RPA process intelligent optimization method based on large model

Abstract

The invention belongs to the technical field of robot process automation optimization, and particularly relates to an RPA process intelligent optimization method based on a large model. According to the method, the process data is obtained through a convenient file import mode, and the bottleneck and the optimization point are accurately identified by means of data acquisition, preprocessing, semantic understanding and scene learning deep analysis process. And then, the large model automatically generates optimization suggestions and displays an expected adjusted process, and after confirmation by a user, the process can be automatically adjusted and a new file is generated. In the verification link, the process execution condition is strictly detected in normal and abnormal environments, and the optimization effect and stability are ensured. According to the technology, the optimization efficiency and quality are effectively improved, the process adaptability is enhanced, the labor cost is reduced, an intelligent, efficient and reliable solution is provided for RPA process optimization of enterprises, the enterprises are assisted in optimizing business processes in digital transformation, and the operation efficiency and competitiveness are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of robotic process automation optimization, and specifically is a RPA process intelligent optimization method based on a large model. Background Art

[0002] In today's digital age, enterprises are facing increasingly fierce market competition and growing business complexity. In order to improve operational efficiency, reduce costs and enhance competitiveness, enterprises are seeking various advanced technology solutions, and RPA technology has emerged.

[0003] RPA aims to automate repetitive and regular business processes by simulating human operations, such as data entry, file processing, report generation and other tasks. It has significant advantages and can greatly reduce the time and energy consumption of manual operations, improve work efficiency, reduce the error rate caused by human factors, and ensure the accuracy and consistency of business processes.

[0004] However, the traditional RPA process has gradually exposed many limitations in practical applications. On the one hand, its process design often relies on manual experience and limited analytical methods, and it is difficult to fully consider various complex business scenarios and potential optimization space. This may lead to some unnecessary steps, waste of resources, and inefficient links in the process. On the other hand, with the continuous development and changes of corporate business, market environment, regulations, policies, customer needs and other factors are in dynamic adjustment. The rigid structure of the traditional RPA process makes it difficult to quickly adapt to these changes, requiring frequent manual intervention and redesign, which undoubtedly increases the operating costs and management difficulty of the enterprise.

[0005] In addition, the performance and scalability of traditional RPA processes are also challenged when facing large-scale data processing and complex business logic. How to further improve its processing speed, optimize resource allocation, and enhance the ability to respond to various abnormal situations while ensuring process accuracy has become a key issue restricting the in-depth application and sustainable development of RPA technology in enterprises. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes an intelligent optimization method for RPA processes based on a large model. The present invention uses a large model to intelligently analyze and optimize the RPA process, can quickly and accurately identify bottlenecks and optimization points in the process, and automatically generate optimization suggestions and adjust the process, thereby improving the optimization efficiency and quality of the RPA process and reducing labor costs. At the same time, through strict verification methods, the stability and reliability of the optimized process are ensured.

[0007] The technical solution of the present invention is as follows:

[0008] A large model-based RPA process intelligent optimization method, including:

[0009] Import the automated business process file drawn by traditional RPA tools by dragging and dropping the process file or clicking "Upload Process" in the large model;

[0010] Click "Start Analysis" in the big model. Using the deep learning capabilities of the big model, first collect and preprocess the process files, then use the semantic understanding of the big model to understand the business process logic, and use the scenario learning capabilities of the big model to evaluate the business process performance under different conditions and identify bottlenecks and potential optimization points in the business process.

[0011] Based on the identified results, the big model automatically generates corresponding optimization suggestions and displays the flowchart graphic steps after the expected adjustments;

[0012] In the big model, the user clicks "Automatically adjust process", and the big model automatically adjusts the process according to the optimization suggestions. The automatically adjusted process will generate a new process file while keeping the original data unchanged. The user uses the newly generated process file for debugging and verification;

[0013] After the user confirms that the debugging and verification results are correct, the user can either use the optimized process file, perform secondary optimization on the optimized process file, or end the optimization operation of the RPA process.

[0014] Furthermore, the preparation work before importing the automated business process file drawn using the traditional RPA tool includes:

[0015] Ensure that the large model platform is running normally and has sufficient resources to process the process files to be imported;

[0016] Check whether the format of the automated business process data file drawn by traditional RPA tools meets the requirements of the large model, and perform format conversion or use corresponding adapter tools;

[0017] Back up the process file to be imported.

