Reconciliation method and system for payment and settlement system, and storage medium

By introducing data collection, abnormal detection, intelligent label management and summary reconciliation modules in the payment and clearing system, combining feature vectors and historical transaction frequency to identify and prioritize abnormal payment records, the existing system's automatic identification under high transaction volume is solved. The problem of insufficient capabilities improves the efficiency and accuracy of the payment and clearing process.

CN118967322AInactive Publication Date: 2024-11-15QINGDAO OTC CLEARING CENT CO LTD

Patent Information

Application Number
CN202411432978.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-15
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing payment and clearing system is difficult to efficiently identify and handle abnormal payment records during the reconciliation process. Especially when the transaction volume is large and the transaction process is complicated, the system's automated recognition ability is weak, resulting in the accumulation of abnormal payment records, which increases the reconciliation work burden and Delay in payment clearing process.

Method used

A payment and clearing system was designed, including data acquisition module, exception detection module, intelligent tag management module, tag priority processing module, summary reconciliation module and evaluation and optimization module. Through the construction of feature vectors and abnormal detection scores, combined with historical transactions Frequency and channel identification, calculate processing priority index, determine the processing order of transaction records, and identify abnormalities by summarizing reconciliation differences values, and perform optimization and repair strategies.

Benefits of technology

The system is intelligent and automated abnormal payment identification and processing is realized, which improves the fluency and efficiency of the reconciliation process, reduces manual intervention, reduces risks and reconciliation workload in payment and clearing, and ensures that in complex transactions efficient and accurate processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967322A_ABST
    Figure CN118967322A_ABST
Patent Text Reader

Abstract

The invention discloses an account checking method and system of a payment clearing system and a storage medium, and relates to the technical field of big data processing. When the system runs, collected transaction data is packaged into a feature vector V, an amount anomaly detection score D1 and a time anomaly detection score D2 are calculated, risk assessment is carried out, and a level label index L is generated; and the evaluation and optimization module performs fitting calculation with stored waiting time information and channel identification information to obtain a processing priority index P so as to determine a processing sequence of transaction records and preferentially process high-risk transactions or transactions with relatively long waiting time, and the evaluation and optimization module further executes optimization and repair strategies according to abnormal conditions so as to improve the processing efficiency. And rule correction is carried out by optimizing the evaluation index Y. According to the whole system, through intelligent and automatic abnormal payment recognition and processing, the defect that a traditional account checking system cannot efficiently mark and process abnormal payment is overcome, and the risk in payment clearing and the account checking workload are reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of big data processing, and in particular to a reconciliation method, system and storage medium of a payment and clearing system. Background Art

[0002] With the rapid growth of global digital transactions, FinTech not only covers emerging technologies such as digital payments, e-wallets, and blockchain, but also includes back-end clearing and reconciliation systems. Payment and clearing are important links in financial transactions, ensuring the flow of funds and the accuracy of accounts between different institutions. The payment and clearing system is responsible for the transfer, settlement, and reconciliation of funds across different payment channels, such as banks and third-party payment platforms.

[0003] In the specific payment and clearing system reconciliation process, the frequent occurrence of abnormal payment records is a major problem in the reconciliation work. These abnormal payment records include: inconsistent amounts, abnormal payment times, repeated payments, and mismatched cross-channel transaction data. The current system mainly relies on rules and manual intervention to identify these anomalies. Reconciliation personnel need to manually troubleshoot problems through complex table comparisons or rely on historical records. This is not only time-consuming, but also prone to missed or misjudgment.

[0004] At the same time, a key disadvantage of the traditional reconciliation method is that it cannot efficiently identify and process these abnormal payment records, especially when the payment transaction volume is large and the transaction process is complex. The system's automatic recognition ability is weak, resulting in the inability to effectively and timely mark abnormal payments. In this case, abnormal payment records are easy to accumulate, increasing the workload of subsequent reconciliation and further slowing down the entire payment and settlement process. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a reconciliation method, system and storage medium for a payment and clearing system, which solve the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a reconciliation system for a payment and clearing system, including a data acquisition module, an anomaly detection module, an intelligent tag management module, a tag priority processing module, a summary reconciliation module and an evaluation and optimization module; The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; The intelligent label management module reflects the risk level of the transaction and obtains the level label index L according to the amount anomaly detection score D1 and the time anomaly detection score D2, and fits the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V; The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation and optimization module determines the execution of optimization strategy and repair strategy according to the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and corrects the rules and adjusts the transactions.

