Maintenance project cost management and control method and system based on dynamic generation and decision tree model

Abstract

The invention provides a maintenance project cost management and control method and system based on dynamic generation and a decision tree model, and belongs to the technical field of project management and informationization, and the method comprises the steps: obtaining project quota data, and obtaining a comprehensive unit price through employing a comprehensive unit price dynamic generation algorithm; carrying out dynamic budgeting on the current maintenance item by adopting a cosine similarity algorithm to obtain a unit price corresponding to historical data with the highest similarity, and judging whether the unit price is an abnormal unit price or not; if no abnormity exists, the unit price obtained through calculation is adopted to calculate the maintenance project cost; and if the unit price is abnormal, analyzing the abnormal unit price data by adopting a gradient boosting decision tree model, and accounting the maintenance project cost according to a unit price analysis result. According to the invention, through collection and analysis of historical maintenance project data, dynamic generation and an intelligent matching algorithm are combined, a maintenance plan and budget are optimized, resource waste is reduced, it is ensured that the cost of the maintenance project is controlled within a reasonable range, and unnecessary expenditure is reduced.

Description

Technical Field

[0001] This invention belongs to the field of engineering management and information technology, and in particular relates to a method and system for cost control of maintenance projects based on dynamic generation and decision tree model. Background Technology

[0002] Traditional maintenance management suffers from the following problems: 1. Information silos: Data is scattered across various stages, lacking a unified platform for integration, resulting in low information transmission efficiency and low data utilization. 2. Manual dependence: Budget preparation, approval processes, and visa management rely on manual operation, leading to low efficiency and a high risk of errors. 3. Lack of intelligent support: The lack of intelligent matching, data analysis, and historical data accumulation functions makes it difficult to achieve efficient budget preparation and cost control. 4. Insufficient management transparency: The lack of transparent management in approval processes, construction processes, and final acceptance makes it difficult to monitor project progress in real time. 5. Low data utilization: Historical maintenance data is not effectively utilized, making it difficult to support future maintenance plans and budget preparation. 6. Existing technologies lack dynamic generation mechanisms, failing to respond to market cost fluctuations in real time; furthermore, the lack of machine learning models such as gradient boosting decision trees makes it difficult to achieve intelligent anomaly detection and cost prediction.

[0003] Most existing technologies remain at the level of traditional information management modules, lacking intelligent tools and automated processes, such as intelligent matching, historical data analysis, and automatic recommendation functions. These technologies typically focus only on a single aspect of maintenance management (such as budget preparation or construction process management), failing to achieve closed-loop management of the entire maintenance project lifecycle. The problems of information fragmentation and decentralized management persist, hindering systematic and transparent management. Historical maintenance data is not fully utilized, lacking data accumulation and analysis capabilities. Data visualization functions are lacking, making it difficult to intuitively display key indicators such as the overall status of maintenance projects, project unit analysis, and construction unit analysis. Users struggle to quickly grasp the progress and key data of maintenance projects. To address these issues, an intelligent and systematic maintenance management system is urgently needed to improve the efficiency, accuracy, and transparency of maintenance management.

[0004] In traditional engineering maintenance management models, budget preparation and cost control have long faced challenges in efficiency and accuracy. Due to a lack of intelligent technology support, existing systems generally only enable basic data storage and simple queries, making it difficult to deeply mine and transform the value of massive amounts of historical maintenance project data. For example, in the budget preparation stage, manually matching historical unit prices is time-consuming, labor-intensive, and prone to errors, failing to dynamically reflect market cost fluctuations. In the cost control stage, the lack of data accumulation and intelligent analysis makes it difficult to identify cost anomalies or predict potential risks. This "data silo" state severely restricts the level of precision in maintenance management, urgently requiring intelligent means to achieve data-driven decision optimization. Summary of the Invention

[0005] The purpose of this invention is to address the problems of insufficient intelligent matching, data analysis, and historical data accumulation, which make it difficult to achieve efficient planning and budgeting and cost control. It provides a maintenance project cost control method based on dynamic generation and a decision tree model. Through a data acquisition module, a dynamic budget generation module, and a unit price analysis module, it achieves cost control of maintenance projects. By collecting and analyzing historical maintenance project data, combined with dynamic generation and intelligent matching algorithms, it optimizes maintenance plans and budgets, reduces resource waste, ensures that maintenance project costs are controlled within a reasonable range, and reduces unnecessary expenditures.

