A work order management and supervisory control method for financial non-performing asset mediation

Abstract

The application belongs to the technical field of financial data processing, and discloses a work order management and supervision control method for financial non-performing asset mediation, which comprises the following steps: S1, collecting multi-source business data of financial non-performing asset mediation cases, performing unified regulation and structured association, and constructing a unified work order data model with a case ID as the core; S2, based on the unified work order data model, performing weighted scoring according to multi-dimensional disposal adaptation indexes, and executing work order intelligent distribution; S3, based on the distributed work order information and the case characteristic vector, performing multi-dimensional search matching and automatically backfilling associated data; S4, according to the search backfilling result and the communication negotiation state, dynamically rendering a work order form and completing information input, and driving the automatic flow of the work order state; S5, recording logs for the whole process of work order input and state flow, calculating a dynamic supervision threshold based on work order processing data, and realizing real-time early warning of abnormal work orders.

Description

Technical Field

[0001] This invention relates to the field of financial data processing technology, specifically to a work order management and supervision control method for the mediation of non-performing financial assets. Background Technology

[0002] Against the backdrop of rapid development in fintech and big data processing technologies, the disposal of non-performing financial assets and dispute mediation are gradually transforming towards digitalization, online processes, and streamlined workflows. To improve case handling efficiency, standardize operational procedures, and strengthen compliance management, the industry widely adopts a collaborative technical model combining work order management systems and call center systems. This utilizes information technology to achieve full-process control, including case information collection, mediator assignment, negotiation and communication, business information entry, and supervision of the disposal process. This transforms the traditional, decentralized, manual processing model into a centralized, standardized, and traceable digital operation mode.

[0003] In the existing data processing of work order management and process supervision, various calculation weights, judgment thresholds, and model parameters all use fixed values ​​that are pre-set manually. The system cannot make real-time dynamic adjustments and adaptive optimizations based on changes in business data distribution, case characteristics, personnel work status, and handling scenarios. As business continues to operate and data accumulates, fixed rules and standards will gradually deviate from actual business characteristics, directly leading to a decrease in the accuracy of work order allocation, a lag in process supervision and early warning, and a decrease in the accuracy of risk identification, making it difficult to maintain stable and efficient processing performance in the long term. Summary of the Invention

[0004] The purpose of this invention is to provide a work order management and supervision control method for the mediation of non-performing financial assets, which solves the problem that the existing technology uses fixed settings for calculating weights and judgment thresholds, which cannot be adaptively adjusted according to the distribution of business data, resulting in a decrease in system processing accuracy and stability.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for managing and monitoring work orders for the reconciliation of non-performing financial assets includes the following steps: S1. Collect multi-source business data of financial non-performing asset mediation cases, perform unified organization and structured association, and construct a unified work order data model with case ID as the core; S2. Based on the unified work order data model, perform weighted scoring according to multi-dimensional processing adaptation indicators, and execute intelligent work order allocation; S3. Based on the allocated work order information and case feature vectors, perform multi-dimensional retrieval and matching and automatically backfill related data; S4. Based on the search and feedback results and communication and negotiation status, dynamically render the work order form and complete the information entry, and drive the automatic flow of the work order status. S5. Log the entire process of work order entry and status transition, calculate dynamic monitoring thresholds based on work order processing data, and realize real-time early warning of abnormal work orders. S6. Based on the early warning results and work order handling performance feedback, incrementally optimize the allocation weights, retrieval weights, and supervision parameters, and iteratively update them to the unified work order data model.

[0006] Preferably, the multi-source business data includes basic case information, debt amount, overdue days, party information, records of dishonesty, information on joint debts, and records of judicial disposal. Step S1 specifically includes: S11. Perform numerical normalization on the two types of continuous business data collected: amount of debt and number of overdue days. S12. For the four discrete attributes corresponding to the normalized data, namely party information, credit record, joint debt information, and judicial disposal record, feature encoding is performed one by one. S13. Take all the data that has been normalized and feature-encoded, establish a multi-dimensional association index with the case ID as the core, and build a unified work order data model.

