Risk assessment method for power infrastructure projects based on multi-factor dynamic weighting algorithm

Abstract

The application discloses a power infrastructure project risk assessment method based on a multi-factor dynamic weighting algorithm, and belongs to the technical field of project management. The steps include: filtering noise from original business flow data, normalizing and compressing the filtered data fields to obtain a standardized data set; stripping factors from the standardized data set to obtain an original risk factor spectrum; assigning weights to the original risk factor spectrum to obtain an initial weighted factor set; calling a weight correction coefficient matrix in a preset benchmark project database, and dynamically and iteratively calibrating the initial weight values to obtain a stage weight vector; weighting and reconstructing the initial weighted factor set to obtain a risk feature vector; matching the risk feature vector with a fuzzy membership degree to obtain an assessment conclusion; and standardizing and packaging the assessment conclusion, the risk feature vector and the stage weight vector to obtain a risk assessment data message. The application can improve the accuracy and efficiency of power infrastructure project risk assessment.

Description

Technical Field

[0001] This invention belongs to the field of project management technology, and in particular relates to a risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm. Background Technology

[0002] Power infrastructure projects are characterized by large investment scale, long construction period, many construction stages and wide range of professional involvement. Throughout the project life cycle, they are prone to various potential risks such as project delays, cost overruns, equipment malfunctions, construction safety accidents, and policy compliance changes. Accurate risk assessment is the core link to ensure the smooth progress of the project and achieve effective risk management. Therefore, it puts forward high requirements for the technical methods of risk assessment.

[0003] The current technical implementation of risk assessment for power infrastructure projects still has some shortcomings. For example, only basic screening operations are performed on the original business flow data. There is no targeted filtering of noise in the data, nor is there any standardization and uniform adjustment of numerical fields with different dimensions. This leads to data distortion and dimensional confusion in the dataset used for subsequent risk assessment, which cannot provide reliable data support for the accurate extraction of risk factors and directly lowers the quality of the basic data for risk assessment. At the same time, the existing technology uses a fixed weight allocation method for risk factors without making dynamic adjustments based on the actual situation at different stages of the project. This ignores the differences in the degree of influence of various risk factors at each stage of the project, resulting in a mismatch between the weight values ​​and the actual risk status of the project, making it difficult to truly reflect the core risk characteristics of each stage.

[0004] Existing assessment methods also have significant shortcomings in the risk feature extraction and assessment result output stages. Risk feature extraction relies solely on a single factor analysis method without in-depth reconstruction and feature enhancement of the weighted risk factors. This results in the extracted risk feature vectors failing to comprehensively and accurately cover various types of project risk information, and the completeness and specificity of feature representation are insufficient, making it difficult to support accurate determination of subsequent risk levels. Furthermore, the output format of risk assessment results is relatively simple, merely presenting the risk level without systematically integrating and encapsulating the assessment conclusions with core data such as risk feature vectors and stage weight vectors. This leads to poor data transferability and reusability, failing to provide a systematic and traceable data analysis basis for risk management in power infrastructure projects, and also failing to support dynamic monitoring, real-time adjustment, and control decision-making for subsequent project risks. Consequently, it cannot meet the actual needs of refined risk management in power infrastructure projects. Summary of the Invention

[0005] To address the shortcomings of the existing technologies, the present invention aims to provide a risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm. This method, for power infrastructure project risk assessment, refines data processing, combines dynamic weighting of risk factors across project milestones, accurately determines risk levels, and standardizes and encapsulates assessment data, significantly improving the accuracy and efficiency of the assessment and providing complete and traceable scientific data for project risk management.

[0006] To achieve the above objectives, the present invention provides a risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm, comprising:

[0007] S1. Noise filtering is performed on the original business flow data of the power infrastructure project to be evaluated, and the filtered data fields are normalized and compressed to obtain a standardized dataset of the power infrastructure project to be evaluated.

[0008] S2. Factor stripping is performed on the standardized dataset to obtain the original risk factor spectrum of the power infrastructure project to be evaluated;

[0009] S3. Based on the standard deviation and coefficient of variation of time series data in the standardized dataset, weights are assigned to the original risk factor spectrum to obtain the initial weighted factor set of the power infrastructure project to be evaluated.

[0010] S4. Based on the mileage stage of the power infrastructure project to be evaluated, retrieve the weight correction coefficient matrix from the preset benchmark project database, and dynamically iterate and calibrate the initial weight values ​​of the power infrastructure project to be evaluated to obtain the stage weight vector of the power infrastructure project to be evaluated.

[0011] S5. Based on the stage weight vector, the initial weighting factor set is reconstructed to obtain the risk characteristic vector of the power infrastructure project to be evaluated.

[0012] S6. Perform fuzzy membership matching on the risk feature vector to obtain the evaluation conclusion of the power infrastructure project to be evaluated.

[0013] S7. Standardize and encapsulate the assessment conclusions, risk feature vectors, and stage weight vectors to obtain the risk assessment data message for the power infrastructure project to be assessed.

[0014] Preferably, in step S1, the process of obtaining the standardized dataset of the power infrastructure project to be evaluated is as follows:

[0015] Signal cleaning is performed on the original business flow data of the power infrastructure project to be evaluated to obtain the cleaned business flow data of the power infrastructure project to be evaluated;

[0016] Based on the average data of adjacent time windows in the cleaning business flow data, numerical values ​​are filled into the cleaning business flow data to obtain the complete business flow data of the power infrastructure project to be evaluated.

[0017] Dimensionless scaling is performed on numerical fields with inconsistent dimensions in the complete business flow data to obtain compressed data fields of the power infrastructure project to be evaluated.

[0018] Based on the data source of the original business flow data, the compressed data fields are classified and merged to obtain a standardized dataset of the power infrastructure projects to be evaluated.

[0019] Preferably, in step S2, the process of obtaining the original risk factor spectrum of the power infrastructure project to be evaluated is as follows:

[0020] Business semantic parsing is performed on the standardized dataset to obtain a cluster of risk indicator fields for the power infrastructure projects to be evaluated.

[0021] Based on the preset dimension classification labels, the risk indicator field clusters are classified and grouped to obtain the original factor group of the power infrastructure project to be evaluated.

[0022] By binding access identifiers to the risk factors in the original factor group, the original risk factor spectrum of the power infrastructure project to be evaluated is obtained.

[0023] Preferably, in step S3, the process of obtaining the initial weighting factor set for the power infrastructure project to be evaluated is as follows:

[0024] Based on the original risk factor spectrum, time series data segments of the original risk factor spectrum are extracted from the standardized dataset to obtain time series data samples of the power infrastructure projects to be evaluated.

[0025] The fluctuation characteristics of the standard deviation and coefficient of variation of the time series data samples are analyzed to obtain the fluctuation characteristic parameters of the power infrastructure project to be evaluated.

