Mn are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the nth time stamp respectively;

[0178] Bp, Tp, Dp, Ep, Fp, Gp, Ip, Kp and Mp are the average values ​​of cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk respectively;

[0179] d1, d2 and d3 are weight values, and the values ​​of d1, d2 and d3 are adjusted and set by the user.

[0180] The second data calculation module 6 calculates and obtains the average cash flow from operating activities Bp, the average cash flow from investment activities Tp, the average cash flow from financing activities Dp, the average short-term debt Ep, the average long-term debt Fp, the average reserve coverage ratio Gp, the average operating efficiency Ip, the average market cost Kp and the average supply chain risk Mp through the following formulas;

[0181]

[0182] Where: Bn, Tn, Dn, En, Fn, Gn, In, Kn and Mn are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the nth time stamp respectively;

[0183] B1, T1, D1, E1, F1, G1, I1, K1 and M1 are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk at the first timestamp respectively.

[0184] In this embodiment: by introducing the calculation formula of multiple reference coefficients, the system can not only identify the individual anomalies of each data point, but also generate more accurate risk judgments through multi-dimensional integrated analysis. This multi-dimensional analysis mechanism significantly improves the accuracy of the risk control system and reduces the possibility of false positives and false negatives. The weight value can be adjusted manually by the user, which means that the user can flexibly adjust the weights of different data dimensions according to the current focus of the enterprise or changes in the market environment. This flexibility ensures that the system can adaptively adjust the analysis focus when facing different business scenarios and risk control requirements, so that the risk control report is more in line with the actual situation. By calculating the reference coefficients for each data dimension separately, the system can efficiently identify the abnormalities of key indicators. In addition, the introduction of maximum and minimum values ​​enables the system to quickly determine the degree of abnormality of current data based on historical data, ensuring the timeliness and accuracy of anomaly detection. This comparative analysis based on historical data and current data helps enterprises quickly locate anomalies and reduces unnecessary analysis time.

[0185] The abnormal judgment coefficient YCX generated by integrated calculation can provide a clear risk picture and flexibly adjust the focus of analysis according to the actual needs of users to ensure that the risk control report is consistent with the strategic needs of the enterprise. This provides a more reliable and accurate risk assessment basis for corporate decision makers, helping them make more targeted decisions in an uncertain market environment.

[0186] By calculating the financial data anomaly coefficient YCC, credit data anomaly coefficient YXC and market data anomaly coefficient YSC respectively, the system can comprehensively detect anomalies from different data dimensions. The data of each dimension is calculated using a specific formula. This comprehensive analysis method improves the system's ability to detect risks and avoids misjudgments or omissions that may result from single-dimensional analysis.

[0187] By comparing the data at the current timestamp with the historical average, the system can quickly detect the amplitude of data fluctuations, thereby determining whether there are anomalies. This comparative analysis not only improves the accuracy of anomaly detection but also enhances the system's response speed to risks, helping enterprises promptly grasp anomaly signals. The weight values in the coefficient formula can be flexibly adjusted according to user needs, which provides the ability for personalized customization. Different enterprises may have different focuses on financial, credit, or market risks. Adjusting the weight values enables the system to dynamically adjust the analysis focus according to specific business requirements, further enhancing the system's adaptability and flexibility.

[0188] The system not only calculates the data anomaly coefficient at a single timestamp but also calculates based on data at different time points, enabling real-time monitoring of the changing trends of financial, credit, and market data. This helps enterprises better predict future risks and adjust their business strategies and risk control measures in a timely manner according to trend changes. This calculation based on the average of historical data and current data helps the system form a more stable benchmark and accurately evaluate long-term trends, and reduces the interference of short-term data fluctuations on anomaly detection, thereby providing more stable risk assessment results.

[0189] Through the hierarchical analysis of multiple anomaly coefficients, the system can provide an early warning mechanism for different risk dimensions. This hierarchical early warning enables enterprises to more intuitively understand whether the risks are caused by financial, credit, or market factors, so that they can take corresponding measures more targeted.

[0190] Example 5: Please refer to Figure 1 , the first data analysis module 7 alarms the first main analysis unit 71 and the first secondary analysis unit 72. The first main analysis unit 71 is used to generate the first comparison result, and the first secondary analysis unit 72 is used to re-analyze the first comparison result;

[0191] The first comparison result is as follows;

[0192] When YCX ≤ Y, it means there is no anomaly in the risk control report. When YCX > Y, it means there is an anomaly in the risk control report;

[0193] The re-analysis result is as follows;

[0194] When YLJ ≤ Y × 60%, it means a first-level anomaly in risk control. When Y × 60% < YLJ ≤ Y × 90%, it means a second-level anomaly in risk control. When YLJ > Y × 90%, it means a third-level anomaly in risk control;

[0195] The risk control anomaly magnitude value YLJ is obtained by calculating through the following formula;

[0196]

[0197] Where: YCX is the abnormality judgment coefficient, and Y is the first threshold.