[0006] According to one aspect of this specification, a method for cost control of maintenance projects based on dynamic generation and decision tree models is provided, including:

[0007] The engineering quota data is obtained, and the comprehensive unit price is obtained by using a comprehensive unit price dynamic generation algorithm. The comprehensive unit price dynamic generation algorithm includes weighted calculation of historical unit price data, correction of the weighted calculation result according to the real-time inflation rate, and output of comprehensive unit price.

[0008] Based on the comprehensive unit price, the cosine similarity algorithm is used to dynamically budget the current maintenance project, obtain the unit price corresponding to the historical data with the highest similarity, and determine whether it is an abnormal unit price; if there is no abnormality, the calculated unit price is used to calculate the cost of the maintenance project.

[0009] If any anomalies are found, a gradient boosting decision tree model is used to analyze the abnormal unit price data, and the maintenance project cost is calculated based on the unit price analysis results. The analysis of abnormal unit price data using the gradient boosting decision tree model includes: loading a pre-trained gradient boosting decision tree model, preprocessing and extracting features from the current abnormal unit price data, inputting the extracted feature data into the pre-trained gradient boosting decision tree model, analyzing the importance of features and calculating weighted values ​​layer by layer through the decision tree ensemble structure generated by multiple iterations, and finally outputting the unit price analysis results.

[0010] Furthermore, the historical unit price data is weighted and calculated, including: retrieving similar historical unit price data from the past three years based on the category of the project to be repaired; and weighting the historical unit price data according to a set time weight coefficient, where the more recent the data, the higher the weight.

[0011] Furthermore, a cosine similarity algorithm is used to dynamically budget the current maintenance project, including:

[0012] Vectorize the current project and the projects in the historical database;

[0013] Calculate the cosine similarity between the current item vector and the corresponding item vector in the historical database;

[0014] Return the unit price corresponding to the historical item with the highest similarity.

[0015] Furthermore, after analyzing the abnormal unit price data, the following was included:

[0016] When the threshold of the analysis result is less than 0.3, the unit price data is normal and will be automatically approved;

[0017] When the threshold of the analysis results is greater than or equal to 0.3 or less than 0.7, the unit price data is then manually reviewed.

[0018] When the threshold of the analysis results is greater than or equal to 0.7, the unit price data is abnormal, triggering an alert.

[0019] Furthermore, the method also includes a visual dashboard for displaying the overall status of maintenance projects.

[0020] According to one aspect of this specification, a maintenance project cost control system based on dynamic generation and decision tree model is provided, comprising:

[0021] The system comprises a data acquisition module, a dynamic budget generation module, and a unit price analysis module. The data acquisition module is used to acquire engineering quota data and calculate the comprehensive unit price. The dynamic budget generation module, in conjunction with the comprehensive unit price, dynamically generates the budget for the current maintenance project. The unit price analysis module is used to analyze and process abnormal data generated in the dynamic budget generation process.

[0022] The data acquisition module includes a quota database access unit and a historical project database unit; the quota database access unit is used to access industry standards; the historical project database unit uses a comprehensive unit price dynamic generation algorithm to generate comprehensive unit prices.

[0023] The dynamic budget generation module includes a historical matching unit and a price comparison report generation unit; the historical matching unit uses a cosine similarity algorithm to perform historical matching and returns the unit price corresponding to the historical data with the highest similarity; the price comparison report generation unit is used to mark abnormal data that deviates from the set price range of the historical average price.

[0024] The unit price analysis module includes a gradient boosting decision tree unit and an analysis unit; the gradient boosting decision tree unit is used to load a pre-trained model to analyze abnormal data; the analysis unit performs corresponding processing based on the analysis results.

[0025] Furthermore, the system also includes:

[0026] The environment support module is used to build the basic operating environment and implement security operations.

[0027] The analysis and decision-making module is used to provide data-driven decision support.

[0028] Furthermore, the data acquisition module also includes a visa unit price database unit, which is used to query project unit prices and supports multi-dimensional queries based on project equipment type and maintenance level.

[0029] Furthermore, the dynamic budget generation module also includes a dual-mode input unit for verifying the integrity of maintenance project data and intercepting records with missing key fields.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] 1. This invention enables the control of maintenance costs for maintenance projects through a data acquisition module, a dynamic budget generation module, and a unit price analysis module.

