Project billing full-process intelligent decision-making auxiliary method and system

Abstract

The invention relates to the technical field of project data management, and particularly discloses a project billing full-process intelligent decision-making auxiliary method and system. Basic data are synchronized in real time, a project billing basic database is formed, project information is automatically called, and budget suggestions are generated based on machine learning; according to the method, automatic filling and intelligent proofreading of the billing application form are realized, a billing application database is constructed, a hierarchical auditing intelligent decision-making model is constructed through the billing application database, project activeness is analyzed for different auditing stages, and auditing decisions of the billing application form are output in combination with historical billing application data. Based on a billing application database, audit and decision linkage mechanism management, loan management and project responsibility budget control, a process progress monitoring board after billing is constructed to monitor a project progress state, and early warning is carried out, so that the automation and intelligence level of a billing process can be improved, and the work efficiency is improved. And scientific basis and decision support are provided for fine project management of enterprises.

Description

Technical Field

[0001] This invention relates to the field of project data management technology, and more specifically, to an intelligent decision support method and system for the entire project invoicing process. Background Technology

[0002] In a company's project management system, project initiation is a core link connecting project establishment with subsequent cost control and fund allocation. The smoothness of the process and the rationality of the decision-making directly affect the overall quality of project progress. The current traditional project initiation process generally relies on manual offline operations and cross-departmental collaboration, which has significant shortcomings: basic data is scattered across different management modules (such as project ledgers, contract files, and supplier information sheets), requiring applicants to manually query and enter data across modules, which is not only cumbersome but also prone to invalidating application materials due to inconsistent data standards or entry errors; the cross-departmental review process lacks a unified data sharing mechanism, requiring the business development department to manually retrieve contract documents when reviewing project compliance, and the finance department to separately compile fund pool data when assessing funding feasibility, leading to information asymmetry that prolongs the review cycle and easily overlooks key risk points (such as supplier historical performance issues and project funding gaps); review decisions rely on the personal experience of reviewers, making it difficult to quickly compare and analyze data from similar historical projects and obtain real-time warnings of potential project risks, resulting in highly subjective decisions; after the initiation review is approved, subsequent processes such as project department creation, internal account opening, and responsibility budget preparation must be triggered manually one by one, requiring repeated data entry and easily causing process gaps; at the same time, the entire process lacks a data accumulation and review mechanism, making it difficult for management to accurately identify process bottlenecks and achieve process optimization and iteration. These problems lead to inefficiencies, high error rates, and a lack of support for decision-making in traditional project initiation processes, failing to meet the needs of enterprises for refined and large-scale project management.

[0003] Therefore, it is necessary to provide an intelligent decision-making support method and system for the entire project invoicing process to solve the above-mentioned technical problems. In order to solve the above problems, a technical solution is provided. Summary of the Invention

[0004] To overcome the aforementioned shortcomings of existing technologies, this invention provides an intelligent decision-making support method and system for the entire project invoicing process. This system addresses the problems in existing project invoicing processes, such as scattered basic data that relies on manual input, information asymmetry in cross-departmental audits, lack of data support for decision-making, and discontinuity in process connections after invoicing. These issues result in low process efficiency, high error rates, and strong decision-making subjectivity, making it difficult to adapt to the needs of refined project management in enterprises.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent decision support throughout the entire project invoicing process includes the following steps: The system synchronizes basic data for project maintenance, contract management, and supplier management in real time via API interface. After preprocessing the basic data, it forms a basic database for project invoicing. The system then uses rule engine technology to automatically verify the basic data. By automatically retrieving project information and generating budget suggestions based on machine learning, the system can automatically fill in and intelligently verify expense requisition forms, and build an expense requisition database. A tiered intelligent decision-making model for review is built using the billing application database. This model analyzes project activity at different review stages and outputs review decisions for billing applications based on historical billing application data. Based on the database of invoice applications and the linkage between the review and decision-making institutions, loan management, and project responsibility budget control, a post-invoice process progress monitoring dashboard is built to monitor the project progress status and issue early warnings. By collecting data from the entire process, intelligent analysis and iterative model optimization are performed to form a closed-loop feedback.

[0006] As a further aspect of the present invention, by automatically retrieving project information and generating budget suggestions based on machine learning, the automatic filling and intelligent verification of invoice application forms are achieved, and an invoice application database is constructed. The specific steps are as follows: The billing application form is automatically populated based on the project billing database. The first clause data of the project contract is extracted through OCR. The data of the first clause data is compared with the billing application form to obtain the information difference coefficient. Based on the information difference coefficient, the difference prompt is triggered to guide the correction of the billing application form and is synchronously stored in the billing application database.

[0007] As a further aspect of the present invention, the data of the first clause is compared with the billing application form to obtain an information difference coefficient. Based on the information difference coefficient, a difference prompt is triggered to guide the correction of the billing application form. The specific steps are as follows: The fixed field data corresponding to the project to be billed and the project billing basic database is automatically retrieved through the data association mapping algorithm, and the extracted fixed field data is filled into the billing application form. By combining machine learning regression models with historical cost structure data of similar projects in the project budget database, current project contract amount and construction period, an initial budget suggestion is intelligently generated, including the planned cost and target cost of various funding categories. The first clause data of the project contract is extracted by OCR. The data of the first clause data is compared with the fixed field data in the billing application form to obtain the information difference coefficient. The difference prompt is triggered based on the information difference coefficient.

[0008] As a further aspect of the present invention, the specific implementation steps for triggering a difference prompt based on the information difference coefficient evaluation are as follows: The information difference coefficient is compared with the preset difference threshold. If the information difference coefficient is greater than or equal to the preset difference threshold, it indicates that there is a content deviation between the billing application and the project contract, and the difference prompt is automatically triggered to guide the applicant to check and correct the billing application. If the information difference coefficient is less than the preset difference threshold, it indicates that there is no content deviation between the billing application and the project contract.

