Financial reconciliation system and method based on AI intelligence

Abstract

The invention relates to the technical field of data processing, and discloses a financial reconciliation system and method based on AI intelligence, and the system and method achieve the comprehensive collection, analysis, calculation and analysis of financial data through the cooperative work of a plurality of modules. The system greatly improves the automation and accuracy of data processing, reduces the risk of human errors, improves the efficiency of anomaly detection, enables enterprises to recognize and cope with financial risks in time and optimize fund flow and management strategies through rapid generation of anomaly judgment and positioning, improves the financial transparency of the enterprises, and improves the safety of the enterprises. In addition, data support is provided for decision making, sustainable and healthy development of enterprises is promoted, and the system can provide a more quantitative and objective anomaly evaluation mode by introducing a calculation formula of the financial anomaly magnitude CYL.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an AI-based financial reconciliation system and method. Background Art

[0002] At present, the reconciliation between banks and enterprises is based on the calculation of relevant expense data by the banks and enterprises respectively, and then reconciliation is carried out according to the calculation results. It is mainly done by manually configuring reconciliation rules and converting the format of the business data to be accounted for to generate reconciliation data. Then, the reconciliation processing is carried out on each reconciliation data according to the configured reconciliation rules.

[0003] Although financial management technology is becoming increasingly advanced, traditional manual reconciliation and data analysis still have many shortcomings. First, manual data processing is prone to human errors, resulting in inaccurate data and affecting the reliability of decision-making. Second, traditional methods are often difficult to quickly and comprehensively identify financial anomalies and cannot respond to potential risks in a timely manner. In addition, the lack of quantitative anomaly assessment means that the financial team often relies on subjective judgment when dealing with anomalies, which increases the ambiguity of decision-making.

[0004] Therefore, we proposed an AI-based financial reconciliation system and method to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a financial reconciliation system and method based on AI intelligence to solve the problem that manual data processing proposed in the above background technology is prone to human errors, resulting in inaccurate data, affecting the reliability of decision-making, and traditional methods are often difficult to quickly and comprehensively identify financial anomalies and cannot respond to potential risks in a timely manner. In addition, the lack of quantitative anomaly assessment means makes the financial team often rely on subjective judgment when dealing with anomalies, which increases the ambiguity of decision-making.

[0006] To achieve the above object, the present invention provides the following technical solutions: a financial reconciliation system based on AI intelligence, comprising a data acquisition module, a data parsing module, a first data calculation module, a first data analysis module, a second data calculation module, a second data analysis module and a feedback module;

[0007] The data acquisition module is used to acquire data and integrate the acquired data into a multi-source data set;

[0008] The data parsing module is used to process the multi-source data sets and reorganize them into a first data set, a second data set, a third data set and a fourth data set;

[0009] The first data calculation module is used to integrate and calculate the first data set, the second data set, the third data set and the fourth data set, so as to generate an abnormality judgment value CYC;

[0010] The first data analysis module is used to compare the financial anomaly judgment value CYC with a preset first threshold value Y, thereby generating a first comparison result, and judging whether an account is abnormal according to the first comparison result. If the first comparison result shows that an abnormality has occurred, the abnormality judgment value and the preset first threshold value Y are integrated and calculated to generate a financial anomaly magnitude value, and the financial anomaly magnitude value is re-compared with the first threshold value Y to generate a first re-comparison result, and the abnormality level of the current finance is judged according to the first re-comparison result;

[0011] The second data calculation module is used to integrate and calculate the first data set, the second data set, the third data set and the fourth data set respectively, so as to generate a first abnormality location value DY, a second abnormality location value DR, a third abnormality location value DS and a fourth abnormality location value DZ;

[0012] The second data analysis module is used to compare the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ with the preset first abnormal location threshold value YY, the second abnormal location threshold value RY, the third abnormal location threshold value SY and the fourth abnormal location threshold value ZY, respectively, so as to generate a second comparison result, and locate the abnormal area of ​​the account according to the second comparison result, and integrate and calculate the abnormal location value corresponding to the abnormal area with the corresponding abnormal location threshold, so as to generate a corresponding abnormal location magnitude value, and re-compare the generated abnormal location magnitude value with the corresponding abnormal location threshold, so as to generate a second re-comparison result, and judge the abnormal level of the corresponding area according to the second re-comparison result;

[0013] The feedback module is used to feed back multiple data to the terminal.

[0014] Preferably, the data collection module includes a transaction data collection unit, an enterprise system data collection unit, a payment gateway data and an external data collection unit;

[0015] The transaction data collection unit is used to collect enterprise transaction data, including transaction amount, transaction quantity, foreign exchange rate and account balance fluctuation range;

[0016] The enterprise system data collection unit is used to collect enterprise system data, including the total bill amount, invoicing tax rate, receivable amount and bad debt reserve;

[0017] The payment gateway data is used to collect enterprise payment gateway data, including total payment amount, payment success rate, refund amount and transaction processing time;

[0018] The external data collection unit is used to collect external data of the enterprise, including invoice amount, supplier discount rate, preferential amount and invoice tax amount;

[0019] The multi-source data set includes transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation, bill total, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, discount amount and invoice tax amount.

[0020] Preferably, the data analysis module includes a data preprocessing unit and a data sorting unit;

[0021] The data preprocessing unit is used to preprocess and normalize the multi-source data set, and the data sorting unit is used to divide the processed multi-source data set into a first data set, a second data set and a third data set;

[0022] The first data set includes transaction amounts, transaction quantities, foreign exchange rates, and account balance fluctuations;

[0023] The transaction amounts are recorded as A1, A2, A3, ..., An according to the number of transactions;

[0024] The transaction quantity is recorded as B1, B2, B3, ..., Bn according to the number of transactions;

[0025] The foreign exchange rates are recorded as C1, C2, C3, ..., Cn according to the number of transactions;

[0026] The fluctuation range of account balance is recorded as D1, D2, D3, ..., Dn according to the number of transactions;

[0027] The second data set includes the total billing amount, invoicing rate, receivable amount, and bad debt provision;

[0028] The total bill amount is recorded as E1, E2, E3, ..., En according to the number of transactions;

[0029] The invoicing tax rates are recorded as F1, F2, F3, ..., Fn according to the number of transactions;

[0030] The receivable amounts are recorded as G1, G2, G3, ..., Gn according to the number of transactions;

[0031] The bad debt reserves are recorded as H1, H2, H3, ..., Hn according to the number of transactions;

[0032] The third data set includes total payment amount, payment success rate, refund amount, and transaction processing time;

[0033] The total payment amount is recorded as I1, I2, I3, ..., In according to the number of transactions;

[0034] The payment success rates are recorded as J1, J2, J3, ..., Jn according to the number of transactions;

[0035] The refund amount is recorded as K1, K2, K3, ..., Kn according to the number of transactions;

[0036] The transaction processing time is recorded as L1, L2, L3, ..., Ln according to the number of transactions;

[0037] The fourth data set includes the invoice amount, supplier discount rate, preferential amount, and invoice tax amount;

[0038] The invoice amount is recorded as M1, M2, M3, ..., Mn according to the number of transactions;

[0039] The supplier discount rates are recorded as O1, O2, O3, ..., On according to the number of transactions;

[0040] The preferential amount is recorded as P1, P2, P3, ..., Pn according to the number of transactions;

[0041] The invoice tax amount is recorded as Q1, Q2, Q3, ..., Qn according to the number of transactions.