[0018] Furthermore, the use of the deep learning capability of the large model to collect and preprocess the process file includes:

[0019] Call the parsing module to extract various business process elements from the file;

[0020] Clean the collected data and remove noise data;

[0021] Missing data are supplemented or marked, data are standardized, and data are classified and labeled.

[0022] Furthermore, the use of semantic understanding of the big model to understand the business process logic further includes:

[0023] Use natural language processing technology and pre-trained semantic understanding models to parse text information such as operation step descriptions and conditional judgment statements in process files;

[0024] Identify the key business concepts, actions, and relationships within them.

[0025] Furthermore, the use of the scenario learning capability of the big model to evaluate the performance of business processes under different conditions further includes:

[0026] Simulate different operating scenarios, including normal business scenarios and various abnormal situations that may occur;

[0027] Calculate key performance indicators of the process;

[0028] Compare and analyze the process performance in different scenarios to find out in which scenarios the process performs well and in which scenarios there are problems.

[0029] Furthermore, the large model automatically generates corresponding optimization suggestions including:

[0030] Generate corresponding optimization suggestions from multiple dimensions, including process structure adjustment, algorithm optimization, resource allocation optimization, and error handling enhancement;

[0031] Make virtual adjustments to the original business process and generate a process model after the expected adjustment;

[0032] Convert the adjusted process model into a graphical representation, so that users can intuitively see the changes in the structure and execution sequence of the process after optimization.

[0033] Furthermore, after the user clicks "Automatically adjust process", the large model automatically adjusts the process according to the optimization suggestions, including:

[0034] Reconfirm the previously generated optimization suggestions to ensure their accuracy and feasibility;

[0035] Perform actual modification operations on the original process files stored in a specific area;

[0036] During the modification process, strictly follow the process modeling specifications and programming best practices;

[0037] Generate a new process file and store it in a different location from the original file, while updating the process file index and related metadata in the system.

[0038] Furthermore, after the user confirms that the debugging and verification results are correct, the following steps are further performed:

[0039] Carefully review debugging and verification results, including execution time comparison under normal environment, output result consistency check results, and exception capture and handling reports under abnormal environment;

[0040] For any results that do not meet expectations, users conduct in-depth analysis of the causes and perform rollbacks or adjustments and re-verifications;

[0041] If the verification result is completely in line with expectations, the user performs confirmation operations on the large model platform, marks the process optimization as successful, and records the optimized process performance indicators and related information.

[0042] The beneficial effects of the present invention are as follows: through this method, users can import process files drawn by traditional RPA tools into the system by simply dragging or clicking to upload, automatically analyze and optimize the process through deep learning, identify redundant and inefficient steps, and make the overall execution time of the optimized process significantly lower than the original process, significantly improving the execution speed of the process. It reduces resource consumption and unnecessary operations, reduces energy consumption and operating costs, and thus achieves efficient use of the process.

[0043] The optimized process is strictly verified under normal and abnormal conditions to ensure that its output is consistent with the original process. This ensures the continuity and accuracy of business logic, improves the quality of process execution, and ensures high quality and high accuracy of output results while improving efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the overall flow chart of the present invention;

[0045] Figure 2 is a flowchart of method S100;

[0046] Figure 3 is a flow chart of method S200;

[0047] Figure 4 is a flow chart of method S300;

[0048] Figure 5 is a flow chart of method S400;

[0049] Figure 6 is a flow chart of method S500. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] See also Figure 1 , which shows an RPA process intelligent optimization method based on a large model provided by an embodiment of the present invention, including:

[0052] S100: In the large model, you can import the automated business process files drawn using traditional RPA tools by dragging and dropping the process files or clicking "Upload Process".

[0053] S200: Click "Start Analysis" in the big model. Using the deep learning capabilities of the big model, first collect and preprocess the process files. Then use the semantic understanding of the big model to understand the business process logic. Use the scenario learning capabilities of the big model to evaluate the business process performance under different conditions and identify bottlenecks and potential optimization points in the business process.

[0054] S300: Based on the recognition results, the large model automatically generates corresponding optimization suggestions and displays the flowchart steps after the expected adjustments.

[0055] S400: In the big model, the user clicks "Automatically adjust process", and the big model automatically adjusts the process according to the optimization suggestions. The automatically adjusted process will generate a new process file while keeping the original data unchanged. The user uses the newly generated process file for debugging and verification.