[0032] 2. This invention optimizes maintenance plans and budgets by collecting and analyzing historical maintenance project data, combined with dynamic generation and intelligent matching algorithms, thereby reducing resource waste, ensuring that maintenance project costs are controlled within a reasonable range, and reducing unnecessary expenditures.

[0033] 3. This invention reduces manual operations and significantly lowers the labor costs of maintenance management through intelligent tools and standardized processes. The intelligentization of budget preparation and visa management reduces human error and repetitive work, thereby improving work efficiency. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 A diagram illustrating a maintenance project cost control method based on dynamic generation and a decision tree model, provided in an embodiment of the present invention.

[0036] Figure 2 This is a schematic diagram of an intelligent maintenance system architecture provided in an embodiment of the present invention.

[0037] Figures 3(a)-(b) are schematic diagrams of the visualization dashboard provided in the embodiments of the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] like Figure 1 As shown, this embodiment of the invention provides a method for cost control of maintenance projects based on dynamic generation and decision tree models, including: acquiring engineering quota data, obtaining a comprehensive unit price using a dynamic comprehensive unit price generation algorithm, wherein the dynamic comprehensive unit price generation algorithm includes weighted calculation of historical unit price data, correcting the weighted calculation result according to the real-time inflation rate, and outputting a comprehensive unit price; based on the comprehensive unit price, using a cosine similarity algorithm to dynamically budget the current maintenance project, obtaining the unit price corresponding to the historical data with the highest similarity, and determining whether it is an abnormal unit price; if there is no abnormality, the cost of the maintenance project is calculated using the calculated unit price; if there is an abnormality, a gradient boosting decision tree model is used to analyze the abnormal unit price data, and the cost of the maintenance project is calculated based on the unit price analysis result; wherein, the analysis of abnormal unit price data using a gradient boosting decision tree model includes: loading a pre-trained gradient boosting decision tree model, preprocessing and extracting features from the current abnormal unit price data, inputting the extracted feature data into the pre-trained gradient boosting decision tree model, analyzing the importance of features and calculating weighted values ​​layer by layer through a decision tree ensemble structure generated by multiple iterations, and finally outputting the unit price analysis result.

[0040] Specifically, the data acquisition module includes: access to the quota database based on industry standards (such as the Ministry of Housing and Urban-Rural Development's engineering quota database), supporting online pre-settlement and final settlement preparation. The historical project database uses a dynamic algorithm for generating comprehensive unit prices, with the specific steps as follows:

[0041] 1. Query historical data: Based on the current project category, retrieve historical project data of the same type over the past three years.

[0042] 2. Weighted average calculation: The historical unit price is weighted by time weight (the weights for the past three years are 0.3, 0.5, and 0.2 respectively), which reflects the dynamic adjustment logic of "more recent data has higher weight".

[0043] 3. Inflation Adjustment: The weighted average unit price is adjusted based on the current inflation rate to ensure that the composite unit price reflects the latest market costs. The final output is the dynamically adjusted composite unit price, which is used for budget preparation and cost control.

[0044] The pseudocode implementation is as follows:

[0045] def update_unit_price(project):

[0046] historical_data = query_similar_projects(project.category)

[0047] avg_price = calculate_weighted_avg(historical_data, weights=[0.3,0.5, 0.2]) # Weights increase over the last 3 years

[0048] return adjust_for_inflation(avg_price).

[0049] The visa unit price database enables intelligent querying and supports multi-dimensional filtering by equipment type and maintenance level.

[0050] Specifically, the dynamic budget generation module includes: dual-mode input; automatic data integrity verification during Excel import, rejecting records lacking "equipment code" or "quantity"; and historical matching using a cosine similarity algorithm, as follows:

[0051] Use TfidfVectorizer to vectorize the descriptions of the current item and the historical database;

[0052] Use cosine_similarity to calculate the cosine similarity between the current item vector and each description vector in the historical database;

[0053] Return the unit price corresponding to the historical record with the highest similarity.

[0054] The pseudocode is as follows:

[0055] def similarity_match(current_item, history_db):

[0056] vectorizer = TfidfVectorizer()

[0057] vectors = vectorizer.fit_transform([current_item] + history_db.descriptions)

[0058] cos_sim = cosine_similarity(vectors[0:1], vectors[1:])

[0059] return history_db.iloc[np.argmax(cos_sim)]['unit_price'].

