Enterprise account receivable data anomaly detection method based on artificial intelligence

Abstract

The invention discloses an enterprise receivable data anomaly detection method based on artificial intelligence, and relates to the technical field of data detection, and the method comprises the steps: building a systematized receivable risk management and control system through customer grading, threshold setting, real-time anomaly detection and early warning management on the whole; according to the method, big data analysis and artificial intelligence technologies are utilized, customer risk levels are identified from historical data, and a detection threshold value is dynamically adjusted, so that repayment behaviors of different customers are accurately monitored, automatic screening and intelligent early warning of abnormal accounts are realized, and financial risks are reduced; the account management efficiency is improved, the workload of manual participation and the false alarm rate are remarkably reduced, enterprises are helped to carry out effective intervention before potential risks occur, and the fund mobility is protected.

Description

Technical Field

[0001] The present invention belongs to the technical field of data detection, and in particular relates to an anomaly detection method for enterprise accounts receivable data based on artificial intelligence. Background Art

[0002] In corporate financial management, accounts receivable, which is part of accounting data, is one of the key indicators for maintaining cash flow and profitability. However, due to the influence of many factors such as customer payment behavior, contract terms, market environment and credit policy, accounts receivable data often show strong sensitivity and volatility. Abnormal accounts receivable data may indicate customer collection risks, long account age or credit problems. If not discovered and handled in time, it is very easy to lead to bad debt risks, which directly affects the financial stability and resource liquidity of the company. Therefore, how to effectively identify and control anomalies in accounts receivable has become an important problem in corporate management.

[0003] Traditional accounts receivable management methods usually rely on manual inspection and fixed rules. This method is not only inefficient, but also prone to omissions or false positives when faced with complex business scenarios and high-frequency data changes, making it difficult to cope with the risks brought about by dynamic changes. With the expansion of enterprise scale and the increasing complexity of the market environment, manual processing alone can no longer meet the management needs of modern enterprises.

[0004] In recent years, the rapid development of artificial intelligence (AI) technology and big data analysis has provided new solutions for anomaly detection of accounts receivable. Using AI technology, we can dynamically adjust the detection threshold, evaluate the customer's credit rating, and quickly warn when anomalies occur through deep learning and intelligent analysis of historical data. The AI-based accounting data anomaly detection method can not only effectively improve the accuracy and efficiency of detection, but also reduce the errors and resource waste caused by human judgment, helping enterprises to seize potential risks in data fluctuations and ensure financial health.

[0005] Therefore, an artificial intelligence-based anomaly detection method for enterprise accounts receivable data is proposed, which combines customer classification, dynamic threshold setting and intelligent anomaly detection to achieve accurate identification and management of high-risk accounts. Summary of the invention

[0006] The purpose of the present invention is to provide an artificial intelligence-based method for detecting anomalies in enterprise accounts receivable data, which solves the problem that traditional accounts receivable management methods usually rely on manual inspection and fixed rules, and are prone to omissions or false alarms when faced with complex business scenarios and high-frequency data changes, making it difficult to deal with the risks brought about by dynamic changes.

[0007] The technical solution adopted by the present invention is an artificial intelligence-based method for detecting anomalies in enterprise accounts receivable data, comprising: Step 1: Identify all the accounts receivable customers of the enterprise, and classify them into first-level accounts receivable customers, second-level accounts receivable customers, and third-level accounts receivable customers based on historical accounts receivable records; Step 2: After classification, establish corresponding abnormal detection thresholds for account customers in each level; Step 3: Perform anomaly detection on the current accounts receivable records through the evaluation model. First, determine the level of accounts receivable customers to which the customers in all current accounts receivable records belong, determine the relevant anomaly detection thresholds according to the different levels of accounts receivable customers, and compare the current accounts receivable records of the accounts receivable customers with the relevant anomaly detection thresholds. The comparison results are evaluated through the evaluation model, and whether the current accounts receivable are abnormal is determined based on the evaluation results. Step 4: Identify current accounts receivable records with abnormally low risk and current accounts receivable records with abnormally high risk, and send them to corporate management for verification and management.