[0056] S500: After the user confirms that the debugging and verification results are correct, the user either uses the optimized process file, performs secondary optimization on the optimized process file, or ends the optimization operation of the RPA process.

[0057] S100 involves preparation and import operations to ensure that the big model platform operates normally and has sufficient resources, check and ensure that the automated business process data file format meets the requirements, back up the process files, and import the files into the big model platform by dragging or clicking "Upload Process" to perform a preliminary file integrity check.

[0058] Please refer to Figure 2 , which shows a flowchart of an exemplary RPA process intelligent optimization method S100 based on a large model of the present application, and its contents include:

[0059] S110: Preparation

[0060] Make sure that the large model platform is running normally and has sufficient resources to process the process files to be imported.

[0061] Check whether the file format of the automated business process data drawn by the traditional RPA tool meets the requirements of the big model. Common supported formats may include but are not limited to specific XML, JSON or proprietary formats. If the format is incompatible, format conversion or the use of corresponding adapter tools is required in advance.

[0062] Back up the process files to be imported to prevent file damage or loss due to unexpected circumstances during the import process. The backup files should be stored in a safe and reliable location, and the backup time and path should be recorded.

[0063] S120: Import Operation

[0064] The import operation is divided into dragging and dropping the process file method and clicking "Upload process" method.

[0065] The operation of dragging and dropping process files is as follows:

[0066] Open the user interface of the large model platform and find the area dedicated to importing process files. This area usually has obvious signs or prompts, such as "Drag and drop files to this area for import."

[0067] Find the automated business process data file that needs to be imported in the local file system, select the file, hold down the left mouse button and drag it to the specified large model platform import area.

[0068] During the dragging process, the large model platform will monitor the movement of the file in real time, and after the file enters the valid import area, it will trigger the file receiving mechanism and start preparing to receive the file data.

[0069] Once the file is successfully received, the large model platform will automatically start the preliminary file integrity check procedure to check whether the file size is consistent with the expectation, whether the file structure is complete, etc. If the check passes, it will enter the next step of data reading; if the check fails, the platform will pop up a detailed error message to inform the user of possible problems with the file, such as file damage, format error, etc. The user needs to fix the problem according to the prompt and re-import the operation.

[0070] Click "Upload Process" to perform the following operations:

[0071] In the user interface of the large model platform, find and click the "Upload Process" button, which is usually located in the menu bar or in a specific process management function area.

[0072] After clicking the button, the platform will pop up a file selection dialog box. The user navigates to the folder where the automated business process data file is stored in the local file system and selects the target file.

[0073] After selecting the file, click the "Open" or "OK" button in the dialog box, and the large model platform will start uploading the file. During the upload process, the platform will display the upload progress bar so that users can understand the upload status in real time.

[0074] After the upload is complete, the file integrity check will also be performed, and the subsequent processing flow is the same as the drag-and-drop import method.

[0075] S200 focuses on data collection and preprocessing, semantic understanding of business process logic, and scenario learning to evaluate process performance. The big model extracts business process elements from imported process files, cleans data, standardizes processing, and uses natural language processing technology to understand business process logic, simulate different operating scenarios, evaluate process performance, and identify bottlenecks and potential optimization points.

[0076] Please refer to Figure 3 , which shows a flowchart of an exemplary RPA process intelligent optimization method S200 based on a large model of the present application, and its contents include:

[0077] S210: Data acquisition and preprocessing

[0078] The big model calls the parsing module according to the format and structure of the imported process file to extract various business process elements from the file, including but not limited to operation steps, input and output data, conditional judgment logic, execution order, etc.

[0079] Clean the collected data and remove the noise data (such as duplicate records, erroneous characters, incomplete data items, etc.) to improve the quality and availability of the data.

[0080] Supplement or mark missing data, standardize data, and classify and label data.

[0081] S220: Semantic understanding of business process logic

[0082] The big model uses natural language processing technology and pre-trained semantic understanding models to parse text information such as operation step descriptions, conditional judgment statements, etc. in process documents to identify key business concepts, actions, and relationships.

[0083] S230: Situational Learning Assessment Process Performance

[0084] The big model simulates different operating scenarios based on the semantic model of the business process and the collected data. These scenarios cover normal business scenarios and various abnormal situations that may occur, such as input data anomalies, external system failures, resource shortages, etc.