[0009] As a further aspect of the present invention, a hierarchical intelligent decision-making model for billing application is constructed through a billing application database. This model analyzes project activity at different review stages and outputs review decisions for billing application forms based on historical billing application data. The specific steps are as follows: Extract historical application data that has been approved from the account opening application database; Based on historical application data, behavioral indicators of similar projects are obtained as the first behavioral indicator, and behavioral indicators of projects applying for account opening are obtained as the second behavioral indicator. Identify the current review stage based on the approval process nodes, and construct an activity analysis model using first-behavior indicators and second-behavior indicators to analyze project activity at different review stages. By combining project activity at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. If the review is passed, the billing application is automatically pushed to the next stage; if the review is not passed, the reason for return is selected, and a list of modification suggestions is automatically generated based on the reason for return, including the types of supplementary materials and the direction of data correction, and pushed to the applicant.

[0010] As a further aspect of the present invention, the behavioral indicators include the number of data modifications, the average duration of modification review, and the number of cross-departmental collaborations.

[0011] As a further aspect of the present invention, by combining the project activity levels at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. The specific steps are as follows: Based on historical application data, obtain a checklist of key points for reviewing account opening applications, historical data comparison results, and financial analysis results; A comprehensive review scoring model is constructed by combining the review checklist, historical data comparison results, and financial analysis results with project activity to obtain a review score, and the review pass rate is determined based on the review score.

[0012] As a further aspect of the present invention, a comprehensive audit scoring model is constructed by combining the audit key points list, historical data comparison results, and financial analysis results with project activity to obtain the audit score. The comprehensive audit scoring model is as follows: At each review stage, a basic review score is calculated based on multiple data points in the review checklist, according to their importance weights. The score is then adjusted by comparing the average risk indicators with historical data and using weighted coefficients. Finally, the score is further adjusted using weighted coefficients based on the funding matching results.

[0013] As a further aspect of the present invention, the review is determined based on the review score. Specifically, the review score is compared with a preset score range. If the review score is higher than the upper limit of the preset score range, the current project is marked as "Review passed". If the review score is within the preset score range, the current project is marked as "Requires supplementary materials". If the review score is lower than the lower limit of the preset score range, the current project is marked as "Review failed, returned".

[0014] An intelligent decision support system for the entire project invoicing process includes a basic data processing and verification module, an intelligent verification module for invoicing applications, an intelligent hierarchical review and decision support module, a monitoring dashboard linkage module, and an iterative optimization module; The basic data processing and verification module is used to synchronize basic data for project maintenance, contract management and supplier management in real time through API interface. After preprocessing the basic data, it forms a basic database for project invoicing and uses rule engine technology to automatically verify the basic data. The billing application intelligent verification module is used to automatically fill in and intelligently verify billing application forms by automatically retrieving project information and generating budget suggestions based on machine learning, and to build a billing application database. The intelligent hierarchical review decision support module is used to build a hierarchical review intelligent decision model through the billing application database, analyze the activity of projects at different review stages, and output the review decision of billing application forms in combination with historical billing application data; The monitoring dashboard linkage module is used to build a post-opening process progress monitoring dashboard based on the opening application database and the linkage of review decision-making institutions, loan management, and project responsibility budget control to monitor the project progress status and issue early warnings. The iterative optimization module is used to perform intelligent analysis and iterative optimization of the model by collecting data from the entire process, forming a closed-loop feedback.

[0015] The technical effects and advantages of this invention, a fully intelligent decision-making support method and system for project invoicing, are as follows: This invention achieves real-time synchronization of basic data for project maintenance, contract management, and supplier management through API interfaces. Combined with preprocessing and automatic verification via a rule engine, it not only reduces manual input errors but also achieves centralized data management, improving the integrity and reliability of basic data and providing a solid data foundation for subsequent invoicing decisions. By automatically retrieving project information and using machine learning to generate budget suggestions, invoicing application forms can be automatically filled and intelligently verified. This significantly improves process efficiency, reduces manual operation costs, and minimizes deviations caused by human error or subjective judgment, making budget suggestions more scientific and accurate. Through a hierarchical review intelligent decision-making model combined with project activity and historical invoicing data, the review process can achieve dynamic, data-driven decision support, thereby optimizing cross-departmental collaboration, reducing information asymmetry, and improving review quality and consistency. The post-invoicing process progress monitoring dashboard can track project status in real time and link with institutional management, loan management, and budget control, achieving transparent management and risk warning throughout the project process. This allows managers to promptly identify problems and take measures to ensure the smooth progress of the project. By collecting and analyzing data throughout the entire process, the model can be continuously iterated and optimized to form a closed-loop feedback loop. This not only improves the automation and intelligence of the billing process, but also provides scientific basis and decision support for the enterprise's refined project management. As a result, the overall process efficiency is significantly improved, the error rate is reduced, and the data-driven decision-making capability is strengthened, thus meeting the enterprise's needs for refined and intelligent project management. Attached Figure Description

[0016] Figure 1 A flowchart illustrating an intelligent decision support method for the entire project invoicing process provided in this embodiment of the invention; Figure 2 This is a system block diagram of an intelligent decision support system for the entire project invoicing process, provided as an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of this invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described technical solutions are only a part of this invention, and not all of it. All other technical solutions obtained by those skilled in the art based on the technical solutions of this invention without inventive effort are within the scope of protection of this invention.