[0007] Preferably, step S2 specifically includes: S21. Based on the corresponding data within the unified work order data model, construct the case priority calculation formula: ,in, Score the case priority. , , These are the weighting coefficients. For the amount of the debt, The number of overdue days This represents the urgency level coefficient. S22. Based on the obtained case priority scores, simultaneously construct the formula for calculating mediator suitability: ,in, For mediator suitability, For historical mediation success rates, For expertise matching, This represents the current pending workload. For online status coefficients; S23. Based on the mediator’s current pending workload, calculate the team’s average workload value and, combined with the mediator’s suitability, restrict new work orders from being received by mediators who exceed the load threshold. S24. Within the range of mediators selected after load constraint screening, select the mediator with the highest comprehensive score to complete the work order assignment and update the work order running status simultaneously.

[0008] Preferably, step S2 further includes: when multiple mediators have the same fit score, constructing a formula for calculating the mediator's fair gain value: ,in, For fair gain value, For the first The mediators are overwhelmed. For the first Success rate of mediators. This represents the team's average success rate.

[0009] Preferably, step S3 specifically includes: S31. Based on the case feature vector corresponding to the assigned work order, construct search constraints in sequence according to case ID, party's name, mobile phone number, and ID number; S32. Construct a formula for calculating the search matching score based on the search constraints: ,in, To retrieve the total matching score, , , For the weight of the matching item, Matching value for case ID, For name similarity, The value is matched to the mobile phone number; S33. Using the character edit distance recursive formula, with character insertion, deletion, and replacement as the basic operations, find the minimum number of operation steps required to convert between the name strings of two parties: , where the string Length is , string Length is , For the two strings before , Minimum number of editing steps for a character. For the corresponding character deletion operation, For the corresponding character insertion operation, For the corresponding character replacement operation; Then, the similarity score of the parties' names is obtained by converting the normalized relation: ,in, This is the final calculated name similarity score. The lengths of the strings representing the names of the two parties involved; S34. Based on the search matching score, sort the search results in descending order and automatically backfill the corresponding case-related data for the first matching result.

[0010] Preferably, step S4 specifically includes: S41. Based on the case-related data obtained from the retrieval and backfilling, match different communication and negotiation statuses and display the corresponding work order form fields in a differentiated manner; S42. Based on the differentiated display of work order form fields, the content draft is stored synchronously during the work order information entry process; S43. Based on the field content entered in the form, construct the formula for calculating the completeness of the work order: ,in, To score for completeness, To ensure the number of fields are filled effectively, Number of required fields Number of optional fields to fill in; S44. After the work order completeness verification is passed, the work order status will automatically flow through each level.

[0011] Preferably, step S5 specifically includes: S51. For all actions related to work order form entry and status transition, execute an immutable log to record each entry, and construct a dynamic monitoring threshold calculation formula based on the work order processing data recorded in the log: ,in, For dynamic monitoring thresholds, The mean of the sliding window. The standard deviation of the sliding window. Confidence coefficient; S52. Calculate the actual early warning rate of work orders over historical periods, and construct an adaptive correction formula for the confidence coefficient based on the preset standard early warning rate: ,in, To correct the confidence coefficient, This is the current confidence coefficient. To adaptively adjust the step size coefficient, To set a pre-defined standard warning rate, This represents the actual statistical early warning rate; S53. Compare the actual processing data of the work order with the corrected dynamic monitoring threshold, mark the work orders that exceed the threshold as abnormal work orders and trigger the warning indicator.

[0012] Preferably, step S5 further includes: identifying the same case, the same mediator, and the same statistical time period, and constructing an aggregation calculation formula for abnormal scoring risk based on multiple abnormal work order records within the time period: ,in, For risk aggregation value, This represents the number of abnormal samples. For a single abnormal score, The polymerization coefficient, This represents the number of abnormal work orders. This represents the total number of work orders. The risk aggregation value is compared with a preset threshold, and cases and mediators that exceed the threshold are marked with a key supervision mark.