[0085] In each simulation scenario, the key performance indicators (KPIs) of the process are calculated, such as process execution time, resource utilization, output result accuracy, error rate, etc. These indicators are used to quantitatively evaluate the performance of the process under different conditions.

[0086] Compare and analyze the process performance in different scenarios to find out in which scenarios the process performs well and in which scenarios there are problems. For example, it is found that a process takes too long to execute under high concurrent data input, or causes a sharp increase in the error rate under specific abnormal input.

[0087] Based on the results of scenario learning, bottlenecks and potential optimization points in the business process are identified. Bottlenecks are usually key steps or resource constraints that have a significant impact on process performance in certain scenarios, while potential optimization points may include areas where process logic can be improved, resource utilization can be increased, or fault tolerance can be enhanced. For example, a data processing step can reduce execution time by adopting a more efficient algorithm, or a more complete error handling mechanism can be added to a place where exceptions are prone to occur.

[0088] Based on the identified business process bottlenecks and potential optimization points, S300 generates multi-dimensional optimization suggestions based on the large model, covering aspects such as process structure adjustment, algorithm optimization, resource allocation optimization, and error handling enhancement. It also applies the optimization suggestions to the original business process, generates the expected adjusted process model, and graphically displays the optimized process structure and execution order.

[0089] Please refer to Figure 4 , which shows a flowchart of an exemplary RPA process intelligent optimization method S300 based on a large model of the present application, and its contents include:

[0090] S310: Generate optimization suggestions

[0091] Based on the identification results of business process bottlenecks and potential optimization points, the big model generates corresponding optimization suggestions from multiple dimensions, including but not limited to process structure adjustment, algorithm optimization, resource allocation optimization, error handling enhancement, etc.

[0092] In terms of process structure adjustment, if it is found that there are steps in the process that are executed sequentially but have no data dependencies, it is recommended to parallelize them to improve the overall execution efficiency of the process; or if there is a complex nested conditional judgment structure, consider simplifying or reconstructing it into a clearer logical structure to reduce unnecessary computing overhead.

[0093] For algorithm optimization, if the algorithm used in a certain data processing step is inefficient, the big model will recommend a more suitable and efficient algorithm based on business needs and data characteristics. For example, the traditional brute force search algorithm can be replaced with a fast search algorithm based on an index or hash table to speed up data retrieval.

[0094] In terms of resource allocation optimization, if it is found that a certain resource is over-allocated or under-allocated during the operation of the process, it will be recommended to adjust the resource allocation strategy to ensure the rational use of resources and avoid resource waste or performance problems caused by insufficient resources.

[0095] Regarding error handling enhancement, if it is found in scenario learning that the process cannot correctly handle errors under certain abnormal circumstances, the large model will suggest adding detailed error logging, improving error prompt information, or introducing a more complete exception recovery mechanism, such as automatic retry, failover, etc., to improve the stability and reliability of the process.

[0096] S320: Flowchart steps showing expected adjustments

[0097] The big model makes virtual adjustments to the original business process based on the generated optimization suggestions and generates a process model after the expected adjustments.

[0098] Convert the adjusted process model into a graphical representation, so that users can intuitively see the structure and execution order of the process after optimization. Graphical representation usually uses standard process modeling symbols, such as flow charts, activity diagrams, etc., to clearly present each step, decision point, data flow, and the relationship between steps.

[0099] In the graphical display, the parts that have changed before and after the optimization are highlighted. For example, new, deleted or modified steps are marked with different colors or icons. Detailed text descriptions are provided to explain the reasons and expected effects of each change, helping users quickly understand the specific impact of the optimization suggestions on the process.

[0100] It provides interactive graphic operation functions. Users can view detailed information of the process through operations such as zooming, panning, and clicking. For example, clicking a step can view its detailed properties, which is convenient for users to deeply analyze whether the optimized process meets business needs.

[0101] S400 includes automatic process adjustment and verification. After the user clicks "Automatically adjust the process", the large model actually modifies the original process file according to the optimization suggestions, generates a new process file, and verifies the adjusted process under normal and abnormal environments to ensure functional integrity and consistency of exception handling logic.

