Enterprise financial process intelligent arrangement monitoring system and method

Abstract

The invention relates to an enterprise financial process intelligent arrangement monitoring system and method, and the system comprises a process modeling module which carries out the modeling based on a BPMN standard, introduces a process meta-model, and achieves the flexible combination and replacement of task units; the large model driving module is used for generating flow path suggestions by utilizing a fine-tuned large language model in the financial field; the flow scheduling optimization module is used for dynamically adjusting the flow priority according to the financial statement data; the reinforcement learning optimization module is used for optimizing a process strategy through a PPO or DQN algorithm; the compliance checking module is used for carrying out semantic-level compliance judgment in combination with the knowledge graph; and the natural language interaction module supports multi-mode interaction and outputs interpretable results. The method comprises the steps of modeling, path suggestion generation, dynamic priority adjustment, strategy optimization, compliance judgment, interactive response and the like. The method improves the flexibility, adaptability and compliance of the process, is suitable for various financial process scenes, and has wide application prospects and commercial values.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of financial process management, and particularly relates to an enterprise financial process intelligent arrangement and monitoring system and method. BACKGROUND

[0002] In modern enterprise operation, financial process management is a key link to ensure efficient operation and compliance operation of enterprises. However, the existing financial process management systems generally have some limitations. For example, the process arrangement method is rigid, and lacks the ability to automatically adjust the process path and task priority according to the real-time business state (such as financial statement data changes, resource availability, external policy changes, etc.), resulting in low efficiency of process execution and difficulty in adapting to complex and changing enterprise operation environment. In addition, most existing systems only focus on whether the process is completed, and ignore the influence of process execution on financial indicators (such as cash flow, cost control, profit structure, etc.), so that there is a lack of linkage optimization between process scheduling and financial performance, and it is difficult for enterprises to optimize financial goals through process management. At the same time, when the financial process is delayed, conflicted or abnormally operated, the traditional system often cannot automatically identify and recommend alternative paths or remedial measures, affecting the stability and integrity of process execution, and even may cause compliance risks. SUMMARY

[0003] (I) Invention purpose

[0004] In order to overcome the above shortcomings, the purpose of the present application is to provide an enterprise financial process intelligent arrangement and monitoring system and method to solve the above technical problems.

[0005] (II) Technical solution

[0006] In order to achieve the above purpose, the technical solution provided by the present application is as follows:

[0007] An enterprise financial process intelligent arrangement and monitoring system, comprising:

[0008] A process modeling module for modeling enterprise financial processes based on the BPMN standard, supporting visual process design, and introducing a process meta-model to abstract process nodes into combinable and replaceable task units, each task unit containing input and output parameters, execution conditions and constraints, optional execution paths and alternative solutions;

[0009] A large model driven module that uses a large language model fine-tuned in the financial field to receive process context information, generates process path suggestions consistent with business logic after analyzing input semantics, and supports natural language generation of process description;

[0010] The process scheduling optimization module periodically obtains the latest financial statement data from the financial system, analyzes the trend of key financial indicators through the large model, and dynamically adjusts the process priority and execution path according to the financial report analysis results.

[0011] The reinforcement learning optimization module records the process execution results, compliance judgment and user feedback, and uses reinforcement learning algorithm to train the large model to optimize the process strategy. The learning goals include minimizing the process execution time, maximizing the process compliance score, and improving the positive impact of the process on financial indicators.

[0012] The compliance checking module uses the large model to make semantic-level compliance judgments during the process running, and judges whether there is risk according to the compliance rules in the knowledge graph. High-risk operations are automatically marked and the process is suspended. At the same time, audit logs and explanatory reports are generated.

[0013] The natural language interaction module allows users to interact with the system through a natural language interface. The large model generates responses according to user intent, and outputs results including conclusions, reasoning processes and data support.

[0014] Preferably, the process modeling module supports the construction and version management of the process template library, facilitating cross-departmental reuse and rapid deployment.

[0015] Preferably, the large model driven module can identify key tasks and recommend optimization paths according to user input natural language instructions.

[0016] Preferably, the process scheduling optimization module automatically triggers the payment condition change process when it finds that the customer's credit rating has decreased, and inserts the budget review process when it finds that the cost has exceeded the standard.

[0017] Preferably, the reinforcement learning optimization module uses PPO or DQN algorithm for training, supporting policy version management and A / B testing.