[0013] Preferably, step S6 specifically includes: S61. Based on the work order warning results and actual handling records, generate multi-dimensional performance indicators. S62. Based on the aforementioned multi-dimensional performance indicators, construct the weight incremental update calculation formula: ,in, For the updated weights, For the weights before the update, To learn step length, This represents the actual performance value. The target performance value; S63. During the incremental weight update process, a single weight fluctuation threshold is introduced, a weight amplitude constraint determination formula is constructed, and the amplitude truncation control of the initial updated weight is performed based on the formula: ,in, For the new weights, As the current weight, The final weights after constraints The maximum fluctuation threshold for a single weight is preset; S64. Synchronously integrate all weights adjusted by fluctuation constraints into the unified work order data model to form a closed-loop business iteration.

[0014] Preferably, step S6 further includes: introducing a historical inertia correction term during the regular weight incremental update process to construct a stable weight update calculation formula: ,in, For the new weights, As the current weight, Weighted according to the previous version This is the inertia correction factor; Based on the updated weights, the dispatch scoring model, retrieval scoring model, and supervision threshold model are reconstructed and synchronously updated to the unified work order data model.

[0015] By adopting the above technical solution, the present invention has the following advantages compared with the prior art: 1. This invention provides a work order management and supervision control method for the mediation of non-performing financial assets. It constructs a unified work order data model with case ID as the core, adopts multi-dimensional weighted scoring and dynamic load constraints to realize intelligent work order allocation, and combines fair gain calculation to ensure the balance of work order allocation. It overcomes the defects of traditional fixed rules or manual assignment such as low matching accuracy, uneven load, and insufficient fairness, and improves the suitability of cases and mediators and the utilization rate of resources.

[0016] 2. This invention provides a work order management and supervision control method for the mediation of non-performing financial assets. It achieves multi-dimensional retrieval and matching based on case feature vectors, improves the accuracy of name similarity calculation by using character edit distance algorithm, completes automatic backfilling of associated data, greatly reduces manual repetitive input, reduces information error rate, and improves the system's automation level and work efficiency.

[0017] 3. This invention provides a work order management and supervision control method for the mediation of non-performing financial assets. The work order form is dynamically rendered according to the communication and negotiation status. The status is automatically transferred through the verification of the completeness of the form. The entire process is traceable by using an immutable log. The method realizes real-time early warning of abnormal work orders based on sliding windows and adaptive thresholds. It also combines multi-dimensional risk aggregation to identify key regulatory targets, thereby improving the timeliness, accuracy and proactive prevention and control capabilities of monitoring.

[0018] 4. This invention provides a work order management and supervision control method for the mediation of non-performing financial assets. Based on the early warning results and disposal performance, it realizes incremental iterative optimization of allocation weight, retrieval weight, and supervision parameters. It introduces weight fluctuation constraints and historical inertia correction terms to form a complete business closed loop, enabling the system to continuously and adaptively adjust with the distribution of business data, maintain stable processing accuracy in the long term, avoid system performance degradation due to data feature deviation, and extend the effective life cycle of the system. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0021] Example Please refer to Figure 1 As shown, this invention discloses a work order management and supervision control method for the mediation of non-performing financial assets, including the following steps: S1. Collect multi-source business data of financial non-performing asset mediation cases, perform unified standardization and structured association, and construct a unified work order data model with case ID as the core.

[0022] The multi-source business data comes from the case management system, call center platform, credit data interface, judicial public information and third-party data platform. The data types cover structured data, semi-structured data and unstructured data. The collection, cleaning, deduplication and format unification are completed through ETL process to provide a high-quality data foundation for subsequent modeling.

[0023] Multi-source business data includes basic case information, debt amount, overdue days, party information, credit default records, joint debt information, and judicial disposal records.