[0007] Preferably, the data acquisition module includes a data acquisition unit and a feature vector construction unit; The data acquisition unit obtains relevant transaction data, including actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID, by connecting with several source channels of payment system, invoice system and bank system, including using API call and database connection mode; The feature vector construction unit sequentially packages the acquired actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID into feature vector parameters V1, namely the actual transaction amount TA, V2, expected amount EA, V3, payment time TT, V4, historical transaction frequency TF, channel identifier TC, V5, V6, exchange rate fluctuation FR, V7, transaction time period identifier TID and V8 to-be-processed time WT, and marks the package time SJ as the package time SJ of V9 to form a feature vector V={V1, V2, V3, V4, V5, V6, V7, V8, V9}, and stores it in the system database.

[0008] Preferably, the anomaly detection module includes a parameter extraction unit and an anomaly detection unit; The parameter extraction unit extracts relevant anomaly detection parameters from the stored feature vector V, including the actual transaction amount TA of V1, the expected amount EA of V2, the payment time TT of V3, and the historical transaction frequency TF of V4; The anomaly detection unit performs an anomaly detection process on the extracted anomaly detection parameters to obtain an amount anomaly detection score D1 and a time anomaly detection score D2; The amount anomaly detection score D1 is obtained by the following calculation formula: ; In the formula, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, The standard deviation of the actual transaction amount TA of V1 specifically reflects the volatility of the transaction amount. Represents the average value of the actual transaction amount TA of V1; The temporal anomaly detection score D2 is obtained by the following calculation formula: ; In the formula, V3 represents the payment time TT, V4 represents the historical transaction frequency TF, represents the average value of the payment time TT of V3, It represents the standard deviation of the payment time TT of V3, and log represents the logarithmic function.

[0009] Preferably, the intelligent label management module reflects the risk level of the transaction and obtains the level label index L by fitting the amount anomaly detection score D1 and the time anomaly detection score D2 with the historical transaction frequency TF of V4 and the channel identifier TC of V5 in the stored feature vector V; The level label index L is obtained by the following calculation formula: ; Where r1 and r2 represent the preset risk weight values ​​of the amount anomaly detection score D1 and the time anomaly detection score D2, respectively, and , V4 represents the historical transaction frequency TF, V5 represents the channel identifier TC, Indicates that the historical transaction frequency TF of V4 is mapped to the range of [0, 1]. Indicates the channel identifier TC of nonlinear adjustment V5, Represents a mathematical constant, specifically pi.

[0010] Preferably, the tag priority processing module includes a priority calculation unit and a sorting unit; The priority calculation unit is based on the level label index L and the stored waiting time information and channel identification information, wherein the waiting time information includes the waiting time WT of V8 in the feature vector V, and the channel identification information includes the channel identification TC of V5 in the feature vector V, and the level label index L, the waiting time WT of V8 and the channel identification TC of V5 are fitted and calculated to obtain the processing priority index P; The processing priority index P is obtained by the following calculation formula: ; Wherein, V8 represents the waiting time WT, V5 represents the channel identifier TC, and tan represents the inverse tangent function.

[0011] Preferably, the sorting unit obtains 1 to n processing priority indexes P according to the fitting calculation to form a priority processing list Pn, and readjusts the positions of 1 to n processing priority indexes P in the priority processing list Pn by bubble sorting the priority processing list Pn. The sorted priority processing list Pn is used to determine the processing order of the transaction records.

[0012] Preferably, the summary reconciliation module includes a difference calculation unit and a summary evaluation unit; The difference calculation unit performs calculations based on the processing priority index P and the transaction amount information stored in the feature vector V, wherein the transaction amount information includes the actual transaction amount TA of V1, the expected amount EA of V2 and the historical transaction frequency TF of V4 in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S; The summary reconciliation difference value S is obtained by the following calculation formula: ; Wherein, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, V4 represents the historical transaction frequency TF, V1i represents the actual transaction amount TA in the i-th record, V2i represents the expected amount EA in the i-th record, V4i represents the historical transaction frequency TF in the i-th record, n represents the total number of records in the priority processing list Pn, i represents the i-th record, Pi represents the processing priority index P of the i-th record in the priority processing list Pn, and log represents the logarithmic function.

