Capital expenditure monitoring and prediction analysis system and method based on big data

Abstract

The invention discloses a capital expenditure monitoring and prediction analysis system and method based on big data, and belongs to the technical field of capital expenditure monitoring and prediction. The system comprises a multi-source data integration module used for extracting multi-source data features according to capital expenditure related data; the business scene feature extraction module is used for extracting dynamic business scene features according to the project basic features; extracting refined external environment features according to the external environment features; the self-adaptive prediction module is used for constructing a prediction model fusing multi-source data features, dynamic business scene features and refined external environment features, optimizing model parameters in real time according to business scene changes through a dynamic weight adjustment mechanism, and generating a multi-cycle capital expenditure prediction result; the abnormal root cause tracing module is used for positioning a core root cause; and the closed-loop iterative optimization module is used for realizing dynamic iterative optimization of the prediction model. According to the invention, the refinement level and decision-making efficiency of capital expenditure management can be improved.

Description

Technical Field

[0001] This invention relates to the field of capital expenditure monitoring and forecasting technology, and in particular to a big data-based capital expenditure monitoring and forecasting analysis system and method. Background Technology

[0002] In the past, corporate capital expenditures were like disassembled jigsaw puzzle pieces. Each department had its own Excel spreadsheet, with fragments scattered everywhere, making it difficult to piece together a complete picture. Each reporting process operated independently, with inconsistent formats, making horizontal comparisons only "visual." Every update relied on manual copying and pasting, which became visually overwhelming with many rows, and a single mistake would require starting the entire spreadsheet anew. Each expenditure was only visible after the month-end summary, and forecasting was often based on intuition. While existing technologies for forecasting corporate capital expenditures can utilize time-series models like ARIMA and SARIMA, or machine learning models such as logistic regression and decision trees, they still suffer from insufficient data fusion depth, remaining at a single-dimensional analysis of historical financial data and lacking accurate extraction and application of dynamic business scenario characteristics. Existing forecasting models have poor adaptability, often being static and fixed, leading to significant forecasting bias. Furthermore, existing forecasting models have weak anomaly handling capabilities, only able to identify abnormal expenditure phenomena but unable to trace the core root causes, thus failing to meet the needs of enterprises for refined capital expenditure management.

[0003] Therefore, there is an urgent need for a capital expenditure monitoring and predictive analysis system that can solve technical problems such as deep fusion of multi-source data and dynamic scene features, adaptive iterative optimization of prediction models, accurate tracing of the root causes of anomalies, and early warning of propagation risks. Summary of the Invention

[0004] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a capital expenditure monitoring and predictive analysis system and method based on big data, which can achieve deep fusion of multi-source data and dynamic scene features, adaptive iterative optimization of the predictive model, accurate tracing of the root causes of anomalies, and early warning of propagation risks.

[0005] One embodiment of the present invention provides a capital expenditure monitoring and predictive analysis system based on big data, comprising: a multi-source data integration module, used to acquire historical financial data, real-time execution data, and project management data related to capital expenditure, and after cleaning and standardizing them respectively, extract corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features, and basic project features; a business scenario feature extraction module, used to extract dynamic business scenario features based on the basic project features; the dynamic business scenario features include project stage, resource allocation, and dependencies; and an adaptive prediction module, used to construct a system that integrates the multi-source data features with the dynamic business scenario. The system includes a feature-based prediction model that uses a dynamic weight adjustment mechanism to optimize model parameters in real time based on changes in business scenarios, generating multi-period capital expenditure prediction results; an anomaly root cause tracing module that uses pre-set multi-dimensional association rules to construct a full-link structured association model of capital expenditure based on monitored abnormal expenditure data, automatically tracing the project links, cost types, approval nodes, and related influencing factors where anomalies occur, and locating the core root cause; and a closed-loop iterative optimization module that compares the prediction results with actual execution data in real time, calculates the deviation rate and analyzes the causes of deviation, generates model parameter adjustment instructions, and feeds them back to the adaptive prediction module to achieve dynamic iterative optimization of the prediction model.

[0006] The embodiments of the present invention achieve at least the following beneficial effects: The embodiments of the present invention solve the pain points of data fragmentation and static prediction in the prior art; by integrating multi-source data features with dynamic business scenario features to construct a prediction model, and combining it with a closed-loop iteration mechanism, the accuracy and adaptability of capital expenditure prediction are significantly improved; the accurate location of abnormal problems is achieved, providing technical support for capital expenditure management decisions and improving the refinement level and decision-making efficiency of capital expenditure management.