[0018] Preferably, the compliance checking module combines the compliance rules in the knowledge graph to determine whether the operation meets the requirements of the internal control matrix, budget control and cost limit.

[0019] Preferably, the natural language interaction module supports Web dialog box, mobile App, and voice assistant interaction modes.

[0020] An enterprise financial process intelligent arrangement and monitoring method, comprising the following steps:

[0021] S1 model the enterprise financial process based on the BPMN standard, and introduce a process meta-model to abstract the process nodes into task units;

[0022] S2 receives process context information using a large language model fine-tuned in the financial field, generates process path suggestions after analyzing input semantics;

[0023] S3 periodically obtains financial statement data from the financial system, analyzes the trend of key financial indicators through a large model, and dynamically adjusts process priority and execution path;

[0024] S4 records process execution results, compliance judgments and user feedback, and uses reinforcement learning algorithm to train large model to optimize process strategy;

[0025] S5 uses a large model to make semantic-level compliance judgments during the process, combines compliance rules in the knowledge graph to judge risks, and automatically marks and suspends the process for high-risk operations;

[0026] S6 users interact with the system through a natural language interface, and the large model generates responses according to user intent, outputting results including conclusions, reasoning processes and data support.

[0027] Preferably, the S1 supports the construction and version management of the process template library, facilitating cross-departmental reuse and rapid deployment;

[0028] Preferably, in S4, the learning objectives include minimizing process execution time, maximizing process compliance score, and improving the positive impact of the process on financial indicators.

[0029] Beneficial effects:

[0030] 1. Improve process flexibility and adaptability: Through process modeling based on BPMN standard and dynamic path generation driven by large model, the system can automatically adjust process path and task priority according to real-time business state (such as financial data changes, resource availability, external policy adjustments, etc.), effectively solving the problem of rigid process arrangement in existing systems, and significantly improving the adaptability of the process to complex and changing enterprise operation environment.

[0031] 2. Realize financial goal-oriented process optimization: The system can analyze financial statement data in real time, dynamically adjust process scheduling strategy, and ensure that process execution is closely linked with financial goals. For example, when a cost exceeds the standard, the system automatically inserts a budget review process to optimize cost control; when cash flow is tight, prioritize key collection tasks to ensure the financial health of the enterprise, achieving coordinated optimization of process and financial performance.

[0032] 3. Enhanced adaptive repair capability for abnormal processes: With the semantic understanding of large models and the strategy optimization of reinforcement learning, the system can automatically identify problems and recommend alternative paths or remedial measures when the process appears to be delayed, conflicted, or abnormally operated. For example, when a key task is delayed due to insufficient resources, the system can dynamically adjust the task priority and recommend the optimal alternative solution to ensure the stability and integrity of the process, reducing the impact of abnormalities on business. DETAILED DESCRIPTION

[0033] To make the objectives, technical solutions, and advantages of the present application clearer, further detailed descriptions of the present application are given below in conjunction with specific embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present application. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concept of the present application.

[0034] The enterprise financial process intelligent orchestration monitoring system provided by the present application comprises:

[0035] A process modeling module is used to model the enterprise financial process based on the BPMN standard, supports visual process design, and introduces a process meta-model to abstract process nodes into combinable and replaceable task units, each task unit containing input and output parameters, execution conditions and constraints, optional execution paths and alternative solutions. The process modeling module supports the construction and version management of the process template library, facilitating cross-department reuse and rapid deployment.

[0036] A large model driven module receives process context information, including user instruction financial report data, and system document, using a large language model fine-tuned in the financial field. After analyzing the input semantics, it generates process path suggestions that conform to business logic, such as:

[0037] User input: "I need to speed up the month-end closing process", the system can identify key tasks that affect closing, such as voucher review and depreciation, and recommend an accelerated path

[0038] When a certain expense expenditure exceeds the budget, the system can automatically insert an additional approval link:

[0039] The large model driven module supports natural language generation of process description, which can identify key tasks and recommend optimization paths according to user input natural language instructions.

