RPA technology-based reimbursement process management method and system

Abstract

The invention discloses a reimbursement process management method and system based on an RPA technology. The method comprises the following steps: constructing a reimbursement system: carding and recording a reimbursement process, creating an RPA automatic process, planning an execution path of a software robot, developing the software robot to automatically grab data and identifying key information; data acquisition: uploading a reimbursement voucher, identifying and acquiring reimbursement data, and performing data cleaning and format conversion; filling reimbursement receipts: filling the converted reimbursement data into a preset template; and intelligent auditing and checking: checking and approving according to the enterprise financial system and the approval rule. Compared with the prior art, the RPA software robot has the advantages that the RPA software robot has the capability of automatically executing various processes, interacts with the system according to a set program, automatically executes the reimbursement process and replaces manual repeated work such as data entry, information processing and form filling, and compared with the prior art, the RPA software robot is higher in efficiency, fewer in error and more convenient to use.

Description

Technical Field

[0001] The present invention belongs to the field of enterprise information management technology, and more specifically, it particularly relates to a reimbursement process management method based on RPA technology. At the same time, the present invention also relates to a reimbursement process management system based on RPA technology. Background Art

[0002] Existing reimbursement management is a comprehensive process involving multiple rules and processes. Enterprises should establish and improve reimbursement management systems based on their actual conditions to improve work efficiency, reduce financial risks, and ensure compliance. At the same time, with the development of technology, many companies have also begun to adopt electronic reimbursement systems to simplify processes and improve efficiency.

[0003] The existing reimbursement process within enterprises usually requires employees to manually fill out reimbursement forms according to the reimbursement template specified by the company; employees need to collect relevant invoices, receipts or other supporting materials, scan or take photos of them for archiving, and then match them with the reimbursement form. After completing the form, the employee submits the reimbursement form and related materials to the superior leader or the financial department for approval. The approval process may involve multiple levels of approval, and each level requires manual review and confirmation of relevant information; finally, after the financial department receives the approved reimbursement application, it will check the information again and enter it into the financial system, and then perform accounting processing, including bookkeeping, payment of reimbursement, etc. And after the reimbursement is completed, if there are any questions or audit tracing is required, it is often necessary to manually check paper documents or electronic records to find historical reimbursement data.

[0004] However, there are some problems with the existing technology: the existing reimbursement process requires employees to manually fill out reimbursement documents and requires manual inspection and approval by accountants, so there are obvious problems of high manpower consumption, low efficiency, high error risk and difficulty in information tracing. At the same time, the main purpose of the innovative application of intelligent RPA technology in the enterprise OA system is to efficiently and automatically solve the problems of high manpower consumption, low efficiency, high error risk and difficulty in information tracing. The reimbursement process management method based on RPA technology is an innovative method proposed to address these problems. It replaces most manual operations with automated processes, greatly improving the efficiency and accuracy of the reimbursement process. Summary of the invention

[0005] In view of the problems existing in the prior art, the purpose of the present invention is to provide a reimbursement process management method and system based on RPA technology. The recommended method utilizes the ability to build an RPA software robot to automatically execute various processes and interact with the system according to the set program, including repetitive tasks such as data entry, information processing, and form filling. Specifically, the use of RPA robots can improve efficiency, reduce errors, facilitate implementation, and facilitate the query of historical data.

[0006] To achieve the above object, the present invention provides the following technical solution: a reimbursement process management method and system based on RPA technology, comprising the following steps:

[0007] S1. Reimbursement system architecture: sort out and record the reimbursement process, create RPA automation process, plan the execution path of the software robot, program automatic login to the OA system, and develop software robots to automatically capture data and identify key information;

[0008] S2. Data collection: Employees upload reimbursement vouchers, and the software robot identifies and obtains the reimbursement data, and performs data cleaning and format conversion;

[0009] S3. Fill in the reimbursement document: The software robot fills the converted reimbursement data into the preset template;

[0010] S4. Intelligent review and verification: Software robots conduct inspections and approvals based on the company’s financial system and approval rules.