[0018] like Figure 1 The diagram shown is a flowchart of an intelligent decision-making support method for the entire project invoicing process provided by an embodiment of the present invention. Figure 1The execution entity of the method shown can be a software and / or hardware device. The execution entity of this application can include, but is not limited to, at least one of the following: user equipment, network equipment, etc. User equipment can include, but is not limited to, computers, smartphones, personal digital assistants (PDAs), and the aforementioned electronic devices. Network equipment can include, but is not limited to, a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers. Cloud computing is a type of distributed computing, consisting of a super virtual computer composed of a group of loosely coupled computers. This embodiment does not limit this. Steps S1 to S5 are detailed as follows: Step S1: Real-time synchronization of basic data for project maintenance, contract management, and supplier management via API interface; preprocessing of basic data to form a basic database for project invoicing; and automatic verification of basic data using rule engine technology. Step S2: By automatically retrieving project information and generating budget suggestions based on machine learning, the automatic filling and intelligent verification of invoice application forms are achieved, and an invoice application database is built. Step S3: Construct a tiered intelligent decision-making model for billing application through the billing application database, analyze project activity at different review stages, and output review decisions for billing application forms based on historical billing application data; Step S4: Based on the opening application database and the management of the review and decision-making linkage institutions, loan management, and project responsibility budget control, build a post-opening process progress monitoring dashboard to monitor the project progress status and issue early warnings; Step S5 involves collecting data from the entire process for intelligent analysis and iterative model optimization, forming a closed-loop feedback loop.

[0019] It should be noted that the automatic validation of basic data using rule engine technology follows these steps: By verifying whether the project to be billed has been registered for compliance in the project maintenance module, a prompt will be automatically triggered if it has not been registered, and a quick access to the project maintenance module will be provided to allow the applicant to complete the data in real time. By using OCR recognition and natural language processing technology to analyze supplier creditworthiness documents, extract key information and compare it with the supplier creditworthiness data stored in the system, if the creditworthiness level does not meet the company's preset standards, a risk warning report is automatically generated, along with a summary of abnormal performance records. Based on collaborative filtering algorithms, according to the type of project to be reimbursed, such as railway reconstruction or building expansion, the system matches basic data templates of similar historical projects and intelligently recommends commonly used funding categories, such as subcontracting costs and material costs in direct costs, office expenses and testing and experimental fees in on-site expenses, as well as the cost structure proportion range. The recommended data is marked with an "intelligent recommendation" label, which allows applicants to adjust according to the actual situation of the project, laying a data foundation for the generation of subsequent reimbursement applications.

[0020] The basic data includes project code, project overview, related department information, contract filing number, settlement terms, contract amount, supplier name, bank account information, credit rating and historical performance records.

[0021] Preferably, by automatically retrieving project information and generating budget suggestions based on machine learning, the system achieves automatic filling and intelligent verification of expense requisition forms and builds an expense requisition database. The specific steps are as follows: The billing application form is automatically populated based on the project billing database. The first clause data of the project contract is extracted through OCR. The data of the first clause data is compared with the billing application form to obtain the information difference coefficient. Based on the information difference coefficient, the difference prompt is triggered to guide the correction of the billing application form and is synchronously stored in the billing application database.

[0022] Preferably, the data in the first clause is compared with the billing application form to obtain an information difference coefficient. Based on the information difference coefficient, a difference prompt is triggered to guide the correction of the billing application form. The specific steps are as follows: The fixed field data corresponding to the project to be billed and the project billing basic database is automatically retrieved through the data association mapping algorithm, and the extracted fixed field data is filled into the billing application form. By combining machine learning regression models with historical cost structure data of similar projects in the project budget database, current project contract amount and construction period, an initial budget suggestion is intelligently generated, including the planned cost and target cost of various funding categories. The first clause data of the project contract is extracted by OCR. The data of the first clause data is compared with the fixed field data in the billing application form to obtain the information difference coefficient. The difference prompt is triggered based on the information difference coefficient.

[0023] In the actual enterprise project billing process of this embodiment of the invention, the infrastructure construction company needs to initiate a billing application for the "Urban Rail Transit Phase III Project, Section 2" project. First, based on the enterprise project billing database, the structured data of the project already registered in the project maintenance module, contract management module, and supplier management module is automatically retrieved, including project number, project name, contract number, contract amount, project manager and affiliated branch information. The data is then directly filled into the billing application form using a data association mapping algorithm, eliminating the need for manual entry.

[0024] Subsequently, the internally deployed machine learning regression model is invoked to retrieve historical cost structure sample data of similar urban rail transit projects from the project's basic accounting database. The model comprehensively analyzes the current project's contract amount and planned construction period, intelligently generating an initial budget suggestion. The budget is divided into three main categories: direct costs, indirect costs, and site expenses, and further refined into subcategories such as subcontracting fees, material costs, labor costs, and office expenses. The model outputs the planned cost and target cost for each expense category, and the results are automatically displayed in the budget column of the accounting application form, marked with the words "Intelligent Recommendation" to prompt the applicant to verify.

[0025] After the application form is automatically generated, the OCR intelligent verification module is activated. After the project manager uploads a scanned copy of the contract, the OCR system recognizes the contract text and parses its "First Clause Data". The recognition result contains fixed field data such as project manager, associated branch, and contract amount. The system automatically compares the recognized "First Clause Data" with the corresponding fixed field data in the invoice application form and calculates the information difference coefficient based on the text similarity calculation formula.

[0026] When a discrepancy is detected between the amount field in the contract file and the record on the application form (e.g., the contract amount is 48 million yuan, while the application form shows 50 million yuan), and the information difference exceeds a preset threshold, the system automatically triggers a discrepancy alert mechanism. On the application form interface, the "Contract Amount" field is highlighted in red, and a discrepancy alert box pops up, displaying "The application amount is inconsistent with the contract file. Do you want to correct it based on the contract data?" It also provides two operation options: "Automatic Correction" and "Manual Confirmation".