[0024] S11. Perform numerical normalization on the collected continuous business data, including the amount of debt and the number of overdue days. This normalization process eliminates differences in magnitude and units between the amount and the number of days, ensuring that different types of features can participate in calculations within a unified numerical range. This avoids model distortion due to excessively large or small values, improving the stability and accuracy of subsequent weighted scoring and matching retrieval. The normalization formula is: ,in, The values ​​are normalized. This is the original data. This is the minimum value for this type of data. This represents the maximum value for this type of data; S12. For the four discrete attributes corresponding to the normalized data—party information, records of dishonesty, information on joint debts, and records of judicial disposal—feature encoding is performed one by one. Discrete attributes are encoded using one-hot encoding, label encoding, or hash encoding to convert textual category information into numerical features that can be recognized and calculated by computers, thereby achieving a quantitative representation of non-numerical attributes and enhancing the model's ability to express case risks and party characteristics.

[0025] S13. Take all the data that has been normalized and feature-encoded, establish a multi-dimensional association index with the case ID as the core, and build a unified work order data model.

[0026] The unified work order data model uses the case ID as the primary key to link and bind case information, party information, handling information, call information, and supervision information, forming a single data source that is traceable, searchable, and scalable, providing unified data support for intelligent processing throughout the entire process.

[0027] S2. Based on a unified work order data model, a weighted score is calculated according to multi-dimensional handling adaptation indicators, and intelligent work order allocation is performed.

[0028] S21. Based on the corresponding data within the unified work order data model, construct the case priority calculation formula: ,in, Score the case priority. , , These are the weighting coefficients. For the amount of the debt, The number of overdue days This represents the urgency level coefficient. By using a logarithmic function to compress the amount and number of days, we can avoid extreme values ​​from having an excessive impact on the priority results, making the priority distribution smoother and more reasonable, and objectively reflecting the urgency of case handling.

[0029] S22. Based on the obtained case priority scores, a formula for calculating mediator suitability is constructed simultaneously. Suitability comprehensively considers the mediator's ability, professional expertise, workload, and online status, making the dispatching of cases more in line with the actual handling scenarios and improving the efficiency of case connection and the success rate of mediation. ,in, For mediator suitability, For historical mediation success rates, For expertise matching, This represents the current pending workload. For online status coefficients; S23. Based on the mediators' current pending workload, calculate the team's average workload. Combined with mediator suitability, restrict new work orders for mediators exceeding the load threshold. Through dynamic load balancing, prevent individual mediators from being overloaded, ensure overall team workload balance, and prevent service quality degradation and process backlog caused by overload. S24. Within the range of mediators selected after load constraint screening, select the mediator with the highest comprehensive score to complete the work order assignment and update the work order running status simultaneously.

[0030] Step S2 also includes: when multiple mediators have the same fit score, constructing a formula for calculating the mediator's fair gain value: ,in, For fair gain value, For the first The mediators are overwhelmed. For the first Success rate of mediators. This represents the team's average success rate.

[0031] Fairness gain is used to achieve balanced task assignment under the same fit, taking into account workload and historical performance, improving the fairness and rationality of the allocation mechanism, and avoiding the long-term concentration of resources in a few people.

[0032] S3. Based on the assigned work order information and case feature vector, perform multi-dimensional retrieval and matching and automatically backfill related data.

[0033] S31. Based on the case feature vector corresponding to the assigned work order, construct search constraints in sequence according to case ID, party's name, mobile phone number, and ID number; S32. Construct a formula for calculating the search matching score based on the search constraints: ,in, To retrieve the total matching score, , , For the weight of the matching item, Matching value for case ID, For name similarity, The value is matched to the mobile phone number; S33. Using the character edit distance recursive formula, with character insertion, deletion, and replacement as the basic operations, find the minimum number of operation steps required to convert between the name strings of two parties: , where the string Length is , string Length is , For the two strings before , Minimum number of editing steps for a character. For the corresponding character deletion operation, For the corresponding character insertion operation, For the corresponding character replacement operation; Then, the similarity score of the parties' names is obtained by converting the normalized relation: ,in, This is the final calculated name similarity score. The lengths of the strings representing the names of the two parties involved; Character edit distance can effectively handle situations such as incorrect name input, homophones, and similar character shapes, improve fuzzy matching capabilities, and make search results closer to the real case subjects.