[0007] According to some embodiments of the present invention, the system further includes: a visualization interaction module for displaying prediction results, root cause analysis reports of anomalies and model optimization effects, and supporting user-defined core analysis dimensions and basic interactive operations; and a permission management module for configuring operation permissions, data access permissions and approval permissions for different users.

[0008] According to some embodiments of the present invention, the multi-source data integration module is further used to acquire external business environment data, perform cleaning and standardization processing, and then extract external environment features, which are included in the multi-source data feature category; the business scenario feature extraction module is used to extract refined external environment features based on the external environment features; the refined external environment features include market fluctuations, policy adjustments, and industry trends; the adaptive prediction module incorporates the refined external environment features into the fusion feature set of the prediction model, and the dynamic weight adjustment mechanism synchronously adjusts the weight of the refined external environment features in real time.

[0009] According to some embodiments of the present invention, the adaptive prediction module is equipped with a time series algorithm and a regression analysis algorithm, and an algorithm that is automatically matched and adapted based on the business scenario type; the dynamic weight adjustment mechanism is based on the frequency of scenario changes and the feature priority ranking results, and adjusts the weight ratio of multi-source data features, dynamic business scenario features, and refined external environment features (if present) in the prediction model in real time, and the weight adjustment magnitude is positively correlated with the magnitude of scenario changes and the feature contribution.

[0010] According to some embodiments of the present invention, the multi-dimensional association rules of the anomaly root cause tracing module include project stage-cost type association rules, approval node-expenditure compliance association rules, and resource allocation-expenditure efficiency association rules; the full-link structured association model includes project stage nodes, cost type nodes, approval nodes, and resource nodes, and the association edges between nodes are constructed based on the quantitative analysis of data association strength; root cause localization is achieved by traversing the node association paths, calculating the influence coefficient of each node, and selecting the node combination with the highest influence coefficient as the core root cause; wherein, influence coefficient = node association strength × path length weight.

[0011] According to some embodiments of the present invention, the closed-loop iterative optimization module includes: a deviation cause classification subunit, used to classify deviation causes into scenario feature missing type, algorithm adaptability type, data quality type, and business mutation type; for scenario feature missing type deviations, triggering the business scenario feature extraction module to supplement the corresponding dynamic business scenario features, and if they exist, supplementing and refining the external environment features; for algorithm adaptability type deviations, switching the algorithm of the adaptive prediction module; for data quality type deviations, starting the secondary data cleaning process of the multi-source data integration module; for business mutation type deviations, marking them as special scenario samples and updating the scenario adaptability rules of the prediction model.

[0012] According to some embodiments of the present invention, the adaptive prediction module supports user input of custom business assumption parameters, including resource allocation adjustment parameters, market environment change parameters, and project schedule adjustment parameters. The parameter input adopts the form of interval values ​​or discrete values. The prediction model simulates the capital expenditure prediction results under different assumption scenarios and generates a multi-scenario prediction comparison report. The report includes the expenditure amount range, deviation risk, and resource demand differences under each scenario.

[0013] According to some embodiments of the present invention, the anomaly root cause tracing module includes: an anomaly propagation path analysis unit, used to identify subsequent project links and related projects that may be affected by abnormal expenditures based on the capital expenditure full-link structured association model, combined with project dependencies and time series characteristics, calculate the propagation probability, wherein the propagation probability = node association strength × time decay coefficient, and generate an anomaly propagation risk heat map.

[0014] According to some embodiments of the present invention, the visualization interaction module supports drill-down data exploration, allowing users to drill down layer by layer from the summarized prediction results to the multi-source data feature details, dynamic business scenario feature details, and refined external environment feature details corresponding to a single project, a single cost type, and a single time period; it also supports custom visualization chart types, with chart data updated synchronously in real time.

[0015] One embodiment of the present invention provides a method for monitoring and predictive analysis of capital expenditures based on big data, comprising the following steps: S100, acquiring historical financial data, real-time execution data, and project management data related to capital expenditures, and after cleaning and standardizing them respectively, extracting corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features, and basic project features; S200, extracting dynamic business scenario features based on the basic project features; the dynamic business scenario features include project stage, resource allocation, and dependencies; S300, constructing a system that integrates the multi-source data features and the dynamic business scenario features. The prediction model uses a dynamic weight adjustment mechanism to optimize model parameters in real time according to changes in business scenarios, generating multi-cycle capital expenditure prediction results; S400, for monitored abnormal expenditure data, based on preset multi-dimensional association rules, a full-link structured association model of capital expenditure is constructed to automatically trace the project links, cost types, approval nodes and related influencing factors where the abnormality occurred, and locate the core root cause; S500, the prediction results are compared with the actual execution data in real time, the deviation rate is calculated and the cause of the deviation is analyzed, a model parameter adjustment instruction is generated and fed back to the adaptive prediction module to realize the dynamic iterative optimization of the prediction model.

