A power grid infrastructure project progress matching evaluation method based on multi-source data fusion

Abstract

The application discloses a power grid infrastructure project progress matching evaluation method based on multi-source data fusion, relates to the technical field of power grid infrastructure project management, and comprises the following steps: collecting four types of progress completion rate data of construction, investment, accounting and materials from a project management system, an investment plan management system, a financial management system and a material management system, forming a standardized four-rate time series data set after standardization processing and time sequence alignment; constructing a four-rate time series data matrix; calculating a project-level matching degree index and a dispersion index; constructing a project group-level evaluation matrix and calculating a comprehensive matching degree index and a comprehensive dispersion index; triggering project-level and project group-level matching abnormal early warning and stability early warning according to a preset threshold value; identifying a dominant deviation rate by a leave-one-out method, positioning an abnormal period by a sliding window analysis, and outputting a root cause tracing report. The power grid infrastructure project progress matching evaluation method based on multi-source data fusion realizes accurate quantitative evaluation of power grid infrastructure project progress matching.

Description

Technical Field

[0001] This invention relates to the field of power grid infrastructure project management technology, specifically to a method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion. Background Technology

[0002] Power grid infrastructure project management involves multiple professional departments, including development, finance, operation and maintenance, infrastructure, materials, and dispatch. Each professional department has different types of progress data. Currently, the industry's common management method is the four-rate integrated data monitoring and analysis system, which monitors the matching degree of the construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate of power grid infrastructure projects, and strengthens process control through dynamic deviation early warning.

[0003] However, existing monitoring technologies that integrate the four rates still have the following significant shortcomings: First, there is a lack of quantitative assessment methods for the matching degree among the four rates. Existing technologies mainly rely on manual experience or simple threshold comparisons to determine whether the four rates match. For example, setting a certain percentage limit for the deviation between each rate, this approach cannot reflect the complex nonlinear relationships and temporal evolution patterns among the four rates, making it difficult to accurately identify abnormal patterns in schedule matching. The assessment results are highly subjective and have low accuracy. Second, existing monitoring systems mainly monitor single projects and lack the comprehensive assessment capability for schedule matching at the project group or project portfolio level. In large-scale power grid infrastructure projects, multiple sub-projects usually need to be promoted in a coordinated manner. The projects have complex dependencies and mutual influences, making it difficult for existing technologies to effectively and uniformly assess the progress matching status of the entire project portfolio. Third, the integration of multi-source data is insufficient. In existing technologies, the four key performance indicators (KPIs, ERP, and GPRS) are stored in different business systems, such as the same source system, ERP system, and online power grid system. This results in inconsistent data formats, varying update frequencies, and inconsistent data quality. The lack of an effective data fusion and alignment mechanism makes it difficult to guarantee the accuracy and timeliness of the assessment results. Fourth, there is a lack of intelligent tracing and analysis capabilities for the causes of progress deviations. When mismatches in the four KPIs are detected, existing technologies can only issue warning signals but cannot automatically locate the specific causes and responsible parties. Managers still need to manually verify each item, which is inefficient, and the warning mechanism mainly relies on static threshold settings, lacking dynamic adjustment and intelligent analysis capabilities. Summary of the Invention

[0004] The purpose of this invention is to provide a method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion, in order to solve the problems mentioned in the background art of the existing monitoring technology that integrates the four rates, lacks a quantitative evaluation method for the degree of matching between the four rates, lacks a comprehensive evaluation capability for the progress matching at the project group or project combination level, has insufficient multi-source data fusion, and lacks intelligent traceability and analysis capabilities for the causes of progress deviations.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion, comprising the following steps:

[0006] Step S1: Collect raw data on construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate from multiple business systems of the power grid infrastructure project, and perform standardization and time-series alignment on the raw data to form a standardized four-rate time-series dataset.

[0007] Step S2: For the first Individual power grid infrastructure projects, including Time window of each time sampling point Internally, construct a four-rate time series data matrix. ;

[0008] Step S3: Based on the time series data matrix Calculate the first The four-rate matching index of each project and dispersion index ;

[0009] Step S4: For a project group containing N projects, construct a project group-level matching evaluation matrix. And calculate the comprehensive matching index of the project group. Comprehensive dispersion index of project groups ;

[0010] Step S5: Set the matching index threshold and the dispersion index threshold ,when < or < When a matching error warning is triggered, > A stability warning is triggered at any time;

[0011] Step S6: When an alert is triggered, the dominant deviation rate is identified by calculating the matching degree index after removing each rate, and the abnormal period is located by analyzing the sliding window, and a root cause tracing report is output.

[0012] Furthermore, the multiple business systems in step S1 include a project management system, an investment plan management system, a financial management system, and a materials management system. The standardization process includes unifying the timestamp format, unifying the calculation method for progress percentage, and removing outliers and handling missing values.

[0013] Furthermore, the four-rate time-series data matrix in step S2 Represented as:

[0014] ,in, Indicates the first The project in the Construction progress completion rate of each time sampling point; Indicates the investment progress completion rate; This indicates the completion rate of the accounting entries process; This indicates the completion rate of material delivery.

[0015] Furthermore, in step S3, the four-rate matching index The calculation method is as follows: First, calculate the Pearson correlation coefficient between any two rates. Then calculate .

[0016] Furthermore, in step S3, the four-rate dispersion index The calculation method is as follows:

[0017] ;in, , , and The first The average completion rate of the four types of progress for each project within the time window T.

[0018] Furthermore, the item group-level matching evaluation matrix in step S4 Represented as: ;in, For the first The weighting coefficients for each project are determined comprehensively based on the project's investment scale, construction period, and importance level.

[0019] Furthermore, in step S4, the comprehensive matching index of the project group... The calculation method is as follows: Project Group Comprehensive Dispersion Index The calculation method is as follows: .

[0020] Furthermore, the method for identifying the dominant deviation rate in step S6 is as follows: calculate the matching degree index after removing the construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate, respectively. ,Will Compared to The rate with the largest increase was identified as the dominant source of deviation.