[0034] S34. Based on the search matching score, sort the search results in descending order and automatically backfill the corresponding case-related data for the first matching result.

[0035] S4. Based on the search and feedback results and communication and negotiation status, dynamically render the work order form, complete the information entry, and drive the automatic flow of work order status.

[0036] S41. Based on the case-related data obtained from the retrieval and backfilling, match different communication and negotiation statuses and display the corresponding work order form fields in a differentiated manner. That is, the form interface is automatically adjusted according to different stages such as initial contact, repayment negotiation, plan confirmation, objection handling, and case closure.

[0037] S42. Based on the differentiated display of work order form fields, the content draft is stored synchronously during the work order information entry process to prevent information loss due to page refresh, network disconnection, or accidental operation.

[0038] S43. Based on the field content entered in the form, construct the formula for calculating the completeness of the work order: ,in, To score for completeness, To ensure the number of fields are filled effectively, Number of required fields Number of optional fields to fill in; S44. After the work order completeness verification is passed, the work order status will automatically flow through each level.

[0039] S5. Log the entire process of work order entry and status transition, calculate dynamic monitoring thresholds based on work order processing data, and realize real-time early warning of abnormal work orders.

[0040] S51. For all actions related to work order form entry and status transition, execute an immutable log to record each entry, and construct a dynamic monitoring threshold calculation formula based on the work order processing data recorded in the log: ,in, For dynamic monitoring thresholds, The mean of the sliding window. The standard deviation of the sliding window. Confidence coefficient; S52. Calculate the actual early warning rate of work orders over historical periods, and construct an adaptive correction formula for the confidence coefficient based on the preset standard early warning rate: ,in, To correct the confidence coefficient, This is the current confidence coefficient. To adaptively adjust the step size coefficient, To set a pre-defined standard warning rate, This represents the actual statistical early warning rate; S53. Compare the actual processing data of the work order with the corrected dynamic monitoring threshold, mark the work orders that exceed the threshold as abnormal work orders and trigger the warning indicator.

[0041] Step S5 also includes: identifying the same case, the same mediator, and the same statistical period; and constructing an aggregation calculation formula for anomaly scoring risk based on multiple abnormal work order records within the period. ,in, For risk aggregation value, This represents the number of abnormal samples. For a single abnormal score, The polymerization coefficient, This represents the number of abnormal work orders. This represents the total number of work orders. By comparing the aggregated risk value with a preset threshold, cases and mediators that exceed the threshold are marked with a key supervision indicator. Risk aggregation enables the identification of overall risks from single-point anomalies, facilitating early intervention and key control by management personnel, and improving the accuracy and effectiveness of supervision.

[0042] S6. Based on the early warning results and work order handling performance feedback, incrementally optimize the allocation weights, retrieval weights, and monitoring parameters, and iteratively update them to the unified work order data model.

[0043] S61. Based on the work order warning results and actual handling records, generate multi-dimensional performance indicators. S62. Based on the aforementioned multi-dimensional performance indicators, construct the weight incremental update calculation formula: ,in, For the updated weights, For the weights before the update, To learn step length, This represents the actual performance value. The target performance value; S63. During the incremental weight update process, a single weight fluctuation threshold is introduced, a weight amplitude constraint determination formula is constructed, and the amplitude truncation control of the initial updated weight is performed based on the formula: ,in, For the new weights, As the current weight, The final weights after constraints The maximum fluctuation threshold for a single weight is preset; S64. Synchronously integrate all weights adjusted by fluctuation constraints into the unified work order data model to form a closed-loop business iteration.