[0102] Please refer to Figure 5, which shows a flowchart of an exemplary RPA process intelligent optimization method S400 based on a large model of the present application, and its contents include:

[0103] S410: Automatic process adjustment

[0104] When the user clicks "Automatically adjust process" in the big model, the big model will first reconfirm the previously generated optimization suggestions to ensure the accuracy and feasibility of the suggestions.

[0105] Based on the confirmed optimization suggestions, the big model actually modifies the original process files stored in a specific area, which involves updating the process definition, configuration parameters, code logic and other parts of the file.

[0106] During the modification process, we strictly follow the process modeling specifications and programming best practices to ensure that the adjusted process file structure is complete, the syntax is correct, and it meets the business logic requirements. For example, when modifying the code logic, we will conduct code review and unit testing to avoid introducing new errors.

[0107] After the adjustment is completed, a new process file is generated and stored in a different location from the original file. At the same time, the process file index and related metadata in the system are updated to reflect the new state of the process. The name of the new process file usually contains information such as an optimized timestamp or version number to facilitate user identification and management.

[0108] S420: Verification under normal conditions

[0109] Prepare the verification environment and ensure that it is as consistent as possible with the actual running environment of the process, including hardware configuration, software dependencies, network settings, etc.

[0110] Under standard operating conditions, the adjusted process is executed using the same input data set as the original process. The input data set should cover various data combinations and value ranges that may appear in normal business situations to fully test the functional integrity of the process.

[0111] Monitor the process execution, record the start time, end time and execution results of each step of the process execution, calculate the overall execution time of the adjusted process file, and compare it with the overall execution time of the original process file to ensure that the execution speed of the adjusted process is improved.

[0112] Perform detailed inspection and comparison of the output results of the process to ensure that the output results are consistent with the output results of the original process under the same input conditions, including comparison of data content, format, accuracy, etc. If differences are found in the output results, timely analysis is carried out to determine whether it is an expected change caused by optimization or an error.

[0113] S430: Verification under abnormal conditions

[0114] Simulate various abnormal situations, such as input data errors, external system failures, and resource exhaustion. The simulation of these abnormal situations should be as close as possible to the situations that may occur in the actual production environment.

[0115] When an exception occurs, observe whether the exception can be correctly captured during process execution and check whether the exception handling mechanism is started as expected, for example, whether the exception information is correctly recorded, whether an alarm notification is sent to relevant personnel, and whether error recovery operations are attempted.

[0116] Ensure that the exception handling logic is consistent with the original process. That is, for the same type of exceptions, the adjusted process processing method should be consistent with the original process in terms of business rules and processing effects to avoid improper exception handling due to optimization, which will affect the stability and reliability of the system.

[0117] S500 involves result confirmation and subsequent operations. The user reviews the debugging and verification results to confirm whether the optimization meets expectations, and performs rollback, secondary optimization, or ends the optimization operation based on the results. If the optimization is successful, the user deploys the optimized process file to the production environment, or performs secondary optimization, or ends the optimization operation of the RPA process, and summarizes and archives the entire optimization process.

[0118] Please refer to Figure 6 , which shows a flowchart of an exemplary RPA process intelligent optimization method S500 based on a large model of the present application, and its contents include:

[0119] S510: Result confirmation

[0120] Users carefully review debugging and verification results, including execution time comparison under normal environment, output result consistency check results, and exception capture and handling reports under abnormal environment.

[0121] For any results that do not meet expectations, users need to analyze the reasons in depth, and may need to view detailed execution logs, system monitoring data, and other information to determine the problem. If the problem is caused by an error or unreasonable adjustment during the optimization process, users can choose to roll back to the original process file, or adjust and re-verify the optimization suggestions according to the specific situation.

[0122] If the verification results are in full compliance with expectations, the user performs confirmation operations on the large model platform, marks the process optimization as successful, and records the optimized process performance indicators and related information, which will serve as a reference for future process management and further optimization.

[0123] S520: Subsequent operations

[0124] Subsequent operations include using the optimized process file, re-optimizing the optimized process file, and ending the optimization operation of the RPA process.

[0125] Use optimized process files: Users deploy optimized process files to the actual production environment, replacing the original process files to ensure that the business process runs normally in the optimized manner.

[0126] Secondary optimization of the optimized process file: If the user believes that the optimized process still has room for further improvement, or the business requirements have changed, the user can start the RPA process intelligent optimization process based on the large model again, repeat steps S100-S400, and perform secondary optimization on the process.