[0042] Preferably, the first data calculation module calculates and obtains the financial abnormality judgment value CYC through the following formula:

[0043]

[0044] Where: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the bill total analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax analysis coefficient;

[0045] a1, a2, a3 and a4 are weight values, and the values ​​of a1, a2, a3 and a4 are adjusted and set by the user.

[0046] Preferably, the first data calculation module calculates and obtains the transaction amount analysis coefficient PA, the transaction quantity analysis coefficient PB, the foreign exchange rate analysis coefficient PC, the account balance fluctuation range analysis coefficient PD, the bill total analysis coefficient PE, the invoicing tax rate analysis coefficient PF, the receivable amount analysis coefficient PG, the bad debt reserve analysis coefficient PH, the total payment amount analysis coefficient PI, the payment success rate analysis coefficient PJ, the refund amount analysis coefficient PK, the transaction processing time analysis coefficient PL, the invoice amount analysis coefficient PM, the supplier discount rate analysis coefficient PO, the preferential amount analysis coefficient PP and the invoice tax amount analysis coefficient PQ respectively through the following formulas;

[0047]

[0048]

[0049]

[0050] Where: An, Bn, Cn, Dn, En, Fn, Gn, Hn, In, Jn, Kn, Ln, Mn, On, Pn and Qn are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the nth transaction;

[0051] A1, B1, C1, D1, E1, F1, G1, H1, I1, J1, K1, L1, M1, O1, P1 and Q1 are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the first transaction;

[0052] Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax and Smax are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount;

[0053] Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin, and Smin are the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount, and invoice tax amount respectively.

[0054] Preferably, the first data analysis module includes a first data comparison unit and a first data re-comparison unit. The first data comparison unit is used to generate a first comparison result, specifically as follows:

[0055] When CYC ≤ Y, it means that the current finance is normal. When CYC > Y, it means that the current finance is abnormal.

[0056] The first data re-comparison unit is used to generate a first re-comparison result, specifically as follows:

[0057] When CYL ≤ Y × 20%, it means that the current finance has a first-level abnormality. When Y × 20% < CYL ≤ Y × 30%, it means that the current finance has a first-level abnormality. When CYL > Y × 30%, it means that the current finance has a first-level abnormality.

[0058] The calculation formula of the financial anomaly judgment value CYL is as follows;

[0059]

[0060] In the formula: CYC is the financial anomaly judgment value, and Y is the first threshold.

[0061] Preferably, the second data calculation module calculates and obtains the first anomaly location value DY, the second anomaly location value DR, the third anomaly location value DS, and the fourth anomaly location value DZ respectively through the following formulas;

[0062]

[0063] In the formula: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the total bill amount analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax amount analysis coefficient.

[0064] Preferably, the second data analysis module includes a second data comparison unit and a second data re-comparison unit;

[0065] The second data comparison unit is used to generate a second comparison result, specifically as follows:

[0066] When DY ≤ YY, it means that the transaction data in the current finance is normal; when DY > YY, it means that the transaction data in the current finance is abnormal;

[0067] When DR ≤ RY, it means that the enterprise system data in the current finance is normal; when DR > RY, it means that the transaction data in the current finance is abnormal;

[0068] When DS ≤ SY, it means that the enterprise payment gateway data in the current finance is normal; when DY > SY, it means that the enterprise payment gateway data in the current finance is abnormal;

[0069] When DZ ≤ ZY, it means that the enterprise external data in the current finance is normal; when DY > SY, it means that the enterprise external data in the current finance is abnormal;

[0070] The second data re-comparison unit is used to generate a second re-comparison result, specifically as follows:

[0071] When SY ≤ YY × 20%, it means that the current transaction has a first-level anomaly; when YY × 20% < SY ≤ JY × 30%, it means that the current finance has a first-level anomaly; when SY > YY × 30%, it means that the current finance has a first-level anomaly;

[0072] When SR ≤ RY × 20%, it means that the current finance has a first-level anomaly; when RY × 20% < SR ≤ RY × 30%, it means that the current finance has a first-level anomaly; when SR > RY × 30%, it means that the current finance has a first-level anomaly;

[0073] When SS ≤ SY × 20%, it means that the current finance has a first-level anomaly; when SY × 20% < SS ≤ SY × 30%, it means that the current finance has a first-level anomaly; when SS > SY × 30%, it means that the current finance has a first-level anomaly;

[0074] When SZ ≤ ZY × 20%, it means that the current finance has a first-level anomaly; when ZY × 20% < SZ ≤ ZY × 30%, it means that the current finance has a first-level anomaly; when SZ > ZY × 30%, it means that the current finance has a first-level anomaly;

[0075] The first anomaly positioning magnitude value SY, the second anomaly positioning magnitude value SR, the third anomaly positioning magnitude value SS, and the fourth anomaly positioning magnitude value SZ are respectively obtained by calculating through the following formulas;

[0076]

[0077] In the formula: DY is the first abnormal location value, DR is the second abnormal location value, DS is the third abnormal location value, DZ is the fourth abnormal location value, YY is the first abnormal location threshold, RY is the second abnormal location threshold, SY is the third abnormal location threshold, and ZY is the fourth abnormal location threshold.

[0078] This application also includes a financial reconciliation method based on AI intelligence, the specific steps are as follows:

[0079] S1. Acquire data through a data acquisition module and integrate the acquired data into a multi-source data set;

[0080] S2, processing the multi-source data sets through a data analysis module, and reorganizing them into a first data set, a second data set, a third data set, and a fourth data set;

[0081] S3, integrating and calculating the first data set, the second data set, the third data set and the fourth data set through the first data calculation module to generate an abnormality judgment value CYC, and comparing the financial abnormality judgment value CYC with a preset first threshold value Y through the first data analysis module to generate a first comparison result, and judging whether the account is abnormal according to the first comparison result;

[0082] S4, integrating and calculating the first data set, the second data set, the third data set and the fourth data set respectively through the second data calculation module, so as to generate the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ, and comparing the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ with the preset first abnormal location threshold value YY, the second abnormal location threshold value RY, the third abnormal location threshold value SY and the fourth abnormal location threshold value ZY respectively through the second data analysis module, so as to generate a second comparison result, and locate the abnormal area of ​​the account according to the second comparison result;

[0083] S5. Feedback multiple data to the terminal through the feedback module.

[0084] Compared with the prior art, the present invention has the following beneficial effects:

[0085] 1. This AI intelligent financial reconciliation system and method realizes the comprehensive collection, parsing, calculation and analysis of financial data through the collaborative work of multiple modules. Compared with traditional financial management methods, the system greatly improves the automation and accuracy of data processing, reduces the risk of human error, and improves the efficiency of anomaly detection. By quickly generating anomaly judgments and positioning, enterprises can identify and respond to financial risks more promptly, optimize capital flows and management strategies. This system not only improves the financial transparency of enterprises, but also provides data support for decision-making, promoting the sustainable and healthy development of enterprises.