[0198] In this embodiment: the first comparison result generated by the first main analysis unit 71 can quickly identify abnormal situations in the risk control report. This rapid identification mechanism ensures that the enterprise can take countermeasures at the early stage of the problem, thereby reducing potential losses and maintaining the financial security of the enterprise. The first analysis unit 72 can subdivide the abnormal situation into three levels: level one, level two, and level three by re-analyzing the first comparison result, providing the enterprise with a more refined abnormal judgment. This grading mechanism helps the management quickly understand the severity of the abnormality and guide subsequent processing measures and resource allocation.

[0199] By providing specific information on graded anomalies, enterprises can formulate targeted risk management strategies. Different levels of anomalies require different processing methods. Accurate analysis can improve the efficiency of risk response and reduce the overall risk level of the enterprise. Comparing the anomalies in the risk control report with the preset thresholds enhances the effectiveness of decision support. Management can quickly adjust decisions based on the analysis results, optimize resource allocation, and ensure that they maintain competitive advantages in a changing market environment. The first data analysis module 7 monitors data in real time to ensure that the system can effectively operate the early warning mechanism. Once an anomaly is detected, the system will issue an alarm in time to help the enterprise take necessary measures quickly to avoid greater losses. Through a systematic analysis process, deviations and errors in manual judgment are reduced, and the accuracy of information processing is improved. This improvement reduces the potential risks caused by improper information processing and improves the overall management level of the enterprise.

[0200] Example 6: Please refer to Figure 1 The second data analysis module 8 includes a second main analysis unit 81 and a secondary analysis unit 82, the second main analysis unit 81 is used to generate a second comparison result, and the secondary analysis unit 82 is used to re-analyze the second comparison result;

[0201] The second comparison results are as follows;

[0202] When YCC≤YC, it means that there is no abnormality in the financial data. When YCC>YC, it means that there is an abnormality in the financial data.

[0203] When YXC≤YX, it means that there is no abnormality in the credit data; when YCC>YX, it means that there is an abnormality in the credit data;

[0204] When YSC≤YS, it means that there is no abnormality in the market data, and when YCC>YS, it means that there is an abnormality in the market data report;

[0205] The reanalysis results are as follows;

[0206] When YCJ ≤ 60% of YC, it represents a first-level anomaly in risk control. When 60% of YC < YCJ ≤ 90% of YC, it represents a second-level anomaly in risk control. When YCJ > 90% of YC, it represents a third-level anomaly in risk control;

[0207] When YXJ ≤ 60% of YX, it represents a first-level anomaly in risk control. When 60% of YX < YXJ ≤ 90% of YX, it represents a second-level anomaly in risk control. When YXJ > 90% of Y, it represents a third-level anomaly in risk control;

[0208] When YSJ ≤ 60% of YS, it represents a first-level anomaly in risk control. When 60% of YS < YSJ ≤ 90% of YS, it represents a second-level anomaly in risk control. When YSJ > 90% of YS, it represents a third-level anomaly in risk control;

[0209] The anomaly magnitude values of financial data YCJ, credit data YXJ, and market data YSJ are obtained by calculating through the following formulas respectively;

[0210]

[0211] In the formula: YCC is the financial data anomaly coefficient, YXC is the credit data anomaly coefficient, YSC is the market data anomaly coefficient, YC is the financial data anomaly threshold, YX is the credit data anomaly threshold, and YS is the market data anomaly threshold.

[0212] In this embodiment: The second main analysis unit 81 can perform independent anomaly detection on financial data, credit data, and market data, and generate a second comparison result. By analyzing multiple data dimensions respectively, it ensures that the enterprise can comprehensively understand the potential risks in all fields and avoid missing anomalies in a certain key field. This all-round detection method significantly improves the comprehensiveness of the risk control report.

[0213] The second analysis unit 82 can perform hierarchical analysis on data anomalies in different dimensions, classify the anomalies into first-level, second-level, and third-level, and perform precise classification. The multi-level analysis provides the enterprise with deeper insights, helps the management understand the specific scope and severity of the anomalies, and thus adjusts the strategy and resource allocation more precisely.