[0026] Based on the volatility characteristic parameter, initial weight coefficients are assigned to the risk factors in the original risk factor spectrum, and the initial weight coefficients are associated and bound with the risk factors to obtain the initial weight factor set of the power infrastructure project to be evaluated.

[0027] Preferably, in step S4, the process of obtaining the stage weight vector of the power infrastructure project to be evaluated is as follows:

[0028] Real-time capture of mileage node trigger signals for power infrastructure projects to be evaluated;

[0029] The mileage node trigger signal is parsed using stage coding to obtain the current mileage stage code of the power infrastructure project to be evaluated.

[0030] Based on the current mileage stage coding, a stage matching search is performed on the preset benchmark project database to obtain the correction coefficient matrix of the power infrastructure projects to be evaluated.

[0031] The correction coefficient matrix is ​​iteratively adjusted item by item with the initial weight values ​​in the initial weighting factor set of the power infrastructure project to be evaluated to obtain the stage weight vector of the power infrastructure project to be evaluated.

[0032] Preferably, the step of iteratively adjusting the correction coefficient matrix and the initial weight values ​​in the initial weighting factor set of the power infrastructure project to be evaluated specifically involves:

[0033] By stripping the coefficients from the correction coefficient matrix, the stage disturbance factor of the power infrastructure project to be evaluated is obtained.

[0034] Based on the stage disturbance factor, the initial weight values ​​in the initial weight factor set are multiplied and fused item by item to obtain the iterative intermediate weight parameters of the power infrastructure project to be evaluated.

[0035] The intermediate weight parameters of the iteration are normalized and constrained for verification, and the intermediate weight parameters of the iteration are proportionally redistributed according to the measured residuals to obtain the stage weight components of the power infrastructure project to be evaluated.

[0036] Based on the order of risk factors in the initial weighting factor set, the stage weight components are vectorized and encapsulated to obtain the stage weight vector of the power infrastructure project to be evaluated.

[0037] Preferably, in step S5, the process of obtaining the risk feature vector of the power infrastructure project to be evaluated is as follows:

[0038] Using the stage weight vector as the modulation signal, the initial weighting factor set is carrier modulated to obtain the risk factor sequence of the power infrastructure project to be evaluated.

[0039] The modulated risk factor sequence is coherently demodulated to obtain the enhanced risk factor sequence of the power infrastructure project to be evaluated;

[0040] By embedding the enhanced risk factors into a manifold, the risk feature vector of the power infrastructure project to be evaluated is obtained.

[0041] Preferably, in step S6, the process of obtaining the evaluation conclusion of the power infrastructure project to be evaluated is as follows:

[0042] Extract risk level cluster centers from the historical completed project database of power infrastructure projects to be evaluated;

[0043] Based on the risk level cluster center points, the spatial distance of the risk feature vectors is determined to obtain the Euclidean distance value of the power infrastructure project to be evaluated.

[0044] Based on the preset membership mapping relationship, the membership values ​​of the Euclidean distance are assigned to obtain the membership value group of the power infrastructure project to be evaluated;

[0045] Extreme value retrieval is performed on the membership value group to capture the risk level pointed to by the maximum membership score in the membership value group, and the risk level is used as the risk level label of the power infrastructure project to be evaluated.

[0046] By associating and binding risk level labels with risk feature vectors, an assessment conclusion with risk level labels is obtained for the power infrastructure project to be evaluated.

[0047] Preferably, the method of obtaining the membership value group of the power infrastructure project to be evaluated specifically includes:

[0048] Extract distance scale parameters and benchmark membership correction coefficients from the historical completed project database;

[0049] The initial membership value of the power infrastructure project to be evaluated is calculated based on the Euclidean distance value, distance scale parameter, and benchmark membership correction coefficient. The formula for calculating the initial membership value is as follows:

[0050] ;

[0051] In the formula, This is the initial membership value. This represents the Euclidean distance between the risk feature vector and the cluster center of the k-th risk level. For the k-th distance scale parameter, This is the baseline membership correction coefficient for the k-th risk level;

[0052] The initial membership values ​​are balanced and allocated to obtain a set of membership values ​​for the power infrastructure projects to be evaluated.

[0053] Preferably, in step S7, the process of obtaining the risk assessment data message of the power infrastructure project to be evaluated is as follows:

[0054] The assessment conclusions are decomposed in a structured manner to obtain the assessment conclusion metadata of the power infrastructure project to be assessed;

[0055] The assessment conclusion metadata is associated and assembled with the risk feature vector to obtain the assessment data carrier of the power infrastructure project to be assessed.

[0056] The assessment data carrier and the stage weight vector are merged and packaged to obtain the risk assessment data message of the power infrastructure project to be assessed.

[0057] The present invention has the following beneficial effects:

[0058] This invention performs refined processing of the original business flow data of power infrastructure projects throughout the entire process. It forms a standardized dataset through noise filtering and normalization compression, and then obtains an accurate original risk factor spectrum through factor stripping. Based on the feature analysis of time series data, it completes scientific weight allocation. At the same time, it realizes dynamic iterative calibration of weight values ​​in combination with project milestone stages, so that the weight settings of risk factors are highly consistent with the actual implementation status of the project. This greatly improves the accuracy and adaptability of risk factor weighting, and strengthens the underlying data support and scientific nature of factor analysis for risk assessment of power infrastructure projects.

[0059] This invention reconstructs the initial weighting factor set based on stage weight vectors, forming a risk feature vector that comprehensively characterizes project risks. It achieves accurate risk level determination through fuzzy membership matching. Furthermore, it standardizes and encapsulates the assessment conclusions, risk feature vectors, and stage weight vectors to obtain risk assessment data messages. This not only improves the accuracy and comprehensiveness of risk level assessment for power infrastructure projects but also creates a systematic and structured carrier for risk assessment data, enhancing the transmissibility and reusability of the assessment data. It provides complete and traceable data analysis basis for the subsequent implementation of risk management in power infrastructure projects, comprehensively improving the overall efficiency and application value of risk assessment for power infrastructure projects. Attached Figure Description

[0060] Figure 1 This is a schematic flowchart of the method of the present invention;

[0061] Figure 2 This is a comparison chart of risk level membership curves in Embodiment 1 of the present invention. Detailed Implementation

[0062] Example 1: As Figure 1 As shown, the risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm includes the following steps:

[0063] S1. Noise filtering is performed on the original business flow data of the power infrastructure project to be evaluated, and the filtered data fields are normalized and compressed to obtain a standardized dataset of the power infrastructure project to be evaluated.

[0064] S2. Factor stripping is performed on the standardized dataset to obtain the original risk factor spectrum of the power infrastructure project to be evaluated;

[0065] S3. Based on the standard deviation and coefficient of variation of time series data in the standardized dataset, weights are assigned to the original risk factor spectrum to obtain the initial weighted factor set of the power infrastructure project to be evaluated.

[0066] S4. Based on the mileage stage of the power infrastructure project to be evaluated, retrieve the weight correction coefficient matrix from the preset benchmark project database, and dynamically iterate and calibrate the initial weight values ​​of the power infrastructure project to be evaluated to obtain the stage weight vector of the power infrastructure project to be evaluated.