[0060] When generating a price comparison report, mark any outliers that deviate from the historical average price by ±10% and send out an alert.

[0061] Specifically, the unit price analysis module includes:

[0062] Model prediction: Load a pre-trained Gradient Boosting Decision Tree (GBDT) model and analyze the feature data (between 0 and 1) of visa matters.

[0063] Classification: Analysis results with a threshold below 0.3 are considered normal unit price data; analysis results with a threshold between 0.3 and 0.7 require manual judgment; analysis results with a threshold above 0.7 are considered abnormal unit price data.

[0064] Automatic triage: Normal unit price data is automatically passed; intermediate thresholds are handed over to manual judgment; abnormal unit price data is frozen and an early warning notification is triggered.

[0065] Specifically, a pre-trained gradient boosting decision tree model is loaded, and feature extraction and standardization (preprocessing) are performed on the current abnormal data. The processed feature data is then input into the pre-trained gradient boosting decision tree model. Through the decision tree ensemble structure generated by multiple iterations, the importance of features is analyzed layer by layer and weighted values ​​are calculated. Finally, the analysis results are output.

[0066] The pseudocode for the Gradient Boosting Decision Tree (GBDT) model is as follows:

[0067] def visa_approval(visa_item):

[0068] gbdt_model = load_model('visa_gbdt.mdl')

[0069] risk_score = gbdt_model.predict(visa_item.features)

[0070] Based on the results after classification, the pseudocode is as follows:

[0071] if risk_score < 0.3: # Automatic approval

[0072] return auto_approve(visa_item)

[0073] elif 0.3 <= risk_score < 0.7: # Transfer to manual assistance

[0074] return route_to_manager(visa_item)

[0075] else: # Freeze and issue a warning

[0076] return freeze_and_alert(visa_item).

[0077] Specifically, this invention also provides Embodiment 1, which integrates aspects such as planning and budgeting, project initiation, contract management, construction process monitoring, final acceptance, and settlement approval. The specific implementation method is as follows:

[0078] 1. System Architecture Design: Design an information platform based on a B / S architecture, supporting multi-role user login and access control. Integrate functional modules such as a basic database, maintenance plan and budget module, project initiation management module, contract management module, construction process management module, completion acceptance module, and settlement approval module.

[0079] 2. Basic Database Construction: Configure a complete quota library, supporting online preparation and review of pre-settlement and final settlement. Accumulate historical maintenance project data and dynamically form a comprehensive unit price library for maintenance projects.

[0080] 3. Intelligent Budgeting for Maintenance Plans: Offers multiple functions such as Excel import, intelligent matching, and historical project settlement, supporting efficient budgeting. It provides an intelligent matching function that matches current work order data from the historical maintenance project database to calculate the highest, lowest, and average prices.

[0081] 4. Approval Process and Access Control: A multi-level approval process is designed, supporting functions such as referral and return. Strict access control is implemented to ensure that sensitive data can only be viewed by authorized personnel.

[0082] 5. Construction Process Management: Provides a construction log entry function, supporting both PC and WeChat mini-program platforms. Offers a function to check and rectify issues related to the construction log to ensure construction quality.

[0083] 6. Completion Acceptance and Settlement Management: Provides completion acceptance approval function to ensure construction quality meets requirements. Provides maintenance visa approval function, supporting intelligent matching and review of visa data. Provides intelligent maintenance settlement approval function, recording the amount changes at each approval stage.

[0084] 7. Data Visualization and Reporting: Provides visual dashboards to display the overall status of maintenance projects. Offers various reporting functions to support data statistics and analysis.

[0085] Specifically, the basic database construction includes: a quota database, a fully configured quota database supporting online preparation and review of pre-settlement and final settlement; a historical maintenance project database, which accumulates historical maintenance project data and dynamically forms a comprehensive unit price database for maintenance projects; and a comprehensive unit price database for maintenance change orders, providing intelligent query and matching functions for comprehensive unit prices of change orders.

[0086] Specifically, the intelligent maintenance plan budget preparation supports both offline preparation and intelligent matching methods. It provides intelligent matching functionality for historical maintenance project data, calculates the highest, lowest, and average prices, and recommends the optimal comprehensive unit price. It supports multiple functions such as Excel import, intelligent matching, and historical project settlement, improving the efficiency and accuracy of budget preparation.