[0214] By independently analyzing data in different dimensions, the enterprise can locate the specific source of the anomaly. Whether it is an anomaly in financial data, or fluctuations in credit or market data, the system can quickly identify and mark it. This function helps the enterprise focus more when dealing with risk control problems and reduces the waste of time and resources.

[0215] The second data analysis module 8 can correlate the impacts between different data through data comparison and re-analysis in multiple dimensions. For example, fluctuations in market data may cause credit or financial anomalies. This correlation analysis function helps enterprises identify potential systemic risks, thereby taking more comprehensive risk control measures to prevent local problems from spreading into global crises. Through precise anomaly detection and graded analysis, management can make reliable decisions based on detailed data and clear anomaly reports. Regardless of whether it is financial, credit or market-level risks, the system can provide clear anomalies and their impact levels, thereby helping management to quickly take countermeasures and reduce decision-making uncertainty. The graded anomaly analysis results can help enterprises optimize emergency response mechanisms. For lower-level anomalies, enterprises can make routine adjustments, while for severe anomalies, emergency measures need to be initiated quickly. This flexible risk management mechanism can help enterprises better respond to sudden risks and improve the flexibility and response speed of overall risk control.

[0216] This application also includes a risk control report analysis method based on AI intelligence, please refer to Figure 2 , the specific steps are as follows:

[0217] S1. Perform multi-source extraction on the financial data, credit data and market data in the risk control report to form a first multi-source data set, and extract the financial data, credit data and market data in the risk control report to form a second multi-source data set;

[0218] S2, preprocessing the first multi-source data set, including data cleaning, denoising, missing value filling and normalization. The processed data will be rearranged into: a first data set, a second data set, and a third data set. The abnormality judgment coefficient YCX is calculated for the preprocessed first data set, the second data set, and the third data set;

[0219] S3. Compare the generated abnormality judgment coefficient YCX with the preset first threshold value Y to determine whether there is an abnormality in the risk control report. If the first comparison result shows that there is an abnormality, the abnormality judgment coefficient YCX and the first threshold value Y are integrated and calculated again to generate an abnormality magnitude value. By comparing the abnormality magnitude value with the first threshold value Y, an abnormality magnitude result of the risk control report is generated to determine the severity of the abnormality.

[0220] S4. After the risk control report is determined to be abnormal, the fourth data set, the fifth data set, and the sixth data set are calculated respectively to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC, and the market data abnormality coefficient YSC, which are used to refine the location and analysis of the abnormal position;

[0221] S5. Compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC, and the market data anomaly coefficient YSC with the preset corresponding anomaly thresholds to generate a second comparison result. According to the comparison result, locate the location of the anomaly and the specific anomaly type, integrate and calculate the located anomaly coefficient with the corresponding anomaly threshold, generate a corresponding anomaly magnitude value, and compare it with the threshold again to output the anomaly magnitude result;

[0222] S6. Send each comparison result and abnormal magnitude result to the terminal for the terminal user to view, analyze and make decisions.

[0223] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

[0224] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An AI-based risk control report analysis system, characterized by: It comprises a first data extraction module (1), a second data extraction module (2), a first data processing module (3), a second data processing module (4), a first data calculation module (5), a second data calculation module (6), a first data analysis module (7) and a second data analysis module (8); The first data extraction module (1) and the second data extraction module (2) are used to extract financial data, credit data and market data in the risk control report, thereby forming a first multi-source data set and a second multi-source data set respectively; The first data processing module (3) is used to preprocess the first multi-source data set, and reorganize the preprocessed first multi-source data set into a first data set, a second data set and a third data set; The first data calculation module (5) is used to integrate and calculate the first data set, the second data set and the third data set, so as to generate an abnormality judgment coefficient YCX; The first data analysis module (7) is used to compare the abnormality judgment coefficient YCX with a preset first threshold value Y, thereby generating a first comparison result, and judging whether the risk control report is abnormal according to the first comparison result. If the first comparison result shows that the risk control report is abnormal, the abnormality judgment coefficient YCX is integrated and calculated with the first threshold value Y to generate a risk control abnormality magnitude value YLJ, and the abnormal magnitude value is compared with the first threshold value Y again, thereby generating a risk control report abnormality magnitude result; When the first comparison result of the first data analysis module (7) shows that the risk control report is abnormal, the second data processing module (4) preprocesses the second multi-source data set and reorganizes the preprocessed second multi-source data set into a fourth data set, a fifth data set and a sixth data set; The second data calculation module (6) is used to calculate the fourth data set, the fifth data set and the sixth data set respectively, so as to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC and the market data abnormality coefficient YSC respectively; The second data analysis module (8) is used to compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC and the market data anomaly coefficient YSC with the preset financial data anomaly threshold YC, the credit data anomaly threshold YX and the market data anomaly threshold YS, respectively, so as to generate a second comparison result, locate the abnormal position in the risk control report according to the second comparison result, and integrate and calculate the corresponding anomaly coefficient and the corresponding anomaly threshold according to the specific abnormal position located, so as to generate a corresponding anomaly magnitude value, and compare the calculated anomaly magnitude value with the corresponding anomaly threshold again, so as to generate an anomaly magnitude result; The feedback module (9) is used to send each comparison result and magnitude result to the terminal.