[0086] 2. This AI intelligent financial reconciliation system and method, by introducing the calculation formula of the financial anomaly magnitude CYL, the system can provide a more quantitative and objective anomaly assessment method. The CYL value is calculated by the difference between the financial anomaly judgment value CYC and the threshold value Y, ensuring that the degree of each anomaly can be accurately quantified. This multi-level management method enables the financial team to handle minor and serious anomalies more effectively, thereby further optimizing the company's financial management process. After the initial anomaly judgment, the system conducts further analysis through the re-comparison unit and makes a secondary judgment based on the calculation results of CYL. This double comparison mechanism effectively reduces false alarms and ensures that after the initial detection of the anomaly, the system can assess a more accurate financial status based on the magnitude, thereby improving the accuracy of the overall analysis.

[0087] 3. The AI ​​intelligent financial reconciliation system and method can comprehensively evaluate the financial health status by comparing and analyzing multiple data sources through the second data comparison unit. This comprehensive analysis enables enterprises to more comprehensively identify and analyze abnormal situations and avoid misjudgments caused by one-sided analysis. The setting of thresholds allows enterprises to flexibly adjust according to actual business conditions, thereby enhancing the adaptability of the system. This dynamic adjustment can better meet the actual operational needs of the enterprise, ensure the accuracy and flexibility of anomaly detection, and through real-time calculation of the abnormal location magnitude value, the enterprise can obtain feedback on abnormal monitoring in real time. This real-time performance improves the company's response capabilities, allowing the financial team to take quick measures to reduce potential risks. By setting a specific abnormal location magnitude value calculation formula, the assessment of financial anomalies is more objective and quantitative. This quantitative analysis provides data support for financial decision-making, reduces the possibility of subjective judgment, and improves the scientific nature of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a system flow chart of the present invention.

[0089] Figure 2 It is a diagram of the method steps of the present invention.

[0090] In the figure: 1. Data acquisition module; 11. Transaction data acquisition unit; 12. Enterprise system data acquisition unit; 13. Payment gateway data acquisition unit; 14. External data acquisition unit; 2. Data analysis module; 21. Data preprocessing unit; 22. Data sorting unit; 3. First data calculation module; 4. First data analysis module; 41. First data comparison unit; 42. First data re-comparison unit; 5. Second data calculation module; 6. Second data analysis module; 61. Second data comparison unit; 62. Second data re-comparison unit; 7. Feedback module. DETAILED DESCRIPTION

[0091] 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.

[0092] Example 1: Please refer to Figure 1 , a financial reconciliation system based on AI intelligence, including a data acquisition module 1, a data parsing module 2, a first data calculation module 3, a first data analysis module 4, a second data calculation module 5, a second data analysis module 6 and a feedback module 7;

[0093] The data acquisition module 1 is used to acquire data and integrate the acquired data into a multi-source data set;

[0094] The data analysis module 2 is used to process the multi-source data sets and reorganize them into a first data set, a second data set, a third data set and a fourth data set;

[0095] The first data calculation module 3 is used to perform integrated calculation on the first data set, the second data set, the third data set and the fourth data set, so as to generate an abnormality judgment value CYC;

[0096] The first data analysis module 4 is used to compare the financial anomaly judgment value CYC with the preset first threshold value Y, so as to generate a first comparison result, and judge whether the account is abnormal according to the first comparison result. If the first comparison result shows that the account is abnormal, the abnormal judgment value and the preset first threshold value Y are integrated and calculated to generate a financial anomaly magnitude value, and the financial anomaly magnitude value is re-compared with the first threshold value Y to generate a first re-comparison result, and the abnormal level of the current finance is judged according to the first re-comparison result;

[0097] The second data calculation module 5 is used to integrate and calculate the first data set, the second data set, the third data set and the fourth data set, respectively, so as to generate a first abnormality location value DY, a second abnormality location value DR, a third abnormality location value DS and a fourth abnormality location value DZ;

[0098] The second data analysis module 6 is used to compare the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ with the preset first abnormal location threshold value YY, the second abnormal location threshold value RY, the third abnormal location threshold value SY and the fourth abnormal location threshold value ZY, respectively, so as to generate a second comparison result, and locate the abnormal area of ​​the account according to the second comparison result, and integrate and calculate the abnormal location value corresponding to the abnormal area with the corresponding abnormal location threshold, so as to generate a corresponding abnormal location magnitude value, and re-compare the generated abnormal location magnitude value with the corresponding abnormal location threshold, so as to generate a second re-comparison result, and judge the abnormal level of the corresponding area according to the second re-comparison result;

[0099] The feedback module 7 is used to feed back multiple data to the terminal.

[0100] In this embodiment: the main function of the data acquisition module 1 is to obtain financial data from multiple data sources and integrate these data into a multi-source data set. This process ensures that the system can obtain comprehensive, accurate and real-time financial information, including bank transaction data, internal enterprise records, payment gateway data, etc. Through the integration of multi-source data, the system can more comprehensively reflect the financial status of the enterprise and lay a solid foundation for subsequent analysis and calculation.

[0101] Data analysis module 2 is responsible for processing and organizing multi-source data sets, converting raw data into usable first data sets, second data sets, third data sets, and fourth data sets. This process includes data cleaning, formatting, and classification to ensure data consistency and accuracy. Through data analysis, the system can extract key financial indicators, provide necessary data support for subsequent calculations, and improve the efficiency of data use.

[0102] The first data calculation module 3 performs comprehensive calculation on the integrated first data set, second data set, third data set and fourth data set to generate an abnormality judgment value CYC. This calculation process can quickly identify potential financial anomalies through the set algorithm and model. By generating CYC, the system can achieve rapid abnormality warning, improve the timeliness and effectiveness of financial monitoring, and reduce the workload of manual review.

[0103] The first data analysis module 4 compares the generated abnormality judgment value CYC with the preset first threshold value Y, thereby generating a first comparison result. This comparison process enables the system to determine whether there is an abnormality in the account. Once an abnormality is found, the system further calculates the financial abnormality magnitude value and re-compares it to determine the abnormality level. This analysis not only improves the accuracy of abnormality detection, but also helps enterprises quickly locate problems and provide support for decision-making.

[0104] The second data calculation module 5 performs independent integration calculations on each data set to generate the first abnormality location value DY, the second abnormality location value DR, the third abnormality location value DS and the fourth abnormality location value DZ. By refining the abnormality location, the module can accurately identify data anomalies from different sources or types, which is convenient for enterprises to prescribe the right medicine and optimize the financial management process, thereby improving the overall financial transparency and risk control capabilities.