[0027] After the applicant selects "Auto Correct," the system immediately updates the corresponding field values ​​on the application form and generates a correction log, recording the values ​​before and after the correction, the operator, and the timestamp information. The corrected application form is re-verified and marked as "Checked," and finally stored in the expense application database in a structured data format. The database synchronously saves complete application information, budget generation records, discrepancy check logs, and manual interaction results for subsequent multi-department intelligent review and historical data retrospective analysis.

[0028] As can be seen from the above examples, the entire process of project invoicing application forms, from automatic data acquisition, intelligent budget generation, contract discrepancy identification to intelligent correction and storage, has been fully automated. This effectively avoids manual input errors and inconsistencies in document information, and significantly improves the accuracy and efficiency of project invoicing.

[0029] Preferably, the data in the first clause is compared with the fixed field data in the billing application form to obtain an information difference coefficient. The formula for calculating the information difference coefficient is as follows: ; ; In the formula: Let be the information discrepancy coefficient for project contract A. This represents the total number of fixed field data points to be verified in the first clause of the data. The weight coefficient for the i-th fixed field data. This is the data for the i-th fixed field of project contract A. For the i-th fixed field data in the billing application form, The text matching degree between the data in the first clause and the data in the i-th fixed field of the billing application form.

[0030] It should be noted that the fixed field data includes project manager, associated branch office, contract amount, etc.

[0031] In one embodiment of the present invention, when a construction company initiates an invoice application for a "Comprehensive Urban Road Improvement Project," the system automatically extracts the text content of the "First Clause Data" of the project contract using OCR technology. This clause clearly states that the project manager is "Li Jianguo," the associated branch is "South China Branch," and the contract amount is "RMB 48 million." Simultaneously, the system extracts corresponding fixed field data from the invoice application form, namely "Li Jianguo," "South China Regional Company," and "RMB 50 million."

[0032] Based on the preset field weight coefficients, the weight of the project leader is 0.2, the weight of the associated branch is 0.3, and the weight of the contract amount is 0.5. The information difference coefficient calculation formula is activated. Among them, the text matching degree is calculated by the semantic similarity algorithm: the matching degree of the project leader field is 1, which is completely consistent; the matching degree of the associated branch field is 0.95, because "South China Branch" and "South China Regional Company" are similar in semantics but different in expression; the matching degree of the contract amount field is 0.96, and the amount difference is 2 million yuan.

[0033] After substituting the values ​​into the information difference coefficient calculation formula, the calculated information difference coefficient is 0.035. Comparing this result with the preset difference threshold of 0.5 indicates that there is no content deviation between the invoice application and the project contract. Alternatively, if the preset difference threshold is 0.03, the coefficient is considered to exceed the standard range, indicating a deviation between the application and the contract. The system will then highlight the deviation field and the corresponding original contract data on the interface, displaying the message: "The contract amount differs from the original contract document. Please check and confirm the correction." Through the above process, semantic-level comparison and difference quantification of multi-field content can be automatically completed in the pre-billing stage. This not only achieves accurate verification of the consistency between project contract and application form information, but also provides a reliable data foundation for subsequent intelligent verification and difference prompting mechanisms.

[0034] Preferably, the specific implementation steps for triggering the difference prompt based on the information difference coefficient are as follows: The information difference coefficient is compared with the preset difference threshold. If the information difference coefficient is greater than or equal to the preset difference threshold, it indicates that there is a content deviation between the billing application and the project contract, and the difference prompt is automatically triggered to guide the applicant to check and correct the billing application. If the information difference coefficient is less than the preset difference threshold, it indicates that there is no content deviation between the billing application and the project contract.

[0035] The discrepancy alerts include the names of the discrepancy fields; fragments of the original contract terms identified by the system; the content of the corresponding fields on the billing application form; and the information discrepancy coefficient.

[0036] In the actual operation of enterprise project accounting, taking the "City A Municipal Infrastructure Improvement Project" as an example, the system extracts the "First Clause Data" of the contract through OCR in the early stage and calculates the information difference coefficient. Comparing this value with the preset difference threshold of 0.02, it was found that... If the threshold is exceeded, it indicates a discrepancy between the invoice request and the project contract. In this case, the discrepancy alert mechanism will be automatically triggered.

[0037] In the billing application system interface, a pop-up window generates a discrepancy prompt message, which includes: the name of the discrepancy field: contract amount; the original text of the contract terms recognized by the system: "First clause data: the total contract amount is RMB 48 million"; the corresponding field content of the billing application form: "contract amount: ¥50,000,000.00"; and the information discrepancy coefficient: 0.042.

[0038] The prompt box will simultaneously display the message "The system has detected a discrepancy between the application form and the contract document. Please check and confirm the correction." Applicants can choose between "automatic correction" or "manual verification." If automatic correction is selected, the system will automatically update the application form fields based on the contract OCR recognition results. If manual verification is selected, the system will mark the relevant fields as "pending confirmation" and prompt the applicant to upload the corrected contract text or remarks.

[0039] Conversely, in another project, "Park Supporting Road Project," the system calculated an information discrepancy coefficient of 0.009, which is lower than the set threshold of 0.02, indicating that the application form and contract information are consistent. The system determines there is no content deviation, does not trigger a discrepancy warning, directly marks the application form as "Contract Information Consistent," and proceeds to the next step.

[0040] This mechanism enables the dynamic assessment of the accuracy of billing application data based on the information discrepancy coefficient. It ensures timely alerts and guidance for manual correction when discrepancies are significant, thereby effectively reducing review delays and financial risks caused by discrepancies between contract and application information, and significantly improving the automation and data consistency of the project billing process.