2. According to claim 1, the risk control report analysis system based on AI intelligence is characterized by: The first data extraction module (1) comprises a first financial data extraction unit (11), a first credit data extraction unit (12) and a first market data extraction unit (13); the second data extraction module (2) comprises a second financial data extraction unit (21), a second credit data extraction unit (22) and a second market data extraction unit (23); The first financial data extraction unit (11) and the second financial data extraction unit (21) are respectively used to extract financial data from the risk control report, wherein the first financial data extraction unit (11) is used to extract the cash flow ratio, net profit and return on assets from the risk control report; and the second financial data extraction unit (21) is used to extract the cash flow from operating activities, the cash flow from investment activities and the cash flow from financing activities from the risk control report; The first credit data extraction unit (12) and the second credit data extraction unit (22) are respectively used to extract credit data from the risk control report, wherein the first credit data extraction unit (12) is used to extract the credit rating change rate, overdue loan ratio and default rate from the risk control report, and the second credit data extraction unit (22) is used to extract short-term debt, long-term debt and reserve coverage ratio from the risk control report; The first market data extraction unit (13) and the second market data extraction unit (23) are respectively used to extract market data in the risk control report, wherein the first market data extraction unit (13) is used to extract interest rate fluctuations, exchange rate fluctuations and industry average return change rate in the risk control report, and the second market data extraction unit (23) is used to extract operating efficiency, market cost and supply chain risk; The first multi-source data set includes cash flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate fluctuation, exchange rate fluctuation and industry average earnings change rate; The second multi-source data set includes cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage ratio, operating efficiency, market cost and supply chain risk.

3. The AI-based risk control report analysis system according to claim 2 is characterized by: The first data processing module (3) is used to preprocess the data contained in the first multi-source data set, and to make the preprocessed data dimensionless, and to reorganize the preprocessed data into a first data set, a second data set, and a third data set; The first data set includes cash flow ratio, net profit, and return on assets; The cash flow ratios are recorded as Z1, Z2, Z3, ..., Zn according to the timestamps; Net profit is recorded in months according to timestamps as J1, J2, J3, ..., Jn; The asset returns are recorded as C1, C2, C3, ..., Cn according to the timestamp; The second data set includes the credit rating change rate, the percentage of overdue loans, and the default rate; The rating change rates are recorded as P1, P2, P3, ..., Pn according to the timestamps; The overdue loan ratios are recorded as Y1, Y2, Y3, ..., Yn according to the timestamps; The default rates are recorded as W1, W2, W3, ..., Wn according to the timestamps; The third data set includes interest rate fluctuations, exchange rate fluctuations, and industry average earnings change rates; The interest rate fluctuations are recorded as L1, L2, L3, ..., Ln according to the timestamps; The exchange rate fluctuations are recorded as H1, H2, H3, ..., Hn according to the timestamps; The industry average return change rate is recorded as S1, S2, S3, ..., Sn according to the timestamp; The second data processing module (4) is used to pre-process the data contained in the second multi-source data set, and to make the pre-processed data dimensionless, and to reorganize the pre-processed data into a fourth data set, a fifth data set, and a sixth data set; The fourth data set includes cash flows from operating activities, investing activities, and financing activities; The cash flows from operating activities are recorded as B1, B2, B3, ..., Bn according to the timestamps; The cash flows from investment activities are recorded as T1, T2, T3, ..., Tn according to the timestamps; The cash flows from financing activities are recorded as D1, D2, D3, ..., Dn according to the timestamps; The fifth data set includes short-term debt, long-term debt, and reserve coverage ratios; Short-term debts are recorded as E1, E2, E3, ..., En according to the timestamp; Long-term debt is recorded as F1, F2, F3, ..., Fn according to the time stamp; The reserve coverage ratios are recorded as G1, G2, G3, ..., Gn according to the timestamps; The sixth data set includes operational efficiency, market costs, and supply chain risks; The operational efficiency is recorded as I1, I2, I3, ..., In according to the timestamp; The market costs are recorded as K1, K2, K3, ..., Kn according to the timestamp; Supply chain risks are recorded as M1, M2, M3, ..., Mn according to timestamps.