[0105] The second data analysis module 6 compares the generated abnormal location value with the preset threshold value to generate a second comparison result to help identify the specific abnormal area. This analysis process ensures the accurate location of accounting anomalies, further integrates and calculates the abnormal location magnitude value, and re-compares it to determine the abnormal level. Through such analysis, enterprises can take timely measures to reduce potential financial risks and improve financial management efficiency.

[0106] Feedback module 7 is used to feed back multiple data to the terminal, so that financial management personnel can obtain system analysis results and abnormal situations in real time. This function improves the convenience of data visualization and operation, allowing financial personnel to respond and make decisions quickly. At the same time, the feedback mechanism enhances the interactivity and user experience of the system, ensuring that enterprises can quickly take countermeasures when faced with abnormalities.

[0107] The AI ​​intelligent financial reconciliation system realizes the comprehensive collection, parsing, calculation and analysis of financial data through the collaborative work of multiple modules. Compared with traditional financial management methods, the system greatly improves the automation and accuracy of data processing, reduces the risk of human error, and improves the efficiency of anomaly detection. By quickly generating anomaly judgments and positioning, enterprises can identify and respond to financial risks more promptly, optimize capital flow and management strategies. This system not only improves the financial transparency of enterprises, but also provides data support for decision-making, promoting the sustainable and healthy development of enterprises.

[0108] Example 2: Please refer to Figure 1 , the data collection module 1 includes a transaction data collection unit 11, an enterprise system data collection unit 12, a payment gateway data collection unit 13 and an external data collection unit 14;

[0109] The transaction data collection unit 11 is used to collect enterprise transaction data, including transaction amount, transaction quantity, foreign exchange rate and account balance fluctuation range;

[0110] The enterprise system data collection unit 12 is used to collect enterprise system data, including the total bill amount, invoicing tax rate, receivable amount and bad debt reserve;

[0111] The payment gateway data collection unit 13 is used to collect enterprise payment gateway data, including total payment amount, payment success rate, refund amount and transaction processing time;

[0112] The external data collection unit 14 is used to collect external data of the enterprise, including invoice amount, supplier discount rate, preferential amount and invoice tax amount;

[0113] The multi-source data set includes transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, discount amount, and invoice tax amount.

[0114] In this embodiment: through multiple collection units, the system not only collects the core transaction data within the enterprise, but also obtains the key financial data, payment gateway data and external invoice-related data in the enterprise system. This all-round data collection enables the enterprise to obtain financial data from more dimensions and different sources, thereby improving the integrity and accuracy of the data.

[0115] The system is not limited to single transaction data, but also integrates the company's financial system information and external data resources. By introducing specific financial indicators such as invoicing tax rates, bad debt reserves, and supplier discount rates, the system can classify and analyze data more finely. This provides a more refined data foundation for subsequent calculations and anomaly analysis, supporting more complex financial models and predictive analysis. Data collected by multiple acquisition units can be aggregated into a comprehensive multi-source data set, including transaction, financial, payment, and external data. This multi-source data set integrates different types of key financial information, providing multi-dimensional support for subsequent analysis and calculations. This data integration capability can improve the system's anomaly detection accuracy and help companies identify anomalies or potential risks more effectively.

[0116] By collecting data from more dimensions, companies can use the system to conduct more comprehensive financial analysis and combine different data sources for correlation analysis, which can help companies identify potential financial risks and optimize their financial strategies and capital management decisions.

[0117] Example 3: Please refer to Figure 1 , the data analysis module 2 includes a data preprocessing unit 21 and a data sorting unit 22;

[0118] The data preprocessing unit 21 is used to preprocess and normalize the multi-source data set, and the data sorting unit 22 is used to divide the processed multi-source data set into a first data set, a second data set, and a third data set;

[0119] The first data set includes transaction amounts, transaction quantities, foreign exchange rates, and account balance fluctuations;

[0120] The transaction amounts are recorded as A1, A2, A3, ..., An according to the number of transactions;

[0121] The transaction quantity is recorded as B1, B2, B3, ..., Bn according to the number of transactions;

[0122] The foreign exchange rates are recorded as C1, C2, C3, ..., Cn according to the number of transactions;

[0123] The fluctuation range of account balance is recorded as D1, D2, D3, ..., Dn according to the number of transactions;

[0124] The second data set includes the total billing amount, invoicing rate, receivable amount, and bad debt provision;

[0125] The total bill amount is recorded as E1, E2, E3, ..., En according to the number of transactions;

[0126] The invoicing tax rates are recorded as F1, F2, F3, ..., Fn according to the number of transactions;

[0127] The receivable amounts are recorded as G1, G2, G3, ..., Gn according to the number of transactions;

[0128] The bad debt reserves are recorded as H1, H2, H3, ..., Hn according to the number of transactions;

[0129] The third data set includes total payment amount, payment success rate, refund amount, and transaction processing time;

[0130] The total payment amount is recorded as I1, I2, I3, ..., In according to the number of transactions;

[0131] The payment success rates are recorded as J1, J2, J3, ..., Jn according to the number of transactions;

[0132] The refund amount is recorded as K1, K2, K3, ..., Kn according to the number of transactions;

[0133] The transaction processing time is recorded as L1, L2, L3, ..., Ln according to the number of transactions;

[0134] The fourth data set includes the invoice amount, supplier discount rate, preferential amount, and invoice tax amount;

[0135] The invoice amount is recorded as M1, M2, M3, ..., Mn according to the number of transactions;

[0136] The supplier discount rates are recorded as O1, O2, O3, ..., On according to the number of transactions;

[0137] The preferential amount is recorded as P1, P2, P3, ..., Pn according to the number of transactions;

[0138] The invoice tax amount is recorded as Q1, Q2, Q3, ..., Qn according to the number of transactions.

[0139] In this embodiment: the data preprocessing unit 21 preprocesses and normalizes the multi-source data set to ensure the consistency and accuracy of various types of data before calculation and analysis. By standardizing data from different sources, the system avoids errors and mismatches caused by different data formats, units or dimensions. This improvement greatly improves the reliability of subsequent calculations and analysis of the system and reduces human intervention and errors. Through the data sorting unit 22, the system scientifically divides the preprocessed multi-source data set into a first data set, a second data set, a third data set and a fourth data set. Each data set is refined and organized according to business logic. For example, the first data set is for transaction information, the second data set focuses on financial indicators, the third data set analyzes payment gateway data, and the fourth data set processes external invoices and supplier data. Such a fine division enables the system to perform more accurate calculations and analyses for different data categories, improving processing efficiency.

[0140] The system marks and numbers the key data points in each data set according to the number of transactions, such as A1, A2, ..., An, etc. This multi-dimensional data organization method provides a structured basis for subsequent mathematical models and complex calculations. This flexible data organization method not only enhances the scalability of the system when processing large amounts of transaction data, but also facilitates in-depth analysis through batch processing or time series analysis, providing enterprises with more in-depth insights and financial forecasting capabilities.