[0008] The technical solution of the present invention is also characterized in that: In step 1, all accounts receivable customers of the enterprise are identified, and the specific method of classifying accounts receivable customers is as follows based on historical accounts receivable records: S1: Select the accounts payable customers determined by the enterprise as target accounts payable customers in turn; S2: Obtain the historical accounts receivable records of the target account customer and determine the specific number of historical accounts receivable records, recorded as ,in, It is to make corresponding changes according to the number of historical accounts receivable records of different target account customers; S3: Then obtain the cooperation time between the target account customer and the enterprise, recorded as ; S4: Get the account amount in each historical accounts receivable record of the target account customer, recorded as ,in, , and the average amount of accounts receivable is calculated by formula (1) , formula (1) is specifically: (1); S5: Calculate the difference between the actual repayment time and the stipulated repayment time in each historical accounts receivable record, recorded as , and the average delay days are calculated by formula (2) , formula (2) is specifically: (2); S6: At the same time, determine the number of delayed repayments of the target account customer in the historical accounts receivable records, and calculate the ratio of the number of delayed repayments to the total number of historical accounts receivable records to obtain the overdue repayment frequency, recorded as F; S7: Then, the target account customer rating formula is used to determine the target account customer rating Sc: (3); in, is the weight parameter; S8: Score the target account customers according to their grade Divide the target account customers into first-level account customers, second-level account customers, and third-level account customers; S9: Repeat steps S2-S8, and classify the accounts payable customers determined by the selected enterprises as target accounts payable customers.

[0009] According to the target account customer rating The target account customers are divided into first-level account customers, second-level account customers, and third-level account customers. The specific method is as follows: If the rating , classify the target account customer into a first-level account customer; If the rating , classify the target account customer into a secondary account customer; If the rating , classify the target account customers into three levels of account customers; in, and S2 are preset values, and .

[0010] In step 2, after classification, the specific contents of establishing corresponding abnormal detection thresholds for account customers in each level include establishing a repayment time limit threshold and establishing an account amount limit threshold.

[0011] The specific method for establishing the repayment time threshold is: AS1: Determine the historical repayment time data set of first-level account customers: ,in, Represents the historical average repayment time of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; AS2: Calculate the average repayment time of all first-tier account customers using formula (4) , formula (4) is specifically: (4); Then, the standard deviation of repayment time for all first-tier account customers is calculated using formula (5): , formula (5) is specifically: (5); AS3: Sort all the data in set D from small to large, and determine the values ​​of Q1 and Q3 by quartile method, where Q1 is the 25th percentile of the data in set D sorted from small to large, and Q3 is the 75th percentile of the data in set D sorted from small to large; AS4: Limit the data in set D to ,in, , the data in set D exceeds Remove the data in the range and get the data set after removal ,in, It is expressed as the number of total account customers in the first-level account customer group after removal; AS5: Then, use formula (6) to establish the repayment time limit threshold for first-tier account customers ; (6); In the formula, and The data sets The mean and standard deviation of is the conservative coefficient; AS6: Similarly, repeat steps AS1-AS5 to establish the repayment time limit threshold for secondary account customers. and the repayment time limit threshold for third-tier account customers .

[0012] The specific method of establishing the account amount limit threshold is as follows: BS1: Determine the historical account amount data set of first-level account customers as ,in, Represents the historical average account amount of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; BS2: Calculate the average amount of accounts receivable of all first-level accounts receivable customer groups using formula (7) , formula (7) is: (7); Then, the standard deviation of the accounts receivable amount of all first-level accounts receivable customer groups is calculated using formula (8): , formula (8) is specifically: (8); BS3: Calculate the fluctuation value of the account amount of the first-level account customer group through formula (9) , formula (9) is specifically: (9); BS4: Then, the threshold of the account amount of the first-level account customers is established through formula (10) ; (10); In the formula, is the risk buffer factor; BS5: Similarly, repeat steps BS1-BS4 to establish the threshold of the account amount of secondary account customers. and the threshold of the amount of the account for the third-tier account customers .

[0013] In step 3, determine which level of accounts receivable customers all the current accounts receivable customers belong to, determine the relevant anomaly detection thresholds according to the different levels of accounts receivable customers, and compare the current accounts receivable records of the accounts receivable customers with the relevant anomaly detection thresholds. The comparison results are evaluated through the evaluation model. The specific content of judging whether the current accounts receivable are abnormal according to the evaluation results is as follows: CS1: Select one account receivable record from all current accounts receivable in turn as the target account receivable; CS2: Determine the account customer level of the target accounts receivable; at the same time determine the anomaly detection threshold of this level including the repayment time threshold and account amount limit threshold ,in, Indicates different levels, ; CS3: Determine the current outstanding period and current account amount through the target accounts receivable, recorded as and ; CS4: Get repayment time threshold and the current outstanding period , calculate the time deviation rate , get the current account amount and account amount limit threshold , calculate the amount deviation ratio ; CS5: Time deviation rate by evaluating the model and amount deviation ratio To evaluate, the evaluation model is evaluated by the following formula: (11); (12); (13); In the formula, Expressed as the risk assessment coefficient, and is the weight coefficient, and ; CS6: Get , determine whether the target accounts receivable are abnormal; CS7: Repeat steps CS1-CS6, and select one account receivable record in turn as the target account receivable to determine whether there is any abnormality.