[0067] S5. Based on the stage weight vector, the initial weighting factor set is reconstructed to obtain the risk characteristic vector of the power infrastructure project to be evaluated.

[0068] S6. Perform fuzzy membership matching on the risk feature vector to obtain the evaluation conclusion of the power infrastructure project to be evaluated.

[0069] S7. Standardize and encapsulate the assessment conclusions, risk feature vectors, and stage weight vectors to obtain the risk assessment data message for the power infrastructure project to be assessed.

[0070] In S1, the process of obtaining the standardized dataset of the power infrastructure projects to be evaluated is as follows:

[0071] Signal cleaning is performed on the original business flow data of the power infrastructure project to be evaluated to obtain the cleaned business flow data of the power infrastructure project to be evaluated;

[0072] Based on the average data of adjacent time windows in the cleaning business flow data, numerical values ​​are filled into the cleaning business flow data to obtain the complete business flow data of the power infrastructure project to be evaluated.

[0073] Dimensionless scaling is performed on numerical fields with inconsistent dimensions in the complete business flow data to obtain compressed data fields of the power infrastructure project to be evaluated.

[0074] Based on the data source of the original business flow data, the compressed data fields are classified and merged to obtain a standardized dataset of the power infrastructure projects to be evaluated.

[0075] Anomalies in the original business flow data are identified and removed field by field and value by value. The criteria for identifying anomalies are set as follows: first, values ​​that exceed the preset threshold range of the power infrastructure project business data. The preset threshold range is defined based on the actual data fluctuation range of each business link of the power infrastructure project; second, discrete data that does not conform to the logic of the data collection time series, that is, isolated values ​​that deviate from the normal data change trend in the same time series and have no reasonable business cause. After completing the identification, marking and removal of all anomalies in the original business flow data, the cleaned business flow data of the power infrastructure project to be evaluated is obtained.

[0076] According to the preset time window division rules for data collection in power infrastructure projects, the cleaning business flow data is divided into multiple time windows of equal duration. The effective data values ​​in each time window are statistically analyzed and summarized one by one, and then the arithmetic mean of the effective data values ​​in each time window is calculated. For the missing data positions in each field of the cleaning business flow data, the adjacent time window before and after the time window to which it belongs is accurately located. The arithmetic mean of the effective data calculated from these two adjacent time windows is accurately filled into the corresponding missing data positions. After completing the numerical filling operation for all missing positions in the cleaning business flow data, the complete business flow data of the power infrastructure project to be evaluated is obtained.

[0077] According to the pre-set dimension conversion rules for power infrastructure projects, a unified dimensionless scaling process is carried out on all numerical fields with inconsistent dimensions in the complete business flow data. The core operation of the scaling process is to map the values ​​of different dimensions to a fixed numerical range of 0 to 1 through a linear transformation. During the transformation process, the relative size relationship of each value in the original field is preserved. After completing the scaling operation of all fields with inconsistent dimensions in the complete business flow data, the compressed data field of the power infrastructure project to be evaluated is obtained.

[0078] A comprehensive review of all data acquisition terminals corresponding to the original business flow data of the power infrastructure project to be evaluated was conducted. At the same time, the pre-set business data type classification standard of the power infrastructure project was retrieved. This standard divides the data into fixed types such as project progress, cost input, equipment status, and construction safety according to the project implementation stage. The compressed data fields were classified in a dual manner according to the source of the original business flow data acquisition terminal and the pre-set data type classification standard. Compressed data fields belonging to the same acquisition terminal and the same data type were systematically integrated and summarized, maintaining the time sequence order of the data within the fields. After completing all classification and integration operations of the compressed data fields, a standardized dataset of the power infrastructure project to be evaluated was obtained.

[0079] Signal cleaning of raw business flow data can effectively remove abnormal and discrete information, ensuring the basic validity of the data. Numerical imputation based on the average of adjacent time windows can fill in missing content in the data, allowing the business flow data to form a complete information system. Dimensionless scaling of numerical fields with inconsistent dimensions can achieve unified processing of different types of data, eliminating data analysis interference caused by differences in dimensions. Classifying and merging compressed data fields according to the data source of the raw business flow data can form a standardized dataset with regular and categorized characteristics. Overall, it realizes refined processing of raw business flow data of power infrastructure projects throughout the entire process, improving the integrity, standardization, and validity of the data. This lays a solid and accurate data foundation for subsequent data processing and analysis related to risk assessment of power infrastructure projects, ensuring the smooth implementation and accurate results of subsequent risk factor stripping, weight allocation, and other processes.

[0080] In S2, the process of obtaining the original risk factor spectrum of the power infrastructure project to be evaluated is as follows:

[0081] Business semantic parsing is performed on the standardized dataset to obtain a cluster of risk indicator fields for the power infrastructure projects to be evaluated.

[0082] Based on the preset dimension classification labels, the risk indicator field clusters are classified and grouped to obtain the original factor group of the power infrastructure project to be evaluated.

[0083] By binding access identifiers to the risk factors in the original factor group, the original risk factor spectrum of the power infrastructure project to be evaluated is obtained.

[0084] Based on the clearly defined business attributes of risk assessment for power infrastructure projects, this definition covers the business scope and data attribute definition standards related to various risk assessments throughout the entire implementation process of power infrastructure projects. For each data field in the standardized dataset, business semantic matching and identification operations are performed sequentially to determine whether each field belongs to the business scope of risk assessment for power infrastructure projects. All data fields directly related to risk assessment for power infrastructure projects are accurately screened out. All screened risk-related data fields are then systematically organized and integrated according to data type to form a complete set of risk-related data fields, resulting in the risk indicator field cluster of the power infrastructure project to be assessed.

[0085] The system retrieves a pre-defined risk dimension classification tag library for power infrastructure projects. This library establishes a complete classification system based on the implementation stage and risk type of the power infrastructure project. It includes exclusive classification tags for each risk dimension throughout the entire process, including project planning, construction, and acceptance. Each classification tag has a clear business semantic definition and data field matching standard. Each data field in the risk indicator field cluster is semantically matched according to the classification tags in the tag library. The risk dimension classification tag to which each data field belongs is determined strictly according to the matching standard. All risk indicator fields that match the same classification tag are centrally integrated to form a clearly categorized set of risk factors according to the tag system, thus obtaining the original factor group of the power infrastructure project to be evaluated.

[0086] According to the preset coding rules for risk factor management of power infrastructure projects, the rules adopt a fixed-digit combination coding form, which includes coding dimensions such as risk dimension, factor category, and sequence number. Each risk factor in the original factor group is assigned a unique and non-repeating access identifier code. The coding content is associated with the dimension and category to which the risk factor belongs. The assigned access identifier code is linked and bound one-to-one with the corresponding risk factor. The access identifier code is used as the exclusive identification information of the risk factor and is annotated. After completing the binding and annotation operation of all risk factors in the original factor group with the corresponding access identifier codes, the original risk factor spectrum of the power infrastructure project to be evaluated is obtained.