[0016] The embodiments of the present invention achieve at least the following beneficial effects: The embodiments of the present invention solve the pain points of data fragmentation and static prediction in the prior art; by integrating multi-source data features with dynamic business scenario features to construct a prediction model, and combining it with a closed-loop iteration mechanism, the accuracy and adaptability of capital expenditure prediction are significantly improved; the accurate location of abnormal problems is achieved, providing technical support for capital expenditure management decisions and improving the refinement level and decision-making efficiency of capital expenditure management.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0019] Figure 1 This is a schematic block diagram of the system modules according to an embodiment of the present invention.

[0020] Figure 2 This is the project management interface according to an embodiment of the present invention.

[0021] Figure 3 This is the project budget management interface according to an embodiment of the present invention.

[0022] Figure 4 This is the capital expenditure monitoring interface according to an embodiment of the present invention.

[0023] Figure 5 This is a flowchart illustrating the method according to an embodiment of the present invention.

[0024] Figure label:

[0025] Multi-source data integration module 100, business scenario feature extraction module 200, adaptive prediction module 300, anomaly root cause tracing module 400, and closed-loop iterative optimization module 500. Detailed Implementation

[0026] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0027] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0028] Reference Figure 1 This invention proposes a capital expenditure monitoring and predictive analysis system based on big data, comprising:

[0029] The multi-source data integration module 100 is used to acquire historical financial data, real-time execution data and project management data related to capital expenditures, and after cleaning and standardization, extract the corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features and basic project features.

[0030] The business scenario feature extraction module 200 is used to extract dynamic business scenario features based on the basic features of the project; the dynamic business scenario features include project stage, resource configuration and dependency relationship.

[0031] The adaptive prediction module 300 is used to build a prediction model that integrates multi-source data features and dynamic business scenario features. The prediction model optimizes the model parameters in real time according to changes in business scenarios through a dynamic weight adjustment mechanism, and generates multi-cycle capital expenditure prediction results.

[0032] The Anomaly Root Cause Tracing Module 400 is used to construct a full-link structured correlation model of capital expenditure based on preset multi-dimensional correlation rules for monitored abnormal expenditure data. It automatically traces the project links, cost types, approval nodes and related influencing factors where the anomaly occurred, and locates the core root cause.

[0033] The closed-loop iterative optimization module 500 is used to compare the prediction results with the actual execution data in real time, calculate the deviation rate and analyze the causes of the deviation, generate model parameter adjustment instructions, and feed them back to the adaptive prediction module to realize the dynamic iterative optimization of the prediction model.

[0034] In some embodiments, the system of this invention further includes: a visualization and interaction module, used to display prediction results, root cause analysis reports, and model optimization effects, supporting user-defined analysis dimensions and interactive operations. Figure 4 As shown, the visualization interaction module of this embodiment of the invention can display the capital expenditure monitoring interface.

[0035] In some embodiments, the system of this invention further includes: a permission management module, used to configure the operation permissions, data access permissions and approval permissions of different users.

[0036] In some embodiments, the system of this invention further includes: an Excel-compatible interactive module, used to provide an Excel-like operation interface, simulating Excel's cell editing, batch selection, formula input / calculation logic, and supporting common operations such as data filtering, sorting, and merging cells, reducing the learning cost for financial personnel. This Excel-compatible interactive module supports importing .xls and .xlsx format files, automatically identifies the mapping relationship between table fields and the system data model, and verifies data format and integrity; it supports exporting budget tables, reports, project details, etc., from the system to Excel format, retaining the original data structure, calculation formulas, and format styles. This Excel-compatible interactive module also provides a bidirectional conversion function between system templates and Excel templates, supporting users to edit Excel templates offline and then re-import them into the system, automatically synchronizing data changes and verifying logical consistency. This Excel-compatible interactive module supports batch importing project information, budget details, expenditure records, etc., from Excel, automatically matching fields and removing duplicates, reducing manual data entry workload. In this embodiment, the data imported by the Excel-compatible interactive module is verified and synchronized to the multi-source data integration module to participate in feature extraction and model training; its export function is linked with the visualization interactive module, supporting the direct export of analysis results and reports to Excel format.