[0014] The specific methods for judging whether the current accounts receivable are abnormal are: like , indicating that no abnormal risk has occurred; like , indicating that there is a low risk of abnormality; like , indicating that there is a high risk of abnormality; in, and These are preset values.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The present invention manages highly sensitive data in accounts receivable in a hierarchical manner through customer classification and dynamic threshold setting, enabling enterprises to accurately identify customers' credit risk levels and potential anomalies in collection behavior; this hierarchical management model not only improves the accuracy of account risk monitoring, but also more effectively ensures the stability of the enterprise's cash flow and profitability; The present invention as a whole establishes a systematic accounts receivable risk management and control system through customer classification, threshold setting, real-time anomaly detection and early warning management; this method uses big data analysis and artificial intelligence technology to identify customer risk levels from historical data and dynamically adjust detection thresholds, thereby accurately monitoring the repayment behavior of different customers, realizing automatic screening and intelligent early warning of abnormal accounts, and reducing financial risks; it improves the efficiency of accounts receivable management, and also significantly reduces the workload and false alarm rate of manual participation, helping enterprises to effectively intervene before potential risks arise and protect capital liquidity. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a flowchart of the steps of the method for detecting anomalies in enterprise accounts receivable data based on artificial intelligence. DETAILED DESCRIPTION

[0017] The present invention is described in detail below in conjunction with specific implementation modes and accompanying drawings.

[0018] 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.

[0019] Embodiment 1 See also Figure 1 ,The present invention provides an enterprise accounts receivable data anomaly detection method based on artificial intelligence, including; Step 1: Identify all the accounts receivable customers of the enterprise and classify them into first-level accounts receivable customers, second-level accounts receivable customers, and third-level accounts receivable customers based on historical accounts receivable records; Step 2: After the account receivable customers are classified into different levels, a corresponding anomaly detection threshold is established for the account receivable customers in each level; Step 3: Then, the current accounts receivable records are detected for anomalies through the evaluation model. First, determine which level of accounts receivable customers the customers in all current accounts receivable records belong to, determine the relevant anomaly detection thresholds according to the different levels of accounts receivable customers, and compare the current accounts receivable records of the accounts receivable customers with the relevant anomaly detection thresholds. The comparison results are evaluated through the evaluation model, and whether the current accounts receivable are abnormal is determined based on the evaluation results. Step 4: Determine the current accounts receivable records with abnormally low risk and the current accounts receivable records with abnormally high risk, and send them to the company management personnel for verification and management.

[0020] Embodiment 2 Based on Example 1, in step 1, the account customer representative enterprise is willing to provide services to this customer first, and the accounts receivable of this customer can be collected later; It should be noted that, for the accounts receivable record, each accounts receivable record includes the customer's name, account start time, specified repayment time, actual repayment time, account amount, cooperation time, etc.; in this embodiment, the accounts receivable record includes historical accounts receivable records and current accounts receivable records, the historical accounts receivable record indicates that the corresponding customer has repaid, and the current accounts receivable record indicates that the corresponding customer has not yet repaid; In step 1, all the accounts receivable customers of the enterprise are identified, and the specific method of classifying the accounts receivable customers according to the historical accounts receivable records is as follows: S1: Select the accounts payable customers determined by the enterprise as target accounts payable customers in turn; S2: Obtain the historical accounts receivable records of the target account customer and determine the specific number of historical accounts receivable records, recorded as ,in, It is to make corresponding changes according to the number of historical accounts receivable records of different target account customers; S3: Then obtain the cooperation time between the target account customer and the enterprise, recorded as ; Specifically, the cooperation time between the target account customer and the enterprise is represented as a period from the time when the target account customer first started to cooperate with the enterprise to the current time; S4: Get the account amount in each historical accounts receivable record of the target account customer, recorded as ,in, , and the average amount of accounts receivable is calculated by formula (1) , formula (1) is specifically: (1); S5: Calculate the difference between the actual repayment time and the stipulated repayment time in each historical accounts receivable record, recorded as , and the average delay days are calculated by formula (2) , formula (2) is specifically: (2); S6: At the same time, determine the number of delayed repayments of the target account customer in the historical accounts receivable records, and calculate the ratio of the number of delayed repayments to the total number of historical accounts receivable records to obtain the overdue repayment frequency, recorded as F; S7: Then, the target account customer rating formula is used to determine the target account customer rating Sc: (3); in, It is a weight parameter, and the specific value is set by professional staff; S8: Score the target account customers according to their grade Classify the target account customers: If the grade score , classify the target account customer into a first-level account customer; If the rating , classify the target account customer into a secondary account customer; If the rating , classify the target account customers into three levels of account customers; in, and S2 are preset values, and the specific parameters are set by professional staff. ; S9: repeating steps S2-S8, and classifying the account receivable customers determined by the selected enterprises as target account receivable customers; It should be noted that, in the above, all accounts payable customers are divided into levels, namely, first-level accounts payable customers, second-level accounts payable customers, and third-level accounts payable customers. The higher the level, the lower the credit rating of the accounts payable customer for the enterprise, and the lower the level, the higher the credit rating of the accounts payable customer for the enterprise.