[0044] Step S6 also includes: introducing a historical inertia correction term during the regular incremental weight update process to construct a stable weight update calculation formula: ,in, For the new weights, As the current weight, Based on the weight of the previous version, This is the inertia correction factor; Based on the updated weights, the dispatch scoring model, retrieval scoring model, and supervision threshold model are reconstructed and synchronously updated to the unified work order data model. Through incremental learning, fluctuation limitation, and inertia correction, the parameter updates are smooth, the system operation is stable, model oscillation is avoided, and the system is guaranteed to adapt to business changes in the long term without decay.

[0045] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for work order management and supervision control for the reconciliation of non-performing financial assets, characterized in that, Includes the following steps: S1. Collect multi-source business data of financial non-performing asset mediation cases, perform unified organization and structured association, and construct a unified work order data model with case ID as the core; S2. Based on the unified work order data model, perform weighted scoring according to multi-dimensional processing adaptation indicators, and execute intelligent work order allocation; S3. Based on the allocated work order information and case feature vectors, perform multi-dimensional retrieval and matching and automatically backfill related data; S4. Based on the search and feedback results and communication and negotiation status, dynamically render the work order form and complete the information entry, and drive the automatic flow of the work order status. S5. Log the entire process of work order entry and status transition, calculate dynamic monitoring thresholds based on work order processing data, and realize real-time early warning of abnormal work orders. S6. Based on the early warning results and work order handling performance feedback, incrementally optimize the allocation weights, retrieval weights, and supervision parameters, and iteratively update them to the unified work order data model.

2. The work order management and supervision control method for the mediation of non-performing financial assets as described in claim 1, characterized in that: The multi-source business data includes basic case information, debt amount, overdue days, party information, records of dishonesty, information on joint debts, and records of judicial disposal. Step S1 specifically involves: S11. Perform numerical normalization on the two types of continuous business data collected: amount of debt and number of overdue days. S12. For the four discrete attributes corresponding to the normalized data, namely party information, credit record, joint debt information, and judicial disposal record, feature encoding is performed one by one. S13. Take all the data that has been normalized and feature-encoded, establish a multi-dimensional association index with the case ID as the core, and build a unified work order data model.

3. The work order management and supervision control method for resolving non-performing financial assets as described in claim 1, characterized in that, Step S2 is as follows: S21. Based on the corresponding data within the unified work order data model, construct the case priority calculation formula: ,in, Score the case priority. , , These are the weighting coefficients. For the amount of the debt, The number of overdue days This represents the urgency level coefficient. S22. Based on the obtained case priority scores, simultaneously construct the formula for calculating mediator suitability: ,in, For mediator suitability, For historical mediation success rates, For expertise matching, This represents the current pending workload. For online status coefficients; S23. Based on the mediator’s current pending workload, calculate the team’s average workload value and, combined with the mediator’s suitability, restrict new work orders from being received by mediators who exceed the load threshold. S24. Within the range of mediators selected after load constraint screening, select the mediator with the highest comprehensive score to complete the work order assignment and update the work order running status simultaneously.

4. The work order management and supervision control method for resolving non-performing financial assets as described in claim 3, characterized in that, Step S2 also includes: when multiple mediators have the same fit score, constructing a formula for calculating the mediator's fair gain value: ,in, For fair gain value, For the first The mediators are overwhelmed. For the first Success rate of mediators. This represents the team's average success rate.

5. The work order management and supervision control method for resolving non-performing financial assets as described in claim 1, characterized in that, Step S3 is as follows: S31. Based on the case feature vector corresponding to the assigned work order, construct search constraints in sequence according to case ID, party's name, mobile phone number, and ID number; S32. Construct a formula for calculating the search matching score based on the search constraints: ,in, To retrieve the total matching score, , , For the weight of the matching item, Matching value for case ID, For name similarity, The value is matched to the mobile phone number; S33. Using the character edit distance recursive formula, with character insertion, deletion, and replacement as the basic operations, find the minimum number of operation steps required to convert between the name strings of two parties: , where the string Length is , string Length is , For the two strings before , Minimum number of editing steps for a character. For the corresponding character deletion operation, For the corresponding character insertion operation, For the corresponding character replacement operation; Then, the similarity score of the parties' names is obtained by converting the normalized relation: ,in, This is the final calculated name similarity score. The lengths of the strings representing the names of the two parties involved; S34. Based on the search matching score, sort the search results in descending order and automatically backfill the corresponding case-related data for the first matching result.