[0037] Reference Figure 2 In some embodiments, the system of this invention further includes: a project team and phase management module, used to support adding project team members, setting member roles (project manager, budget administrator, executor, approver, etc.) and operation permissions; recording member contact information (name, email, phone number), allocating workload (percentage or specific tasks) and tracking completion progress; supporting the division of projects into multiple phases (such as requirements analysis, design, implementation, testing, and launch), setting start / end times, milestone events, dependencies (such as the implementation phase depending on the design phase), and completion standards for each phase; supporting the addition, editing, deletion, and order adjustment of phases; the system automatically tracks the progress of each phase, associates the corresponding budget allocation amount with the actual expenditure, and automatically triggers an alert when the phase progress is delayed or the expenditure is exceeded; supports viewing correlation analysis charts between phase progress and budget execution rate; allows uploading project-related documents (requirement documents, design schemes, contracts, invoices, etc.), storing them by phase and type, supporting online document preview, download, and version management, and linking with project team member permissions (only authorized members can view / edit).

[0038] Reference Figure 3In some embodiments, the system of this invention further includes a budget adjustment and multi-level approval module, used to support customizable multi-level budget adjustment thresholds, establish a mapping relationship between threshold amounts and approval levels (department heads, financial directors, general managers, etc.), and adapt to the enterprise's financial approval system. After a budget adjustment application is submitted, the system automatically matches the corresponding approval node according to the adjustment amount and pushes the application form to the approver; it supports setting approval time limits, expiration reminders (system messages, emails), and an automatic upgrade mechanism for timeouts. It also manages the approval process, allowing approvers to view application details online, enter approval opinions, and upload attachments (such as adjustment instructions, quotations); it allows applications to be rejected with reasons noted, and applicants can resubmit after supplementing materials, forming a complete approval loop. It is also used to trace approval history, fully recording the application information, approval process, approval opinions, and adjustment results of each budget adjustment, supporting searches by project, department, time, and other dimensions to meet audit and compliance requirements. In this embodiment, the approval results are fed back to the closed-loop iterative optimization module, serving as the basis for budget adjustments to update prediction model parameters. The approval threshold and permission configuration are linked to the permission management module to ensure that different approval levels can only process applications within their corresponding scope.

[0039] In some embodiments, the system of this invention further includes: a report lifecycle management module, used to support automatic distribution of reports according to preset rules (recipient, department, distribution cycle, distribution method), including in-system messages, emails, and push notifications from enterprise instant messaging tools; supporting the setting of recipient permissions (view only, annotation allowed, export allowed); classifying and archiving reports by report type (budget report, expenditure report, progress analysis report), cycle (daily / weekly / monthly / quarterly / annual report), department, project, and other dimensions; supporting multi-condition combined retrieval (report name, time, department, keywords) to quickly locate historical reports. Allowing users to add text annotations and highlight key content in reports, supporting @mentioning specific personnel to trigger feedback reminders; annotations are stored in association with the report, retaining all interaction records. Recording report generation versions and update records (such as data corrections and content additions), supporting version comparison and historical version recovery, ensuring the traceability of report data.

[0040] In some embodiments, the multi-source data integration module is further used to acquire external business environment data, perform cleaning and standardization processing, and extract external environment features, which are then incorporated into the multi-source data feature category. The business scenario feature extraction module is used to extract and refine external environment features based on these features. These refined external environment features include market fluctuations, policy adjustments, and industry trends. The adaptive prediction module incorporates the refined external environment features into the fusion feature set of the prediction model, and a dynamic weight adjustment mechanism synchronously adjusts the weights of these refined external environment features in real time. Existing technologies often limit capital expenditure prediction to internal enterprise data, failing to fully incorporate the impact of the external business environment, resulting in prediction results that are difficult to adapt to external changes such as market fluctuations and policy adjustments. This embodiment, by adding external business environment data access and refining feature extraction, and incorporating it into the prediction model and weight adjustment system, enables the prediction model to consider both internal and external influencing factors, significantly improving the comprehensiveness and accuracy of the prediction results and meeting the differentiated needs of different enterprises for external data.