[0021] Embodiment 3 Based on the second embodiment, step 2: after the account receivable customers are classified into different levels, a corresponding abnormal detection threshold is established for each account receivable customer in each level; The anomaly detection thresholds include the repayment time limit thresholds and the account amount limit thresholds for accounts receivable. Generally, the higher the level of the account customers, the shorter the repayment time limit thresholds and the smaller the account amount set. The lower the level of the account customers, the longer the repayment time limit thresholds and the larger the account amount set. After the account customers are classified into different levels, the specific contents of establishing the corresponding abnormal detection threshold for each account customer in each level include establishing the repayment time limit threshold and the account amount limit threshold: The specific method for establishing the repayment time threshold is: AS1: Determine the historical repayment time data set of first-level account customers: ,in, Represents the historical average repayment time of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; AS2: Calculate the average repayment time of all first-tier account customers using formula (4) , formula (4) is specifically: (4); Then, the standard deviation of repayment time for all first-tier account customers is calculated using formula (5): , formula (5) is specifically: (5); AS3: Sort all the data in set D from small to large, and determine the values ​​of Q1 and Q3 by quartile method, where Q1 is the 25th percentile of the data in set D sorted from small to large, and Q3 is the 75th percentile of the data in set D sorted from small to large; AS4: Limit the data in set D to ,in, , the data in set D exceeds Remove the data in the range and get the data set after removal ,in, It is expressed as the number of total account customers in the first-level account customer group after removal; AS5: Then, use formula (6) to establish the repayment time limit threshold for first-tier account customers ; (6); In the formula, and The data sets The mean and standard deviation of is a conservative coefficient, and its value range is , which can be adjusted according to the enterprise’s risk tolerance; AS6: Similarly, repeat steps AS1-AS5 to establish the repayment time limit threshold for secondary account customers. and the repayment time limit threshold for third-tier account customers .

[0022] Embodiment 4 Based on the third embodiment, the specific method of establishing the account amount limit threshold is as follows: BS1: Determine the historical account amount data set of first-level account customers as ,in, Represents the historical average account amount of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; BS2: Calculate the average amount of accounts receivable of all first-level accounts receivable customer groups using formula (7) , formula (7) is: (7); Then, the standard deviation of the accounts receivable amount of all first-level accounts receivable customer groups is calculated using formula (8): , formula (8) is specifically: (8); BS3: Calculate the fluctuation value of the account amount of the first-level account customer group through formula (9) , formula (9) is specifically: (9); BS4: Then, the threshold of the account amount of the first-level account customers is established through formula (10) ; (10); In the formula, is the risk buffer factor; BS5: Similarly, repeat steps BS1-BS4 to establish the threshold of the account amount of secondary account customers. and the threshold of the amount of the account for the third-tier account customers .