[0041] Preferably, a tiered intelligent decision-making model for billing application review is constructed using the billing application database. This model analyzes project activity at different review stages and outputs review decisions for billing applications based on historical billing application data. The specific steps are as follows: Extract historical application data that has been approved from the account opening application database; Based on historical application data, behavioral indicators of similar projects are obtained as the first behavioral indicator, and behavioral indicators of projects applying for account opening are obtained as the second behavioral indicator. Identify the current review stage based on the approval process nodes, and construct an activity analysis model using first-behavior indicators and second-behavior indicators to analyze project activity at different review stages. By combining project activity at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. If the review is passed, the billing application is automatically pushed to the next stage; if the review is not passed, the reason for return is selected, and a list of modification suggestions is automatically generated based on the reason for return, including the types of supplementary materials and the direction of data correction, and pushed to the applicant.

[0042] It should be noted that the behavioral indicators include the number of times data was modified, the average time for modification review, and the number of times cross-departmental collaboration was conducted.

[0043] In the practice of enterprise project billing management, taking the "Urban Rail Transit Phase IV Construction Project" as an example, the system constructs a hierarchical intelligent decision-making model through the billing application database to conduct multi-stage intelligent review and activity analysis of the project billing process.

[0044] First, the system extracts historical application data that has been reviewed and approved from the account opening application database. This includes project information such as project size, contract amount, construction period, etc., review information such as approval results of each review node, the functional department of the approver, approval time, and process characteristic data such as the number of review stages, number of returns, and material supplement records. The system then selects historical sample data that meet the criteria from similar projects as the training set.

[0045] Secondly, representative behavioral indicators are extracted from historical data as "first behavioral indicators," such as the average number of data modifications for similar projects in the past being 2, the number of cross-departmental collaborations being 3, and the average modification review time being 1.8 days. At the same time, the system obtains the behavioral indicators of the current account opening application project as "second behavioral indicators," such as the current project having 4 data modifications, 5 cross-departmental collaborations, and the average modification review time being 3.2 days.

[0046] When the system identifies that the current review node belongs to the "Business Development Department Review Stage", it automatically calls the activity analysis model and calculates the project activity level as 0.78, which is higher than the average of 0.55 for similar projects. This indicates that the current project has a lot of frequent modifications and multi-departmental interactions during the review process, reflecting that the project has a high degree of complexity and risk exposure in the review process.

[0047] Next, the activity results for this phase are compared with historical samples of similar projects that have passed review, and the review decision is output using a machine learning model. If the model determines that a project's activity level is higher than the warning threshold and the number of modifications exceeds three, manual review is required, and the system generates a review prompt: "The project's behavioral characteristics deviate from the average of similar samples; manual verification of the reasonableness of contract terms and budget adjustments is recommended."

[0048] After the applicant completes the revised materials according to the prompts and resubmits them, the behavioral indicators are updated simultaneously, and the activity level and review results are recalculated for the next submission. Through this closed-loop mechanism, the project opening review process achieves intelligent judgment, early risk warning, and self-learning optimization, significantly reducing the burden of manual review and the time spent on process round trips.

[0049] Preferably, an activity analysis model is constructed using first and second behavioral indicators to analyze project activity at different review stages. The formula for the activity analysis model is: ; In the formula: The activity level of project q during the review phase. The number of data types for behavioral indicators. The weight coefficient corresponding to the j-th data category. This is the second-row indicator corresponding to the j-th data type in the review phase. For projects of the same type, this represents the minimum value of the first row of indicators corresponding to the j-th data type in the review phase q. This refers to the maximum value of the first row of the indicator corresponding to the j-th data type in the review phase of projects of the same type.

[0050] It should be noted that, This is a dynamically adjustable weighting coefficient. If the current project involves frequent data modifications and cross-departmental collaborations, and the average modification and review time is relatively long, it indicates that although the activity is frequent, the processing cycle is long, suggesting communication friction or slow response. In this case, the project activity score is high, but it exhibits a "high activity - low efficiency" characteristic. The current project is very active due to the high level of participation, communication, and adjustments, but the active behavior is more focused on "repeated modifications and coordination" rather than efficient progress. Therefore, the system may still mark it as a "high activity project" in the output results, but the annotation will be: "high activity - low flow efficiency project" or "active but with risk of delayed review response".

[0051] In this embodiment of the invention, taking the "Urban Integrated Pipe Gallery Construction Project" as an example, a multi-level review process for enterprise project invoicing management needs to be quantitatively analyzed to assess the smoothness and collaborative efficiency of the project's performance at different review stages. Based on historical project data accumulated in the invoicing application database, an activity analysis model is constructed to compare the differences between the current project and similar projects in multiple behavioral indicators, thereby calculating the project activity at each stage.

[0052] Specifically, three core behavioral indicators were first selected: number of data modifications, average review time, and number of cross-departmental collaborations. For projects currently in the "Finance Department Review Stage," the system extracted the second behavioral indicator data for this stage from the account opening application database: 4 data modifications, 2.8 days average review time, and 5 cross-departmental collaborations. Simultaneously, historical statistical data for similar projects, such as municipal infrastructure projects, at the same review stage were retrieved to obtain the minimum and maximum values ​​for the first behavioral indicator: the maximum number of data modifications was 6, the minimum was 1, the maximum average review time was 4, the minimum was 1.5, the maximum number of cross-departmental collaborations was 7, and the minimum was 2. The project activity level for this stage was calculated using an activity analysis model, with a weighting coefficient of 0.4 for the number of data modifications, 0.35 for the average review time, and 0.25 for the number of cross-departmental collaborations. The final project activity level for this stage was 0.56. Comparing the project's activity level with the average activity level of 0.45 for similar projects, we can see that the current project's activity level is slightly higher. This indicates that there are more data modifications and cross-departmental collaborative operations at this stage, reflecting a high frequency of communication and dynamic responsiveness in the project's review process.