[0087] Business semantic parsing of standardized datasets can accurately filter out field information related to risk assessment of power infrastructure projects, eliminate irrelevant data interference to ensure the relevance of risk-related data, and classify and collect risk indicator field clusters based on preset dimension classification labels to achieve systematic sorting of risk factors, forming a clear classification system for risk-related data. Access identification binding of risk factors in the original factor group can assign exclusive and unique identification information to each risk factor, enabling accurate location and rapid retrieval of risk factors. The whole process completes the accurate separation from the standardized dataset to the original risk factor spectrum, making the risk factors of power infrastructure projects present structured, systematic, and identifiable characteristics. This provides a clear, standardized, and accurate factor foundation for subsequent risk factor weight allocation, ensuring that the subsequent weight allocation process can be carried out in an orderly manner around effective risk factors, and improving the overall efficiency of risk factor analysis and processing.

[0088] In S3, the process of obtaining the initial weighting factor set for the power infrastructure project to be evaluated is as follows:

[0089] Based on the original risk factor spectrum, time series data segments of the original risk factor spectrum are extracted from the standardized dataset to obtain time series data samples of the power infrastructure projects to be evaluated.

[0090] The fluctuation characteristics of the standard deviation and coefficient of variation of the time series data samples are analyzed to obtain the fluctuation characteristic parameters of the power infrastructure project to be evaluated.

[0091] Based on the volatility characteristic parameter, initial weight coefficients are assigned to the risk factors in the original risk factor spectrum, and the initial weight coefficients are associated and bound with the risk factors to obtain the initial weight factor set of the power infrastructure project to be evaluated.

[0092] By referring to the unique access identifier code corresponding to each risk factor in the original risk factor spectrum, and using this identifier code as the retrieval basis, a precise search and location is performed in the standardized dataset. All time series data corresponding to each risk factor throughout the entire project implementation process are searched and extracted one by one. According to the preset time slicing rules, the effective data segments related to risk assessment in the time series data corresponding to each risk factor are extracted with the key business nodes of the project as the dividing points. Redundant data segments that are not related to risk feature analysis are removed. All effective data segments corresponding to all risk factors are uniformly integrated and summarized according to the arrangement order of the original risk factor spectrum to obtain the time series data sample of the power infrastructure project to be evaluated.

[0093] For each risk factor in the time series data sample, the effective data segments corresponding to each risk factor are analyzed independently. First, all values ​​within each data segment are statistically calculated to obtain the standard deviation, which can intuitively reflect the dispersion of the values ​​within the data segment. Then, based on the standard deviation, the mean of the corresponding data segment is combined to calculate the coefficient of variation, which can accurately reflect the relative fluctuation of the values ​​within the data segment. The standard deviation and coefficient of variation for each risk factor are combined and organized to extract the core feature information that can completely characterize the data fluctuation state of the risk factor. The core feature information of the fluctuation of all risk factors is integrated into a unified and standardized feature set to obtain the fluctuation feature parameters of the power infrastructure project to be evaluated.

[0094] Based on the pre-defined mapping rules for volatility characteristics and weight coefficients in the risk assessment of power infrastructure projects, these rules clearly define the weight coefficient ranges corresponding to different degrees of dispersion and relative volatility. The volatility characteristic information of each risk factor in the volatility characteristic parameters is precisely matched with the mapping rules to determine the corresponding weight coefficient value. According to the arrangement order of the original risk factor spectrum, a matched weight coefficient value is assigned to each risk factor. The unique access identifier code, the specific content of the factor, and the assigned initial weight coefficient of each risk factor are associated one-to-one and fixedly recorded. After completing the binding and association operation of all risk factors with the initial weight coefficients, the initial weighting factor set of the power infrastructure project to be evaluated is obtained.

[0095] By selectively extracting time-series data segments corresponding to the original risk factor spectrum from standardized datasets, subsequent volatility analysis can be conducted around effective data related to risk factors. Eliminating irrelevant data ensures the accuracy and relevance of the analysis. Analyzing the standard deviation and coefficient of variation of time-series data samples reveals volatility characteristics, accurately identifying the volatility features of risk factors from both data dispersion and relative volatility dimensions. This results in a more comprehensive representation of volatility features that closely reflects the actual data state. Assigning initial weight coefficients to the risk factors in the original risk factor spectrum based on volatility feature parameters and establishing correlations allows for scientific weighting based on actual data volatility characteristics. This ensures that the weights closely match the actual volatility of the risk factors. The resulting initial weighted factor set possesses data-supported scientific validity and rationality, laying a precise and standardized foundation for the dynamic calibration of risk factor weights in subsequent power infrastructure projects, and improving the overall scientific rigor and adaptability of risk factor weight allocation.

[0096] In S4, the process of obtaining the stage weight vector of the power infrastructure project to be evaluated is as follows:

[0097] Real-time capture of mileage node trigger signals for power infrastructure projects to be evaluated;

[0098] The mileage node trigger signal is parsed using stage coding to obtain the current mileage stage code of the power infrastructure project to be evaluated.

[0099] Based on the current mileage stage coding, a stage matching search is performed on the preset benchmark project database to obtain the correction coefficient matrix of the power infrastructure projects to be evaluated.

[0100] The correction coefficient matrix is ​​iteratively adjusted item by item with the initial weight values ​​in the initial weighting factor set of the power infrastructure project to be evaluated, resulting in the stage weight vector of the power infrastructure project to be evaluated, specifically:

[0101] By stripping the coefficients from the correction coefficient matrix, the stage disturbance factor of the power infrastructure project to be evaluated is obtained.

[0102] Based on the stage disturbance factor, the initial weight values ​​in the initial weight factor set are multiplied and fused item by item to obtain the iterative intermediate weight parameters of the power infrastructure project to be evaluated.

[0103] The intermediate weight parameters of the iteration are normalized and constrained for verification, and the intermediate weight parameters of the iteration are proportionally redistributed according to the measured residuals to obtain the stage weight components of the power infrastructure project to be evaluated.

[0104] Based on the order of risk factors in the initial weighting factor set, the stage weight components are vectorized and encapsulated to obtain the stage weight vector of the power infrastructure project to be evaluated.

[0105] Based on the pre-set mileage node division standard for the entire process of power infrastructure projects from preliminary preparation, construction to final acceptance, this standard clearly defines the node conditions and judgment criteria for each key implementation stage of the project. The project implementation progress, process completion status, and key node achievement status are continuously monitored in real time. When the actual implementation status of the project fully reaches the pre-set mileage node judgment threshold, the status trigger information corresponding to that node is immediately collected. This information is used as an effective identifier for stage switching, thus obtaining the mileage node trigger signal of the power infrastructure project to be evaluated.