[0021] Furthermore, the method for locating abnormal time periods in step S6 is as follows: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Perform sliding window analysis, calculate the matching degree index within each sub-window, and identify the time intervals in which the matching degree significantly decreases. This interval is marked as an abnormal period.

[0022] Furthermore, the system modules of the evaluation method include:

[0023] The data acquisition module is used to collect raw data on the four rates from multiple business systems;

[0024] The data preprocessing module is used to standardize and time-series align the collected multi-source heterogeneous data.

[0025] The matrix construction module is used to construct a four-rate time series data matrix;

[0026] The matching degree calculation module is used to calculate the matching degree index and dispersion index at the project level and project group level.

[0027] The early warning judgment module is used to trigger matching anomaly warnings and stability warnings based on preset thresholds;

[0028] The root cause tracing module is used to identify the dominant deviation rate and locate abnormal periods, and output a root cause tracing report.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] This method for evaluating the progress matching of power grid infrastructure projects, based on multi-source data fusion, achieves accurate quantitative evaluation and intelligent early warning of the progress matching degree of power grid infrastructure projects by constructing a four-rate time-series data matrix, calculating a multi-dimensional matching degree index, establishing a dynamic deviation early warning model, and automatically tracing the root causes.

[0031] 1. Furthermore, by constructing a four-rate time-series data matrix and calculating the matching degree index and dispersion index, the traditional qualitative judgment relying on human experience and static thresholds is transformed into a quantifiable mathematical evaluation. The evaluation results are objective, accurate, and reproducible, significantly improving the scientificity and consistency of the four-rate matching status judgment.

[0032] 2. Furthermore, by constructing a project group-level evaluation matrix and introducing project weight coefficients to assign differentiated weights to multiple projects, a multi-level matching evaluation from a single project to a project group is achieved, filling the gap in existing technologies that cannot uniformly quantify and judge the overall four-rate matching status of a project group.

[0033] 3. Furthermore, by standardizing and aligning the heterogeneous data from four independent business systems, obstacles caused by inconsistent data formats and update frequencies are eliminated, enabling the four rates data scattered across different systems to be effectively integrated and utilized, thus significantly improving data utilization efficiency.

[0034] 4. Furthermore, by using the leave-one-out method to compare and identify the dominant deviation rate and the sliding window analysis to locate abnormal periods, the abstract early warning signal is transformed into a specific investigation guide for which type of rate deviates during which period. This narrows the scope of problem investigation from four links × the entire project period to one link × one period, significantly shortening the response cycle from early warning to root cause confirmation. Attached Figure Description

[0035] Figure 1 This is an overall flowchart of the power grid infrastructure project schedule matching evaluation method based on multi-source data fusion, as described in this invention.

[0036] Figure 2 This is a schematic diagram illustrating the construction of the four-rate time-series data matrix of the present invention;

[0037] Figure 3 This is a schematic diagram illustrating the hierarchical relationship between project-level and project group-level matching evaluation in this invention;

[0038] Figure 4 This is a flowchart of the dynamic deviation early warning and root cause tracing process of the present invention. Detailed Implementation

[0039] 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.

[0040] This invention provides a method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion. Its core lies in collecting progress completion rate data for four categories—construction, investment, accounting, and materials—from project management systems, investment planning systems, financial management systems, and material management systems. After standardization and time-series alignment, a standardized four-rate time-series dataset is formed. A four-rate time-series data matrix is ​​constructed; project-level matching degree index and dispersion index are calculated; a project group-level evaluation matrix is ​​constructed, and a comprehensive matching degree index and comprehensive dispersion index are calculated; project-level and project group-level matching anomaly warnings and stability warnings are triggered based on preset thresholds; the dominant deviation rate is identified through leave-one-out comparison; abnormal time periods are located through sliding window analysis; and a root cause tracing report is output. This invention achieves quantitative evaluation of the four-rate matching degree and intelligent root cause tracing.

[0041] Example 1: To better understand the above technical solution, the following will provide a detailed description of the technical solution in conjunction with the accompanying drawings and specific implementation methods. (Refer to...) Figure 1 As shown in the figure, this is the overall flowchart of the power grid infrastructure project schedule matching assessment method based on multi-source data fusion. The power grid infrastructure project schedule matching assessment method based on multi-source data fusion includes the following steps:

[0042] Step S1: Collect raw data from multiple business systems of the power grid infrastructure project, including: Construction progress completion rate (obtained from the project management system, reflecting the completion rate of the physical construction progress relative to the milestone plan); Investment progress completion rate (obtained from the investment planning management system, reflecting the completion rate of fixed asset investment relative to the annual investment plan); Accounting progress completion rate (obtained from the financial management system, reflecting the completion rate of financial accounting amount relative to the annual budget); and Material progress completion rate (obtained from the material management system and ERP system, reflecting the completion rate of material supply and requisition relative to the material demand plan). Standardize and time-series align the raw data to form a standardized four-rate time-series dataset. Specifically, the multiple business systems include a project management system, an investment planning management system, a financial management system, and a materials management system. The standardization process includes unifying the timestamp format: aligning scattered date fields into a unified time coordinate; unifying the calculation method for progress percentages: ensuring that each rate is represented by the same percentage definition; and using conventional data cleaning methods to remove obvious outliers and fill in missing values. On this basis, the data sources are aligned according to a preset time granularity, such as day, week, and month, so that the four types of data are comparable at the same time sampling point, thereby forming a standardized four-rate time series dataset, providing standardized and unified data input for subsequent construction of time series data matrices and calculation of matching degree indices.

[0043] The role of step S1 in the overall technical solution: Step S1 is the data foundation layer of the entire evaluation method. The four rates data of power grid infrastructure projects are stored in different business systems, and have typical characteristics of being multi-source, heterogeneous, and asynchronous. If the unprocessed raw data is used directly, the subsequent time series correlation analysis and matching degree calculation will lose accuracy due to inconsistent data caliber and misaligned time axis. Through the standardization and time series alignment in this step, the heterogeneity and quality problems between multi-source data are eliminated, and the scattered and heterogeneous raw data are transformed into a standardized dataset with unified structure and time synchronization. This lays a reliable data foundation for the construction of the four rate time series data matrix in step S2, and also ensures the accuracy and repeatability of the entire evaluation method from the source.