[0087] Specifically, regarding approval processes and access control, a standardized approval process is designed to support multi-level approval, referral, and return functions. Strict access control is implemented to ensure that sensitive data can only be viewed by authorized personnel.

[0088] Specifically, the construction process management includes a maintenance and construction log entry function, supporting both PC and WeChat mini-program platforms to record construction progress in real time. It also provides inspection and rectification functions for the construction logs to ensure construction quality. Furthermore, it supports construction management functions such as commencement reports, hazardous operation factor notification forms, and project lists.

[0089] Specifically, the project completion acceptance and settlement management system provides completion acceptance approval functionality to ensure construction quality meets requirements. It also provides maintenance visa approval functionality, supporting intelligent matching and review of visa data. Furthermore, it offers intelligent maintenance settlement approval functionality, recording amount changes at each approval stage and supporting visa comparison and settlement detail analysis.

[0090] Specifically, the present invention also provides a second embodiment, a smart maintenance system, such as... Figure 2 As shown, the system includes: an environment support module for building the basic operating environment and implementing security measures; a basic data module for collecting, storing, and updating maintenance management data; a business module for realizing a closed-loop management system for the entire intelligent maintenance process; and an analysis and decision-making module for providing data-driven decision support. The intelligent maintenance system proposed in this invention provides standardized data interfaces, supporting data integration and sharing with other systems.

[0091] Specifically, the environment support module is used to build the basic operating environment and implement security operations, including:

[0092] The data server centrally stores all core structured data, including but not limited to user accounts, business documents, multi-level approval process statuses, detailed permission configuration tables, and data synchronized from third-party systems such as ERP through standardized interfaces. It supports high-concurrency front-end (PC / mini-program) access through high-performance database services; its rigorous data table structure design forms the basis for row-level / column-level secure isolation of multi-level approval process status transitions and sensitive data access permissions at the database module level.

[0093] The application service library encapsulates all reusable business logic and technical services. It provides a unified Web API interface for both PC browsers and WeChat mini-programs, achieving "one backend, multi-platform adaptation." It includes an API gateway and protocol adaptation components, crucial for implementing standardized interfaces and unified management of internal and external requests. An integrated identity authentication and access control engine provides unified identity verification and fine-grained access control for all requests. A workflow engine is provided to drive and configure the business logic of multi-level approval processes.

[0094] The data backup server is dedicated to executing regular, automated data backup plans. It implements encrypted backup and off-site storage strategies for sensitive data, ensuring that even in extreme circumstances, data security audit requirements are met and rapid recovery is achieved. This is an important supplement to the "sensitive data access permission isolation" strategy in the data persistence module.

[0095] The network environment is designed to build and manage channels for all data flow. By configuring load balancing, HTTPS / WSS encryption, and other features, it provides secure and efficient network access capabilities for B / S architectures and WeChat mini-programs. Using firewalls, virtual private networks (VPNs), and network isolation policies, security domains are defined at the network module level. For example, access to internal data servers and application service libraries is isolated from areas providing external services, and dedicated communication links with third-party systems such as ERP are strictly controlled. This is the first line of defense for achieving overall system security and preventing unauthorized access.

[0096] Specifically, the basic data module is used to store maintenance management data assets, including:

[0097] The Steel Maintenance Engineering Quota Library stores and manages standard quota data for various tasks in steel industry maintenance projects, including standard values ​​for labor, materials, and machinery consumption. This quota data is a crucial basis for preparing maintenance project budgets, controlling costs, and settling accounts.

[0098] The maintenance project settlement database records and stores actual settlement data for maintenance projects, including project name, settlement amount, settlement time, and settlement basis. This database helps companies track the financial status of maintenance projects, conduct cost analysis and benefit assessments, and provides experience for subsequent maintenance projects.

[0099] The Railway Maintenance Engineering Quota Database is specifically designed for maintenance engineering in the railway transportation industry, storing and managing standard quota data for various maintenance tasks. This data helps railway transportation companies accurately prepare maintenance budgets, rationally allocate maintenance resources, and ensure the safety and smooth operation of railway transportation.

[0100] The maintenance work quantity comprehensive unit price database is used to store and manage comprehensive unit price data for various tasks in maintenance projects. This data typically includes all relevant costs such as labor, materials, and machinery. This database helps companies quickly and accurately calculate the total cost of maintenance projects, providing strong support for project bidding and contract signing.