[0041] In one specific embodiment, a manufacturing company uses this system to manage a production line expansion project. In addition to acquiring historical financial data, real-time execution data, and project management data, the multi-source data integration module also needs to acquire external business environment data such as raw material market price data and industry environmental policy documents. After cleaning and standardization, external environment characteristics such as quarterly raw material price fluctuations and new equipment requirements under environmental policies are extracted. The business scenario feature extraction module further extracts detailed external environment characteristics from these characteristics, such as a 5% monthly increase in raw material prices and the need for additional investment in environmental protection equipment. The adaptive prediction module integrates these detailed external environment characteristics with historical financial characteristics and dynamic business scenario characteristics (the project is in the equipment procurement stage). The dynamic weight adjustment mechanism adjusts the weight of the detailed external environment characteristics from the initial 10% to 25% based on the impact of raw material price fluctuations on expenditures. Ultimately, this reduces the project capital expenditure prediction deviation rate from 12% before external data integration to 6%, significantly improving prediction accuracy.

[0042] In some embodiments, the business scenario feature extraction module includes a feature priority ranking unit, used to uniformly prioritize dynamic business scenario features, refined external environment features (if present), and other multi-source data features based on business impact quantification indicators, providing a quantitative basis for weight allocation in the prediction model; the quantification indicators include two core indicators: the correlation between features and expenditure deviations and the frequency of feature occurrence. This embodiment solves the problem of strong subjectivity in feature ranking in existing technologies by clearly defining the two core quantification indicators of feature-expenditure deviation correlation and feature occurrence frequency, and through a feature priority ranking mechanism, ensures that the prediction model prioritizes high-impact features, which improves prediction accuracy, reduces computational redundancy caused by irrelevant features, and is adaptable to different scenarios without external refined environment features, possessing good flexibility.

[0043] In some embodiments, the business scenario feature extraction module includes: a feature priority ranking subunit, which evaluates feature importance by combining mutual information entropy and decision tree gain ratio, and performs unified priority ranking of dynamic business scenario features, refined external environment features and multi-source data features based on capital expenditure impact quantification indicators, providing a quantitative basis for the weight allocation of the prediction model; the capital expenditure impact quantification indicators include the correlation between features and expenditure deviation, feature occurrence frequency and feature contribution to prediction results.

[0044] In some embodiments, the adaptive prediction module includes a time series algorithm and a regression analysis algorithm, automatically matching the appropriate algorithm based on the business scenario type. A dynamic weight adjustment mechanism adjusts the weight proportions of multi-source data features, dynamic business scenario features, and, if present, refined external environment features in the prediction model in real time based on the frequency of scenario changes and feature priority ranking. The magnitude of the weight adjustment is positively correlated with the magnitude of scenario changes and the contribution of features. This invention, by limiting the two mainstream adaptation algorithms (time series and regression analysis) and achieving automatic matching, combined with dynamic weight adjustment based on the frequency of scenario changes and feature priority, enables the prediction model to adapt to changes in business scenarios in real time, significantly improving the model's flexibility and adaptability.

[0045] In some embodiments, the adaptive prediction module includes a preset basic algorithm library for automatically matching and adapting basic algorithms based on business scenario types, and integrating the prediction results of multiple algorithms through a weighted voting fusion strategy; the dynamic weight adjustment mechanism adjusts the weight ratio of multi-source data features, dynamic business scenario features, and refined external environment features in the prediction model in real time based on the frequency of scenario changes and the evaluation results of feature importance, and the weight adjustment magnitude is positively correlated with the magnitude of scenario changes and feature contribution; the preset basic algorithm library includes time series algorithms, regression analysis algorithms, and machine learning classification algorithms.

[0046] In some embodiments, the multi-dimensional association rules of the anomaly root cause tracing module include project stage-cost type association rules, approval node-expenditure compliance association rules, and resource allocation-expenditure efficiency association rules; the full-link structured association model includes project stage nodes, cost type nodes, approval nodes, and resource nodes, and the association edges between nodes are constructed based on the quantitative analysis of data association strength; root cause localization involves traversing the node association paths, calculating the influence coefficient of each node, and selecting the node combination with the highest influence coefficient as the core root cause; wherein, influence coefficient = node association strength × path length weight.

[0047] In some embodiments, the multi-dimensional association rules in the anomaly root cause tracing module include project stage-cost type association rules, approval node-expenditure compliance association rules, and resource allocation-expenditure efficiency association rules; the capital expenditure full-link data graph is constructed using a graph neural network algorithm, and the graph nodes include project stage nodes, cost type nodes, approval nodes, and resource nodes, with the association edges between nodes constructed based on the quantification of data association strength; root cause localization involves traversing the association paths of graph nodes, calculating the influence coefficient of each node, and selecting the node combination with the highest influence coefficient as the core root cause; wherein, influence coefficient = node association strength × path length weight.