[0013] The summary evaluation unit matches the summary reconciliation difference value S with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation results of the anomaly marking process are generated by the following matching method: When the summary reconciliation difference value S> the summary reconciliation difference threshold CZ, the abnormal marking process evaluation result is generated as an abnormal result; When the summary reconciliation difference value S ≤ the summary reconciliation difference threshold CZ, the evaluation result of the abnormal marking process is generated as no abnormal result.

[0014] Preferably, the evaluation and optimization module includes an abnormality judgment unit and an optimization identification unit; The abnormality judgment unit judges and executes the optimization strategy and the repair strategy according to the evaluation result of the abnormality marking process. When the evaluation result of the abnormality marking process is an abnormal result, the optimization strategy and the repair strategy are triggered. The optimization strategy includes regenerating invoices, correcting payment data and notifying relevant personnel to perform manual intervention. The repair strategy includes adjusting the tolerance of transaction amount differences and optimizing the rules in the reconciliation process; The optimization identification unit processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and performs rule modification and transaction adjustment, including determining the reconciliation rules according to the optimization evaluation index Y, and performing rule modification and transaction adjustment, wherein the rule modification includes updating anomaly detection parameters and redefining the calculation method of reconciliation differences; The optimization evaluation index Y is The calculation formula is obtained, where exp represents the exponential function.

[0015] A reconciliation method for a payment and clearing system comprises the following steps: Step 1: The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; Step 2: The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; Step 3: The intelligent label management module reflects the risk level of the transaction based on the amount anomaly detection score D1 and the time anomaly detection score D2, and fits them with the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V, and obtains the level label index L; Step 4: The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; Step 5: The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information, summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; Step 6: The evaluation and optimization module determines the execution of optimization strategies and repair strategies based on the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategies and repair strategies to obtain the optimization evaluation index Y, and then corrects the rules and adjusts the transactions.

[0016] A reconciliation storage medium of a payment and clearing system, wherein the storage medium stores a computer program, and when the computer program is executed, a reconciliation system of the payment and clearing system is implemented.

[0017] The present invention provides a payment and clearing system reconciliation method, system and storage medium, which have the following beneficial effects: (1) When the system is running, the collected transaction data is packaged into a feature vector V, providing a unified and efficient data foundation for subsequent modules. The anomaly detection module extracts the transaction data in the feature vector V, calculates the amount anomaly detection score D1 and the time anomaly detection score D2, combines the historical transaction frequency TF and the channel identifier TC to conduct risk assessment on the transaction, generates the level label index L, and performs fitting calculation with the stored waiting time information and channel identifier information to obtain the processing priority index P, thereby determining the processing order of transaction records, giving priority to high-risk or long-waiting transactions, and improving the response speed of the system. Finally, the summary reconciliation module generates a summary reconciliation difference value S, and identifies anomalies by matching it with the preset summary reconciliation difference threshold CZ. The evaluation and optimization module further executes optimization and repair strategies based on the abnormal situation, and corrects the rules by optimizing the evaluation index Y. The overall system solves the problem that the traditional reconciliation system cannot efficiently mark and process abnormal payments through intelligent and automated abnormal payment identification and processing, ensuring a smoother and more efficient reconciliation process, reducing manual intervention, and reducing the risk and reconciliation workload in payment settlement.

[0018] (2) By combining the amount anomaly detection score D1 and the time anomaly detection score D2, and combining with the feature vector V, the risk level of each transaction is evaluated and the level label index L is generated. By using the level label index L and the feature vector V, the processing priority index P of each transaction is comprehensively calculated, so that the system can not only identify high-risk transactions, but also reasonably arrange the processing order according to the waiting time and channel characteristics. The inverse tangent function is used to smooth the processing time WT to avoid excessive inflation of the priority due to long waiting time, ensuring that the processing order is more scientific and reasonable. Combined with the use of the bubble sort algorithm, the system can dynamically adjust the priority processing list of transaction records and quickly respond to the processing needs of different transactions.