[0106] The mileage stage coding rule library for power infrastructure projects is retrieved. This rule library pre-establishes a one-to-one correspondence between all mileage node trigger signals and unique stage codes. Each code corresponds to an independent and clear implementation stage of the project. The collected mileage node trigger signals are precisely matched item by item with all information in the rule library to determine the stage category to which the signal belongs and extract the unique coding information corresponding to the matching result to obtain the current mileage stage code of the power infrastructure project to be evaluated.

[0107] The pre-built benchmark project database is retrieved. This database is trained using a large amount of historical completed power infrastructure project data. It contains a complete set of exclusive weight correction coefficient matrices corresponding to different mileage stage codes. The current mileage stage code is used as the core search keyword. A precise matching search is performed in the benchmark project database. The coefficient matrix information that completely corresponds to the stage code in the search results is located and extracted to obtain the correction coefficient matrix of the power infrastructure project to be evaluated.

[0108] According to the pre-defined matrix row and column arrangement rules of the correction coefficient matrix, all correction coefficients contained in the matrix are extracted and split row by row and column by column. The independent coefficients obtained by splitting are arranged in an orderly manner according to the original matrix arrangement order to form a set of independent coefficients that correspond one-to-one with the risk factors, thus obtaining the stage disturbance factors of the power infrastructure project to be evaluated.

[0109] Each independent coefficient in the stage disturbance factor and each initial weight value in the initial weighting factor set are fused one by one according to a preset one-to-one correspondence. This ensures that each independent coefficient of the stage disturbance factor corresponds to only one initial weight value. The fusion operation of all coefficients and weight values ​​is completed in sequence to obtain the iterative intermediate weight parameters of the power infrastructure project to be evaluated.

[0110] According to the normalized constraint range of the risk factor weights of the power infrastructure project, the compliance of all values ​​in the intermediate weight parameters of the iteration is verified item by item. The total value of all values ​​is calculated and compared with the preset constraint benchmark value. The difference between the two is determined as the residual value. Based on the specific value of the residual value, the intermediate weight parameter values ​​of each iteration are evenly adjusted and distributed according to their proportion in the whole, so that the sum of all adjusted values ​​completely matches the preset constraint benchmark value. After all the values ​​are adjusted and verified, the stage weight components of the power infrastructure project to be evaluated are obtained.

[0111] The original order of each risk factor in the initial weighting factor set is retrieved. This order is completely consistent with the original risk factor spectrum. Each value in the stage weight component is arranged in strict accordance with this original order to ensure that the correspondence between the weight values ​​and the risk factors does not change. The arranged set of values ​​is standardized and encapsulated into a vector structure to fully preserve the order of each value and the corresponding relationship, thus obtaining the stage weight vector of the power infrastructure project to be evaluated.

[0112] Real-time capture of mileage node trigger signals in power infrastructure projects and parsing of stage codes enable precise location of the current implementation stage, ensuring a high degree of alignment between weight calibration and actual project progress. Based on stage codes, the system matches and retrieves correction coefficient matrices from a benchmark project database, providing a scientific and practical reference for weight calibration using mature data from benchmark projects. The system extracts stage disturbance factors from the correction coefficient matrix and integrates them with the initial weight values ​​through item-by-item multiplication, achieving refined, stage-specific adjustment of the initial weight values. Normalization constraint verification of intermediate weight parameters during iterations and proportional redistribution based on residuals ensure that the adjusted weights... The values ​​conform to the preset numerical constraint standards, ensuring the rationality and standardization of the weight values. The stage weight components are vectorized and encapsulated according to the risk factor order of the initial weight factor set, which can retain the corresponding relationship between risk factors and weight values. This gives the stage weight vector structured and ordered characteristics, and realizes dynamic iterative calibration of weights based on project mileage stages. This allows the weight values ​​of risk factors to be dynamically adapted with the project implementation stage, greatly improving the matching degree between weight values ​​and the actual risk status of the project. This provides a staged weight basis that fits the actual project for the subsequent refactoring of risk factors, and further enhances the accuracy and adaptability of risk assessment for power infrastructure projects.

[0113] In S5, the process of obtaining the risk characteristic vector of the power infrastructure project to be evaluated is as follows:

[0114] Using the stage weight vector as the modulation signal, the initial weighting factor set is carrier modulated to obtain the risk factor sequence of the power infrastructure project to be evaluated.

[0115] The modulated risk factor sequence is coherently demodulated to obtain the enhanced risk factor sequence of the power infrastructure project to be evaluated;

[0116] By embedding the enhanced risk factors into a manifold, the risk feature vector of the power infrastructure project to be evaluated is obtained.

[0117] Using the stage weight vector as the core modulation signal, and following the pre-set modulation rules for risk factor weighting in power infrastructure projects, the unique identifier of each risk factor is used as the matching basis. Each calibrated weight value in the stage weight vector is then assembled and modulated one by one with the corresponding risk factors in the initial weighted factor set. This process fully loads the project mileage stage characteristics and dynamic calibration characteristics carried by the stage weights onto each risk factor in the initial weighted factor set, so that each risk factor carries both its original characteristics and stage weight characteristics. After completing the modulation processing of all risk factors, the risk factor sequence of the power infrastructure project to be evaluated is obtained.

[0118] Based on the pre-set coherent demodulation rules, the modulated risk factor sequence is analyzed to accurately extract the stage weight feature information loaded in it. The modulated and fused risk factor sequence is then reversed and analyzed. During the restoration process, the core risk features that are highly relevant to the current stage of the project are emphasized, while redundant features that are irrelevant to the current stage are filtered out and weakened. This makes the effective features of the risk factors more prominent and stable. After completing all demodulation operations and enhancing the risk features, the enhanced risk factor sequence of the power infrastructure project to be evaluated is obtained.

[0119] According to the pre-defined manifold embedding rules for extracting risk features of power infrastructure projects, the enhanced risk factor sequence is mapped as a whole to a pre-constructed high-dimensional feature space. Within this high-dimensional space, the feature points are arranged in an orderly spatial manner based on the inherent correlations, influence relationships, and business logic between risk factors. This allows the risk factors to form a distribution structure in the space that conforms to actual business rules. Then, the high-dimensional feature points that have been arranged in the space are uniformly vectorized and integrated to fully preserve the feature correlations between risk factors and the project stage attribute features. After completing all manifold embedding operations, the risk feature vector of the power infrastructure project to be evaluated is obtained.

[0120] By using the stage weight vector as the modulation signal to carrier modulate the initial weighted factor set, the weight features of the current stage of the project can be accurately loaded into the risk factors. This allows the risk factor sequence to fully match the risk characteristics of the actual implementation stage of the project. Coherent demodulation of the modulated risk factor sequence can effectively strengthen the core risk features related to the current stage of the project, weaken the interference of irrelevant features, and improve the feature identification and targeting of the risk factors. The resulting enhanced risk factor sequence can more accurately represent the actual risk status of the project. Manifold embedding of the enhanced risk factors can map the risk factors to a high-dimensional feature space and complete the feature arrangement based on the inherent correlation, making the representation of risk features more comprehensive and spatially correlated. The final risk feature vector can systematically and structurally integrate the risk factor features of the project stages, accurately and comprehensively reflecting the overall risk characteristics of the current stage of the power infrastructure project. This provides a highly accurate and highly consistent feature foundation for subsequent risk level assessment, significantly improving the accuracy and scientific nature of subsequent risk assessments.