[0040] The process scheduling optimization module periodically obtains the latest financial statement data from the financial system, analyzes the change trend of key financial indicators through the large model, the change trend includes income decline, cash flow tension, overdue accounts receivable increase, dynamically adjusts the process priority and execution path according to the financial report analysis result, for example: when it is found that the credit rating of a certain customer decreases, the system automatically triggers the payment condition change process; when it is found that a certain type of cost is continuously over-standard, the system inserts the budget review process, the process scheduling optimization module automatically triggers the payment condition change process when it is found that the credit rating of a customer decreases, inserts the budget review process when it is found that the cost is over-standard, realizes the linkage of process scheduling and financial target, and improves the financial sensitivity of process execution.

[0041] The reinforcement learning optimization module records the association relationship between process execution results, compliance judgment and user feedback, trains the large model to optimize the process strategy by using the reinforcement learning algorithm, and the learning goals include minimizing the process execution time, maximizing the process compliance score, and improving the positive influence of the process on the financial indicators, the reinforcement learning optimization module adopts PPO or DQN algorithm for training, supports strategy version management and A / B testing.

[0042] The compliance checking module uses the large model to make semantic-level compliance judgment during the process running, judges whether there is risk in combination with the compliance rules in the knowledge graph, specifically including: whether it violates the internal control matrix; whether there is suspicion of repeated payment or false transaction; whether it meets the budget control and cost limit,

[0043] High-risk operations are automatically marked and the process is suspended, and audit logs and explanatory reports are generated.

[0044] The natural language interaction module allows users to interact with the system through a natural language interface, and the large model generates responses according to user intent, outputs results including conclusions, reasoning processes and data support, and improves transparency and credibility. The natural language interaction module supports Web dialog box, mobile App, and voice assistant interaction modes.

[0045] By introducing the process modeling module, the modularity and scalability of process arrangement are realized; unlike the rigid structure of traditional process alert, the present application supports flexible combination and dynamic replacement of task units, and improves process adaptability.

[0046] Through the understanding ability of the large model to the semantics of the financial statements, the automatic generation of process scheduling strategy and path recommendation based on the change of financial indicators are realized; the key indicator trends in the financial statements (such as income decline, cost increase, and liquidity tension) can be automatically identified, and the process priority and execution path are dynamically adjusted accordingly,

[0047] By constructing a financial process and compliance knowledge graph and integrating it with a large model, the semantic logic of the process and compliance rules are deeply modeled and reasoned to support semantic understanding and compliance judgment of complex processes, improving the accuracy and compliance of process scheduling.

[0048] Through natural language interface and large model dialogue interaction, the explainability output of process suggestion and compliance conclusion and user intent understanding are realized; users can ask "why this task is marked as high risk" through natural language, and the system will give clear explanation combining financial data changes and compliance rules.

[0049] Through reinforcement learning mechanism to continuously optimize process arrangement strategy, realize the self-adaptive adjustment ability of the system; the system not only formulates strategies according to historical data, but also continuously optimizes the model through actual execution effect feedback, improves the accuracy and efficiency of process scheduling.

[0050] Through semantic analysis of unstructured text (such as contracts, approval opinions, audit reports) by large model, compliance consistency check and process adaptation across documents are realized. Contract terms, approval reasons and other information can be automatically extracted and compared with process execution data for consistency, improving the comprehensiveness and automation level of process compliance review.

[0051] An intelligent enterprise financial process arrangement and monitoring method, comprising the following steps:

[0052] S1: Model the enterprise financial process based on the BPMN standard, and introduce a process meta-model to abstract the process nodes into task units;

[0053] S2: Use a large language model fine-tuned in the financial field to receive process context information, and generate process path suggestions after analyzing the input semantics;

[0054] S3: Obtain financial statement data from the financial system regularly, analyze the trend of key financial indicators through the large model, and dynamically adjust the process priority and execution path;

[0055] S4: Record process execution results, compliance judgments and user feedback, and use reinforcement learning algorithm to train the large model to optimize process strategy;

[0056] S5: Use the large model to make semantic-level compliance judgments during the process running, combine the compliance rules in the knowledge graph to judge the risk, and automatically mark and suspend the process for high-risk operations;

[0057] S6: Users interact with the system through a natural language interface, and the large model generates responses according to user intent, outputs results including conclusions, reasoning process and data support.

[0058] Preferably, the S1 supports the construction and version management of process template library, facilitating cross-department reuse and rapid deployment;

[0059] Preferably, in S4, the learning target includes minimizing the process execution time, maximizing the process compliance score, and improving the positive impact of the process on financial indicators.