[0011] Optionally, in step S1, the reimbursement system framework includes:

[0012] S11. Determine the reimbursement process, including logging into the OA system, data capture, entering reimbursement forms, reviewing and verifying;

[0013] S12. Design RPA software robots and plan the execution paths of software robots;

[0014] S13, data integration;

[0015] S14. Audit rule definition

[0016] S15. Configuration approval process;

[0017] S16. Test and test run the RPA software robot, optimize it based on the feedback from the test run results, and continuously monitor and maintain it.

[0018] Optionally, the OA system is an information system platform for daily management and operation of an enterprise, which is used to help store, retrieve and manage documents and data within the enterprise, automate business processes and collaborate in office work;

[0019] The execution path of the software robot is the reimbursement process;

[0020] The data integration includes reading reported data, writing data capture rules, writing data cleaning rules and presetting reimbursement document templates;

[0021] The audit rules include establishing a rule base for verifying the compliance of reimbursement documents;

[0022] The approval process uses the preset approval process template of the OA system and is associated with the RPA software robot. After the RPA software robot checks the compliance of the submitted reimbursement vouchers, it pushes the approval request along the established path.

[0023] Optionally, in step S2, the specific steps of data collection include:

[0024] S21. Upload reimbursement vouchers: Submit reimbursement vouchers through the upload interface provided by the system. The reimbursement vouchers may be in the form of paper invoices, PDF files, and electronic invoices.

[0025] S22. The software robot identifies and obtains reimbursement data: The software robot uses OCR optical character recognition technology to read the text information on the voucher; and uses NLP natural language processing technology to parse the unstructured text data in the reimbursement voucher to help the software robot identify and process;

[0026] S23, data preprocessing: data cleaning and data conversion for the data recognized by the software robot;

[0027] S24, data post-processing: check the pre-processed text data again, keep the records of the original data and the processed data, and finally store the processed data text in the database and send it to the next process.

[0028] Optionally, in step S2, the data cleaning includes:

[0029] Missing value handling: Identify and handle missing data by filling in default values, estimating using statistical methods, and deleting;

[0030] Outlier handling: Use statistical tests, machine learning algorithms, and rule engines to identify outliers and handle them;

[0031] Duplicate data processing: identifying and removing duplicate records in the data;

[0032] Consistency check: Check the read data through cross-table association to ensure that the read data is consistent in different data sets;

[0033] The data conversion content includes:

[0034] Standardization: Convert the read data into a standard format, including a unified format for date and time and a unified unit for numerical values;

[0035] Data type conversion: converting data from one type to another according to rules;

[0036] Feature encoding: Convert non-numeric data into numeric data so that software robots can perform calculations.

[0037] Optionally, in the outlier detection, the machine learning algorithm includes a gradient boosting algorithm;

[0038] The gradient boosting algorithm formula is as follows:

[0039] F m (x) = F m-1 (x)+p·h(x;α m )

[0040] Among them, F m (x) is the model prediction at step m;

[0041] F m-1 (x) is the model prediction at step m-1;

[0042] p is the learning rate (step size);

[0043] h(x; α m ) is the prediction result of the base learner added in the mth step;

[0044] α m are the parameters of the base learner.

[0045] Optionally, step S3 of filling in the reimbursement document includes:

[0046] S31. The software robot loads the preset reimbursement document template through the OA system;

[0047] S32, the software robot distinguishes and locates the filling area by identifying specific marks in the reimbursement document template;

[0048] S33, the software robot inputs the processed data text into the corresponding filling area respectively according to the type of the data text;

[0049] S34. The software robot compares the data text in the reimbursement document with the processed data text to ensure the data is consistent and sends it to the next process.