[0121] In S6, the process of obtaining the evaluation conclusion for the power infrastructure project to be evaluated is as follows:

[0122] Extract risk level cluster centers from the historical completed project database of power infrastructure projects to be evaluated;

[0123] Based on the risk level cluster center points, the spatial distance of the risk feature vectors is determined to obtain the Euclidean distance value of the power infrastructure project to be evaluated.

[0124] Based on the preset membership mapping relationship, membership values ​​are assigned to the Euclidean distance values ​​to obtain the membership value group of the power infrastructure project to be evaluated, specifically:

[0125] Extract distance scale parameters and benchmark membership correction coefficients from the historical completed project database;

[0126] The initial membership value of the power infrastructure project to be evaluated is calculated based on the Euclidean distance value, distance scale parameter, and benchmark membership correction coefficient. The formula for calculating the initial membership value is as follows:

[0127] ;

[0128] In the formula, This is the initial membership value. This represents the Euclidean distance between the risk feature vector and the cluster center of the k-th risk level. For the k-th distance scale parameter, This is the baseline membership correction coefficient for the k-th risk level;

[0129] The initial membership values ​​are balanced and allocated to obtain a set of membership values ​​for the power infrastructure projects to be evaluated.

[0130] Extreme value retrieval is performed on the membership value group to capture the risk level pointed to by the maximum membership score in the membership value group, and the risk level is used as the risk level label of the power infrastructure project to be evaluated.

[0131] By associating and binding risk level labels with risk feature vectors, an assessment conclusion with risk level labels is obtained for the power infrastructure project to be evaluated.

[0132] The database of historical completed power infrastructure projects that have been constructed and passed acceptance and archiving is retrieved. This database contains complete risk assessment data for power infrastructure projects of the same type and scale. According to the common and preset risk level classification standards in the power infrastructure industry, cluster analysis is performed on the risk feature vectors of all historical projects in the database. Multiple independent risk level clusters are formed by aggregating similar risk features. Then, feature vectors at the spatial distribution center are selected from each risk level cluster. The feature vector at the center is defined as the core reference benchmark for the corresponding risk level. Finally, the core reference features of all risk levels are summarized to obtain the risk level cluster center points of the power infrastructure projects to be evaluated.

[0133] The risk feature vector of the power infrastructure project to be evaluated and the cluster centers of each risk level obtained earlier are placed in the same high-dimensional feature space to maintain a consistent coordinate system. According to the spatial geometric measurement rules, the straight-line distance between the risk feature vector and the cluster center of each risk level is calculated. This distance is used to characterize the similarity between the two. The calculation results of the straight-line distances corresponding to all risk levels are recorded and standardized to obtain the Euclidean distance value of the power infrastructure project to be evaluated.

[0134] The membership mapping table, which was trained in advance based on a large amount of historical project data, is retrieved. This table clearly divides different Euclidean distance value intervals and assigns a unique corresponding membership score to each interval, forming a fixed matching correspondence. Each Euclidean distance value obtained earlier is matched to the distance interval to which it belongs in the membership mapping table. The standard membership score corresponding to the interval is extracted, and then all scores are sorted in order from low to high risk level to obtain the membership value group of the power infrastructure project to be evaluated.

[0135] Each membership score within a membership value group is compared and its magnitude is searched to accurately locate the membership score with the largest value. Based on the position of the largest score in the membership value group, the corresponding risk level is queried in reverse. This risk level is determined as the exclusive level determination identifier for the project to be evaluated. After the level labeling is completed, the risk level label of the power infrastructure project to be evaluated is obtained.

[0136] The text, code, and other identifying information of the risk level label are precisely linked one-to-one with the risk feature vector of the power infrastructure project to be evaluated. The risk level label is embedded as a formal feature attribute into the data structure of the risk feature vector. A comprehensive data body containing complete risk feature information and clear risk level determination results is constructed to obtain the evaluation conclusion of the power infrastructure project to be evaluated with risk level label.

[0137] The historical completed project database of the power infrastructure projects to be evaluated is retrieved again. Based on the risk level number and attribute information corresponding to the cluster center point of each risk level, the distance scale parameters and benchmark membership correction coefficients pre-statistically completed under each risk level are extracted. Both types of values ​​are fixed reference values ​​determined after a large number of sample statistics and verifications under the corresponding risk level in the historical completed project database. They have industry universality and authority. After completing the extraction of parameters and coefficients corresponding to all risk levels, the distance scale parameters and benchmark membership correction coefficients matching the power infrastructure projects to be evaluated are obtained.

[0138] The Euclidean distance value for each risk level of the power infrastructure project to be evaluated is fused with the distance scale parameter of the same risk level in a standardized manner. The intermediate result obtained by the fusion is then further combined with the benchmark membership correction coefficient of the risk level. This process is applied one by one according to the order of risk levels to complete the calculation process of all risk levels and obtain the initial membership value corresponding to each risk level of the power infrastructure project to be evaluated.

[0139] According to the pre-set membership degree equalization and allocation rules of the power infrastructure risk assessment, all initial membership degree values ​​are put into a unified numerical allocation system for overall verification. The distribution of each initial membership degree value is checked to see if it meets the preset rationality benchmark. Values ​​that exceed the benchmark range are finely adjusted evenly according to their proportion in the whole, so that the distribution of adjusted membership degree values ​​is highly consistent with the actual risk characteristics of the project. After the equalization and compliance verification of all initial membership degree values ​​are completed, the membership degree value group of the power infrastructure project to be assessed is obtained.

[0140] Distance scale parameters are extracted from the historical completed project database of the power infrastructure projects to be evaluated. These parameters are fixed reference values ​​that correspond to the cluster centers of each risk level in the historical completed project database after statistical processing, and form a one-to-one correlation with each risk level.

[0141] The benchmark membership correction coefficient is extracted from the historical completed project database of the power infrastructure projects to be evaluated. This coefficient is a fixed reference value after statistical processing for each risk level in the historical completed project database, and forms a one-to-one correlation with each risk level.

[0142] The spatial distance between the risk feature vector of the power infrastructure project to be evaluated and the cluster center point of each risk level is measured. The resulting straight-line distance is the Euclidean distance value, which forms a one-to-one correspondence with each risk level.

[0143] By processing Euclidean distance values, distance scale parameters, and benchmark membership correction coefficients through numerical fusion and combination, the results can accurately reflect the correlation between the risk feature vector of the power infrastructure project to be evaluated and the cluster centers of each risk level.