[0044] Step S2: For the first Individual power grid infrastructure projects, including Time window of each time sampling point Internally, construct a four-rate time series data matrix. Specifically, the four-rate time series data matrix Represented as:

[0045] ,in, Indicates the first The project in the Construction progress completion rate of each time sampling point; Indicates the investment progress completion rate; This indicates the completion rate of the accounting entries process; This indicates the completion rate of material progress. The specific steps are as follows: After completing the collection and standardized preprocessing of multi-source data, the four types of dispersed progress data need to be organized into a unified data structure for subsequent time-series correlation analysis and matching degree calculation. Therefore, for the first... Each power grid infrastructure project, within the preset time window Within this time window, The number of sampling points is determined by the time window length and the preset time granularity, constructing a four-rate time series data matrix for this project; the four-rate time series data matrix is... The matrix is ​​a two-dimensional matrix with four rows and four columns. Each row corresponds to a time sampling point, and each column corresponds to a type of progress completion rate. The rows of the matrix are arranged from top to bottom in chronological order to ensure the correctness of the time sequence. Specifically, the first column of the matrix is ​​the construction progress completion rate sequence, the second column is the investment progress completion rate sequence, the third column is the accounting progress completion rate sequence, and the fourth column is the material progress completion rate sequence. The element in the first column of the row represents the first element. The project in the Construction progress completion rate at each time sampling point, the first The element in the second column of the row represents the investment progress completion rate at the corresponding time point. The element in the third column of the row represents the completion rate of the accounting progress. The element in row 4 represents the material progress completion rate. Time window. The selection strategy can be flexibly set according to the actual management needs of the project: for projects in the peak construction period, a shorter time window, such as the most recent 3 months, is appropriate to capture rapid changes in progress; for projects in a stable progress phase, a longer time window, such as the most recent 6 months or 12 months, can be used to reflect the long-term synergistic trend of the four rates. The time granularity can also be set to the daily, weekly, or monthly level according to the control precision requirements; in the above way, each power grid infrastructure project is represented as a A 4×4 time-series data matrix completely preserves all information about the four types of progress completion rates over time, including the magnitude of each rate, its changing trends, and the temporal correspondence between them. This provides standardized data input for the calculation of the matching degree index and the dispersion index in step S3. (Refer to...) Figure 2 As shown.

[0046] The role of step S2 in the overall technical solution: Step S2 serves as a bridge connecting data preprocessing and matching degree evaluation. After standardization in step S1, the four rates still exist as independent sequences, failing to directly reflect the correlation between them. This step constructs a unified time-series data matrix, integrating the originally scattered four types of progress data into a structured mathematical object. This establishes a precise temporal correspondence between the four rates under the same coordinate system. This matrix structure is the foundational data carrier for all subsequent calculations in this solution, including correlation coefficient calculation, matching degree index calculation, dispersion analysis, and sliding window anomaly location. Compared to existing technologies that only perform independent statistics or simple difference comparisons on each rate, the time-series correlation matrix constructed in this step provides data structure support for uncovering the deep linkage between the four rates. This is a key data organization method that distinguishes this solution from existing technologies.

[0047] Step S3: Based on the time series data matrix Calculate the first The four-rate matching index of each project and dispersion index Specifically, the four-rate matching index The calculation method is as follows: First, calculate the Pearson correlation coefficient between any two rates. Then calculate Four-rate dispersion index The calculation method is as follows:

[0048] ;in, , , and The first The completion rates of the four types of progress for each project within the time window The mean within.

[0049] The specific operation is as follows: After completing the construction of the four-rate time series data matrix, it is necessary to extract quantitative indicators from the matrix that can reflect the correlation and synergistic stability among the four types of progress completion rates in order to determine whether the four rates of the project match and to what extent they match. To this end, this step proposes two complementary evaluation indicators: the matching degree index and the dispersion index. The two indicators respectively characterize the intrinsic correlation state of the four rates from the two dimensions of trend consistency and fluctuation synergy. The calculation of the four-rate matching index: The core idea of ​​the matching index is that in a power grid infrastructure project with normal progress coordination, the construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate should show a trend of fluctuating in the same direction. When the project construction progresses, the investment is completed accordingly, the financial accounting follows simultaneously, and the supply and requisition of materials also occur. There should be a positive temporal linkage relationship among the four. Conversely, if the trend of a certain rate deviates significantly from the other rates, it indicates that there may be an anomaly in the progress execution of that stage. Based on the above idea, we first conduct a temporal correlation analysis between each pair of the four rates, and use conventional indicators in statistics to measure the degree of linear correlation between two variables to quantitatively characterize any two types of progress completion rates within a time window. To assess the consistency of trends within the project, specific correlation coefficients were calculated between the construction progress completion rate and the investment progress completion rate, the construction progress completion rate and the accounting progress completion rate, the construction progress completion rate and the material progress completion rate, the investment progress completion rate and the accounting progress completion rate, the investment progress completion rate and the material progress completion rate, and the accounting progress completion rate and the material progress completion rate, resulting in six correlation coefficients. The arithmetic mean of these six correlation coefficients was taken as the comprehensive matching index of the four rates for the project, with a value range of [-1, 1]. When the index value is close to 1, it indicates that there is a strong positive correlation among the four types of progress completion rates, the time-series change trends are highly consistent, and the four rates are well matched. When the index value is close to -1, it indicates that there is an overall negative correlation or inverse relationship among the four types of progress completion rates, and the four rates are seriously mismatched. When the index value is close to 0, it indicates that there is no obvious linear correlation among the four types of progress completion rates, and the synergy of the four rates is insufficient. Calculation of the four-rate dispersion index: The matching index focuses on characterizing the trend consistency among the four rates, but it cannot reflect the coordinated stability of the four rates in terms of numerical fluctuation. For example, the four rates may all show the same upward trend and the matching index may be high, but their respective increases may differ greatly, with some increasing by 10 percentage points and others by only 1 percentage point. This situation also indicates that the progress pace of different stages is not coordinated. Therefore, this step further introduces the dispersion index. The core idea of ​​the dispersion index is that in projects with good schedule matching, the four types of schedule completion rates should not only have the same trend direction, but also have roughly the same deviation from their own historical averages. There should not be a situation where one rate deviates significantly while the other rates remain stable. Specifically, for each time sampling point, the four types of schedule completion rates at that moment are calculated relative to their respective values ​​within the time window. The deviation of the internal mean is calculated, and then the square mean of the four deviations is taken. Finally, the square root of the overall mean of this value is calculated over all time sampling points to obtain the four-rate dispersion index of the project. This index reflects the overall fluctuation dispersion of the four types of progress completion rates over time. The smaller the index value, the closer the fluctuation range of the four types of progress completion rates is around their respective means, and the better the coordination stability of the four rates. The larger the index value, the more significant the fluctuation difference of the four types of progress completion rates, and the more uncoordinated the fluctuations among the four rates. There may be a risk of local loss of control in project schedule management. The synergistic evaluation significance of the two indices: The matching degree index and the dispersion index complement each other from different dimensions to characterize the matching status of the four rates. A high matching degree and low dispersion indicate that the four rates are in sync and fluctuate stably, which is the most ideal progress matching state. A low matching degree and high dispersion indicate that the four rates lack both trend consistency and fluctuation differences, which is a serious mismatch state. When one is high and the other is low, it is necessary to make a comprehensive judgment based on the specific values. For example, a high matching degree but a high dispersion indicates that although the four rates have the same trend direction, the fluctuation amplitude is large, which may indicate significant differences in the progress speed of different links, and this also needs to be paid attention to.