[0050] Optionally, in step S4, the intelligent review and verification includes:

[0051] S41. The software robot performs a budget limit check on the expenses of the reimbursement documents according to the enterprise financial system in the preset rule base;

[0052] S42. The software robot compares historical data with the database based on the audit rules in the preset rule base and marks the anomalies in the reimbursement documents;

[0053] S43, the software robot generates an audit report, including inspection results and abnormal data;

[0054] S44. The software robot sends the reimbursement documents that meet the standards to the next process, and sends the reimbursement documents with abnormal data to manual approval.

[0055] A reimbursement process management system based on RPA technology, the system is used to implement the above-mentioned reimbursement process management method based on RPA technology, including:

[0056] A server, in which the RPA software robot is deployed and run, and includes a memory bar, a CPU, a GPU, and a hard disk space;

[0057] A storage device, in which a database is deployed for storing electronic documents, invoice images and expense receipts;

[0058] Terminal devices include PCs, laptops and mobile devices, which are used to access the OA system for employees to interact with the system.

[0059] Technical effects and advantages of the present invention: The present invention provides a reimbursement process management method based on RPA technology. The present invention configures an RPA software robot, presets the software robot execution path, database and rule base, and uses the RPA software robot's ability to automatically execute various processes. It interacts with the system according to the set program and automatically executes the reimbursement process, replacing manual data entry, information processing, form filling and other repetitive tasks. Compared with the prior art, the RPA software robot is more efficient, has fewer errors and is more convenient to use;

[0060] Secondly, during the data cleaning process of the software robot during data collection, the learning algorithm of the RPA software robot adopts a gradient boosting algorithm that gradually adds models, which improves the crawling accuracy of the software robot to a certain extent and makes the data collection results more accurate and correct.

[0061] Further features and advantages of the present invention will become apparent from the following detailed description of exemplary embodiments of the present invention with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the method provided by the present invention;

[0063] Figure 2 This is the RPA software robot architecture flow chart provided by the present invention;

[0064] Figure 3 It is a system flow chart provided by the present invention. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0066] like Figures 1 to 3 As shown, an embodiment of the present invention provides a reimbursement process management method and system based on RPA technology, including the following steps:

[0067] S1. Reimbursement system architecture: sort out and record the reimbursement process, create RPA automation process, plan the execution path of the software robot, program automatic login to the OA system, and develop software robots to automatically capture data and identify key information;

[0068] S2. Data collection: Employees upload reimbursement vouchers, and the software robot identifies and obtains the reimbursement data, and performs data cleaning and format conversion;

[0069] S3. Fill in the reimbursement document: The software robot fills the converted reimbursement data into the preset template;

[0070] S4. Intelligent review and verification: Software robots conduct inspections and approvals based on the company’s financial system and approval rules.

[0071] Usually, the reimbursement process within an enterprise is divided into manual form filling, bill collection and sorting, approval, financial processing, follow-up tracking and inquiry; first, employees need to manually fill out the reimbursement form according to the reimbursement template specified by the company, including but not limited to the reimbursement date, item details, amount, invoice number, invoice image file, etc. This stage takes a long time and is prone to errors or missing information;

[0072] Then the employee needs to provide relevant invoices, receipts or other supporting materials, scan or take photos of them for archiving, and then match them with the reimbursement form. This process is not only cumbersome, but also time-consuming to manage and find a large number of bills; after completing the form, the employee submits the reimbursement form and related materials to the superior leader or the financial department for approval. The approval process may involve multiple levels of approval, and each level requires manual review and confirmation of relevant information; after the financial department receives the approved reimbursement application, it will check the information again and enter it into the financial system, and then perform accounting processing, including bookkeeping, payment of reimbursements, etc.;

[0073] After the final reimbursement is completed, if there are any questions or audit tracing is required, it is often necessary to manually check paper documents or electronic records to find historical reimbursement data.