[0053] Further analysis based on preset rules indicates the following: High activity levels and a significantly higher number of modifications compared to the average for similar projects may indicate issues such as repeated document revisions or inefficient workflows. Low activity levels may suggest stagnation in the review phase or insufficient collaboration. For this project, the system generates the following message: "High activity levels in the finance department's review phase. It is recommended to review the budget details and fund matching to prevent duplicate modifications." This example demonstrates that the activity analysis model can achieve dynamic quantitative evaluation of projects at each review stage. It not only reflects the operational status of the project process but also provides quantifiable support for intelligent review decisions, helping to identify process bottlenecks and improve overall billing efficiency.

[0054] Preferably, by combining project activity at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. The specific steps are as follows: Based on historical application data, obtain a checklist of key points for reviewing account opening applications, historical data comparison results, and financial analysis results; A comprehensive review scoring model is constructed by combining the review checklist, historical data comparison results, and financial analysis results with project activity to obtain a review score, and the review score determines whether the review is passed. The review score is compared with the preset score range. If the review score is higher than the upper limit of the preset score range, the current project is marked as "Review passed"; if the review score is within the preset score range, the current project is marked as "Supplementary materials required"; if the review score is lower than the lower limit of the preset score range, the current project is marked as "Review failed, returned".

[0055] Preferably, the audit score is obtained by constructing a comprehensive audit scoring model by combining the audit checklist, historical data comparison results, and financial analysis results with project activity. The formula for the comprehensive audit scoring model is as follows: ; In the formula: For review and scoring, This represents the total number of audit stages. The activity level of project q during the review phase. To verify the total number of data in the checklist, To review the weight coefficient corresponding to the b-th data point in the checklist, To review the b-th data point in the checklist, The weighting coefficients for comparing the average risk indicators of the data. To compare the data with the average risk indicators, This is a weighting coefficient for the degree of matching funds. This represents the funding matching result.

[0056] In the enterprise project opening intelligent review system of this invention, taking the "urban underground integrated pipe gallery construction project" as an example, the system uses a hierarchical review intelligent decision-making model to comprehensively score and automatically decide on the review results of the project at each stage.

[0057] First, retrieve historical project data that has been reviewed from the account opening application database, extract a list of key review points for similar projects, and also extract corresponding historical data comparison results and financial analysis results. Based on this, establish a comparison framework for the current project.

[0058] Next, combining the project activity results calculated in the previous step for each review stage (e.g., 0.63 for the Business Development Department and 0.56 for the Finance Department), the comprehensive review scoring model is used to calculate the overall review score for the project. The model comprehensively considers the importance of each indicator in the review checklist, sets weighting coefficients, quantifies each review point, and combines historical data on the average risk indicators and funding matching results to arrive at a final review score of 0.76 for the project.

[0059] The system then compares the review score with a preset score range: upper limit 0.75, lower limit 0.45. Because... The system will automatically mark the project as "approved" and push the account opening application to the next process node, "Fund Disbursement and Responsibility Budget Confirmation".

[0060] If the audit score for another project, "Industrial Park Supporting Road Renovation Project," is calculated to be 0.62, which is within the preset score range, the system will automatically mark it as "requiring supplementary materials" and generate an intelligent prompt list, indicating that "the budget reasonableness score is too low, and it is recommended to supplement the material cost breakdown table and supplier quotation comparison table." At the same time, the prompt will be pushed to the project manager.

[0061] For example, when a project's review score is detected to be only 0.41, which is lower than the lower threshold of 0.45, the system automatically determines that "the review is not passed and the project is returned," and automatically generates a return reason from the standardized reason library: "The contract compliance is lacking and the funding matching is insufficient." The system also outputs a list of modification suggestions, indicating the types of supplementary materials and the direction of correction.

[0062] This mechanism enables a shift from data-driven audit scoring calculations to automated audit decision outputs, ensuring that each audit result is supported by quantitative evidence and traceable data, thereby significantly improving the transparency, objectivity, and automation of enterprise project invoicing audits.

[0063] It should be noted that the specific steps for obtaining the audit checklist, historical data comparison results, and financial analysis results are as follows: By obtaining project feasibility study report summaries, contract compliance clauses, and project leader qualification certificates from historical application data, and integrating them with knowledge graph technology to generate a structured review checklist, the data in the review checklist includes project compliance indicators, contract clause completeness indicators, and the matching degree of the leader's qualifications, etc. By retrieving review records of similar projects from historical application data, key project parameters are obtained. A data comparison risk indicator is then generated by calculating the similarity of key parameters between the current project and historical projects. Key project parameters include project size, cooperating suppliers, and cost structure. It should be noted that if the current project has special circumstances, such as newly added special provisions in the contract, a risk warning card is automatically generated, recommending review and handling solutions for similar historical situations for reviewers' reference.

[0064] The funding analysis results are connected to the funding management module through a real-time data interface to obtain the current balance of the enterprise's cash pool and the amount of funds already used for each project. Using a cash flow simulation algorithm, combined with the budget amount of the projects to be opened and the funding needs calculated based on the construction period, a funding feasibility analysis report is generated to clarify the funding supply and demand matching degree of the projects. If there is a funding gap, the funding allocation plan is automatically recommended, such as allocating idle project funds or applying for special loans.

[0065] Preferably, based on the invoice application database and the management of the review and decision-making linkage institutions, loan management, and project responsibility budget control, a post-invoice process progress monitoring dashboard is constructed to monitor the project progress status and issue early warnings. The specific steps are as follows: In the organization management module, the corresponding project department organization is automatically created based on the project name and associated branch information in the billing application through API call technology. Basic information such as the project department leader and contact number are pre-filled, the project department status is set to "enabled" at the same time, and the information is pushed to the personnel management module to support subsequent personnel configuration and permission allocation.

[0066] In the loan management module, based on the project budget amount and funding requirements in the application form, the internal account creation application is automatically initialized, and information such as the account name, associated project department, and recommended bank is pre-filled. After the account opening application form is generated, it is pushed to the finance staff's to-do list.