[0019] (3) By combining the processing priority index P with the feature vector V, the reconciliation difference information of each transaction is effectively calculated, and the summary reconciliation difference value S is generated by accumulating these difference information, ensuring accurate reconciliation calculation for all transactions. Subsequently, the summary evaluation unit generates an abnormal mark and triggers the corresponding evaluation and processing process by matching the summary reconciliation difference value S with the summary reconciliation difference threshold CZ. This ensures that efficient and accurate reconciliation processing is maintained in the face of complex and high-frequency transactions, effectively reducing transaction risks and improving the reliability and efficiency of overall operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of a reconciliation system flow of a payment and clearing system of the present invention; Figure 2 The present invention is a schematic diagram of the steps of a reconciliation method of a payment and clearing system. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0022] Example 1 The present invention provides a reconciliation system for a payment and clearing system, see Figure 1 , including data collection module, anomaly detection module, intelligent label management module, label priority processing module, summary reconciliation module and evaluation and optimization module; The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; The intelligent label management module reflects the risk level of the transaction and obtains the level label index L according to the amount anomaly detection score D1 and the time anomaly detection score D2, and fits the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V; The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation and optimization module determines the execution of optimization strategy and repair strategy according to the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and corrects the rules and adjusts the transactions.

[0023] In this embodiment, the data collection module is connected with multiple source channels of the payment system, invoice system and bank system, and the collected transaction data is packaged into a feature vector V, providing a unified and efficient data basis for subsequent modules. The anomaly detection module extracts the transaction data in the feature vector V, calculates the amount anomaly detection score D1 and the time anomaly detection score D2, realizes the ability to automatically identify abnormal transactions, and avoids the inefficiency of manual reconciliation under traditional methods. Then, the intelligent label management module conducts risk assessment on the transaction based on the amount anomaly detection score D1 and the time anomaly detection score D2, combined with the historical transaction frequency TF and the channel identification TC, and generates a level label index L to ensure that the system can quickly and accurately judge and mark high-risk transactions. Subsequently, the label priority processing module performs fitting calculations based on the level label index L and the stored waiting time information and channel identification information to obtain the processing priority index P, thereby determining the processing order of transaction records, giving priority to high-risk or long-waiting transactions, and improving the response speed of the system. Finally, the summary reconciliation module calculates the reconciliation difference based on the processing priority index P and the transaction amount information stored in the feature vector V, generates a summary reconciliation difference value S, and identifies anomalies by matching it with the preset summary reconciliation difference threshold CZ. The evaluation and optimization module further executes optimization and repair strategies based on the abnormal situation, and corrects the rules by optimizing the evaluation index Y. The overall system solves the problem that the traditional reconciliation system cannot efficiently mark and process abnormal payments through intelligent and automated abnormal payment identification and processing, ensuring a smoother and more efficient reconciliation process, reducing manual intervention, and reducing the risks and reconciliation workload in payment settlement.

[0024] Example 2 This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a data acquisition unit and a feature vector construction unit; The data acquisition unit obtains relevant transaction data, including actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID, by connecting with several source channels of payment system, invoice system and bank system, including using API call and database connection mode; The feature vector construction unit sequentially packages the acquired actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID into feature vector parameters V1, namely the actual transaction amount TA, V2 expected amount EA, V3 payment time TT, V4 historical transaction frequency TF, V5 channel identifier TC, V6 exchange rate fluctuation FR, V7 transaction time period identifier TID and V8 pending time WT, and marks the packaging time SJ as the packaging time SJ of V9 to form a feature vector V={V1, V2, V3, V4, V5, V6, V7, V8, V9}, and represents specific transaction data by marking V1 to V9, thereby protecting privacy and reducing the occurrence of data leakage-related situations, and storing the data in the system database.

[0025] The anomaly detection module includes a parameter extraction unit and an anomaly detection unit; The parameter extraction unit extracts relevant anomaly detection parameters from the stored feature vector V, including the actual transaction amount TA of V1, the expected amount EA of V2, the payment time TT of V3, and the historical transaction frequency TF of V4; The anomaly detection unit performs an anomaly detection process on the extracted anomaly detection parameters to obtain an amount anomaly detection score D1 and a time anomaly detection score D2; The amount anomaly detection score D1 is obtained by the following calculation formula: ; In the formula, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, The standard deviation of the actual transaction amount TA of V1 specifically reflects the volatility of the transaction amount. Represents the average value of the actual transaction amount TA of V1; The temporal anomaly detection score D2 is obtained by the following calculation formula: ; In the formula, V3 represents the payment time TT, V4 represents the historical transaction frequency TF, represents the average value of the payment time TT of V3, represents the standard deviation of V3’s payment time TT, log represents the logarithmic function, The historical transaction frequency TF is smoothed to ensure that users with high transaction frequency will not overly amplify their influence in the anomaly detection score.