4. According to claim 3, the risk control report analysis system based on AI intelligence is characterized by: The first data calculation module (5) calculates and obtains the abnormality judgment coefficient YCX through the following formula: YCX=a1×CK1+a2×CK2+a3×CK3; In the formula: CK1 is the first reference coefficient, which is obtained by integrating and calculating the first data set; CK2 is the second reference coefficient, which is obtained by integrating and calculating the second data set; CK3 is the third reference coefficient, which is obtained by integrating and calculating the third data set; a1, a2 and a3 are weight values, and the values ​​of a1, a2 and a3 are adjusted and set by the user.

5. According to claim 4, the risk control report analysis system based on AI intelligence is characterized by: The first data calculation module (5) calculates and obtains the first reference coefficient CK1, the second reference coefficient CK2 and the third reference coefficient CK3 respectively through the following formulas; CK1=b1×CZ+b1×CJ+b1×CC; CK2=b1×CP+b1×CY+b1×CW; CK3 = b1 × CL + b1 × CH + b1 × CS; Where: CZ is the intervention value of cash flow ratio, CJ is the intervention value of net profit, CC is the intervention value of return on assets, CP is the intervention value of credit rating change rate, CY is the intervention value of overdue loan ratio, CW is the intervention value of default rate, CL is the intervention value of interest rate fluctuation, CH is the intervention value of exchange rate fluctuation, and CS is the intervention value of industry average return change rate, which are calculated and obtained by integrating the first data set, the second data set and the third data set respectively; Zn, Jn, Cn, Pn, Yn, Wn, Ln, Hn, and Sn are the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate at the nth timestamp respectively; Z1, J1, C1, P1, Y1, W1, L1, H1, and S1 are the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate at the first timestamp respectively; Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax, Smax are the maximum values of the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate respectively; Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin, Smin are the minimum values of the capital flow ratio, net profit, return on assets, credit rating change rate, overdue loan ratio, default rate, interest rate volatility, exchange rate volatility, and industry average earnings change rate respectively; b1, b2, and b3 are weight values, and the values of b1, b2, and b3 are adjusted and set by the user.

6. The AI-based risk control report analysis system according to claim 5 is characterized by: The first data analysis module (7) alarms the first main analysis unit (71) and the first analysis unit (72). The first main analysis unit (71) is used to generate the first comparison result, and the first analysis unit (72) is used to re-analyze the first comparison result; The first comparison result is as follows; When YCX ≤ Y, it means that the risk control report is normal. When YCX > Y, it means that the risk control report is abnormal; The re-analysis result is as follows; When YLJ ≤ Y × 60%, it means that the risk control has a first-level abnormality. When Y × 60% < YLJ ≤ Y × 90%, it means that the risk control has a second-level abnormality. When YLJ > Y × 90%, it means that the risk control has a third-level abnormality; The risk control abnormality magnitude value YLJ is obtained by calculating through the following formula; In the formula: YCX is the abnormality judgment coefficient, and Y is the first threshold.

7. The AI-based risk control report analysis system according to claim 6 is characterized by: The second data calculation module (6) calculates and obtains the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC, and the market data abnormality coefficient YSC respectively through the following formula; In the formula: Bn, Tn, Dn, En, Fn, Gn, In, Kn, and Mn are the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage rate, operating efficiency, market cost, and supply chain risk at the nth timestamp respectively; Bp, Tp, Dp, Ep, Fp, Gp, Ip, Kp, and Mp are the average values of the cash flow from operating activities, cash flow from investing activities, cash flow from financing activities, short-term debt, long-term debt, reserve coverage rate, operating efficiency, market cost, and supply chain risk respectively; d1, d2, and d3 are weight values, and the values of d1, d2, and d3 are adjusted and set by the user.