[0101] The Metallurgical Construction Engineering Quota Database stores and manages standard quota data for various construction projects in the metallurgical industry. This data covers all stages of the metallurgical construction process, such as mining, smelting, and rolling, providing crucial information for metallurgical enterprises to prepare project budgets, control costs, and settle accounts.

[0102] The overhaul and maintenance project database stores and manages information on a company's overhaul and maintenance projects, including project name, project type, project scale, project budget, and project schedule. This database helps companies gain a comprehensive understanding of the status of their overhaul and maintenance projects, allocate project resources rationally, and ensure the smooth progress and timely completion of projects.

[0103] The enterprise supplementary quota database is used to store and manage supplementary quota data developed by enterprises based on their own actual conditions and specific needs. This data may involve the use of specific processes, special materials, or special equipment, helping enterprises to more accurately prepare project budgets and control costs, thereby improving their competitiveness and profitability.

[0104] Organizational structure is used to store and manage an enterprise's organizational information, including department setup, job responsibilities, and staffing. This system helps enterprises optimize their organizational structure, clarify the responsibilities and authority of each department and position, and improve management efficiency and operational effectiveness.

[0105] Specifically, the business module is used to achieve a closed-loop management of the entire maintenance process, including:

[0106] The intelligent editing and review of maintenance plans and budgets enables the automated generation and compliance pre-review of plans and budgets. Based on historical equipment data, maintenance procedures, and quota standards, it automatically recommends maintenance projects, estimates resources and costs, and conducts multi-level online editing and review through a built-in rule engine to form scientific and controllable annual or special maintenance plans and budgets.

[0107] The system assists in the creation of project dashboards, establishing a global visual monitoring and decision-making center for projects. After the plan is approved, projects are formally created, and key indicators, stage results, and resource loads of all projects in progress are centrally displayed on graphical dashboards, providing managers with real-time and intuitive data views for overall decision-making and priority scheduling.

[0108] Receive project initiation contract information, establish the legal and commercial benchmark files for the project, and systematically receive, parse and structure the approved maintenance contracts or task books to solidify the core terms such as scope of work, schedule, cost, quality and safety, which serve as the sole legal basis for all subsequent execution, acceptance and settlement.

[0109] Maintenance project work orders enable the professional decomposition and instruction-based issuance of work tasks. They are used to break down macro-level projects into specific, executable "work orders" by profession, region, or system, clearly defining the work content, standards, planned working hours, safety requirements, and the executing team. They are the sole authoritative instruction to begin on-site operations.

[0110] The maintenance commencement report approval process is used for the final condition verification and safety authorization before construction. After the work order is issued and before physical construction begins, the construction unit submits an online commencement application, which includes the construction plan, personnel and equipment preparation, safety technical briefing, and emergency plan. The application can only commence after the supervisor and the owner approve it online at each level, ensuring that all preconditions for construction are in place.

[0111] The maintenance work ticket log submission and rectification record enable refined, traceable, real-time recording and problem closure of the construction process. It is used by the work teams to fill in the actual progress, working hours, resource consumption and problems encountered online every day. At the same time, the quality inspection process can record the non-conformities found and track their rectification process until closure, forming a complete and tamper-proof digital construction log and quality archive.

[0112] The system enables precise and compliant control over the requisition of maintenance materials, linking them to costs. It facilitates online application and approval for materials, spare parts, and tools required for maintenance. The system automatically verifies the rationality of requisitions based on work orders and quota databases, and links them to project costs in real time, achieving refined management of material requisition and proactive cost control.

[0113] The maintenance visa approval process standardizes the management of work outside the contract scope, design changes, and cost adjustments. When work outside the contract scope occurs on-site or the original plan needs to be changed, a visa application can be initiated online, including the reason, plan, workload, and cost estimate. This drives a standardized multi-level approval process, and the approval result serves as a legal attachment to the contract price adjustment.

[0114] The intelligent approval system for maintenance settlement enables automated verification, early warning, and efficient approval of project settlements. After project completion, the system automatically aggregates all cost data, including approved work orders, visas, and material requisition forms, compares them with the contract and budget, automatically reviews compliance and reasonableness, and drives a pre-set, role-based, standardized online approval process, significantly improving settlement efficiency and accuracy.