[0141] By organizing data sets of different dimensions such as transaction amount, foreign exchange rate, total bill amount, payment success rate, etc. in an orderly manner, the system can better support association analysis. The correlation between different data sets, such as transaction amount and payment success rate, total bill amount and bad debt reserve, can help the system identify abnormal patterns. In this way, enterprises can quickly locate and analyze the root cause of the problem when potential anomalies or risks are detected, greatly improving the accuracy of anomaly detection. The way different data sets are marked according to the number of transactions enables the system to accurately track the detailed information of each transaction. This refined data processing method provides rich granularity for the company's financial analysis and supports deeper analysis and prediction. For example, by analyzing data such as the processing time and refund amount of multiple transactions, enterprises can optimize processes, improve payment efficiency, and thus improve financial operation strategies.

[0142] Through data preprocessing and organization, the system has significantly improved data quality and processing efficiency, and can flexibly process multi-dimensional and multi-source financial data, ensuring the accuracy and precision of the system's subsequent analysis and calculations. Compared with traditional financial reconciliation systems, this system not only enhances the ability to process big data, but also improves the efficiency of data organization and supports complex correlation analysis and anomaly detection. In this way, enterprises can more accurately identify financial problems, optimize cash flow management, and thus improve overall financial operational efficiency and the intelligence level of management decisions.

[0143] Example 4: Please refer to Figure 1 The first data calculation module 3 calculates and obtains the financial abnormality judgment value CYC through the following formula;

[0144]

[0145] Where: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the bill total analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax analysis coefficient;

[0146] a1, a2, a3 and a4 are weight values, and the values ​​of a1, a2, a3 and a4 are adjusted and set by the user.

[0147] In this embodiment: the formula combines and calculates multiple key financial parameters such as transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation, total bill, invoice tax rate, receivable amount, bad debt reserve, etc., ensuring that the system can comprehensively evaluate the financial status from multiple dimensions. By introducing different analysis coefficients PA, PB, PC, etc., the system can accurately analyze different categories of financial data, thereby improving the comprehensiveness and accuracy of anomaly detection.

[0148] The weight values ​​a1, a2, a3, and a4 in the formula are adjustable. Users can flexibly set each weight according to the actual situation of the enterprise, thereby enhancing the adaptability of the system. This design allows enterprises to adjust the weights of various types of data in anomaly detection for different business scenarios or risk preferences to ensure that the system can meet the needs of the enterprise. By adjusting the weights, enterprises can give priority to certain key indicators and improve the detection capabilities of the system in specific areas.

[0149] The formula effectively reduces false positives in anomaly detection by multiplying the coefficients of each financial parameter and performing weighted calculations after adding multiple factors. In particular, the comprehensive consideration of dynamic factors such as foreign exchange rate fluctuations, bad debt reserves, and payment success rates enables the system to distinguish between normal fluctuations and abnormal situations, thereby more accurately identifying financial anomalies and reducing unnecessary alarms.

[0150] This formula incorporates the company's internal transaction data PA, PB, PC, etc., financial data PE, PF, PG, etc., payment gateway data PI, PJ, PK, etc., and external invoice data PM, PO, PP, etc. into a unified calculation model. This integrated calculation of multi-source data ensures that the company can conduct a unified analysis of each financial data point and accurately identify possible financial anomalies. This comprehensive analysis covering all aspects of financial parameters can help companies better grasp capital flows and risk points.

[0151] Through a set of standardized formulas, the system can process a large amount of financial data in a short period of time, quickly generate anomaly judgment values ​​CYC, and compare them with preset thresholds. This automated calculation reduces manual intervention and improves the efficiency of the system in processing large amounts of data. At the same time, the design of the formula ensures the intelligent level of financial anomaly detection, allowing enterprises to respond to potential financial risks more promptly and accurately.

[0152] Example 5: Please refer to Figure 1, the first data calculation module 3 respectively calculates and obtains the transaction amount analysis coefficient PA, the transaction quantity analysis coefficient PB, the foreign exchange rate analysis coefficient PC, the account balance fluctuation range analysis coefficient PD, the bill total analysis coefficient PE, the invoicing tax rate analysis coefficient PF, the receivable amount analysis coefficient PG, the bad debt reserve analysis coefficient PH, the total payment amount analysis coefficient PI, the payment success rate analysis coefficient PJ, the refund amount analysis coefficient PK, the transaction processing time analysis coefficient PL, the invoice amount analysis coefficient PM, the supplier discount rate analysis coefficient PO, the preferential amount analysis coefficient PP and the invoice tax amount analysis coefficient PQ through the following formulas;

[0153]

[0154]

[0155] Where: An, Bn, Cn, Dn, En, Fn, Gn, Hn, In, Jn, Kn, Ln, Mn, On, Pn and Qn are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the nth transaction;

[0156] A1, B1, C1, D1, E1, F1, G1, H1, I1, J1, K1, L1, M1, O1, P1 and Q1 are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the first transaction;

[0157] Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax and Smax are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount;

[0158] Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin and Smin are respectively transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount.

[0159] In this embodiment: Each analysis coefficient, such as PA, PB, PC, etc., has carefully calculated key financial indicators such as transaction amount, transaction quantity, foreign exchange rate, and the fluctuation range of account balance. The calculation formulas of these coefficients introduce multiple factors such as maximum value, minimum value, and number of transactions, ensuring that the system can dynamically monitor the fluctuations of each transaction data and comprehensively capture the changes in financial data. This enables the system to more accurately detect abnormal fluctuations and potential risks, effectively improving the comprehensiveness of detection.

[0160] The calculation of each coefficient introduces maximum values such as Amax, Bmax, etc. and minimum values such as Amin, Bmin, etc., and combines them with the data of each transaction for analysis. This design allows the system to effectively identify extreme fluctuations in the data, such as sudden increases or decreases, so that when judging financial anomalies, it can better distinguish normal business fluctuations from abnormal events and reduce the possibility of false alarms.

[0161] Each analysis formula is calculated based on the number of transactions n. By considering the change trend of historical data, data normalization and consistency processing are achieved. This method improves the comparability of financial data in different time periods for the system, helps enterprises conduct more accurate comparative analysis of the financial health status at each time point, and further ensures the accuracy of anomaly detection.

[0162] Multiple key parameters such as transaction amount, total bill amount, and payment success rate are integrated into the formula for analysis and calculation, comprehensively evaluating the financial status from multiple perspectives. This multi-dimensional analysis model can more comprehensively evaluate the financial data of enterprises, ensuring that anomaly detection does not rely solely on a single indicator and enhancing the overall risk detection ability of the system.

[0163] Embodiment Six: Please refer to Figure 1 , the first data analysis module 4 includes a first data comparison unit 41 and a first data re-comparison unit 42. The first data comparison unit 41 is used to generate a first comparison result, specifically as follows:

[0164] When CYC ≤ Y, it represents that there is no anomaly in the current finance. When CYC > Y, it represents that there is an anomaly in the current finance;

[0165] The first data re-comparison unit 42 is used to generate a first re-comparison result, specifically as follows:

[0166] When CYL ≤ 20% × Y, it represents that the current finance has a first-level anomaly. When 20% × Y < CYL ≤ 30% × Y, it represents that the current finance has a first-level anomaly. When CYL > 30% × Y, it represents that the current finance has a first-level anomaly;

[0167] The calculation formula of the financial anomaly judgment value CYL is as follows;

[0168]

[0169] Where: CYC is the financial anomaly judgment value, and Y is the first threshold.