[0048] In some embodiments, the closed-loop iterative optimization module includes: a deviation cause classification subunit, used to classify deviation causes into scenario feature missing type, algorithm adaptability type, data quality type, and business mutation type; for scenario feature missing type deviations, triggering the business scenario feature extraction module to supplement the corresponding dynamic business scenario features, and if they exist, supplementing and refining external environment features; for algorithm adaptability type deviations, switching the algorithm of the adaptive prediction module; for data quality type deviations, initiating the secondary data cleaning process of the multi-source data integration module; for business mutation type deviations, marking them as special scenario samples and updating the scenario adaptation rules of the prediction model. Existing technologies mostly handle prediction deviations with general adjustments, while this embodiment, by subdividing deviation causes into four specific types and designing targeted differentiated processing strategies, ensures that different types of deviations can be accurately resolved, significantly improving the efficiency and effectiveness of the closed-loop iteration of the prediction model, accelerating the adaptive optimization process of the model, and enabling the model to quickly adapt to business changes and data quality fluctuations.

[0049] In some embodiments, the adaptive prediction module supports user input of custom business assumption parameters. The parameter input can be in the form of interval values ​​or discrete values. The prediction model integrates custom parameters, multi-source data features, dynamic business scenario features, and refined external environment features to simulate capital expenditure prediction results under different assumption scenarios and generate a multi-scenario prediction comparison report. The multi-scenario prediction comparison report includes the expenditure amount range, deviation risk, and resource requirement differences under each scenario. The custom business assumption parameters include resource allocation adjustment parameters, market environment change parameters, and project schedule adjustment parameters.

[0050] In some embodiments, the anomaly root cause tracing module includes: an anomaly propagation path analysis unit, used to identify subsequent project links and related projects that may be affected by abnormal expenditures based on the capital expenditure full-link structured association model, combined with project dependencies and time series characteristics, calculate the propagation probability, propagation probability = node association strength × time decay coefficient, and generate an anomaly propagation risk heat map.

[0051] In some embodiments, the anomaly root cause tracing module includes: an anomaly propagation path analysis subunit, which is used to identify subsequent project links and related projects that may be affected by abnormal expenditures based on the full-link data map of capital expenditures, using a graph neural network node influence prediction algorithm, combined with project dependencies and time series characteristics, calculate the propagation probability, generate an anomaly propagation risk heat map, and provide early warning of chain risks.

[0052] In some embodiments, the visualization interaction module supports drill-down data exploration, allowing users to drill down layer by layer from the summarized prediction results to the multi-source data feature details, dynamic business scenario feature details, and refined external environment feature details corresponding to a single project, a single cost type, and a single time period; it also supports custom visualization chart types, with chart data updated synchronously in real time.

[0053] Reference Figure 5 This invention proposes a method for monitoring and predicting capital expenditures based on big data, comprising the following steps:

[0054] S100. Obtain historical financial data, real-time execution data, and project management data related to capital expenditures, and after cleaning and standardizing them respectively, extract the corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features, and basic project features.

[0055] S200. Extract dynamic business scenario features based on the basic characteristics of the project; dynamic business scenario features include project stage, resource configuration, and dependencies.

[0056] S300. Construct a prediction model that integrates multi-source data features and dynamic business scenario features. The prediction model optimizes the model parameters in real time according to changes in business scenarios through a dynamic weight adjustment mechanism, and generates multi-cycle capital expenditure prediction results.

[0057] S400: For abnormal expenditure data monitored, based on preset multi-dimensional correlation rules, a full-link structured correlation model of capital expenditure is constructed to automatically trace the project links, cost types, approval nodes and related influencing factors where the abnormality occurred, and locate the core root cause.

[0058] S500 compares the prediction results with the actual execution data in real time, calculates the deviation rate and analyzes the causes of the deviation, generates model parameter adjustment instructions, and feeds them back to the adaptive prediction module to realize dynamic iterative optimization of the prediction model.

[0059] Although specific embodiments are described herein, those skilled in the art will recognize that many other modifications or alternative embodiments are also within the scope of this disclosure. For example, any of the functions and / or processing capabilities described in connection with a particular device or component can be performed by any other device or component. Furthermore, while various exemplary embodiments and architectures have been described according to embodiments of this disclosure, those skilled in the art will recognize that many other modifications to the exemplary embodiments and architectures described herein are also within the scope of this disclosure.