[0026] Example 3 This embodiment is explained in Example 1, please refer to Figure 1Specifically: the intelligent label management module reflects the risk level of the transaction by fitting the amount anomaly detection score D1 and the time anomaly detection score D2 with the historical transaction frequency TF of V4 and the channel identifier TC of V5 in the stored feature vector V, and obtains the level label index L; The level label index L is obtained by the following calculation formula: ; Where r1 and r2 represent the preset risk weight values ​​of the amount anomaly detection score D1 and the time anomaly detection score D2, respectively, and , V4 represents the historical transaction frequency TF, V5 represents the channel identifier TC, Indicates that the historical transaction frequency TF of V4 is mapped to the range of [0, 1]. Indicates the channel identifier TC of nonlinear adjustment V5, Represents a mathematical constant, specifically pi, Use the Sigmoid function to map the historical transaction frequency TF. The higher the transaction frequency, the more stable the transaction behavior and the lower the risk. The channel identifier TC is processed nonlinearly to reflect the risk differences of different transaction channels.

[0027] The tag priority processing module includes a priority calculation unit and a sorting unit; The priority calculation unit is based on the level label index L and the stored waiting time information and channel identification information, wherein the waiting time information includes the waiting time WT of V8 in the feature vector V, and the channel identification information includes the channel identification TC of V5 in the feature vector V, and the level label index L, the waiting time WT of V8 and the channel identification TC of V5 are fitted and calculated to obtain the processing priority index P; The processing priority index P is obtained by the following calculation formula: ; In the formula, V8 represents the waiting time WT, V5 represents the channel identifier TC, and tan represents the inverse tangent function. The processing priority is dynamically calculated by combining the level label index L, the waiting time WT and the channel identifier TC. The waiting time WT increases the priority, while the square root of the channel identifier TC balances the impact of different channels. Reflects the pending time WT and further impact on priority, using the inverse tangent function To smooth the impact of waiting time and avoid excessive inflation of the priority of transactions that have been waiting for a long time.

[0028] The sorting unit obtains 1 to n processing priority indexes P according to the fitting calculation to form a priority processing list Pn. By bubble sorting the priority processing list Pn, the positions of 1 to n processing priority indexes P in the priority processing list Pn are readjusted. The sorted priority processing list Pn is used to determine the processing order of the transaction records.

[0029] In this embodiment, by combining the amount anomaly detection score D1 and the time anomaly detection score D2, and combining the historical transaction frequency TF and the channel identifier TC in the feature vector V, the risk level of each transaction is evaluated to generate a level label index L. This dynamic risk assessment that combines historical data and channel characteristics ensures that high-risk transactions can be effectively marked, and further improves the system's perception of potential abnormal transactions. At the same time, the label priority processing module uses the level label index L, the waiting time WT in the feature vector V, and the channel identifier TC to comprehensively calculate the processing priority index P of each transaction, so that the system can not only identify high-risk transactions, but also reasonably arrange the processing order according to the waiting time and channel characteristics. The processing time WT is smoothed by the inverse tangent function to avoid excessive expansion of the priority due to long waiting time, ensuring that the processing order is more scientific and reasonable. Combined with the use of the bubble sort algorithm, the system can dynamically adjust the priority processing list of transaction records, quickly respond to the processing requirements of different transactions, ensure that high-priority transactions can be processed in a timely manner, and significantly improve the processing efficiency of the system and the security of transactions.

[0030] Example 4 This embodiment is explained in Example 1, please refer to Figure 1 , specifically: the summary reconciliation module includes a difference calculation unit and a summary evaluation unit; The difference calculation unit performs calculations based on the processing priority index P and the transaction amount information stored in the feature vector V, wherein the transaction amount information includes the actual transaction amount TA of V1, the expected amount EA of V2 and the historical transaction frequency TF of V4 in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S; The summary reconciliation difference value S is obtained by the following calculation formula: ; Wherein, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, V4 represents the historical transaction frequency TF, V1i represents the actual transaction amount TA in the i-th record, V2i represents the expected amount EA in the i-th record, V4i represents the historical transaction frequency TF in the i-th record, n represents the total number of records in the priority processing list Pn, i represents the i-th record, Pi represents the processing priority index P of the i-th record in the priority processing list Pn, log represents the logarithmic function, The historical transaction frequency TF is smoothed to ensure that users with high transaction frequency will not overly amplify their influence in the anomaly detection score.