[0115] Specifically, the analysis and decision-making module is used to provide data-driven decision support, including:

[0116] The comprehensive maintenance and management dashboard provides a holistic strategic view and health monitoring of enterprise-level maintenance operations. It aggregates and visualizes the core KPIs of all maintenance projects across the company in real time, serving as a one-stop command center for senior managers to make strategic decisions, coordinate resources, and evaluate performance.

[0117] The comprehensive maintenance cost dashboard enables real-time, transparent monitoring and analysis of maintenance costs across the company. It dynamically tracks and displays the budget execution status, cost composition, overspending warnings, and cost trends of all maintenance projects, making it an important tool for the finance and planning departments to control costs and optimize budgets.

[0118] Each production plant's maintenance dashboard focuses on the maintenance execution performance and responsibility management of each production unit (plant / workshop). It is used to display indicators such as the status of maintenance projects, plan fulfillment rate, downtime, and maintenance cost input-output ratio under each production plant, and supports horizontal comparison. It is the core workbench for production plant managers to manage their unit's equipment assets and implement maintenance responsibilities.

[0119] The construction unit analysis dashboard establishes a quantitative evaluation and refined management system for cooperating construction units. It is used to conduct performance statistics, ranking, and comparative analysis of each cooperating construction unit from multiple dimensions such as safety, quality, schedule, and cost, providing data basis for optimizing contractor selection and implementing contract rewards and penalties.

[0120] The specific maintenance project dashboard enables transparent management of individual maintenance projects at the micro, all-element, and full-process level. It is used to focus on displaying detailed plans, Gantt charts, real-time progress, resource input, cost consumption, pending approvals, issue lists, and documents for a specific project. It serves as a "battle map" for project managers, supervisors, and owner representatives to track and collaborate on daily projects.

[0121] Specifically, as shown in Figure 3, the data visualization and reporting features include visual dashboards that display the overall status of maintenance projects, project unit analysis, and construction unit analysis. It provides various reporting functions, including individual outsourced maintenance contract ledgers, settlement approval ledgers, and construction log reports, supporting data statistics and analysis.

[0122] Specifically, the environment support module construction includes: deploying a B / S architecture platform, building backend services using the Spring Boot framework, and building the frontend interface using Vue.js. It also involves developing a WeChat mini-program client, achieving real-time data synchronization with the PC client via the WebSocket protocol. Finally, it configures an OAuth 2.0 access control module, isolating data access permissions by role (e.g., project managers can view all data, while construction workers are limited to the construction log module).

[0123] Specifically, this invention proposes an intelligent budget preparation and matching technology. Through intelligent matching and historical data analysis, it achieves intelligent preparation of maintenance plan budgets, significantly improving the efficiency and accuracy of budget preparation. The system can intelligently match current work order data from the historical maintenance project database, calculate the highest price, lowest price, and average price, and recommend the optimal comprehensive unit price, solving the problems of reliance on manual experience and low efficiency in traditional budget preparation. A closed-loop feedback mechanism between the historical project database and the settlement module ensures real-time synchronization of construction monitoring data across both PC and mobile terminals.

[0124] Specifically, this invention also provides a method for cost control of maintenance projects based on dynamic generation and decision tree models. The method involves acquiring engineering quota data, using a dynamic comprehensive unit price generation algorithm to obtain a comprehensive unit price, which includes weighted calculation of historical unit price data, correcting the weighted calculation result based on the real-time inflation rate, and outputting the comprehensive unit price. Based on the comprehensive unit price, a cosine similarity algorithm is used to dynamically budget the current maintenance project, obtaining the unit price corresponding to the historical data with the highest similarity, and determining whether it is an abnormal unit price. If there is no abnormality, the calculated unit price is used to calculate the maintenance project cost. If there is an abnormality, a gradient boosting decision tree model is used to analyze the abnormal unit price data, and the maintenance project cost is calculated based on the unit price analysis results. The analysis of abnormal unit price data using the gradient boosting decision tree model includes: loading a pre-trained gradient boosting decision tree model, preprocessing and extracting features from the current abnormal unit price data, inputting the extracted feature data into the pre-trained gradient boosting decision tree model, analyzing the importance of features and calculating weighted values ​​layer by layer through a multi-round iteratively generated decision tree ensemble structure, and finally outputting the unit price analysis result.