[0060] The foregoing description, with reference to block diagrams and flowcharts of systems, methods, systems, and / or computer program products according to exemplary embodiments, has described certain aspects of this disclosure. It should be understood that one or more blocks in the block diagrams and flowcharts, as well as combinations of blocks in the block diagrams and flowcharts, can be implemented by executing computer-executable program instructions, respectively. Similarly, according to some embodiments, some blocks in the block diagrams and flowcharts may not need to be executed in the order shown, or may not all need to be executed. Furthermore, additional components and / or operations beyond those shown in the blocks in the block diagrams and flowcharts may exist in some embodiments.

[0061] Therefore, blocks in block diagrams and flowcharts support combinations of means for performing a specified function, combinations of elements or steps for performing a specified function, and program instruction means for performing a specified function. It should also be understood that each block in a block diagram and flowchart, and combinations of blocks in block diagrams and flowcharts, can be implemented by a dedicated hardware computer system or a combination of dedicated hardware and computer instructions that performs a specific function, element, or step.

[0062] The program modules, applications, etc., described herein may include one or more software components, including, for example, software objects, methods, data structures, etc. Each such software component may include computer-executable instructions that, in response to execution, cause at least a portion of the functionality described herein (e.g., one or more operations of the exemplary methods described herein) to be performed.

[0063] Software components can be coded using any of a variety of programming languages. An exemplary programming language could be a low-level programming language, such as assembly language associated with a specific hardware architecture and / or operating system platform. Software components including assembly language instructions may need to be converted into executable machine code by an assembler before being executed by the hardware architecture and / or platform. Another exemplary programming language could be a higher-level programming language that is portable across multiple architectures. Software components including higher-level programming languages ​​may need to be converted into an intermediate representation by an interpreter or compiler before execution. Other examples of programming languages ​​include, but are not limited to, macro languages, shell or command languages, job control languages, scripting languages, database query or search languages, or report writing languages. In one or more exemplary embodiments, a software component containing instructions from one of the above-described programming language examples can be executed directly by the operating system or other software components without first being converted into another form.

[0064] Software components can be stored as files or other data storage structures. Software components of similar type or related function can be stored together in a specific directory, folder, or library. Software components can be static (e.g., pre-defined or fixed) or dynamic (e.g., created or modified at runtime).

[0065] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A capital expenditure monitoring and predictive analysis system based on big data, characterized in that, include: The multi-source data integration module is used to acquire historical financial data, real-time execution data, and project management data related to capital expenditures, and after cleaning and standardization, extract the corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features, and basic project features; The business scenario feature extraction module is used to extract dynamic business scenario features based on the basic features of the project. The dynamic business scenario features include project phases, resource allocation, and dependencies; An adaptive prediction module is used to construct a prediction model that integrates the features of the multi-source data and the features of the dynamic business scenario. The prediction model optimizes the model parameters in real time according to the changes in the business scenario through a dynamic weight adjustment mechanism, and generates multi-period capital expenditure prediction results. The anomaly root cause tracing module is used to construct a full-link structured correlation model of capital expenditure based on preset multi-dimensional correlation rules for monitored abnormal expenditure data. It automatically traces the project links, cost types, approval nodes and related influencing factors where the anomaly occurred, and locates the core root cause. The closed-loop iterative optimization module is used to compare the prediction results with the actual execution data in real time, calculate the deviation rate and analyze the causes of the deviation, generate model parameter adjustment instructions, and feed them back to the adaptive prediction module to realize the dynamic iterative optimization of the prediction model.

2. The capital expenditure monitoring and predictive analysis system based on big data according to claim 1, characterized in that, The system also includes: The visualization and interaction module is used to display prediction results, root cause analysis reports of anomalies, and model optimization effects, and supports users to customize core analysis dimensions and basic interactive operations; The permissions management module is used to configure the operation permissions, data access permissions, and approval permissions for different users.

3. The capital expenditure monitoring and predictive analysis system based on big data according to claim 1, characterized in that, The multi-source data integration module is also used to acquire external business environment data, perform cleaning and standardization processing, and extract external environment features, which are included in the multi-source data feature category; the business scenario feature extraction module is used to extract refined external environment features based on the external environment features; the refined external environment features include market fluctuations, policy adjustments, and industry trends; the adaptive prediction module incorporates the refined external environment features into the fusion feature set of the prediction model, and the dynamic weight adjustment mechanism synchronously adjusts the weight of the refined external environment features in real time.