[0050] The role of step S3 in the overall technical solution: Step S3 is the core computational layer of the entire evaluation method. It completes the mathematical mapping from structured data to quantifiable evaluation indicators. The four-rate time-series data matrix constructed in step S2 contains rich time-series information, but its high information density makes direct interpretation difficult. It must be refined into concise and clear quantitative indicators before it can be used for early warning judgment and decision support. This step uses two indicators, the matching degree index and the dispersion index, to transform the complex time-series correlation between the four rates into intuitive numerical evaluations, providing direct quantitative basis for the early warning judgment in step S5, and also providing a data foundation for deviation analysis for root cause tracing in step S6. Compared with the existing technology that relies on manual experience or simple threshold comparison, the dual-indicator evaluation system proposed in this step can more comprehensively and objectively reflect the true state of the four-rate matching, which is one of the core technical contributions of this invention.

[0051] Step S4: For a project group containing N projects, construct a project group-level matching evaluation matrix. And calculate the comprehensive matching index of the project group. Comprehensive dispersion index of project groups Specifically, the project cluster-level matching evaluation matrix. Represented as: ;in, For the first The weighting coefficients for each project are determined comprehensively based on the project's investment scale, construction period, and importance level; the project group comprehensive matching index... The calculation method is as follows: Project Group Comprehensive Dispersion Index The calculation method is as follows: When constructing a project cluster-level schedule matching assessment model, the specific steps are as follows: In actual power grid infrastructure management scenarios, multiple power grid infrastructure projects are typically promoted simultaneously in a region or year, forming a project cluster. These projects may have relationships of resource competition, coordination, and schedule constraints. Simply conducting independent four-rate matching assessments on individual projects cannot meet the management's need to grasp the schedule matching status from the overall perspective of the project cluster. Therefore, based on the project-level assessment in step S3, this step further constructs a comprehensive assessment model at the project cluster level, converging the matching degree index and dispersion index of each project into a comprehensive indicator at the project cluster level, supporting a unified assessment of the overall four-rate matching status of the project cluster from a macro perspective. When constructing the project cluster-level matching assessment matrix: for projects containing... For a project group under construction, the matching degree index and dispersion index of each project calculated in step S3 are first aggregated. Considering that different projects have different importance and management priorities in the project group, a project weight coefficient is introduced to assign differentiated weights to each project. This weight coefficient is determined based on multiple dimensions of the project, including the project's investment scale (the larger the investment, the higher the economic importance of the project, and the corresponding increase in weight), the project's construction period (the longer the construction period, the higher the complexity of the project's schedule management, and the importance level of the project, for example, different importance coefficients are assigned according to different levels such as key projects, general projects, or supporting projects in the power grid planning). The above three dimensions are integrated into the final project weight coefficient through a conventional multi-factor weighted comprehensive method, so that the weight coefficient can comprehensively reflect the actual management priority of each project in the project group. The matching degree index, dispersion index, and corresponding weight coefficient of each project are arranged in project order to form a project group-level matching evaluation matrix. This matrix is... The structure consists of three rows and three columns, with each row corresponding to a project, recording the matching degree index, dispersion index, and weight coefficient for that project. When calculating the overall matching degree index of the project group: the overall matching degree index of the project group represents the overall level of the four-rate matching status of the entire project group. The calculation method is as follows: the matching degree indices of each project are weighted and summed according to their weight coefficients, and then divided by the sum of the weight coefficients. This calculation is essentially a weighted average of the matching degree indices of each project, so that projects with higher weights have a greater impact on the overall matching degree of the project group, reflecting the management's focus on key and complex projects. The calculated overall matching degree index also takes values ​​in the range [-1, 1]. The closer the value is to 1, the better the overall four-rate matching status of the project group; the closer the value is to -1 or 0, the greater the risk of mismatch in the overall project group. When calculating the comprehensive dispersion index of a project cluster: The comprehensive dispersion index of a project cluster represents the overall level of the coordinated stability of the four rates of the entire project cluster. The calculation method is as follows: the dispersion indices of each project are weighted and summed according to their weight coefficients, and then divided by the sum of the weight coefficients. This calculation also uses a weighted average method, so that projects with higher weights contribute more to the overall dispersion of the project cluster. The calculated comprehensive dispersion index is positive; the smaller the value, the better the coordinated stability of the four rates of the project cluster; the larger the value, the more significant the local fluctuations in the project cluster, which require management attention. The management significance of the two comprehensive indices: The comprehensive matching index and the comprehensive dispersion index of the project cluster will reduce the dispersion of... The information from each project is compressed into two macro-level indicators, enabling management to quickly assess the overall matching status of the four key performance indicators (KPIs) across the entire project portfolio without having to review the evaluation results of each individual project. When the composite index reaches the warning threshold set in step S5, it indicates that there is a risk of mismatch in the overall project portfolio. At this point, the management can drill down from the macro to the micro level to pinpoint specific problem projects and deviations, thereby achieving a complete management loop of macro-level situation awareness, meso-level project identification, and micro-level root cause tracing. (Refer to...) Figure 3 As shown.