[0067] In the project responsibility budget module, the generated initial budget suggestions are reused to automatically generate a responsibility budget preparation framework, clarify the budget control amount for each funding category, and associate the payment subject dictionary, such as mapping "subcontracting costs" to the "direct costs" subject. A budget preparation timeout warning mechanism is also set up, and the time node for triggering the warning is based on the enterprise project management specifications.

[0068] Meanwhile, a real-time data stream processing technology is used to build a progress monitoring dashboard for the post-account opening process, which displays the progress status of project creation, account opening, and responsibility budget preparation in real time, such as "completed", "pending", and "timeout warning". If a certain step is not completed within the preset time, a system message and SMS warning will be automatically sent to the corresponding person in charge, such as the project creation specialist or the financial account opening personnel. Abnormal steps are marked on the dashboard, which supports management to track the progress of the process in real time and ensures seamless connection of the process after account opening.

[0069] Preferably, intelligent analysis and iterative model optimization are performed by collecting data from the entire process to form a closed-loop feedback loop. The specific steps are as follows: Collect operational data for the entire project initiation process, including time spent on basic data verification, application form completion time, departmental review time, reasons for and frequency of review rejections, and completion status and time of related processes, and store this data in a data warehouse. Employ data mining algorithms, such as decision trees and cluster analysis, to analyze the entire process data, identifying bottlenecks, such as excessively long review times in certain departments or high rejection rates for certain types of project applications. Based on reinforcement learning algorithms, combine review results with subsequent project progress, such as post-initiation fund utilization efficiency and project cost control, to continuously optimize the project initiation database, budget recommendations, initiation application database, and review decisions, such as adjusting the weight of key review points and optimizing risk warning thresholds. Generate a full-process analysis report on initiation using business intelligence tools, including core indicators such as process efficiency trends, review risk distribution, and the connection status of related processes. The report data feeds back into the preprocessing of basic data and the hierarchical review intelligent decision-making model, enabling continuous iterative optimization of the initiation process and improving the overall level of intelligence.

[0070] An intelligent decision support system for the entire project invoicing process includes a basic data processing and verification module, an intelligent verification module for invoicing applications, an intelligent hierarchical review and decision support module, a monitoring dashboard linkage module, and an iterative optimization module; The basic data processing and verification module is used to synchronize basic data for project maintenance, contract management and supplier management in real time through API interface. After preprocessing the basic data, it forms a basic database for project invoicing and uses rule engine technology to automatically verify the basic data. The billing application intelligent verification module is used to automatically fill in and intelligently verify billing application forms by automatically retrieving project information and generating budget suggestions based on machine learning, and to build a billing application database. The intelligent hierarchical review decision support module is used to build a hierarchical review intelligent decision model through the billing application database, analyze the activity of projects at different review stages, and output the review decision of billing application forms in combination with historical billing application data; The monitoring dashboard linkage module is used to build a post-opening process progress monitoring dashboard based on the opening application database and the linkage of review decision-making institutions, loan management, and project responsibility budget control to monitor the project progress status and issue early warnings. The iterative optimization module is used to perform intelligent analysis and iterative optimization of the model by collecting data from the entire process, forming a closed-loop feedback.

[0071] like Figure 2 The diagram shown is a system block diagram of an intelligent decision support system for the entire project invoicing process according to an embodiment of the present invention, which can be used to execute... Figure 1 The steps in the method embodiments shown are implemented in a similar manner and have similar technical effects, and will not be repeated here.

[0072] Through the above embodiments, this invention achieves real-time synchronization of basic data for project maintenance, contract management, and supplier management via API interfaces. Combined with preprocessing and automatic verification via a rule engine, this not only reduces manual input errors but also achieves centralized data management, improving the integrity and reliability of basic data and providing a solid data foundation for subsequent billing decisions. By automatically retrieving project information and using machine learning to generate budget suggestions, billing applications can be automatically filled and intelligently verified. This significantly improves process efficiency, reduces manual operation costs, and minimizes deviations caused by human error or subjective judgment, making budget suggestions more scientific and accurate. Through a hierarchical review intelligent decision-making model combined with project activity and historical billing data, the review process can achieve dynamic, data-driven decision support, thereby optimizing cross-departmental collaboration, reducing information asymmetry, and improving review quality and consistency. The post-billing progress monitoring dashboard can track project status in real time and link with institutional management, loan management, and budget control, achieving transparent management and risk warning throughout the project process. This allows managers to promptly identify problems and take measures to ensure smooth project progress. By collecting and analyzing data throughout the entire process, the model can be continuously iterated and optimized to form a closed-loop feedback loop. This not only improves the automation and intelligence of the billing process, but also provides scientific basis and decision support for the enterprise's refined project management. As a result, the overall process efficiency is significantly improved, the error rate is reduced, and the data-driven decision-making capability is strengthened, thus meeting the enterprise's needs for refined and intelligent project management.

[0073] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

[0074] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent decision support throughout the entire project initiation process, characterized in that, Includes the following steps: The system synchronizes basic data for project maintenance, contract management, and supplier management in real time via API interface. After preprocessing the basic data, it forms a basic database for project invoicing. The system then uses rule engine technology to automatically verify the basic data. By automatically retrieving project information and generating budget suggestions based on machine learning, the system can automatically fill in and intelligently verify expense requisition forms, and build an expense requisition database. A tiered intelligent decision-making model for review is built using the billing application database. This model analyzes project activity at different review stages and outputs review decisions for billing applications based on historical billing application data. Based on the database of invoice applications and the linkage between the review and decision-making institutions, loan management, and project responsibility budget control, a post-invoice process progress monitoring dashboard is built to monitor the project progress status and issue early warnings. By collecting data from the entire process, intelligent analysis and iterative model optimization are performed to form a closed-loop feedback.