[0170] In this embodiment: the first data comparison unit 41 quickly determines whether the financial data is abnormal by comparing it with the preset threshold value Y. When the CYC financial abnormality judgment value is greater than Y, the system immediately identifies the abnormal situation and triggers further analysis. This mechanism provides a real-time abnormality warning function, ensuring that the system can respond to abnormal situations in the first time, reducing the risk of financial problems being ignored or delayed.

[0171] The first data re-comparison unit 42 can classify financial abnormalities into different levels by further analyzing the abnormal magnitude value CYL. When the CYL value falls into different intervals, such as within Y×20%, Y×20%-30%, and greater than Y×30%, the system can make a more accurate judgment based on the severity of the abnormality. This enables the system to not only determine whether an abnormality has occurred, but also to distinguish the severity of the abnormality, making it easier for enterprises to formulate different response strategies based on the abnormality level.

[0172] By introducing the calculation formula of the financial anomaly level CYL, the system can provide a more quantitative and objective anomaly assessment method. The CYL value is calculated by the difference between the financial anomaly judgment value CYC and the threshold value Y, ensuring that the degree of each anomaly can be accurately quantified. This multi-level management method enables the financial team to handle minor and serious anomalies more effectively, thereby further optimizing the company's financial management process.

[0173] After the initial abnormality judgment, the system further analyzes the first data re-comparison unit 42 and performs a secondary judgment in combination with the calculation result of CYL. This double comparison mechanism effectively reduces false positives and ensures that after the initial abnormality is detected, the system can evaluate a more accurate financial status based on the magnitude, thereby improving the accuracy of the overall analysis.

[0174] Example 7: Please refer to Figure 1 The second data calculation module 5 calculates and obtains the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ respectively through the following formulas;

[0175]

[0176] In the formula: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the bill total amount analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax amount analysis coefficient.

[0177] In this embodiment, by calculating the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ respectively, the system can analyze the abnormal situation of financial data from different angles. This multi-dimensional analysis enables the financial team to understand the reasons behind the data fluctuations more comprehensively, so as to locate the problem more accurately.

[0178] In the calculation formula of each abnormal positioning value, the interdependence between parameters such as PA, PB, PC, etc. makes the abnormal analysis not limited to a single indicator. This kind of correlation analysis can reveal the mutual influence between different financial indicators and help enterprises identify potential sources of financial risks. For example, the relationship between the transaction amount and the foreign exchange rate and the fluctuation range of the account balance can reveal deeper abnormalities.

[0179] By dynamically calculating the abnormal positioning value, enterprises can track the changes in financial data in real time and identify potential risk points in a timely manner. This dynamic adaptability enables the system to quickly respond to changes in the market environment and internal business, improving the flexibility and adaptability of financial management.

[0180] The calculation of each abnormal positioning value is based on a clear mathematical model, making the assessment of financial abnormalities quantitative and objective. This quantitative analysis can provide strong support for financial decision-making, helping management make more informed decisions and effectively avoid financial risks.

[0181] Example 8: Please refer to Figure 1 , the second data analysis module 6 includes a second data comparison unit 61 and a second data re-comparison unit 62;

[0182] The second data comparison unit 61 is used to generate a second comparison result, which is as follows:

[0183] When DY≤YY, it means that there is no abnormality in the transaction data in the current finance. When DY>YY, it means that there is an abnormality in the transaction data in the current finance.

[0184] When DR ≤ RY, it means that there is no abnormality in the enterprise system data in the current finance. When DR > RY, it means that there is no abnormality in the transaction data in the current finance;

[0185] When DS ≤ SY, it means that there is no abnormality in the enterprise payment gateway data in the current finance. When DY > SY, it means that there is an abnormality in the enterprise payment gateway data in the current finance;

[0186] When DZ ≤ ZY, it means that there is no abnormality in the enterprise external data in the current finance. When DY > SY, it means that there is an abnormality in the enterprise external data in the current finance;

[0187] The second data re - comparison unit 62 is used to generate a second re - comparison result, specifically as follows:

[0188] When SY ≤ YY × 20%, it means that the current transaction has a first - level abnormality. When YY × 20% < SY ≤ JY × 30%, it means that the current finance has a first - level abnormality. When SY > YY × 30%, it means that the current finance has a first - level abnormality;

[0189] When SR ≤ RY × 20%, it means that the current finance has a first - level abnormality. When RY × 20% < SR ≤ RY × 30%, it means that the current finance has a first - level abnormality. When SR > RY × 30%, it means that the current finance has a first - level abnormality;

[0190] When SS ≤ SY × 20%, it means that the current finance has a first - level abnormality. When SY × 20% < SS ≤ SY × 30%, it means that the current finance has a first - level abnormality. When SS > SY × 30%, it means that the current finance has a first - level abnormality;

[0191] When SZ ≤ ZY × 20%, it means that the current finance has a first - level abnormality. When ZY × 20% < SZ ≤ ZY × 30%, it means that the current finance has a first - level abnormality. When SZ > ZY × 30%, it means that the current finance has a first - level abnormality;

[0192] The first abnormality positioning magnitude value SY, the second abnormality positioning magnitude value SR, the third abnormality positioning magnitude value SS, and the fourth abnormality positioning magnitude value SZ are respectively obtained by calculating through the following formulas;

[0193]

[0194] In the formula: DY is the first abnormality positioning value, DR is the second abnormality positioning value, DS is the third abnormality positioning value, DZ is the fourth abnormality positioning value, YY is the first abnormality positioning threshold, RY is the second abnormality positioning threshold, SY is the third abnormality positioning threshold, and ZY is the fourth abnormality positioning threshold.

[0195] In this embodiment, by setting different thresholds YY, RY, SY, and ZY, abnormal situations can be classified. This classification mechanism enables the financial team to more clearly understand the severity of the abnormality, facilitates the priority treatment of high-risk issues, and thus improves the effectiveness of risk management.

[0196] By comparing and analyzing multiple data sources through the second data comparison unit 61, the financial health status can be comprehensively evaluated. This comprehensive analysis enables the enterprise to more comprehensively identify and analyze abnormal situations, avoiding misjudgments caused by one-sided analysis. The setting of thresholds allows the enterprise to flexibly adjust according to actual business conditions, thereby enhancing the adaptability of the system. This dynamic adjustment can better meet the actual operational needs of the enterprise and ensure the accuracy and flexibility of abnormality detection.

[0197] Through real-time calculation of the abnormal location magnitude value, enterprises can obtain feedback on abnormal monitoring immediately. This real-time performance improves the enterprise's response capabilities, allowing the financial team to take quick measures to reduce potential risks. By setting a specific abnormal location magnitude value calculation formula, the assessment of financial anomalies is more objective and quantitative. This quantitative analysis provides data support for financial decision-making, reduces the possibility of subjective judgment, and improves the scientific nature of decision-making.