[0052] The role of step S4 in the overall technical solution: Step S4 is the key step in this solution's leap from project-level assessment to project group-level assessment. Existing technologies can only monitor the four rates of a single project, while this step, by constructing a project group-level assessment matrix, introducing weight coefficients, and calculating a comprehensive index, expands the assessment scale from a single project to the entire project group, filling the gap in existing technologies at the project group management level. The project group comprehensive matching degree index and comprehensive dispersion index output by this step serve as the direct judgment basis for project group-level early warning in step S5, making the early warning system more complete at the logical level. It supports both refined early warning for a single project and macro-level early warning for the entire project group, achieving multi-level coverage from micro to macro, and significantly improving the applicability and flexibility of this solution in different management scenarios.

[0053] Step S5: Set the matching index threshold and the dispersion index threshold ,when < or < When a matching error warning is triggered, > Stability warnings are triggered in a timely manner. Specifically, after completing the calculation of the matching degree index and dispersion index at the project level and project group level, a scientific early warning judgment mechanism needs to be established to transform quantitative indicators into actionable early warning signals, promptly detect anomalies in the four ratios, and notify management personnel to intervene. This step achieves automatic identification and hierarchical output of matching anomalies by setting reasonable early warning thresholds and constructing a hierarchical and classified early warning rule system. Method for setting the warning threshold: The warning threshold is the dividing line for determining whether the four-rate matching status is normal. The rationality of its setting directly determines the accuracy and practicality of the warning. This step adopts a dynamic threshold setting method. Compared with the existing technology that uses fixed empirical values, such as directly setting the deviation of each rate to not exceed 5%, the dynamic threshold can adapt to the actual management characteristics of different regions, different periods, and different types of projects, and has stronger adaptability and accuracy. The setting of the matching degree index threshold is based on the statistical distribution of the four-rate matching degree index of historical completed projects of the same region, type, and scale. Specifically, the time series data of the four rates of multiple historical projects in the power grid infrastructure field with the same type and similar voltage level as the project to be evaluated during the normal progress stage are collected, and their matching degree indices are calculated to form a reference distribution. The lower quartile or the quantile at a certain confidence level of this distribution is used as the benchmark value of the matching degree index threshold. The threshold can be fine-tuned based on actual management practices. If the current period is a peak construction season, such as the third quarter of each year, and the control requirements are high, the threshold can be appropriately increased to improve the sensitivity of the early warning. If the current period is a low season for construction, the threshold can be appropriately decreased to avoid excessive warnings. The setting of the dispersion index threshold adopts a similar method: based on the statistical distribution of the dispersion index of similar historical projects under normal progress, the upper quartile or the quantile at a certain confidence level is used as the benchmark value of the dispersion index threshold, and then fine-tuned based on the management requirements of the current stage. In addition, the same project faces different progress control priorities at different construction stages, such as the preliminary preparation stage, civil construction stage, equipment installation stage, and commissioning stage. Different types of projects, such as substation projects, line projects, and distribution network projects, also have different progress coordination modes. Therefore, the early warning threshold can be set according to the project type and construction stage, rather than using a uniform threshold for all projects.After setting the thresholds, the following rules will be used to automatically trigger warnings: The first type is a project-level matching anomaly warning. When the matching index of a project falls below the set matching index threshold, it indicates that the temporal correlation between the four types of progress completion rates for that project is weak, and the consistency of the trends of the four rates is insufficient, triggering a project-level warning. This warning points to the specific responsible project, prompting the relevant department managing the project to pay close attention to the matching status of the four rates. The second type is a project group-level matching anomaly warning. When the overall matching index of the project group falls below the matching index threshold, it indicates that the overall matching status of the four rates in the project group is poor. This is not a problem of individual projects, but rather a widespread matching deviation across multiple projects, triggering a warning. The first type of warning is a project-wide alert, which prompts management to conduct a systematic review of the overall management model or external environmental factors of the project group, rather than addressing only individual projects. The second type is a stability warning, which is triggered when the dispersion index of a project exceeds the dispersion index threshold. This indicates that the four types of progress completion rates of the project fluctuate significantly over time, resulting in insufficient coordination and stability. This stability warning can be triggered independently or simultaneously with the project-level matching anomaly warning. When a project triggers both the matching anomaly warning and the stability warning, it means that the project not only has inconsistent trends in the four rates, but also experiences severe fluctuations and lacks coordination, which is the most serious abnormal state and requires priority handling. Based on the severity of the warnings, they are divided into different levels. When a project triggers only a stability warning but not a matching anomaly warning, it is judged as a general warning, indicating that there is some inconsistency in the fluctuation of the four rates but the trend is still acceptable, and it is recommended to pay attention. When a project triggers only a matching anomaly warning but not a stability warning, it is judged as a relatively serious warning, indicating that the trend of the four rates has deviated and timely intervention is required. When a project triggers two types of warnings at the same time, it is judged as a serious warning, indicating that the overall matching status of the four rates is out of control and an immediate investigation and rectification is required. Project group-level warnings, due to their macro-level implications, are uniformly judged as important warnings, prompting management to conduct a systematic analysis. After each type of warning is triggered, it is sent to the corresponding management personnel and business departments according to the preset push rules. At the same time, the warning information, including the warning project name, warning type, warning level, trigger time, specific value of the matching index or dispersion index, and degree of deviation, is stored in the system's warning history for subsequent tracking and statistical analysis. The purpose of triggering an early warning is to facilitate subsequent problem handling. To this end, this step provides preliminary operational guidance based on the warning type while issuing the warning: When a matching anomaly warning is triggered, it prompts you to perform root cause analysis according to step S6 to locate the dominant deviation rate and abnormal period; when a stability warning is triggered, it prompts you to focus on checking the sources of fluctuations in the four types of progress data, paying attention to whether there are abnormal data entry, changes in statistical caliber, or sudden, rushed accounting; when two types of warnings are triggered simultaneously, it prompts you to prioritize root cause tracing, and then organize multi-disciplinary joint investigations based on the tracing results. Figure 4 As shown.