2. The intelligent decision-making support method for the entire project invoicing process according to claim 1, characterized in that, By automatically retrieving project information and generating budget suggestions based on machine learning, the system automatically fills in and intelligently proofreads expense requisition forms, and builds an expense requisition database. The specific steps are as follows: The billing application form is automatically populated based on the project billing database. The first clause data of the project contract is extracted through OCR. The data of the first clause data is compared with the billing application form to obtain the information difference coefficient. Based on the information difference coefficient, the difference prompt is triggered to guide the correction of the billing application form and is synchronously stored in the billing application database.

3. The intelligent decision-making support method for the entire project invoicing process according to claim 2, characterized in that, The data in the first clause is compared with the billing request form to obtain an information difference coefficient. Based on the information difference coefficient, a difference warning is triggered to guide the correction of the billing request form. The specific steps are as follows: The fixed field data corresponding to the project to be billed and the project billing basic database is automatically retrieved through the data association mapping algorithm, and the extracted fixed field data is filled into the billing application form. By combining machine learning regression models with historical cost structure data of similar projects in the project budget database, current project contract amount and construction period, an initial budget suggestion is intelligently generated, including the planned cost and target cost of various funding categories. The first clause data of the project contract is extracted by OCR. The data of the first clause data is compared with the fixed field data in the billing application form to obtain the information difference coefficient. The difference prompt is triggered based on the information difference coefficient.

4. The intelligent decision-making support method for the entire project invoicing process according to claim 3, characterized in that, The specific steps for triggering a difference prompt based on the information difference coefficient assessment are as follows: The information difference coefficient is compared with the preset difference threshold. If the information difference coefficient is greater than or equal to the preset difference threshold, it indicates that there is a content deviation between the billing application and the project contract, and the difference prompt is automatically triggered to guide the applicant to check and correct the billing application. If the information difference coefficient is less than the preset difference threshold, it indicates that there is no content deviation between the billing application and the project contract.

5. The intelligent decision-making support method for the entire project invoicing process according to claim 1, characterized in that, A tiered intelligent decision-making model for billing application review is built using a database of billing applications. This model analyzes project activity at different review stages and outputs review decisions for billing applications based on historical billing application data. The specific steps are as follows: Extract historical application data that has been approved from the account opening application database; Based on historical application data, behavioral indicators of similar projects are obtained as the first behavioral indicator, and behavioral indicators of projects applying for account opening are obtained as the second behavioral indicator. Identify the current review stage based on the approval process nodes, and construct an activity analysis model using first-behavior indicators and second-behavior indicators to analyze project activity at different review stages. By combining project activity at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. If the review is passed, the billing application is automatically pushed to the next stage. If the review fails, select the reason for return. Based on the reason for return, an automatic list of modification suggestions will be generated, which includes the types of supplementary materials and the direction of data correction, and will be pushed to the applicant.

6. The intelligent decision-making support method for the entire project invoicing process according to claim 5, characterized in that, Behavioral metrics include the number of times data was modified, the average time for modification review, and the number of times cross-departmental collaboration was conducted.

7. The intelligent decision-making support method for the entire project invoicing process according to claim 5, characterized in that, By combining project activity at different review stages with historical billing application data, a review decision is made to determine whether the current project should pass the review. The specific steps are as follows: Based on historical application data, obtain a checklist of key points for reviewing account opening applications, historical data comparison results, and financial analysis results; A comprehensive review scoring model is constructed by combining the review checklist, historical data comparison results, and financial analysis results with project activity to obtain a review score, and the review pass rate is determined based on the review score.

8. The intelligent decision-making support method for the entire project invoicing process according to claim 7, characterized in that, A comprehensive review scoring model is constructed by combining a checklist of key review points, historical data comparison results, and financial analysis results with project activity to obtain the review score. The comprehensive review scoring model is as follows: At each audit stage, a basic audit score is calculated based on the multiple data points in the audit checklist, according to their importance weights; the score is then adjusted by comparing the average risk indicators with historical data and using weighting coefficients. Based on the funding matching results, adjustments are made again using weighting coefficients.

9. The intelligent decision-making support method for the entire project invoicing process according to claim 7, characterized in that, The approval or rejection of an audit is determined based on the audit score. Specifically, the audit score is compared with a preset score range. If the audit score is higher than the upper limit of the preset score range, the current project is marked as "audit approved". If the audit score is within the preset score range, the current project is marked as "materials required". If the audit score is lower than the lower limit of the preset score range, the current project is marked as "audit rejected and returned".

10. An intelligent decision support system for the entire project invoicing process, applied to the intelligent decision support method for the entire project invoicing process as described in any one of claims 1-9, characterized in that, It includes a basic data processing and verification module, an intelligent verification module for account opening applications, an intelligent hierarchical review and decision support module, a monitoring dashboard linkage module, and an iterative optimization module; The basic data processing and verification module is used to synchronize basic data for project maintenance, contract management and supplier management in real time through API interface. After preprocessing the basic data, it forms a basic database for project invoicing and uses rule engine technology to automatically verify the basic data. The billing application intelligent verification module is used to automatically fill in and intelligently verify billing application forms by automatically retrieving project information and generating budget suggestions based on machine learning, and to build a billing application database. The intelligent hierarchical review decision support module is used to build a hierarchical review intelligent decision model through the billing application database, analyze the activity of projects at different review stages, and output the review decision of billing application forms in combination with historical billing application data. The monitoring dashboard linkage module is used to build a post-opening process progress monitoring dashboard based on the opening application database and the linkage of review decision-making institutions, loan management, and project responsibility budget control to monitor the project progress status and issue early warnings. The iterative optimization module is used to perform intelligent analysis and iterative optimization of the model by collecting data from the entire process, forming a closed-loop feedback.

Images