6. The work order management and supervision control method for resolving non-performing financial assets as described in claim 1, characterized in that, Step S4 is as follows: S41. Based on the case-related data obtained from the retrieval and backfilling, match different communication and negotiation statuses and display the corresponding work order form fields in a differentiated manner; S42. Based on the differentiated display of work order form fields, the content draft is stored synchronously during the work order information entry process; S43. Based on the field content entered in the form, construct the formula for calculating the completeness of the work order: ,in, To score for completeness, To ensure the number of fields are filled effectively, Number of required fields Number of optional fields to fill in; S44. After the work order completeness verification is passed, the work order status will automatically flow through each level.

7. The work order management and supervision control method for the mediation of non-performing financial assets as described in claim 1, characterized in that, Step S5 is as follows: S51. For all actions related to work order form entry and status transition, execute an immutable log to record each entry, and construct a dynamic monitoring threshold calculation formula based on the work order processing data recorded in the log: ,in, For dynamic monitoring thresholds, The mean of the sliding window. The standard deviation of the sliding window. Confidence coefficient; S52. Calculate the actual early warning rate of work orders over historical periods, and construct an adaptive correction formula for the confidence coefficient based on the preset standard early warning rate: ,in, To correct the confidence coefficient, This is the current confidence coefficient. To adaptively adjust the step size coefficient, To set a pre-defined standard warning rate, This represents the actual statistical early warning rate; S53. Compare the actual processing data of the work order with the corrected dynamic monitoring threshold, mark the work orders that exceed the threshold as abnormal work orders and trigger the warning indicator.

8. The work order management and supervision control method for resolving non-performing financial assets as described in claim 7, characterized in that, Step S5 also includes: identifying the same case, the same mediator, and the same statistical period; and constructing an aggregation calculation formula for anomaly scoring risk based on multiple abnormal work order records within the period. ,in, For risk aggregation value, This represents the number of abnormal samples. For a single abnormal score, The polymerization coefficient, This represents the number of abnormal work orders. This represents the total number of work orders. The risk aggregation value is compared with a preset threshold, and cases and mediators that exceed the threshold are marked with a key supervision mark.

9. The work order management and supervision control method for resolving non-performing financial assets as described in claim 1, characterized in that, Step S6 is as follows: S61. Based on the work order warning results and actual handling records, generate multi-dimensional performance indicators. S62. Based on the aforementioned multi-dimensional performance indicators, construct the weight incremental update calculation formula: ,in, For the updated weights, For the weights before the update, To learn step length, This represents the actual performance value. The target performance value; S63. During the incremental weight update process, a single weight fluctuation threshold is introduced, a weight amplitude constraint determination formula is constructed, and the amplitude truncation control of the initial updated weight is performed based on the formula: ,in, For the new weights, As the current weight, The final weights after constraints The maximum fluctuation threshold for a single weight is preset; S64. Synchronously integrate all weights adjusted by fluctuation constraints into the unified work order data model to form a closed-loop business iteration.

10. The work order management and supervision control method for resolving non-performing financial assets as described in claim 9, characterized in that, Step S6 also includes: introducing a historical inertia correction term during the regular incremental weight update process to construct a stable weight update calculation formula: ,in, For the new weights, As the current weight, Weighted according to the previous version This is the inertia correction factor; Based on the updated weights, the dispatch scoring model, retrieval scoring model, and supervision threshold model are reconstructed and synchronously updated to the unified work order data model.

Images