[0127] During the secondary optimization process, users can refer to the experience and data accumulated in the previous optimization process, such as the bottleneck analysis results of the previous optimization, the evaluation of the effectiveness of optimization suggestions, etc., to make targeted improvements and improve the optimization efficiency and quality.

[0128] End the optimization operation of the RPA process: If the user is satisfied with the performance and functions of the current optimized process and there is no significant change in business needs in the short term, the user can choose to end the optimization operation of the RPA process.

[0129] After the optimization is completed, the entire optimization process is summarized and archived, and the process file versions, optimization suggestions, verification results, performance improvement and other information involved in the optimization process are recorded to form an optimization report to provide reference for internal knowledge management and future similar projects. At the same time, temporary resources related to this optimization in the large model platform are released to maintain the efficient operation of the platform.

[0130] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0131] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0132] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0133] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features of the present invention.

[0134] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A RPA process intelligent optimization method based on a large model, characterized in that: include: In the large model, drag and drop the process file or click "Upload process" to import the automated business process file drawn using traditional RPA tools; Click "Start Analysis" in the big model. Using the deep learning capabilities of the big model, first collect and preprocess the process files, then use the semantic understanding of the big model to understand the business process logic, and use the scenario learning capabilities of the big model to evaluate the business process performance under different conditions and identify bottlenecks and potential optimization points in the business process. Based on the identified results, the big model automatically generates corresponding optimization suggestions and displays the flowchart graphic steps after the expected adjustments; In the big model, the user clicks "Automatically adjust process", and the big model automatically adjusts the process according to the optimization suggestions. The automatically adjusted process will generate a new process file while keeping the original data unchanged. The user uses the newly generated process file for debugging and verification; After the user confirms that the debugging and verification results are correct, the user can either use the optimized process file, perform secondary optimization on the optimized process file, or end the optimization operation of the RPA process.

2. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: The preparation work before importing the automated business process file drawn by traditional RPA tools includes: Ensure that the large model platform is running normally and has sufficient resources to process the process files to be imported; Check whether the format of the automated business process data file drawn by traditional RPA tools meets the requirements of the large model, and perform format conversion or use corresponding adapter tools; Back up the process file to be imported.

3. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: The use of the deep learning capability of the large model to collect data and preprocess the process files includes: Call the parsing module to extract various business process elements from the file; Clean the collected data and remove noise data; Missing data are supplemented or marked, data are standardized, and data are classified and labeled.

4. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: The semantic understanding of the big model to understand the business process logic further includes: Use natural language processing technology and pre-trained semantic understanding models to parse text information such as operation step descriptions and conditional judgment statements in process files; Identify the key business concepts, actions, and relationships within them.

5. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: Using the scenario learning capabilities of the big model to evaluate the performance of business processes under different conditions further includes: Simulate different operating scenarios, including normal business scenarios and various abnormal situations that may occur; Calculate key performance indicators of the process; Compare and analyze the process performance in different scenarios to find out in which scenarios the process performs well and in which scenarios there are problems.

6. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: The large model automatically generates corresponding optimization suggestions including: Generate corresponding optimization suggestions from multiple dimensions, including process structure adjustment, algorithm optimization, resource allocation optimization, and error handling enhancement; Make virtual adjustments to the original business process and generate a process model after the expected adjustment; Convert the adjusted process model into a graphical representation, so that users can intuitively see the changes in the structure and execution sequence of the process after optimization.

7. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: After the user clicks "Automatically adjust process", the large model automatically adjusts the process according to the optimization suggestions, including: Reconfirm the previously generated optimization suggestions to ensure their accuracy and feasibility; Perform actual modification operations on the original process files stored in a specific area; During the modification process, strictly follow the process modeling specifications and programming best practices; Generate a new process file and store it in a different location from the original file, while updating the process file index and related metadata in the system.

8. According to claim 1, a large model-based RPA process intelligent optimization method is characterized in that: After the user confirms that the debugging and verification results are correct, the following further steps are included: Carefully review debugging and verification results, including execution time comparison under normal environment, output result consistency check results, and exception capture and handling reports under abnormal environment; For any results that do not meet expectations, users conduct in-depth analysis of the causes and perform rollbacks or adjustments and re-verifications; If the verification result is completely in line with expectations, the user performs confirmation operations on the large model platform, marks the process optimization as successful, and records the optimized process performance indicators and related information.

Images