[0198] This application also includes a financial reconciliation method based on AI intelligence, please refer to Figure 2 , the specific steps are as follows:

[0199] S1, acquiring data through the data acquisition module 1, and integrating the acquired data into a multi-source data set;

[0200] S2, processing the multi-source data set through the data analysis module 2, and reorganizing it into a first data set, a second data set, a third data set, and a fourth data set;

[0201] S3, integrating and calculating the first data set, the second data set, the third data set and the fourth data set through the first data calculation module 3, thereby generating an abnormality judgment value CYC, and comparing the financial abnormality judgment value CYC with a preset first threshold value Y through the first data analysis module 4, thereby generating a first comparison result, and judging whether the account is abnormal according to the first comparison result;

[0202] S4, integrating and calculating the first data set, the second data set, the third data set and the fourth data set respectively through the second data calculation module 5, so as to generate the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ, and comparing the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ with the preset first abnormal location threshold value YY, the second abnormal location threshold value RY, the third abnormal location threshold value SY and the fourth abnormal location threshold value ZY respectively through the second data analysis module 6, so as to generate a second comparison result, and locate the abnormal area of ​​the account according to the second comparison result;

[0203] S5. Feedback multiple data to the terminal through the feedback module 7.

[0204] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

[0205] 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 financial reconciliation system, characterized by: It comprises a data acquisition module (1), a data parsing module (2), a first data calculation module (3), a first data analysis module (4), a second data calculation module (5), a second data analysis module (6) and a feedback module (7); The data acquisition module (1) is used to acquire data and integrate the acquired data into a multi-source data set; The data analysis module (2) is used to process the multi-source data sets and reorganize them into a first data set, a second data set, a third data set and a fourth data set; The first data calculation module (3) is used to perform integrated calculation on the first data set, the second data set, the third data set and the fourth data set, so as to generate an abnormality judgment value CYC; The first data analysis module (4) is used to compare the financial anomaly judgment value CYC with a preset first threshold value Y, thereby generating a first comparison result, and judging whether an account is abnormal based on the first comparison result. If the first comparison result shows that an abnormality has occurred, the abnormality judgment value and the preset first threshold value Y are integrated and calculated to generate a financial anomaly magnitude value, and the financial anomaly magnitude value is re-compared with the first threshold value Y to generate a first re-comparison result, and judging the abnormality level of the current finance based on the first re-comparison result; The second data calculation module (5) is used to integrate and calculate the first data set, the second data set, the third data set and the fourth data set respectively, so as to generate a first abnormality location value DY, a second abnormality location value DR, a third abnormality location value DS and a fourth abnormality location value DZ; The second data analysis module (6) is used to compare the first abnormal location value DY, the second abnormal location value DR, the third abnormal location value DS and the fourth abnormal location value DZ with the preset first abnormal location threshold value YY, the second abnormal location threshold value RY, the third abnormal location threshold value SY and the fourth abnormal location threshold value ZY respectively, so as to generate a second comparison result, and locate the abnormal area of ​​the account according to the second comparison result, and integrate and calculate the abnormal location value corresponding to the abnormal area with the corresponding abnormal location threshold, so as to generate a corresponding abnormal location magnitude value, and re-compare the generated abnormal location magnitude value with the corresponding abnormal location threshold, so as to generate a second re-comparison result, and judge the abnormal level of the corresponding area according to the second re-comparison result; The feedback module (7) is used to feed back multiple data to the terminal.

2. According to claim 1, the AI-based financial reconciliation system is characterized by: The data collection module (1) comprises a transaction data collection unit (11), an enterprise system data collection unit (12), a payment gateway data collection unit (13) and an external data collection unit (14); The transaction data collection unit (11) is used to collect enterprise transaction data, including transaction amount, transaction quantity, foreign exchange rate and account balance fluctuation range; The enterprise system data collection unit (12) is used to collect enterprise system data, including the total bill amount, invoicing tax rate, receivable amount and bad debt reserve; The payment gateway data collection unit (13) is used to collect enterprise payment gateway data, including total payment amount, payment success rate, refund amount and transaction processing time; The external data collection unit (14) is used to collect external data of the enterprise, including invoice amount, supplier discount rate, preferential amount and invoice tax amount; The multi-source data set includes transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation, bill total, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, discount amount and invoice tax amount.

3. The AI-based financial reconciliation system according to claim 2, characterized in that: The data analysis module (2) comprises a data preprocessing unit (21) and a data sorting unit (22); The data preprocessing unit (21) is used to preprocess and normalize the multi-source data set, and the data sorting unit (22) is used to divide the processed multi-source data set into a first data set, a second data set, and a third data set; The first data set includes transaction amounts, transaction quantities, foreign exchange rates, and account balance fluctuations; The transaction amounts are recorded as A1, A2, A3, ..., An according to the number of transactions; The transaction quantity is recorded as B1, B2, B3, ..., Bn according to the number of transactions; The foreign exchange rates are recorded as C1, C2, C3, ..., Cn according to the number of transactions; The fluctuation range of account balance is recorded as D1, D2, D3, ..., Dn according to the number of transactions; The second data set includes the total billing amount, invoicing rate, receivable amount, and bad debt provision; The total bill amount is recorded as E1, E2, E3, ..., En according to the number of transactions; The invoicing tax rates are recorded as F1, F2, F3, ..., Fn according to the number of transactions; The receivable amounts are recorded as G1, G2, G3, ..., Gn according to the number of transactions; The bad debt reserves are recorded as H1, H2, H3, ..., Hn according to the number of transactions; The third data set includes total payment amount, payment success rate, refund amount, and transaction processing time; The total payment amount is recorded as I1, I2, I3, ..., In according to the number of transactions; The payment success rates are recorded as J1, J2, J3, ..., Jn according to the number of transactions; The refund amount is recorded as K1, K2, K3, ..., Kn according to the number of transactions; The transaction processing time is recorded as L1, L2, L3, ..., Ln according to the number of transactions; The fourth data set includes the invoice amount, supplier discount rate, preferential amount, and invoice tax amount; The invoice amount is recorded as M1, M2, M3, ..., Mn according to the number of transactions; The supplier discount rates are recorded as O1, O2, O3, ..., On according to the number of transactions; The preferential amount is recorded as P1, P2, P3, ..., Pn according to the number of transactions; The invoice tax amount is recorded as Q1, Q2, Q3, ..., Qn according to the number of transactions.

4. The AI-based financial reconciliation system according to claim 3 is characterized by: The first data calculation module (3) calculates and obtains the financial abnormality judgment value CYC through the following formula; Where: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the bill total analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax analysis coefficient; a1, a2, a3 and a4 are weight values, and the values ​​of a1, a2, a3 and a4 are adjusted and set by the user.