[0074] The existing reimbursement process has obvious problems such as high labor consumption, low efficiency, high risk of errors, and difficulty in information tracing. The "Innovative Application of Intelligent RPA Technology in Enterprise OA System - Creating an Efficient and Automated Reimbursement Form Filling Solution" is an innovative method proposed to address these problems. It replaces most manual operations with automated processes, greatly improving the efficiency and accuracy of the reimbursement process.

[0075] Furthermore, the embodiments of the present invention construct a reimbursement system framework based on the advantages of RPA technology in automating repetitive tasks and the accuracy and stability of robotic process automation, thereby performing the reimbursement process through robotic automation operations, reducing labor costs, shortening business processing time, and greatly reducing negligence and errors caused by human operations.

[0076] In step S1, the reimbursement system framework includes:

[0077] S11. Determine the reimbursement process, including logging into the OA system, data capture, entering reimbursement forms, reviewing and verifying;

[0078] S12. Design RPA software robots and plan the execution paths of software robots;

[0079] S13, data integration;

[0080] S14. Audit rule definition

[0081] S15. Configuration approval process;

[0082] S16. Test and test run the RPA software robot, optimize it based on the feedback from the test run results, and continuously monitor and maintain it.

[0083] OA system is an information system platform for daily management and operation of an enterprise, which is used to help store, retrieve and manage documents and data within the enterprise, automate business processes and collaborate in office work;

[0084] The execution path of the software robot is the reimbursement process;

[0085] Data integration includes reading reported data, writing data capture rules, writing data cleaning rules and presetting reimbursement document templates;

[0086] Audit rules include establishing a rule base to verify the compliance of reimbursement documents;

[0087] The approval process uses the preset approval process template in the OA system and is associated with the RPA software robot. After the RPA software robot checks the compliance of the submitted reimbursement vouchers, it pushes the approval request along the established path.

[0088] It should be noted that the automated reimbursement process adopts the basic reimbursement process of the enterprise. Based on the intelligent RPA software, when compiling the execution path of the software robot, the process that can perform simple and repetitive operations is selected for automated compilation;

[0089] Furthermore, the confirmed reimbursement process is that first, employees need to upload reimbursement vouchers on their own, the RPA robot automatically logs into the OA system, and then scans the reimbursement vouchers through the recognition module. Based on the scanned data, the reimbursement form is filled in with data, and then the filled-in reimbursement form is verified according to the rules and sent for approval;

[0090] At the same time, the RPA robot monitors the reporting status in real time. Once the reporting is passed, the robot will continue to perform subsequent operations.

[0091] In addition, comprehensive functional testing, performance testing, and exception handling testing are performed on the developed RPA process to ensure that the process is stable and reliable.

[0092] After the test is passed, it will be piloted within the specified scope, and fine-tuned and optimized based on feedback, and finally officially launched across the company; RPA reimbursement process training will be organized for employees to let them understand the new reimbursement process and learn to work with RPA robots; and continuous technical support and maintenance will be provided to deal with possible technical problems and business change requirements.

[0093] Furthermore, the use of RPA software has the following advantages:

[0094] The degree of automation has been greatly improved: the entire process has been automated, from capturing raw data, automatically filling in reimbursement documents, triggering the approval process to feedback on the final reimbursement results, which has greatly reduced the number of manual steps involved and significantly improved work efficiency.

[0095] Enhanced accuracy and compliance: Through precise data capture and intelligent verification mechanisms, we ensure that reimbursement information is accurate, strictly abide by the company's internal financial system and relevant regulations, and reduce compliance risks caused by human errors.

[0096] Optimal resource allocation: It saves employees a lot of time and energy, allowing them to focus on higher-value work. At the same time, it frees up the human resources of the finance department, helping to optimize the internal human resource allocation of the enterprise.

[0097] Faster data processing speed: Since the automated process is not restricted by time and space, reimbursement applications can be quickly processed within any time period, shortening the reimbursement cycle and improving capital turnover efficiency.