[0125] Finally, it should be noted that the above specific embodiments are merely representative examples of the present invention. Obviously, the present invention is not limited to the above specific embodiments and many variations are possible. Any simple modifications, equivalent changes, and alterations made to the above specific embodiments based on the technical essence of the present invention should be considered within the protection scope of the present invention.

Claims

1. A method for cost control of maintenance projects based on dynamic generation and decision tree models, characterized in that, include: The engineering quota data is obtained, and the comprehensive unit price is obtained by using a comprehensive unit price dynamic generation algorithm. The comprehensive unit price dynamic generation algorithm includes weighted calculation of historical unit price data, correction of the weighted calculation result according to the real-time inflation rate, and output of comprehensive unit price. Based on the comprehensive unit price, the cosine similarity algorithm is used to dynamically budget the current maintenance project, obtain the unit price corresponding to the historical data with the highest similarity, and determine whether it is an abnormal unit price. If there are no abnormalities, the cost of the maintenance project shall be calculated using the calculated unit price. If any anomalies are found, a gradient boosting decision tree model is used to analyze the abnormal unit price data, and the maintenance project cost is calculated based on the unit price analysis results. The analysis of abnormal unit price data using the gradient boosting decision tree model includes: loading a pre-trained gradient boosting decision tree model, preprocessing and extracting features from the current abnormal unit price data, inputting the extracted feature data into the pre-trained gradient boosting decision tree model, analyzing the importance of features and calculating weighted values ​​layer by layer through the decision tree ensemble structure generated by multiple iterations, and finally outputting the unit price analysis results.

2. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 1, characterized in that, The historical unit price data is weighted and calculated, including: retrieving similar historical unit price data from the past three years based on the category of the project to be repaired; and weighting the historical unit price data according to a set time weight coefficient, where the more recent the data, the higher the weight.

3. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 1, characterized in that, The cosine similarity algorithm is used to perform dynamic budgeting for the current maintenance project, including: Vectorize the current project and the projects in the historical database; Calculate the cosine similarity between the current item vector and the corresponding item vector in the historical database; Return the unit price corresponding to the historical item with the highest similarity.

4. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 1, characterized in that, After analyzing the abnormal unit price data, the following was included: When the threshold of the analysis result is less than 0.3, the unit price data is normal and will be automatically approved; When the threshold of the analysis results is greater than or equal to 0.3 or less than 0.7, the unit price data is then manually reviewed. When the threshold of the analysis results is greater than or equal to 0.7, the unit price data is abnormal, triggering an alert.

5. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 1, characterized in that, The method also includes a visual dashboard for displaying the overall status of maintenance projects.

6. A maintenance project cost control system based on dynamic generation and decision tree model, characterized in that, include: The system comprises a data acquisition module, a dynamic budget generation module, and a unit price analysis module. The data acquisition module is used to acquire engineering quota data and calculate the comprehensive unit price. The dynamic budget generation module, in conjunction with the comprehensive unit price, dynamically generates the budget for the current maintenance project. The unit price analysis module is used to analyze and process abnormal data generated in the dynamic budget generation process. The data acquisition module includes a quota database access unit and a historical project database unit; the quota database access unit is used to access industry standards; the historical project database unit uses a comprehensive unit price dynamic generation algorithm to generate comprehensive unit prices. The dynamic budget generation module includes a historical matching unit and a price comparison report generation unit; the historical matching unit uses a cosine similarity algorithm to perform historical matching and returns the unit price corresponding to the historical data with the highest similarity; the price comparison report generation unit is used to mark abnormal data that deviates from the set price range of the historical average price. The unit price analysis module includes a gradient boosting decision tree unit and an analysis unit; the gradient boosting decision tree unit is used to load a pre-trained model to analyze abnormal data; the analysis unit performs corresponding processing based on the analysis results.

7. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 6, characterized in that, The system also includes: The environment support module is used to build the basic operating environment and implement security operations. The analysis and decision-making module is used to provide data-driven decision support.

8. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 6, characterized in that, The data acquisition module also includes a visa unit price database unit, which is used to query project unit prices and supports multi-dimensional queries based on project equipment type and maintenance level.

9. The method for cost control of maintenance projects based on dynamic generation and decision tree model according to claim 6, characterized in that, The dynamic budget generation module also includes a dual-mode input unit, used to verify the integrity of maintenance project data and intercept records with missing key fields.

Images