5. The AI-based financial reconciliation system according to claim 4, characterized in that: The first data calculation module (3) calculates and obtains the transaction amount analysis coefficient PA, the transaction quantity analysis coefficient PB, the foreign exchange rate analysis coefficient PC, the account balance fluctuation range analysis coefficient PD, the bill total analysis coefficient PE, the invoicing tax rate analysis coefficient PF, the receivable amount analysis coefficient PG, the bad debt reserve analysis coefficient PH, the total payment amount analysis coefficient PI, the payment success rate analysis coefficient PJ, the refund amount analysis coefficient PK, the transaction processing time analysis coefficient PL, the invoice amount analysis coefficient PM, the supplier discount rate analysis coefficient PO, the preferential amount analysis coefficient PP and the invoice tax amount analysis coefficient PQ respectively through the following formulas; Where: An, Bn, Cn, Dn, En, Fn, Gn, Hn, In, Jn, Kn, Ln, Mn, On, Pn and Qn are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the nth transaction; A1, B1, C1, D1, E1, F1, G1, H1, I1, J1, K1, L1, M1, O1, P1 and Q1 are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount under the first transaction; Zmax, Jmax, Cmax, Pmax, Ymax, Wmax, Lmax, Hmax and Smax are respectively the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount and invoice tax amount; Zmin, Jmin, Cmin, Pmin, Ymin, Wmin, Lmin, Hmin, Smin are the transaction amount, transaction quantity, foreign exchange rate, account balance fluctuation range, total bill amount, invoicing tax rate, receivable amount, bad debt reserve, total payment amount, payment success rate, refund amount, transaction processing time, invoice amount, supplier discount rate, preferential amount, and invoice tax amount respectively.

6. The AI-based financial reconciliation system according to claim 5, characterized in that: The first data analysis module (4) includes a first data comparison unit (41) and a first data re-comparison unit (42). The first data comparison unit (41) is used to generate a first comparison result, specifically as follows: When CYC ≤ Y, it means that the current finance is normal. When CYC > Y, it means that the current finance is abnormal. The first data re-comparison unit (42) is used to generate a first re-comparison result, specifically as follows: When CYL ≤ 20% × Y, it means that the current finance has a first-level abnormality. When 20% × Y < CYL ≤ 30% × Y, it means that the current finance has a first-level abnormality. When CYL > 30% × Y, it means that the current finance has a first-level abnormality. The calculation formula of the financial anomaly judgment value CYL is as follows; In the formula: CYC is the financial anomaly judgment value, and Y is the first threshold.

7. The AI-based financial reconciliation system according to claim 6, characterized in that: The second data calculation module (5) calculates and obtains a first anomaly positioning value DY, a second anomaly positioning value DR, a third anomaly positioning value DS, and a fourth anomaly positioning value DZ respectively through the following formulas; In the formula: PA is the transaction amount analysis coefficient, PB is the transaction quantity analysis coefficient, PC is the foreign exchange rate analysis coefficient, PD is the account balance fluctuation range analysis coefficient, PE is the total bill amount analysis coefficient, PF is the invoicing tax rate analysis coefficient, PG is the receivable amount analysis coefficient, PH is the bad debt reserve analysis coefficient, PI is the total payment amount analysis coefficient, PJ is the payment success rate analysis coefficient, PK is the refund amount analysis coefficient, PL is the transaction processing time analysis coefficient, PM is the invoice amount analysis coefficient, PO is the supplier discount rate analysis coefficient, PP is the preferential amount analysis coefficient, and PQ is the invoice tax amount analysis coefficient.

8. The AI-based financial reconciliation system according to claim 7, characterized in that: The second data analysis module (6) includes a second data comparison unit (61) and a second data re-comparison unit (62); The second data comparison unit (61) is used to generate a second comparison result, specifically as follows: When DY ≤ YY, it means that the transaction data in the current finance is normal. When DY > YY, it means that the transaction data in the current finance is abnormal. When DR ≤ RY, it means that the enterprise system data in the current finance is normal. When DR > RY, it means that the transaction data in the current finance is normal. When DS ≤ SY, it means that the enterprise payment gateway data in the current finance is normal. When DY > SY, it means that the enterprise payment gateway data in the current finance is abnormal. When DZ ≤ ZY, it means that the enterprise external data in the current finance is normal. When DY > SY, it means that the enterprise external data in the current finance is abnormal. The second data re-comparison unit (62) is used to generate a second re-comparison result, specifically as follows: When SY ≤ 20% of YY, it represents a first-level anomaly in the current transaction. When 20% of YY < SY ≤ 30% of JY, it represents a first-level anomaly in the current finance. When SY > 30% of YY, it represents a first-level anomaly in the current finance; When SR ≤ 20% of RY, it represents a first-level anomaly in the current finance. When 20% of RY < SR ≤ 30% of RY, it represents a first-level anomaly in the current finance. When SR > 30% of RY, it represents a first-level anomaly in the current finance; When SS ≤ 20% of SY, it represents a first-level anomaly in the current finance. When 20% of SY < SS ≤ 30% of SY, it represents a first-level anomaly in the current finance. When SS > 30% of SY, it represents a first-level anomaly in the current finance; When SZ ≤ 20% of ZY, it represents a first-level anomaly in the current finance. When 20% of ZY < SZ ≤ 30% of ZY, it represents a first-level anomaly in the current finance. When SZ > 30% of ZY, it represents a first-level anomaly in the current finance; The first anomaly positioning magnitude value SY, the second anomaly positioning magnitude value SR, the third anomaly positioning magnitude value SS, and the fourth anomaly positioning magnitude value SZ are respectively obtained through the following formulas; In the formula: DY is the first anomaly positioning value, DR is the second anomaly positioning value, DS is the third anomaly positioning value, DZ is the fourth anomaly positioning value, YY is the first anomaly positioning threshold, RY is the second anomaly positioning threshold, SY is the third anomaly positioning threshold, and ZY is the fourth anomaly positioning threshold.

9. A financial reconciliation method based on AI intelligence, characterized by: The specific steps are as follows: S1. Obtain data through the data acquisition module (1), and integrate the obtained data into a multi-source data set; S2. Process the multi-source data set through the data parsing module (2), and reorganize it into a first data set, a second data set, a third data set, and a fourth data set; S3. Integrate and calculate the first data set, the second data set, the third data set, and the fourth data set through the first data calculation module (3) to generate an anomaly judgment value CYC, and compare the financial anomaly judgment value CYC with a preset first threshold Y through the first data analysis module (4) to generate a first comparison result. According to the first comparison result, judge whether there is an anomaly in the accounts; S4. Integrate and calculate the first data set, the second data set, the third data set, and the fourth data set respectively through the second data calculation module (5) to generate the first anomaly positioning value DY, the second anomaly positioning value DR, the third anomaly positioning value DS, and the fourth anomaly positioning value DZ, and compare the first anomaly positioning value DY, the second anomaly positioning value DR, the third anomaly positioning value DS, and the fourth anomaly positioning value DZ with the preset first anomaly positioning threshold YY, the second anomaly positioning threshold RY, the third anomaly positioning threshold SY, and the fourth anomaly positioning threshold ZY respectively through the second data analysis module (6) to generate a second comparison result, and according to the second comparison result, locate the anomaly area of the accounts; S5. Feed back multiple data to the terminal through the feedback module (7).

Images