[0054] The role of step S5 in the overall technical solution: Step S5 is the decision output layer of the entire evaluation method. The preceding steps S1 to S4 completed the transformation from multi-source data to quantitative evaluation indicators. However, these indicators themselves cannot be directly used by managers. They must be transformed into clear and actionable warning signals through an early warning mechanism. This step directly links the matching degree index and dispersion index with control actions by constructing a dynamic threshold setting method, multi-level early warning rules, and a hierarchical output mechanism. This achieves a complete closed loop of indicator calculation, anomaly identification, early warning output, and handling guidance. Compared with the early warning method that relies on static fixed thresholds in the existing technology, the dynamic threshold setting method in this step makes the early warning more accurate and realistic. The hierarchical early warning mechanism enables managers to quickly distinguish the severity and urgency of problems and rationally allocate handling resources. The post-early warning operation guidance effectively shortens the response time from problem discovery to problem handling, significantly improving the practical application value of the early warning mechanism.

[0055] Step S6: When an alert is triggered, the dominant deviation rate is identified by calculating the matching degree index after removing each rate. Specifically, the method for identifying the dominant deviation rate is as follows: calculate the matching degree index after removing the construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate. ,Will Compared to The rate with the largest increase was identified as the dominant source of deviation; and anomaly periods were located using sliding window analysis, generating a root cause analysis report. Specifically, the method for locating anomaly periods was as follows: [Analysis of the time series data matrix...] Perform sliding window analysis, calculate the matching degree index within each sub-window, and identify the time intervals in which the matching degree significantly decreases. This interval is marked as an abnormal period.

[0056] The value of an early warning mechanism lies not only in identifying problems, but also in assisting managers to quickly pinpoint the root cause and take precise corrective measures. Current technology only outputs a warning signal after an alert is triggered; managers still need to manually verify data from each business process, which is inefficient and prone to overlooking crucial clues. This step automatically initiates root cause analysis after an alert is triggered, precisely locating the deviation from two dimensions: first, identifying which of the four types of rates is the dominant factor causing the decline in matching accuracy, and identifying the problematic factor; second, pinpointing when the matching accuracy began to deviate significantly on the timeline, and when the problem started, thus providing managers with clear and specific directions for investigation. Identification of the dominant deviation rate: When a project triggers a matching anomaly alert, a key question needs to be answered further: among the four types of progress completion rates, which rate anomaly caused the overall decline in matching accuracy? The core idea of ​​identifying the dominant deviation rate is to use the leave-one-out comparative analysis logic: First, after removing one type of rate, reassess the matching degree of the remaining three types of rates, comparing the changes in the matching degree index after each type of rate is removed. Specifically, remove the columns for construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate from the project's four-rate time-series data matrix. Recalculate the matching degree index for the remaining three types of rates, obtaining four new matching degree indices. Compare these four new matching degree indices with the original matching degree indices, calculating their respective increases. The rate type that causes the largest increase in the matching degree index is the dominant deviation rate that caused the original matching degree to decrease. The principle behind the identification logic is as follows: if removing a certain type of rate significantly improves the matching degree of the remaining three types of rates, it indicates that the removed rate deviates severely from the time-series trend of the other three types of rates, and is a key factor lowering the overall matching degree. For example, if removing the accounting progress completion rate significantly increases the matching degree index of the remaining three types of rates (construction, investment, and materials), it indicates that the accounting progress completion rate has the greatest difference in trend from the other three types of rates, and the financial accounting stage is the dominant source of deviation. Through this analysis, managers can directly pinpoint the responsible link of the problem—whether it is inadequate construction progress execution, delayed financial accounting, or a disconnect in material supply—thus avoiding the inefficient operation of checking each business department one by one. Locating abnormal periods: After identifying the dominant deviation rate, another key question needs to be answered: When did this deviation begin? The purpose of identifying anomalous time periods is to help managers focus their attention on specific timeframes, review management decisions, external environmental changes, or data recordings within those periods, and thus accurately identify the specific events or operations that caused the deviations. The identification of anomalous time periods employs a sliding window analysis method, the basic operation of which involves: within the time window covered by the time series data matrix... Within a fixed-length sub-window, for example containing three time sampling points, the system gradually slides along the time axis from the start time to the end time. At each time step, the four-rate matching index within the current sub-window is calculated, thus obtaining the continuous change trajectory of the matching index on the time axis. The matching index of each sub-window is then compared with the normal reference value, or the entire time window can be used. The average matching index or a preset threshold is compared to identify a continuous sub-window sequence in which the matching index continues to decline and falls below the normal reference value. The time interval covered by this sequence is marked as an abnormal period. This analysis can accurately reveal when the deviation began to accumulate: if the abnormal period begins in a specific month, managers can trace back what happened in the project before and after that month, whether the construction team was changed, whether there was a delay in the supply of materials, whether there were any special operations such as centralized financial accounting, etc. At the same time, the length of the abnormal period provides a reference for judging the nature of the problem. Short-term abnormalities may be caused by occasional factors, while long-term continuous abnormalities indicate the existence of systemic problems. Root Cause Analysis Report Generation and Output: After completing the above two analyses, the system automatically integrates the analysis results and generates a structured root cause analysis report. The report includes the following core elements: 1. Warning Trigger Information: the project name that triggered the warning, the warning type, whether it is a project-level matching anomaly warning or a stability warning, and the warning trigger time; 2. Matching Degree Assessment Data: the current matching degree index, the dispersion index, and the degree of deviation from their respective thresholds; 3. Dominant Deviation Rate Identification Results: indicating which type of progress completion rate is the dominant factor causing the decline in matching degree, and providing corresponding suggested investigation directions. If it is the construction progress completion rate, it is recommended to investigate... The report examines deviations between project execution progress and milestone plans, involving the construction management department. If it concerns investment progress completion rate, it recommends checking the investment plan completion status, involving the development planning department. If it concerns accounting progress completion rate, it recommends checking the financial accounting progress, involving the financial management department. If it concerns material progress completion rate, it recommends checking the material supply and requisition status, involving the materials management department. Fourth, it identifies abnormal period location results, indicating the time interval with a significant decrease in matching accuracy, and indicating whether any events or changes that may have caused the deviation occurred before or after this period. Fifth, it provides preliminary corrective action recommendations based on the warning level and root cause analysis results. This report is automatically pushed to the terminals of relevant management personnel in a structured format and simultaneously stored in the system's historical root cause tracing database for subsequent statistical analysis and management decision-making reference. Figure 4 As shown.