[0144] The processing results can serve as an initial reference for determining the risk level of the power infrastructure project to be evaluated, providing accurate initial numerical support for the formation of subsequent membership value groups, ensuring that the risk level determination has numerical basis that fits the actual situation of the project, and guaranteeing the scientificity and accuracy of the subsequent extreme value search to determine the risk level label.

[0145] Figure 2 A comparison chart of the relationship curves between the membership degree of different risk levels and the Euclidean distance is presented. The horizontal axis is the Euclidean distance between the risk feature vector and the cluster center of the risk level, and the vertical axis is the membership degree score. Figure 2 The text displays the membership degree change curves corresponding to three risk levels: low risk, medium risk, and high risk (light blue fills for low risk, light red fills for medium risk, and light beige fills for high risk). It intuitively presents the changing trend of membership degree scores under various risk levels as Euclidean distance increases, clearly reflecting the correlation between Euclidean distance and membership degree of different risk levels. This provides a visual reference for the membership degree matching process of determining the risk level of a project through Euclidean distance.

[0146] Extracting risk level cluster centers from the historical completed project database provides a standardized reference for matching risk feature vectors. Determining Euclidean distance values ​​based on these centers can accurately quantify the spatial correlation between risk feature vectors and each risk level. Extracting distance scale parameters and benchmark membership correction coefficients from the historical completed project database and combining them with Euclidean distance values ​​to calculate initial membership values ​​ensures that the assignment of membership values ​​has solid historical data support and quantitative basis. Balancing the initial membership values ​​allows the numerical distribution of membership value groups to better match the actual risk characteristics of the project, ensuring the rationality of the values.

[0147] By performing extreme value retrieval on membership value groups to determine risk level labels, the risk level with the highest matching degree to project risk characteristics can be accurately identified. The risk level labels are then associated and bound with risk feature vectors to obtain labeled assessment conclusions. This ensures that the assessment conclusions not only contain accurate risk level determination results but also integrate complete risk feature information of the project. Through the entire process of fuzzy membership matching, the scientific and accurate determination of the risk level of power infrastructure projects is achieved. This ensures that the assessment conclusions have both quantitative basis and feature completeness, providing accurate, comprehensive, and traceable judgment results for risk management of power infrastructure projects. This significantly enhances the reference value and application value of risk assessment conclusions for power infrastructure projects.

[0148] In S7, the process of obtaining the risk assessment data message for the power infrastructure project to be evaluated is as follows:

[0149] The assessment conclusions are decomposed in a structured manner to obtain the assessment conclusion metadata of the power infrastructure project to be assessed;

[0150] The assessment conclusion metadata is associated and assembled with the risk feature vector to obtain the assessment data carrier of the power infrastructure project to be assessed.

[0151] The assessment data carrier and the stage weight vector are merged and packaged to obtain the risk assessment data message of the power infrastructure project to be assessed.

[0152] According to the pre-defined metadata decomposition standard for risk assessment of power infrastructure projects, this standard clearly stipulates the dimensions of assessment conclusion decomposition, data unit division standards, and identification specifications. Assessment conclusions with risk level labels are systematically decomposed into three independent basic data units: risk level label information, risk feature association information, and assessment judgment basis information. The risk level label information includes the level code, level name, and judgment description; the risk feature association information includes the corresponding identifier of the risk feature vector and feature matching details; and the assessment judgment basis information includes core judgment data such as Euclidean distance values ​​and membership value groups. Each basic data unit after decomposition is assigned a unique standardized identifier according to the pre-defined coding rules and is classified and archived according to data type to ensure that each data unit can be accurately identified and individually accessed. After decomposition, identification, and classification, the assessment conclusion metadata of the power infrastructure project to be assessed is obtained.

[0153] Based on the pre-defined association and assembly rules for risk data in power infrastructure projects, these rules clarify the correlation and correspondence between assessment conclusion metadata and risk feature vectors, as well as the assembly process and format requirements. Using data identification as the core matching basis, each basic data unit in the assessment conclusion metadata is associated one-to-one with the corresponding feature item in the risk feature vector. The standardized identification information and data content of the assessment conclusion metadata are embedded one by one into the feature data system of the risk feature vector, achieving deep integration of the basic information of the assessment conclusion and the risk feature information. This constructs an integrated data body that combines the core information of the assessment conclusion with a complete risk feature representation, ensuring that the correlation between data is clear and traceable. After the association and assembly are completed, the assessment data carrier of the power infrastructure project to be assessed is obtained.

[0154] According to the pre-defined packaging specifications for risk assessment data of power infrastructure projects, the specifications clearly define the data encapsulation framework, format standards, coding rules, and verification requirements. The assessment data carrier and the stage weight vector are included in the same standardized data encapsulation framework. The internal data structure of the assessment data carrier, the correlation between each basic data unit, and the weight order of the stage weight vector and its correspondence with risk factors are strictly preserved without changing the original core content of the two types of data. All data within the encapsulation framework undergoes item-by-item format verification to identify problems such as data format inconsistencies and misalignments. After verification, standardized encoding is performed according to the pre-defined coding rules to ensure that the data can be transmitted, read, and parsed across systems. After completing the encoding and overall encapsulation of all data, the risk assessment data message of the power infrastructure project to be assessed is obtained.

[0155] The assessment conclusions are decomposed into structured metadata, breaking down integrated assessment conclusions into standardized basic data units. This makes the information presentation of the assessment conclusions more organized and standardized, facilitating subsequent data association and reuse. Associating and assembling the assessment conclusion metadata with risk feature vectors yields an assessment data carrier, achieving deep integration of assessment results and core risk characteristic information. This allows the data carrier to possess the dual attributes of risk level determination and risk characteristic representation. Merging and packaging the assessment data carrier with stage weight vectors yields a risk assessment data message, which standardizes, integrates, and encapsulates the core data of the entire project risk assessment process. It preserves the correlation and correspondence between data points and the original data structure, forming a systematic and structured whole of risk assessment data. This significantly improves the transmissibility, storability, and reusability of risk assessment data, providing complete, standardized, and traceable full-data support for subsequent dynamic risk monitoring and control decision-making in power infrastructure projects, as well as for risk assessment reference in similar projects. This further expands the application scenarios and practical value of risk assessment data for power infrastructure projects.

[0156] Example 2: A risk assessment device for power infrastructure projects based on a multi-factor dynamic weighting algorithm, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the method in Example 1 is implemented by executing the program through the processor.