[0031] The summary evaluation unit matches the summary reconciliation difference value S with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation results of the anomaly marking process are generated by the following matching method: When the summary reconciliation difference value S> the summary reconciliation difference threshold CZ, the abnormal marking process evaluation result is generated as an abnormal result; When the summary reconciliation difference value S ≤ the summary reconciliation difference threshold CZ, the evaluation result of the abnormal marking process is generated as no abnormal result.

[0032] The evaluation and optimization module includes an abnormality judgment unit and an optimization identification unit; The abnormality judgment unit judges and executes the optimization strategy and the repair strategy according to the evaluation result of the abnormality marking process. When the evaluation result of the abnormality marking process is an abnormal result, the optimization strategy and the repair strategy are triggered. The optimization strategy includes regenerating invoices, correcting payment data and notifying relevant personnel to perform manual intervention. The repair strategy includes adjusting the tolerance of transaction amount differences and optimizing the rules in the reconciliation process; The optimization identification unit processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and performs rule modification and transaction adjustment, including determining the reconciliation rules according to the optimization evaluation index Y, and performing rule modification and transaction adjustment, wherein the rule modification includes updating anomaly detection parameters and redefining the calculation method of reconciliation differences; The optimization evaluation index Y is The calculation formula is obtained, where exp represents the exponential function. Convert the result to a value between 0 and 1.

[0033] In this embodiment, by combining the processing priority index P with the actual transaction amount TA in the feature vector V, the expected amount EA in the feature vector V, and the historical transaction frequency TF in the feature vector V, the reconciliation difference information of each transaction is effectively calculated, and the summary reconciliation difference value S is generated by accumulating these difference information, so as to ensure accurate reconciliation calculation for all transactions. Subsequently, the summary evaluation unit generates an abnormality mark and triggers the corresponding evaluation and processing flow by matching the summary reconciliation difference value S with the summary reconciliation difference threshold CZ. If an abnormality is found, the abnormality judgment unit will automatically execute the corresponding optimization and repair strategy, including regenerating invoices or correcting payment data, thereby greatly improving the automatic repair capability in the system reconciliation process. At the same time, the system provides real-time feedback on the optimization results through the optimization identification unit, generates an optimization evaluation index Y, and corrects and adjusts the rules, so that the system can self-optimize after each reconciliation operation. Through this continuous automatic optimization and rule correction mechanism, the system can dynamically update the abnormality detection parameters and the calculation method of the reconciliation difference, ensuring that efficient and accurate reconciliation processing is maintained in the face of complex and high-frequency transactions, effectively reducing transaction risks and improving the reliability and efficiency of the overall operation.

[0034] Example 5 A payment and clearing system reconciliation method, please refer to Figure 2 , specifically: including the following steps: Step 1: The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; Step 2: The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; Step 3: The intelligent label management module reflects the risk level of the transaction based on the amount anomaly detection score D1 and the time anomaly detection score D2, and fits them with the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V, and obtains the level label index L; Step 4: The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; Step 5: The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information, summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; Step 6: The evaluation and optimization module determines the execution of optimization strategies and repair strategies based on the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategies and repair strategies to obtain the optimization evaluation index Y, and then corrects the rules and adjusts the transactions.

[0035] Example 6 A reconciliation storage medium for a payment and clearing system, specifically: the storage medium stores a computer program, and when the computer program is executed, a reconciliation system for the payment and clearing system is implemented.

[0036] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A reconciliation system for a payment and clearing system, characterized by: It includes data collection module, anomaly detection module, intelligent label management module, label priority processing module, summary reconciliation module and evaluation and optimization module; The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; The intelligent label management module reflects the risk level of the transaction and obtains the level label index L according to the amount anomaly detection score D1 and the time anomaly detection score D2, and fits the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V; The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation and optimization module determines the execution of optimization strategy and repair strategy according to the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and corrects the rules and adjusts the transactions.