[0057] The role of this step in the overall technical solution: Step S6 is one of the core innovative links that distinguishes this solution from existing technologies. It undertakes the key transformation function from problem discovery to problem localization. Existing technologies stop at issuing early warning signals. Managers still need to conduct a lot of blind investigation and reasoning when faced with early warning signals, resulting in long response cycles and low efficiency. This step, through analysis of two dimensions—leading deviation rate identification and abnormal period location—concretizes the abstract matching anomaly early warning into which type of rate has what kind of deviation in what time period. It narrows the scope of problem investigation from four links × the entire project period to one link × one time period, significantly shortening the time from early warning to root cause confirmation. The root cause tracing report output by this step not only provides managers with clear investigation directions and verification suggestions, but also provides data support for subsequent corrective decisions and process optimization. This makes the entire evaluation method form a complete closed loop of data collection, indicator calculation, early warning triggering, root cause tracing, and handling suggestions, significantly improving the practical application value of the early warning mechanism and the efficiency of problem handling.

[0058] The system modules of the evaluation method include:

[0059] The data acquisition module is used to collect raw data on the four rates from multiple business systems. It establishes connections with the project management system, investment planning management system, financial management system, materials management system, and ERP system through standardized data interfaces. According to preset collection frequencies and scopes, it automatically extracts raw data related to the four rates from each business system, including timestamp records and values ​​for construction progress completion rate, investment progress completion rate, accounting progress completion rate, and materials progress completion rate. The effect is that automated interface collection replaces the traditional manual system-by-system export and summarization method, avoiding data omissions and entry errors that may be introduced by manual operation, ensuring the traceability of the four data sources and the efficiency of the collection process. The modular interface design allows for good scalability of new data sources; when new business systems need to be included in the monitoring scope, only the corresponding interface adaptation needs to be added, without modifying the overall architecture.

[0060] The data preprocessing module is used to standardize and time-series align the collected multi-source heterogeneous data. This standardization process includes: converting different timestamp formats from various systems into a standard time coordinate; normalizing and unifying the different calculation methods for progress percentages across systems; identifying and removing obvious outliers and noisy data using conventional data cleaning methods; interpolating and imputing missing values; and finally, aligning all data sources according to a unified time granularity, such as daily, weekly, or monthly, to form a time-comparable standardized four-rate time-series dataset. The effect is that, due to the varying data formats, update frequencies, and quality levels across different business systems, this module eliminates the heterogeneity between multi-source data through standardization, transforming the originally incomparable and unusable raw data into a standardized dataset with a unified structure and reliable quality. This fundamentally ensures the accuracy and comparability of the calculation results from subsequent modules. The time-series alignment process establishes a precise correspondence between the four types of data on the same time coordinate, laying the necessary data foundation for subsequent time-series correlation analysis.

[0061] The matrix construction module is used to construct a four-rate time-series data matrix. For each power grid infrastructure project, within a specified time window, it receives a standardized four-rate time-series dataset output from the data preprocessing module, organizes it according to two dimensions: project number and time sampling point, and arranges the four types of progress completion rate sequences for each project as follows: The module is a two-dimensional time-series data matrix with four rows and four columns. Each row corresponds to a time sampling point, and each column corresponds to a type of progress completion rate. The rows are arranged strictly in chronological order to ensure the correctness of the time sequence relationship. The effect is that this module integrates the four types of progress data, which originally existed as independent sequences, into a unified data structure. This establishes a precise time-dimensional correspondence between the four rates within the same mathematical object, providing a standardized data carrier for all subsequent calculations. The matrix-based data structure facilitates vectorized operations and batch calculations, significantly improving computational efficiency, especially when there are a large number of projects.

[0062] The matching degree calculation module is used to calculate the matching degree index and dispersion index at the project level and project group level. Based on the four-rate time series data matrix output by the matrix construction module, it first calculates the time series correlation coefficient between each pair of rates, and then takes the arithmetic mean of the six correlation coefficients as the matching degree index of the project. At the same time, it calculates the overall fluctuation dispersion of the four types of progress completion rates relative to their respective historical averages at each time sampling point, as the dispersion index of the project. For project group scenarios, this module further receives the weight coefficients of each project and uses a weighted average method to combine the matching degree index and dispersion index of each project. The data converges into a comprehensive matching index and a comprehensive dispersion index for the project group. The effect is that this module extracts the complex and high-dimensional temporal correlation between the four rates into two concise quantitative indicators: the matching index and the dispersion index. This completes the transformation from raw data to evaluable information, clearly quantifying the synergistic state of the four rates that was originally hidden in the data. The matching index characterizes the matching status of the four rates from the perspective of trend consistency, while the dispersion index supplements this characterization from the perspective of fluctuation synergy. The two indicators corroborate each other, forming a more comprehensive and objective evaluation system than a single indicator, providing a reliable quantitative basis for early warning judgments.