[0023] Embodiment 5 Based on the fourth embodiment, in step 3, it is determined which level of accounts receivable customers the current customers of all accounts receivable belong to, and the relevant abnormality detection threshold is determined according to the accounts receivable customers of different levels, and the current accounts receivable records of the accounts receivable customers are compared with the relevant abnormality detection threshold, and the comparison results are evaluated by the evaluation model. The specific content of judging whether the current accounts receivable are abnormal according to the evaluation results is as follows: CS1: Select one account receivable record from all current accounts receivable in turn as the target account receivable; CS2: Determine the account customer level of the target accounts receivable; at the same time determine the anomaly detection threshold of this level including the repayment time threshold and account amount limit threshold ,in, Indicates different levels, ; CS3: Determine the current outstanding period and current account amount through the target accounts receivable, recorded as and ; CS4: Get repayment time threshold and the current outstanding period , calculate the time deviation rate , get the current account amount and account amount limit threshold , calculate the amount deviation ratio ; CS5: Time deviation rate by evaluating the model and amount deviation ratio To evaluate, the evaluation model is evaluated by the following formula: (11); (12); (13); In the formula, Expressed as the risk assessment coefficient, and is the weight coefficient, and ; CS6: Get , determine whether there are any abnormalities in the target accounts receivable.

[0024] Embodiment 6 Based on the fifth embodiment, the specific method for determining whether the current accounts receivable are abnormal is as follows: like , indicating that no abnormal risk has occurred; like , indicating that there is a low risk of abnormality; like , indicating that there is a high risk of abnormality; in, and These are all preset values, and the specific parameters are set by the staff; CS7: Repeat steps CS1-CS6, and select one account receivable record in turn as the target account receivable to determine whether there is any abnormality; This assessment model combines the time deviation rate and the amount deviation ratio to comprehensively assess the degree of abnormality of account records, and divides abnormal risks into no risk, low risk and high risk according to the preset risk assessment coefficient. In this way, not only can a quick response be made to abnormal situations, but also a quantitative risk rating can be provided for each customer's accounts receivable. This real-time anomaly detection capability enhances the company's risk perception of account recovery, ensuring that financial personnel can intervene in time when risks first appear, thereby effectively preventing potential bad debt risks and helping companies to respond more flexibly to the collection behaviors of different customers. Step 4: Determine the current accounts receivable records with abnormally low risk and the current accounts receivable records with abnormally high risk, and send them to the company management personnel for verification and management.

[0025] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0026] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An artificial intelligence-based method for detecting anomalies in enterprise accounts receivable data, characterized in that: include: Step 1: Identify all the accounts receivable customers of the enterprise, and classify them into first-level accounts receivable customers, second-level accounts receivable customers, and third-level accounts receivable customers based on historical accounts receivable records; Step 2: After classification, establish corresponding abnormal detection thresholds for account customers in each level; Step 3: Perform anomaly detection on the current accounts receivable records through the evaluation model. First, determine the level of accounts receivable customers to which the customers in all current accounts receivable records belong, determine the relevant anomaly detection thresholds according to the different levels of accounts receivable customers, and compare the current accounts receivable records of the accounts receivable customers with the relevant anomaly detection thresholds. The comparison results are evaluated through the evaluation model, and whether the current accounts receivable are abnormal is determined based on the evaluation results. Step 4: Identify current accounts receivable records with abnormally low risk and current accounts receivable records with abnormally high risk, and send them to corporate management for verification and management.

2. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 1 is characterized in that: In the step 1, the specific method of determining all the accounts receivable customers of the enterprise and grading the accounts receivable customers according to the historical accounts receivable records is as follows: S1: Select the accounts payable customers determined by the enterprise as target accounts payable customers in turn; S2: Obtain the historical accounts receivable records of the target account customer and determine the specific number of historical accounts receivable records, recorded as ,in, It is to make corresponding changes according to the number of historical accounts receivable records of different target account customers; S3: Then obtain the cooperation time between the target account customer and the enterprise, recorded as ; S4: Get the account amount in each historical accounts receivable record of the target account customer, recorded as ,in, , and the average amount of accounts receivable is calculated by formula (1) , formula (1) is specifically: （1）； S5: Calculate the difference between the actual repayment time and the stipulated repayment time in each historical accounts receivable record, recorded as , and the average delay days are calculated by formula (2) , formula (2) is specifically: （2）； S6: At the same time, determine the number of delayed repayments of the target account customer in the historical accounts receivable records, and calculate the ratio of the number of delayed repayments to the total number of historical accounts receivable records to obtain the overdue repayment frequency, recorded as F; S7: Then, the target account customer rating formula is used to determine the target account customer rating Sc: （3）； in, is the weight parameter; S8: Score the target account customers according to their grade Divide the target account customers into first-level account customers, second-level account customers, and third-level account customers; S9: Repeat steps S2-S8, and classify the accounts payable customers determined by the selected enterprises as target accounts payable customers.

3. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 2 is characterized in that: The target account customer rating is based on The target account customers are divided into first-level account customers, second-level account customers, and third-level account customers. The specific method is as follows: If the rating , classify the target account customer into a first-level account customer; If the rating , classify the target account customer into a secondary account customer; If the rating , classify the target account customers into three levels of account customers; in, and S2 are preset values, and .

4. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 1, characterized in that: In the step 2, the specific contents of establishing corresponding abnormal detection thresholds for account customers in each level after classification include establishing a repayment time limit threshold and establishing an account amount limit threshold.

5. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 4 is characterized in that: The specific method of establishing the repayment time limit threshold is: AS1: Determine the historical repayment time data set of first-level account customers: ,in, Represents the historical average repayment time of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; AS2: Calculate the average repayment time of all first-tier account customers using formula (4) , formula (4) is specifically: （4）； Then, the standard deviation of repayment time for all first-tier account customers is calculated using formula (5): , formula (5) is specifically: （5）； AS3: Sort all the data in set D from small to large, and determine the values ​​of Q1 and Q3 by quartile method, where Q1 is the 25th percentile of the data in set D sorted from small to large, and Q3 is the 75th percentile of the data in set D sorted from small to large; AS4: Limit the data in set D to ,in, , the data in set D exceeds Remove the data in the range and get the data set after removal ,in, It is expressed as the number of total account customers in the first-level account customer group after removal; AS5: Then, use formula (6) to establish the repayment time limit threshold for first-tier account customers ; （6）； In the formula, and The data sets The mean and standard deviation of is the conservative coefficient; AS6: Similarly, repeat steps AS1-AS5 to establish the repayment time limit threshold for secondary account customers. and the repayment time limit threshold for third-tier account customers .

6. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 4 is characterized in that: The specific method of establishing the account amount limit threshold is: BS1: Determine the historical account amount data set of first-level account customers as ,in, Represents the historical average account amount of each first-level account customer, It is expressed as the total number of accounts payable customers in the first-level accounts payable customer group; BS2: Calculate the average amount of accounts receivable of all first-level accounts receivable customer groups using formula (7) , formula (7) is: （7）； Then, the standard deviation of the accounts receivable amount of all first-level accounts receivable customer groups is calculated using formula (8): , formula (8) is specifically: （8）； BS3: Calculate the fluctuation value of the account amount of the first-level account customer group through formula (9) , formula (9) is specifically: （9）； BS4: Then, the threshold of the account amount of the first-level account customers is established through formula (10) ; （10）； In the formula, is the risk buffer factor; BS5: Similarly, repeat steps BS1-BS4 to establish the threshold of the account amount of secondary account customers. and the threshold of the amount of the account for the third-tier account customers .

7. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 1, characterized in that: In the step 3, the determination is to which level of account customers the customers of all current accounts receivable belong, determine the relevant abnormality detection threshold according to the accounts receivable customers of different levels, and compare the current accounts receivable records of the accounts receivable customers with the relevant abnormality detection threshold, evaluate the comparison results through the evaluation model, and judge whether the current accounts receivable are abnormal according to the evaluation results. The specific content is: CS1: Select one account receivable record from all current accounts receivable in turn as the target account receivable; CS2: Determine the account customer level of the target accounts receivable; at the same time determine the anomaly detection threshold of this level including the repayment time threshold and account amount limit threshold ,in, Indicates different levels, ; CS3: Determine the current outstanding period and current account amount through the target accounts receivable, recorded as and ; CS4: Get repayment time threshold and the current outstanding period , calculate the time deviation rate , get the current account amount and account amount limit threshold , calculate the amount deviation ratio ; CS5: Time deviation rate by evaluating the model and amount deviation ratio To evaluate, the evaluation model is evaluated by the following formula: （11）； （12）； （13）； In the formula, Expressed as the risk assessment coefficient, and is the weight coefficient, and ; CS6: Get , determine whether the target accounts receivable are abnormal; CS7: Repeat steps CS1-CS6, and select one account receivable record in turn as the target account receivable to determine whether there is any abnormality.

8. The method for detecting anomaly in enterprise accounts receivable data based on artificial intelligence according to claim 7 is characterized in that: The specific method for determining whether the current accounts receivable are abnormal is: like , indicating that no abnormal risk has occurred; like , indicating that there is a low risk of abnormality; like , indicating that there is a high risk of abnormality; in, and These are preset values.

Images