2. The reconciliation system of a payment and clearing system according to claim 1, characterized in that: The data acquisition module includes a data acquisition unit and a feature vector construction unit; The data acquisition unit obtains relevant transaction data, including actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID, by connecting with several source channels of payment system, invoice system and bank system, including using API call and database connection mode; The feature vector construction unit sequentially packages the acquired actual transaction amount TA, expected amount EA, payment time TT, historical transaction frequency TF, channel identifier TC, exchange rate fluctuation FR and transaction time period identifier TID into feature vector parameters V1, namely, actual transaction amount TA, V2, expected amount EA, V3, payment time TT, V4, historical transaction frequency TF, V5, channel identifier TC, V6, exchange rate fluctuation FR, V7, transaction time period identifier TID and V8, pending time WT, and marks the package time SJ as the package time SJ of V9. The feature vector V={V1, V2, V3, V4, V5, V6, V7, V8, V9} is formed and stored in the system database.

3. The reconciliation system of a payment and clearing system according to claim 2, characterized in that: The anomaly detection module includes a parameter extraction unit and an anomaly detection unit; The parameter extraction unit extracts relevant anomaly detection parameters from the stored feature vector V, including the actual transaction amount TA of V1, the expected amount EA of V2, the payment time TT of V3, and the historical transaction frequency TF of V4; The anomaly detection unit performs an anomaly detection process on the extracted anomaly detection parameters to obtain an amount anomaly detection score D1 and a time anomaly detection score D2; The amount anomaly detection score D1 is obtained by the following calculation formula: ; In the formula, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, The standard deviation of the actual transaction amount TA of V1 specifically reflects the volatility of the transaction amount. Represents the average value of the actual transaction amount TA of V1; The temporal anomaly detection score D2 is obtained by the following calculation formula: ; In the formula, V3 represents the payment time TT, V4 represents the historical transaction frequency TF, represents the average value of the payment time TT of V3, It represents the standard deviation of the payment time TT of V3, and log represents the logarithmic function.

4. The reconciliation system of a payment and clearing system according to claim 3, characterized in that: The intelligent label management module reflects the risk level of the transaction and obtains the level label index L by fitting the amount anomaly detection score D1 and the time anomaly detection score D2 with the historical transaction frequency TF of V4 and the channel identifier TC of V5 in the stored feature vector V; The level label index L is obtained by the following calculation formula: ; Where r1 and r2 represent the preset risk weight values ​​of the amount anomaly detection score D1 and the time anomaly detection score D2, respectively, and , V4 represents the historical transaction frequency TF, V5 represents the channel identifier TC, Indicates that the historical transaction frequency TF of V4 is mapped to the range of [0, 1]. Indicates the channel identifier TC of nonlinear adjustment V5, Represents a mathematical constant, specifically pi.

5. The reconciliation system of a payment and clearing system according to claim 4, characterized in that: The tag priority processing module includes a priority calculation unit and a sorting unit; The priority calculation unit is based on the level label index L and the stored waiting time information and channel identification information, wherein the waiting time information includes the waiting time WT of V8 in the feature vector V, and the channel identification information includes the channel identification TC of V5 in the feature vector V, and the level label index L, the waiting time WT of V8 and the channel identification TC of V5 are fitted and calculated to obtain the processing priority index P; The processing priority index P is obtained by the following calculation formula: ; Wherein, V8 represents the waiting time WT, V5 represents the channel identifier TC, and tan represents the inverse tangent function.

6. The reconciliation system of a payment and clearing system according to claim 5, characterized in that: The sorting unit obtains 1 to n processing priority indexes P according to the fitting calculation to form a priority processing list Pn. By bubble sorting the priority processing list Pn, the positions of 1 to n processing priority indexes P in the priority processing list Pn are readjusted. The sorted priority processing list Pn is used to determine the processing order of the transaction records.