[0098] Improved user experience: Through a simple and intuitive user interface and fool-proof operation, the employee reimbursement process is simplified, which improves employee satisfaction and sense of identity with the company.

[0099] Enhanced decision support: The detailed data records and real-time reports generated by the automated reimbursement process provide business managers with rich decision support data, helping to refine financial management.

[0100] Good system integration and scalability: This technology achievement is easy to integrate with other enterprise information systems (such as ERP, CRM, etc.), has good scalability and portability, and can flexibly respond to the changing needs of enterprise business development.

[0101] In general, the OA system automatic reimbursement form filling solution based on RPA technology has brought significant process optimization effects to enterprises, improved the efficiency and level of financial management, and promoted the digital transformation of enterprises and the modernization of business processes.

[0102] In step S2, the specific steps of data collection include:

[0103] S21. Upload reimbursement vouchers: Submit reimbursement vouchers through the upload interface provided by the system. The reimbursement vouchers can be in the form of paper invoices, PDF files, and electronic invoices.

[0104] S22. The software robot identifies and obtains reimbursement data: The software robot uses OCR optical character recognition technology to read the text information on the voucher; and uses NLP natural language processing technology to parse the unstructured text data in the reimbursement voucher to help the software robot identify and process;

[0105] S23, data preprocessing: data cleaning and data conversion for the data recognized by the software robot;

[0106] S24, data post-processing: check the pre-processed text data again, keep the records of the original data and the processed data, and finally store the processed data text in the database and send it to the next process;

[0107] Furthermore, employees upload reimbursement vouchers according to the reimbursement voucher upload interface provided by the OA system, directly upload PDF files and electronic invoices to the system, and paper invoices can be uploaded after taking a photo;

[0108] OCR optical character recognition technology is a technology that converts text information in images into editable and searchable text data.

[0109] In step S2, data cleaning includes:

[0110] 1) Missing value processing: Identify and process missing data by filling in default values, using statistical methods to estimate and delete;

[0111] 2) Outlier processing: Use statistical tests, machine learning algorithms, and rule engines to identify outliers and handle them.

[0112] Furthermore,

[0113] In outlier detection, machine learning algorithms include gradient boosting algorithms;

[0114] The formula of the gradient boosting algorithm is as follows:

[0115] F m (x) = F m-1 (x)+p·h(x;α m )

[0116] Among them, F m (x) is the model prediction at step m;

[0117] F m-1 (x) is the model prediction at step m-1;

[0118] p is the learning rate (step size);

[0119] h(x; α m ) is the prediction result of the base learner added in the mth step;

[0120] α m are the parameters of the base learner;

[0121] Learning rate: The learning rate or step size p controls the update amplitude of the model in each iteration. A smaller learning rate can make the model more stable, but it may take more iterations to converge; a larger learning rate may make the model converge quickly, but it may also cause overfitting or instability.

[0122] Iteration number: Iteration number determines the number of base learners in the model. Increasing the number of iterations can increase the complexity of the model, thereby improving the performance of the model, but it may also lead to overfitting.

[0123] Tree depth: Limiting the depth of base learners can prevent the model from learning overly complex relationships, thereby reducing the risk of overfitting.

[0124] For example, suppose the model prediction result at step m-1 is F m-1 (x) = 0.6.

[0125] Learning rate (step size) p = 0.1;

[0126] At the mth step, the base learner h(x; α m ) The predicted result for input x is 0.5, where α m are the parameters of the base learner;

[0127] According to the update rule of the gradient boosting algorithm, we can calculate the model prediction at step m as follows:

[0128] F m (x) = F m-1 (x)+p·h(x;α m )

[0129] F m (x) = 0.6 + 0.1 0.5

[0130] F m (x) = 0.6 + 0.05

[0131] F m (x) = 0.65

[0132] Therefore, the model prediction result at step m is F m (x) = 0.65. This means that after considering the base learner added in the mth step, the new prediction probability of the model is 0.65, indicating that for a given input x, the probability estimate of being correctly classified has increased;

[0133] 3) Duplicate data processing: identify and remove duplicate records in the data;

[0134] 4) Consistency check: Check the read data through cross-table association to ensure that the read data is consistent in different data sets;

[0135] The content of data conversion includes:

[0136] 1) Standardization: Convert the read data into a standard format, including a unified format for date and time and a unified unit for values;

[0137] 2) Data type conversion: converting data from one type to another according to rules;

[0138] 3) Feature encoding: Convert non-numeric data into numeric data so that software robots can perform calculations.