[0060] The present application significantly improves the flexibility and adaptability of process execution by implementing modular arrangement and dynamic scheduling of financial processes; supports process optimization and risk early warning based on financial report data, enhancing the agility of enterprise financial operations; at the same time, provides natural language interaction interface and interpretable output, reduces the threshold of use, and improves the user experience. In addition, the system has good scalability, which is suitable for various financial process scenarios and enterprises of different scales, and introduces a reinforcement learning mechanism, so that the system has the ability of continuous learning and strategy optimization, which can adapt to the changing business environment. The present application also supports compliance consistency checking of unstructured text, further improving the integrity and automation of process compliance review. In summary, the present application provides a new intelligent arrangement and compliance monitoring integrated solution for financial processes, which has wide application prospect and significant commercial value.

[0061] It should be noted that, in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0062] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit it; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent orchestration and monitoring system for enterprise financial processes, characterized in that, include: The process modeling module is used to model enterprise financial processes based on the BPMN standard. It supports visual process design and introduces a process meta-model to abstract process nodes into composable and replaceable task units. Each task unit includes input and output parameters, execution conditions and constraints, optional execution paths and alternative solutions. The large model-driven module uses a large language model fine-tuned for the financial domain to receive process context information, parse the input semantics to generate process path suggestions that conform to business logic, and supports the generation of process descriptions in natural language. The process scheduling optimization module regularly retrieves the latest financial statement data from the financial system, analyzes the changing trends of key financial indicators through a large model, and dynamically adjusts process priorities and execution paths based on the financial statement analysis results. The reinforcement learning optimization module records process execution results, compliance judgments, and user feedback. It uses reinforcement learning algorithms to train a large model to optimize process strategies. The learning objectives include minimizing process execution time, maximizing process compliance scores, and improving the positive impact of processes on financial indicators. The compliance check module uses a large model to make semantic-level compliance judgments during the process, combines compliance rules in the knowledge graph to determine whether there are risks, automatically marks high-risk operations and suspends the process, and generates audit logs and explanatory reports. The natural language interaction module allows users to interact with the system through a natural language interface. The large model generates responses based on user intent and outputs results that include conclusions, reasoning processes, and data support.

2. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, The process modeling module supports the construction and version management of process template libraries, facilitating cross-departmental reuse and rapid deployment.

3. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, The large model-driven module can identify key tasks and recommend optimized paths based on the natural language instructions input by the user.

4. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, When the process scheduling optimization module detects a decline in a customer's credit rating, it automatically triggers a payment terms change process. If expenses are found to exceed the budget, insert a budget re-review process.

5. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, The reinforcement learning optimization module is trained using the PPO or DQN algorithm and supports policy version management and A / B testing.

6. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, The compliance check module combines compliance rules in the knowledge graph to determine whether the operation complies with the requirements of the internal control matrix, budget control, and expense limits.

7. The intelligent orchestration and monitoring system for enterprise financial processes according to claim 1, characterized in that, The natural language interaction module supports interaction methods such as web-based dialog boxes, mobile apps, and voice assistants.

8. A method for intelligent orchestration and monitoring of enterprise financial processes, characterized in that, Includes the following steps: S1 models enterprise financial processes based on the BPMN standard, and introduces a process meta-model to abstract process nodes into task units; S2 uses a large language model fine-tuned for the financial domain to receive process context information, and generates process path suggestions after parsing the input semantics. S3 periodically retrieves financial statement data from the financial system, analyzes the changing trends of key financial indicators through large models, and dynamically adjusts process priorities and execution paths. S4 records process execution results, compliance judgments, and user feedback, and uses reinforcement learning algorithms to train large models to optimize process strategies. S5 uses a large model to make semantic-level compliance judgments during the process, combines compliance rules in the knowledge graph to judge risks, and automatically marks high-risk operations and suspends the process. S6 users interact with the system through a natural language interface. The large model generates responses based on user intent and outputs results that include conclusions, reasoning processes, and data support.

9. The intelligent orchestration and monitoring method for enterprise financial processes according to claim 8, characterized in that, The S1 supports the construction and version management of process template libraries, which facilitates cross-departmental reuse and rapid deployment.

10. The intelligent orchestration and monitoring method for enterprise financial processes according to claim 8, characterized in that, In S4, the learning objectives include minimizing process execution time, maximizing process compliance scores, and improving the positive impact of processes on financial metrics.