4. The capital expenditure monitoring and predictive analysis system based on big data according to any one of claims 1 or 3, characterized in that, The business scenario feature extraction module includes: The feature priority ranking unit is used to uniformly prioritize dynamic business scenario features, detailed external environment features (if present), and other multi-source data features based on business impact quantification indicators, providing a quantitative basis for the weight allocation of the prediction model; the quantification indicators include two core indicators: the correlation between features and expenditure deviations and the frequency of feature occurrence.

5. The capital expenditure monitoring and predictive analysis system based on big data according to any one of claims 1 or 3, characterized in that, The adaptive prediction module is equipped with time series algorithm and regression analysis algorithm, and automatically matches and adapts the algorithm based on the business scenario type; the dynamic weight adjustment mechanism adjusts the weight ratio of multi-source data features, dynamic business scenario features and refined external environment features (if present) in the prediction model in real time based on the frequency of scenario changes and the priority ranking results of features. The weight adjustment magnitude is positively correlated with the magnitude of scenario changes and the contribution of features.

6. The capital expenditure monitoring and predictive analysis system based on big data according to claim 1, characterized in that, The multi-dimensional association rules of the anomaly root cause tracing module include project stage-cost type association rules, approval node-expenditure compliance association rules, and resource allocation-expenditure efficiency association rules; the full-link structured association model includes project stage nodes, cost type nodes, approval nodes, and resource nodes, and the association edges between nodes are constructed based on the quantitative analysis of data association strength; root cause localization involves traversing the node association paths, calculating the influence coefficient of each node, and selecting the node combination with the highest influence coefficient as the core root cause; where, influence coefficient = node association strength × path length weight.

7. The capital expenditure monitoring and predictive analysis system based on big data according to any one of claims 1 or 3, characterized in that, The closed-loop iterative optimization module includes: The deviation cause classification subunit is used to divide deviation causes into four categories: missing scenario features, algorithm adaptability, data quality, and business mutation. For deviations of missing scenario features, the business scenario feature extraction module is triggered to supplement the corresponding dynamic business scenario features, and if they exist, the external environment features are supplemented and refined. For deviations of algorithm adaptability, the algorithm of the adaptive prediction module is switched. For deviations of data quality, the secondary data cleaning process of the multi-source data integration module is initiated. For deviations of business mutation, they are marked as special scenario samples and the scenario adaptation rules of the prediction model are updated.

8. The capital expenditure monitoring and predictive analysis system based on big data according to any one of claims 1 or 3, characterized in that, The adaptive prediction module supports user input of custom business assumption parameters, including resource allocation adjustment parameters, market environment change parameters, and project schedule adjustment parameters. The parameters are input in the form of interval values ​​or discrete values. The prediction model simulates the capital expenditure prediction results under different assumption scenarios and generates a multi-scenario prediction comparison report. The report includes the expenditure amount range, deviation risk, and resource demand differences under each scenario.

9. The capital expenditure monitoring and predictive analysis system based on big data according to claim 1, characterized in that, The anomaly root cause tracing module includes: The abnormal propagation path analysis unit is used to identify subsequent project links and related projects that may be affected by abnormal expenditures based on the full-link structured association model of capital expenditures, combined with project dependencies and time series characteristics, and calculate the propagation probability, wherein the propagation probability = node association strength × time decay coefficient, and generate an abnormal propagation risk heat map.

10. A method for monitoring and predictive analysis of capital expenditures based on big data, used in the system as described in any one of claims 1 to 9, characterized in that, Includes the following steps: S100. Obtain historical financial data, real-time execution data, and project management data related to capital expenditures, and after cleaning and standardizing them respectively, extract the corresponding multi-source data features; the multi-source data features include historical financial features, real-time execution features, and basic project features; S200. Extract dynamic business scenario features based on the project's basic features; the dynamic business scenario features include project stage, resource configuration, and dependencies. S300. Construct a prediction model that integrates the features of the multi-source data and the features of the dynamic business scenario. The prediction model optimizes the model parameters in real time according to the changes in the business scenario through a dynamic weight adjustment mechanism, and generates multi-cycle capital expenditure prediction results. S400: For abnormal expenditure data monitored, based on preset multi-dimensional correlation rules, a full-link structured correlation model of capital expenditure is constructed to automatically trace the project links, cost types, approval nodes and related influencing factors where the abnormality occurred, and locate the core root cause. S500 compares the prediction results with the actual execution data in real time, calculates the deviation rate and analyzes the causes of the deviation, generates model parameter adjustment instructions, and feeds them back to the adaptive prediction module to realize dynamic iterative optimization of the prediction model.

Images