[0139] Feature encoding is the process of converting raw data into a numerical form that can be understood by machine learning algorithms, including: One-hot encoding: convert each categorical variable into a binary vector in which only one element is 1 and the rest are 0; Label encoding: map each categorical variable to an integer, usually starting from 0; Binary encoding: convert each categorical variable into a binary number in which each bit indicates whether the category appears; Count encoding: count the number of times each categorical variable appears in the data set and use it as the encoding of the category; Target encoding: encode the categorical variable according to the mean of the target variable;

[0140] The role of feature coding in reimbursement data collection is reflected in processing non-numerical data, optimizing model performance, and refining data processing. Through feature coding, it can ensure that reimbursement data is effectively used to train machine learning models, thereby improving the accuracy and efficiency of automatic review and auditing.

[0141] Step S3: Filling in the reimbursement document includes:

[0142] S31. The software robot loads the preset reimbursement document template through the OA system;

[0143] S32, the software robot distinguishes and locates the filling area by identifying specific marks in the reimbursement document template;

[0144] S33, the software robot inputs the processed data text into the corresponding filling area respectively according to the type of the data text;

[0145] S34. The software robot compares the data text in the reimbursement document with the processed data text to ensure the data is consistent and sends it to the next process.

[0146] Furthermore, the preset document template is a document filling template provided by the OA system, which makes it easier for the RPA robot to identify data and fill in the data. The specific tags refer to the data tags in the document template, such as date, amount and item.

[0147] In step S4, intelligent review and verification includes:

[0148] S41. The software robot performs a budget limit check on the expenses of the reimbursement documents according to the enterprise financial system in the preset rule base;

[0149] S42. The software robot compares historical data with the database based on the audit rules in the preset rule base and marks the anomalies in the reimbursement documents;

[0150] S43, the software robot generates an audit report, including inspection results and abnormal data;

[0151] Furthermore, after the software robot performs a budget limit check on the reimbursement document according to the preset rule base, it generates an audit report containing the check results and abnormal data. This report provides a clear overview for the reviewers, allowing them to quickly understand which reimbursement documents have passed the review, which have problems, and what the specific problems are. During the inspection process, the software robot needs to collect data on all reimbursement documents, including the reimbursement amount, expense type, reimbursement person information, reimbursement time, etc., and compare them one by one according to the preset rule base. For each document, the software robot will record whether it has passed the inspection and the specific reasons for failure.

[0152] First, you need to design a report template, which contains the following basic elements:

[0153] 1) Title and date: clearly state the subject of the report and the date it was generated;

[0154] 2) Summary: A brief summary of the inspection results, such as the total number of documents, the number of documents that passed the review, the number of documents that failed the review, etc.;

[0155] 3) Inspection result list: List the inspection results of each reimbursement document in detail, including document number, reimbursement person, reimbursement amount, expense type, inspection result (pass / fail) and specific reasons for failure;

[0156] 4) Abnormal data summary: Summarize the documents that have not passed the review and classify them by expense type, reimbursement person or other relevant dimensions so that the reviewer can quickly locate the problem;

[0157] 5) Suggestions and actions: Propose improvement suggestions or follow-up actions for the problems found;

[0158] After the inspection process is completed, the software robot automatically generates an audit report based on the collected data and the designed template, which includes several steps:

[0159] 1) Data filling: fill the inspection results and abnormal data into the corresponding positions of the report template;