8. The AI-based risk control report analysis system according to claim 7 is characterized by: The second data calculation module (6) calculates and obtains the average operating cash flow Bp, the average investment cash flow Tp, the average financing cash flow Dp, the average short-term debt Ep, the average long-term debt Fp, the average reserve coverage rate Gp, the average operating efficiency Ip, the average market cost Kp, and the average supply chain risk Mp respectively through the following formulas; Where: Bn, Tn, Dn, En, Fn, Gn, In, Kn, and Mn are the operating cash flow, investment cash flow, financing cash flow, short-term debt, long-term debt, reserve coverage rate, operating efficiency, market cost, and supply chain risk at the nth timestamp respectively; B1, T1, D1, E1, F1, G1, I1, K1, and M1 are the operating cash flow, investment cash flow, financing cash flow, short-term debt, long-term debt, reserve coverage rate, operating efficiency, market cost, and supply chain risk at the first timestamp respectively.

9. The AI-based risk control report analysis system according to claim 8 is characterized by: The second data analysis module (8) includes a second main analysis unit (81) and a second secondary analysis unit (82). The second main analysis unit (81) is used to generate a second comparison result, and the second secondary analysis unit (82) is used to re-analyze the second comparison result; The second comparison result is as follows; When YCC ≤ YC, it represents that the financial data is normal. When YCC > YC, it represents that the financial data is abnormal; When YXC ≤ YX, it represents that the credit data is normal. When YCC > YX, it represents that the credit data is abnormal; When YSC ≤ YS, it represents that the market data is normal. When YCC > YS, it represents that the market data report is abnormal; The re-analysis result is as follows; When YCJ ≤ YC × 60%, it represents that the risk control has a first-level abnormality. When YC × 60% < YCJ ≤ YC × 90%, it represents that the risk control has a second-level abnormality. When YCJ > YC × 90%, it represents that the risk control has a third-level abnormality; When YXJ ≤ YX × 60%, it represents that the risk control has a first-level abnormality. When YX × 60% < YXJ ≤ YX × 90%, it represents that the risk control has a second-level abnormality. When YXJ > Y × 90%, it represents that the risk control has a third-level abnormality; When YSJ ≤ YS × 60%, it represents that the risk control has a first-level abnormality. When YS × 60% < YSJ ≤ YS × 90%, it represents that the risk control has a second-level abnormality. When YSJ > YS × 90%, it represents that the risk control has a third-level abnormality; The abnormal magnitude values YCJ of the financial data, the abnormal magnitude values YXJ of the credit data, and the abnormal magnitude values YSJ of the market data are calculated and obtained respectively through the following formulas; Where: YCC is the abnormal coefficient of the financial data, YXC is the abnormal coefficient of the credit data, YSC is the abnormal coefficient of the market data, YC is the abnormal threshold of the financial data, YX is the abnormal threshold of the credit data, and YS is the abnormal threshold of the market data.

10. A risk control report analysis method based on AI intelligence, characterized by: The specific steps are as follows: S1. Perform multi-source extraction on the financial data, credit data and market data in the risk control report to form a first multi-source data set, and extract the financial data, credit data and market data in the risk control report to form a second multi-source data set; S2, preprocessing the first multi-source data set, including data cleaning, denoising, missing value filling and normalization. The processed data will be rearranged into: a first data set, a second data set, and a third data set. The abnormality judgment coefficient YCX is calculated for the preprocessed first data set, the second data set, and the third data set; S3. Compare the generated abnormality judgment coefficient YCX with the preset first threshold value Y to determine whether there is an abnormality in the risk control report. If the first comparison result shows that there is an abnormality, the abnormality judgment coefficient YCX and the first threshold value Y are integrated and calculated again to generate an abnormality magnitude value. By comparing the abnormality magnitude value with the first threshold value Y, an abnormality magnitude result of the risk control report is generated to determine the severity of the abnormality. S4. After the risk control report is determined to be abnormal, the fourth data set, the fifth data set, and the sixth data set are calculated respectively to generate the financial data abnormality coefficient YCC, the credit data abnormality coefficient YXC, and the market data abnormality coefficient YSC, which are used to refine the location and analysis of the abnormal position; S5. Compare the financial data anomaly coefficient YCC, the credit data anomaly coefficient YXC, and the market data anomaly coefficient YSC with the preset corresponding anomaly thresholds to generate a second comparison result. According to the comparison result, locate the location of the anomaly and the specific anomaly type, integrate and calculate the located anomaly coefficient with the corresponding anomaly threshold, generate a corresponding anomaly magnitude value, and compare it with the threshold again to output the anomaly magnitude result; S6. Send each comparison result and abnormal magnitude result to the terminal for the terminal user to view, analyze and make decisions.

Images