[0063] The early warning judgment module is used to trigger matching anomaly warnings and stability warnings based on preset thresholds. It receives the matching degree index and dispersion index of each project and project group from the matching degree calculation module, compares them with preset matching degree index thresholds and dispersion index thresholds, and triggers a project-level matching anomaly warning when the matching degree index of a project is lower than the threshold; a project group-level matching anomaly warning when the overall matching degree index of a project group is lower than the threshold; and a stability warning when the dispersion index of a project is higher than the threshold. Simultaneously, it automatically determines the warning level (general warning, moderate warning, severe warning, or important warning) based on the combination of warning trigger types, and pushes the warning information along with preliminary operational guidance to the corresponding management personnel. Effects: This module transforms quantified matching degree indicators into intuitive warning signals, achieving a key leap from data calculation to management decision-making. The multi-level (project-level and project group-level) and multi-type warning rule system for matching anomaly warnings and stability warnings enables managers to accurately judge the nature, scope, and severity of problems. Clear warning level classifications help managers rationally allocate resources, prioritize serious issues, and significantly improve problem response efficiency.

[0064] The root cause analysis module identifies the dominant deviation rate and locates abnormal periods, outputting a root cause analysis report. When the early warning judgment module triggers a matching anomaly warning, root cause analysis is automatically initiated. First, using a leave-one-out comparative analysis, the matching degree index is recalculated after removing the four types of rates to identify the dominant deviation rate causing the decrease in matching degree, clarifying the specific business process to which the problem points—construction, investment, accounting, or materials. Then, using a sliding window analysis method, the trajectory of the matching degree index change within each consecutive sub-window is calculated along the time axis to identify the continuous interval of continuous decline in matching degree, pinpointing the abnormal period. Finally, the above analysis is... The results are integrated into a structured root cause analysis report and pushed to relevant management personnel. The effect: This module elevates early warning from simply identifying problems to locating them, transforming abstract warning signals into concrete, actionable investigation instructions, filling the gap in intelligent analysis following existing early warning technologies. Leave-one-out comparative analysis narrows the problem investigation scope from four stages to one, while sliding window analysis reduces the investigation timeframe from the entire project period to a specific time period. The combination of these two methods significantly improves investigation efficiency, allowing management personnel to directly conduct targeted verification based on the report, greatly shortening the response cycle from early warning to problem handling.

[0065] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0066] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion, characterized in that, Includes the following steps: Step S1: Collect raw data on construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate from multiple business systems of the power grid infrastructure project, and perform standardization and time-series alignment on the raw data to form a standardized four-rate time-series dataset. Step S2: For the first Individual power grid infrastructure projects, including Time window of each time sampling point Internally, construct a four-rate time series data matrix. ; Step S3: Based on the time series data matrix Calculate the first The four-rate matching index of each project and dispersion index ; Step S4: For a project group containing N projects, construct a project group-level matching evaluation matrix. And calculate the comprehensive matching index of the project group. Comprehensive dispersion index of project groups ; Step S5: Set the matching index threshold and the dispersion index threshold ,when < or < When a matching error warning is triggered, > A stability warning is triggered at any time; Step S6: When an alert is triggered, the dominant deviation rate is identified by calculating the matching degree index after removing each rate, and the abnormal period is located by analyzing the sliding window, and a root cause tracing report is output.

2. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The multiple business systems in step S1 include a project management system, an investment planning management system, a financial management system, and a materials management system. The standardization process includes unifying the timestamp format, unifying the calculation method for progress percentage, and removing outliers and handling missing values.

3. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The four-rate time-series data matrix in step S2 Represented as: ,in, Indicates the first The project in the Construction progress completion rate of each time sampling point; Indicates the investment progress completion rate; Indicates the completion rate of the accounting entry process; This indicates the completion rate of material delivery.

4. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The four-rate matching index in step S3 The calculation method is as follows: First, calculate the Pearson correlation coefficient between any two rates. Then calculate ; where subscript , , , These represent the completion rates of construction progress, investment progress, accounting entries, and materials progress, respectively.

5. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The four-rate dispersion index in step S3 The calculation method is as follows: ;in, , , and The first The average completion rate of the four types of progress for each project within the time window T.

6. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The project cluster-level matching evaluation matrix in step S4 Represented as: ;in, For the first The weighting coefficients for each project are determined comprehensively based on the project's investment scale, construction period, and importance level.

7. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The project cluster comprehensive matching index in step S4 The calculation method is as follows: ; Project Group Comprehensive Dispersion Index The calculation method is as follows: .

8. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The method for identifying the dominant deviation rate in step S6 is as follows: using the leave-one-out method, calculate the matching degree index after removing the construction progress completion rate, investment progress completion rate, accounting progress completion rate, and material progress completion rate, respectively. ,Will Compared to The rate with the largest increase was identified as the dominant source of deviation.

9. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The method for locating abnormal time periods in step S6 is as follows: [Analyze the time series data matrix...] Perform sliding window analysis, calculate the matching degree index within each sub-window, and identify the time intervals in which the matching degree significantly decreases. This interval is marked as an abnormal period.

10. The method for evaluating the progress matching of power grid infrastructure projects based on multi-source data fusion according to claim 1, characterized in that: The system modules of the evaluation method include: The data acquisition module is used to collect raw data on the four rates from multiple business systems; The data preprocessing module is used to standardize and time-series align the collected multi-source heterogeneous data. The matrix construction module is used to construct a four-rate time series data matrix; The matching degree calculation module is used to calculate the matching degree index and dispersion index at the project level and project group level. The early warning judgment module is used to trigger matching anomaly warnings and stability warnings based on preset thresholds; The root cause tracing module is used to identify the dominant deviation rate and locate abnormal periods, and output a root cause tracing report.

Images