[0160] 2) Format adjustment: ensure that the report format is neat, the font is consistent, the numbering is orderly, etc.;

[0161] 3) File generation: Save the report in an appropriate file format (such as PDF, Word document, etc.) for review by the reviewer;

[0162] After the final audit report is generated, the software robot can automatically distribute it to the designated reviewer or department, which can be achieved through email, internal corporate communication platform or other automated channels;

[0163] After receiving the report, the approver will review the content and make further processing as needed. If the report points out issues that need improvement, the approver can contact the reimbursement person for more information or request to modify the reimbursement document. At the same time, they can also provide feedback to the software robot or IT department to optimize the inspection process and report template.

[0164] S44. The software robot sends the reimbursement documents that meet the standards to the next process, and sends the reimbursement documents with abnormal data to manual approval.

[0165] This embodiment also designs a reimbursement process management system based on RPA technology. The system is used to implement the above-mentioned reimbursement process management method based on RPA technology, including: a server, a storage device and a terminal device. The RPA software robot is deployed and run in the server, and includes a memory bar, a CPU, a GPU and a hard disk space; a database is deployed in the storage device for storing electronic documents, invoice images and reimbursement documents; the terminal devices include PCs, laptops and mobile devices, which are used to access the OA system for interactive operations between employees and the system.

[0166] It should be noted that high-performance servers used to deploy and run RPA robots should have sufficient memory, CPU, and hard disk space to support multiple concurrent robot operations, which involve a large amount of data processing and OCR recognition and may require GPU support to accelerate image recognition and processing speed.

[0167] At the same time, a fast and stable network environment is also required. RPA robots need to access various online systems (such as OA systems, financial systems, email systems, third-party APIs, etc.) to ensure smooth data transmission; and the storage device has a large memory requirement. For the large number of electronic documents, invoice images, etc. involved, sufficient storage space is required to store these temporary or permanent data.

[0168] In summary, the present invention, by constructing an RPA software robot, presets the software robot execution path, database and rule base, and uses the RPA software robot's ability to automatically execute various processes, interacts with the system according to the set program, and automatically executes the reimbursement process, replacing manual data entry, information processing, form filling and other repetitive tasks. Compared with the prior art, the RPA software robot is more efficient, has fewer errors, and is more convenient to use;

[0169] Secondly, during the data cleaning process of the software robot during data collection, the learning algorithm of the RPA software robot adopts a gradient boosting algorithm that gradually adds models, which improves the crawling accuracy of the software robot to a certain extent and makes the data collection results more accurate and correct.

[0170] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A reimbursement process management method based on RPA technology, characterized in that: The steps include: S1. Reimbursement system architecture: sort out and record the reimbursement process, create RPA automation process, plan the execution path of the software robot, program automatic login to the OA system, and develop software robots to automatically capture data and identify key information; S2. Data collection: Employees upload reimbursement vouchers, and the software robot identifies and obtains the reimbursement data, and performs data cleaning and format conversion; S3. Fill in the reimbursement document: The software robot fills the converted reimbursement data into the preset template; S4. Intelligent review and verification: Software robots conduct inspections and approvals based on the company’s financial system and approval rules.

2. According to claim 1, a reimbursement process management method based on RPA technology is characterized in that: In step S1, the reimbursement system framework includes: S11. Determine the reimbursement process, including logging into the OA system, data capture, entering reimbursement forms, reviewing and verifying; S12. Design RPA software robots and plan the execution paths of software robots; S13, data integration; S14, Audit rules definition; S15. Configuration approval process; S16. Test and test run the RPA software robot, optimize it based on the feedback from the test run results, and continuously monitor and maintain it.