7. The reconciliation system of a payment and clearing system according to claim 6, characterized in that: The summary reconciliation module includes a difference calculation unit and a summary evaluation unit; The difference calculation unit performs calculations based on the processing priority index P and the transaction amount information stored in the feature vector V, wherein the transaction amount information includes the actual transaction amount TA of V1, the expected amount EA of V2 and the historical transaction frequency TF of V4 in the feature vector V, obtains the reconciliation difference information and summarizes it to obtain the summary reconciliation difference value S; The summary reconciliation difference value S is obtained by the following calculation formula: ; Wherein, V1 represents the actual transaction amount TA, V2 represents the expected amount EA, V4 represents the historical transaction frequency TF, V1i represents the actual transaction amount TA in the i-th record, V2i represents the expected amount EA in the i-th record, V4i represents the historical transaction frequency TF in the i-th record, n represents the total number of records in the priority processing list Pn, i represents the i-th record, Pi represents the processing priority index P of the i-th record in the priority processing list Pn, and log represents the logarithmic function; The summary evaluation unit matches the summary reconciliation difference value S with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; The evaluation results of the anomaly marking process are generated by the following matching method: When the summary reconciliation difference value S> the summary reconciliation difference threshold CZ, the abnormal marking process evaluation result is generated as an abnormal result; When the summary reconciliation difference value S ≤ the summary reconciliation difference threshold CZ, the evaluation result of the abnormal marking process is generated as no abnormal result.

8. The reconciliation system of a payment and clearing system according to claim 7, characterized in that: The evaluation and optimization module includes an abnormality judgment unit and an optimization identification unit; The abnormality judgment unit judges and executes the optimization strategy and the repair strategy according to the evaluation result of the abnormality marking process. When the evaluation result of the abnormality marking process is an abnormal result, the optimization strategy and the repair strategy are triggered. The optimization strategy includes regenerating invoices, correcting payment data and notifying relevant personnel to perform manual intervention. The repair strategy includes adjusting the tolerance of transaction amount differences and optimizing the rules in the reconciliation process; The optimization identification unit processes the execution feedback information of the optimization strategy and the repair strategy to obtain the optimization evaluation index Y, and performs rule modification and transaction adjustment, including determining the reconciliation rules according to the optimization evaluation index Y, and performing rule modification and transaction adjustment, wherein the rule modification includes updating anomaly detection parameters and redefining the calculation method of reconciliation differences; The optimization evaluation index Y is The calculation formula is obtained, where exp represents the exponential function.

9. A reconciliation method for a payment and clearing system, applied to a reconciliation system for a payment and clearing system according to any one of claims 1 to 8, characterized in that: The following steps are involved: Step 1: The data collection module obtains relevant transaction data by connecting with several source channels such as the payment system, invoice system and bank system, and packages the collected transaction data into a feature vector V and stores it; Step 2: The anomaly detection module extracts relevant anomaly detection parameters from the stored feature vector V, and executes the anomaly detection process to obtain the amount anomaly detection score D1 and the time anomaly detection score D2; Step 3: The intelligent label management module reflects the risk level of the transaction based on the amount anomaly detection score D1 and the time anomaly detection score D2, and fits them with the historical transaction frequency TF of V4 and the channel identifier TC of V5 stored in the feature vector V, and obtains the level label index L; Step 4: The tag priority processing module performs fitting calculation based on the level tag index L and the waiting time information and channel identification information stored in the feature vector V to obtain the processing priority index P, which is used to determine the processing order of the transaction records; Step 5: The summary reconciliation module calculates based on the processing priority index P and the transaction amount information stored in the feature vector V, obtains the reconciliation difference information, summarizes it to obtain the summary reconciliation difference value S, and matches it with the preset summary reconciliation difference threshold CZ to generate an abnormal marking process evaluation result; Step 6: The evaluation and optimization module determines the execution of optimization strategies and repair strategies based on the evaluation results of the abnormal marking process, and processes the execution feedback information of the optimization strategies and repair strategies to obtain the optimization evaluation index Y, and then corrects the rules and adjusts the transactions.

10. A reconciliation storage medium for a payment and clearing system, characterized in that: The storage medium stores a computer program, and when the computer program is executed, a reconciliation system of a payment and clearing system according to any one of claims 1 to 8 is implemented.

Citation Information

Patent Citations

  • Automatic payment and settlement method and system

    CN114418559A

  • Bill payment system based on business configuration

    CN118761773A

  • System and method for optimized funding of electronic transactions

    US20030055783A1

  • Managed real-time transaction fraud analysis and decisioning

    US20100305993A1

Cited By

  • Bank flow verification method and system based on OCR (Optical Character Recognition) technology

    CN119624613A

  • An AI-based automatic reconciliation optimization and processing method and system

    CN119741140A

  • An AI-based Automatic Reconciliation Optimization and Processing Method and System

    CN119741140B

  • Reconciliation and liquidation method and system based on intelligence

    CN120374284A