[0157] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

Claims

1. A power infrastructure project risk assessment method based on a multi-factor dynamic weighting algorithm, characterized in that, Includes the following steps: S1. Noise filtering is performed on the original business flow data of the power infrastructure project to be evaluated, and the filtered data fields are normalized and compressed to obtain a standardized dataset of the power infrastructure project to be evaluated. S2. Factor stripping is performed on the standardized dataset to obtain the original risk factor spectrum of the power infrastructure project to be evaluated; S3. Based on the standard deviation and coefficient of variation of time series data in the standardized dataset, weights are assigned to the original risk factor spectrum to obtain the initial weighted factor set of the power infrastructure project to be evaluated. S4. Based on the mileage stage of the power infrastructure project to be evaluated, retrieve the weight correction coefficient matrix from the preset benchmark project database, and dynamically iterate and calibrate the initial weight values ​​of the power infrastructure project to be evaluated to obtain the stage weight vector of the power infrastructure project to be evaluated. S5. Based on the stage weight vector, the initial weighting factor set is reconstructed to obtain the risk characteristic vector of the power infrastructure project to be evaluated. The process is as follows: Using the stage weight vector as the modulation signal, the initial weighting factor set is carrier modulated to obtain the risk factor sequence of the power infrastructure project to be evaluated. The modulated risk factor sequence is coherently demodulated to obtain the enhanced risk factor sequence of the power infrastructure project to be evaluated; By manifold embedding of the enhanced risk factors, the risk feature vector of the power infrastructure project to be evaluated is obtained; S6. Perform fuzzy membership matching on the risk feature vector to obtain the evaluation conclusion of the power infrastructure project to be evaluated. S7. Standardize and encapsulate the assessment conclusions, risk feature vectors, and stage weight vectors to obtain the risk assessment data message for the power infrastructure project to be assessed.

2. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S1, the process of obtaining the standardized dataset of the power infrastructure project to be evaluated is as follows: Signal cleaning is performed on the original business flow data of the power infrastructure project to be evaluated to obtain the cleaned business flow data of the power infrastructure project to be evaluated; Based on the average data of adjacent time windows in the cleaning business flow data, numerical values ​​are filled into the cleaning business flow data to obtain the complete business flow data of the power infrastructure project to be evaluated. Dimensionless scaling is performed on numerical fields with inconsistent dimensions in the complete business flow data to obtain compressed data fields of the power infrastructure project to be evaluated. Based on the data source of the original business flow data, the compressed data fields are classified and merged to obtain a standardized dataset of the power infrastructure projects to be evaluated.

3. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S2, the process of obtaining the original risk factor spectrum of the power infrastructure project to be evaluated is as follows: Business semantic parsing is performed on the standardized dataset to obtain a cluster of risk indicator fields for the power infrastructure projects to be evaluated. Based on the preset dimension classification labels, the risk indicator field clusters are classified and grouped to obtain the original factor group of the power infrastructure project to be evaluated. By binding access identifiers to the risk factors in the original factor group, the original risk factor spectrum of the power infrastructure project to be evaluated is obtained.

4. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S3, the process of obtaining the initial weighting factor set for the power infrastructure project to be evaluated is as follows: Based on the original risk factor spectrum, time series data segments of the original risk factor spectrum are extracted from the standardized dataset to obtain time series data samples of the power infrastructure projects to be evaluated. The fluctuation characteristics of the standard deviation and coefficient of variation of the time series data samples are analyzed to obtain the fluctuation characteristic parameters of the power infrastructure project to be evaluated. Based on the volatility characteristic parameter, initial weight coefficients are assigned to the risk factors in the original risk factor spectrum, and the initial weight coefficients are associated and bound with the risk factors to obtain the initial weight factor set of the power infrastructure project to be evaluated.

5. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S4, the process of obtaining the stage weight vector of the power infrastructure project to be evaluated is as follows: Real-time capture of mileage node trigger signals for power infrastructure projects to be evaluated; The mileage node trigger signal is parsed using stage coding to obtain the current mileage stage code of the power infrastructure project to be evaluated. Based on the current mileage stage coding, a stage matching search is performed on the preset benchmark project database to obtain the correction coefficient matrix of the power infrastructure projects to be evaluated. The correction coefficient matrix is ​​iteratively adjusted item by item with the initial weight values ​​in the initial weighting factor set of the power infrastructure project to be evaluated to obtain the stage weight vector of the power infrastructure project to be evaluated.

6. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 5, characterized in that, The process of iteratively adjusting the correction coefficient matrix and the initial weight values ​​in the initial weighting factor set of the power infrastructure project to be evaluated involves the following steps: By stripping the coefficients from the correction coefficient matrix, the stage disturbance factor of the power infrastructure project to be evaluated is obtained. Based on the stage disturbance factor, the initial weight values ​​in the initial weight factor set are multiplied and fused item by item to obtain the iterative intermediate weight parameters of the power infrastructure project to be evaluated. The intermediate weight parameters of the iteration are normalized and constrained for verification, and the intermediate weight parameters of the iteration are proportionally redistributed according to the measured residuals to obtain the stage weight components of the power infrastructure project to be evaluated. Based on the order of risk factors in the initial weighting factor set, the stage weight components are vectorized and encapsulated to obtain the stage weight vector of the power infrastructure project to be evaluated.

7. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S6, the process of obtaining the evaluation conclusion of the power infrastructure project to be evaluated is as follows: Extract risk level cluster centers from the historical completed project database of power infrastructure projects to be evaluated; Based on the risk level cluster center points, the spatial distance of the risk feature vectors is determined to obtain the Euclidean distance value of the power infrastructure project to be evaluated. Based on the preset membership mapping relationship, the membership values ​​of the Euclidean distance are assigned to obtain the membership value group of the power infrastructure project to be evaluated; Extreme value retrieval is performed on the membership value group to capture the risk level pointed to by the maximum membership score in the membership value group, and the risk level is used as the risk level label of the power infrastructure project to be evaluated. By associating and binding risk level labels with risk feature vectors, an assessment conclusion with risk level labels is obtained for the power infrastructure project to be evaluated.

8. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 7, characterized in that, The specific details of obtaining the membership value group of the power infrastructure project to be evaluated are as follows: Extract distance scale parameters and benchmark membership correction coefficients from the historical completed project database; The initial membership value of the power infrastructure project to be evaluated is calculated based on the Euclidean distance value, distance scale parameter, and benchmark membership correction coefficient. The formula for calculating the initial membership value is as follows: ； In the formula, This is the initial membership value. This represents the Euclidean distance between the risk feature vector and the cluster center of the k-th risk level. For the k-th distance scale parameter, This is the baseline membership correction coefficient for the k-th risk level; The initial membership values ​​are balanced and allocated to obtain a set of membership values ​​for the power infrastructure projects to be evaluated.

9. The risk assessment method for power infrastructure projects based on a multi-factor dynamic weighting algorithm as described in claim 1, characterized in that, In S7, the process of obtaining the risk assessment data message of the power infrastructure project to be evaluated is as follows: The assessment conclusions are decomposed in a structured manner to obtain the assessment conclusion metadata of the power infrastructure project to be assessed; The assessment conclusion metadata is associated and assembled with the risk feature vector to obtain the assessment data carrier of the power infrastructure project to be assessed. The assessment data carrier and the stage weight vector are merged and packaged to obtain the risk assessment data message of the power infrastructure project to be assessed.

Images