3. According to claim 2, a reimbursement process management method based on RPA technology is characterized in that: The OA system is an information system platform for daily management and operation of an enterprise, which is used to help store, retrieve and manage documents and data within the enterprise, automate business processes and collaborate on office work; The execution path of the software robot is the reimbursement process; The data integration includes reading reported data, writing data capture rules, writing data cleaning rules and presetting reimbursement document templates; The audit rules include establishing a rule base for verifying the compliance of reimbursement documents; The approval process uses the preset approval process template of the OA system and is associated with the RPA software robot. After the RPA software robot checks the compliance of the submitted reimbursement vouchers, it pushes the approval request along the established path.

4. According to claim 1, a reimbursement process management method based on RPA technology is characterized in that: In step S2, the specific steps of data collection include: S21. Upload reimbursement vouchers: Submit reimbursement vouchers through the upload interface provided by the system. The reimbursement vouchers may be in the form of paper invoices, PDF files, and electronic invoices. S22. The software robot identifies and obtains reimbursement data: The software robot uses OCR optical character recognition technology to read the text information on the voucher; and uses NLP natural language processing technology to parse the unstructured text data in the reimbursement voucher to help the software robot identify and process; S23, data preprocessing: data cleaning and data conversion for the data recognized by the software robot; S24, data post-processing: check the pre-processed text data again, keep the records of the original data and the processed data, and finally store the processed data text in the database and send it to the next process.

5. According to claim 1, a reimbursement process management method based on RPA technology is characterized in that: In step S2, the data cleaning includes: Missing value handling: Identify and handle missing data by filling in default values, estimating using statistical methods, and deleting; Outlier handling: Use statistical tests, machine learning algorithms, and rule engines to identify outliers and handle them; Duplicate data processing: identifying and removing duplicate records in the data; Consistency check: Check the read data through cross-table association to ensure that the read data is consistent in different data sets; The data conversion content includes: Standardization: Convert the read data into a standard format, including a unified format for date and time and a unified unit for numerical values; Data type conversion: converting data from one type to another according to rules; Feature encoding: Convert non-numeric data into numeric data so that software robots can perform calculations.

6. According to claim 5, a reimbursement process management method based on RPA technology is characterized in that: In the outlier detection, the machine learning algorithm includes a gradient boosting algorithm; The gradient boosting algorithm formula is as follows: F m (x)=F m-1 (x)+p·h(x;α m ) Among them, F m (x) is the model prediction at step m; F m-1 (x) is the model prediction at step m-1; p is the learning rate (step size); h(x; α m ) is the prediction result of the base learner added in the mth step; α m are the parameters of the base learner.

7. According to claim 1, a reimbursement process management method based on RPA technology is characterized in that: Step S3: Filling in the reimbursement document includes: S31. The software robot loads the preset reimbursement document template through the OA system; S32, the software robot distinguishes and locates the filling area by identifying specific marks in the reimbursement document template; S33, the software robot inputs the processed data text into the corresponding filling area respectively according to the type of the data text; S34. The software robot compares the data text in the reimbursement document with the processed data text to ensure the data is consistent and sends it to the next process.

8. According to claim 1, a reimbursement process management method based on RPA technology is characterized in that: In step S4, intelligent review and verification includes: S41. The software robot performs a budget limit check on the expenses of the reimbursement documents according to the enterprise financial system in the preset rule base; S42. The software robot compares historical data with the database based on the audit rules in the preset rule base and marks the anomalies in the reimbursement documents; S43, the software robot generates an audit report, including inspection results and abnormal data; S44. The software robot sends the reimbursement documents that meet the standards to the next process, and sends the reimbursement documents with abnormal data to manual approval.

9. A reimbursement process management system based on RPA technology, characterized by: The system is used to implement the reimbursement process management method based on RPA technology as described in any one of claims 1 to 8, including: A server, in which the RPA software robot is deployed and run, and includes a memory bar, a CPU, a GPU, and a hard disk space; A storage device, in which a database is deployed for storing electronic documents, invoice images and expense receipts; Terminal devices, including PCs, laptops and mobile devices, are used to access the OA system for employees to interact with the system.

Images