Maintenance decision-making method and system based on high-voltage cable state evaluation and life prediction

Abstract

The application discloses a maintenance decision-making method and system based on high-voltage cable state evaluation and life prediction, and belongs to the technical field of intelligent operation and maintenance and state management and control of high-voltage equipment of a power system.The method is as follows: an initial evaluation index system is constructed; whenever it is detected that an index in the initial evaluation index system deviates from a preset baseline threshold or an abnormal working condition occurs, the evaluation index system is dynamically adjusted to obtain a dynamic evaluation index system, the weight of each index in the system is determined based on a preset weighting algorithm, each index in the system is adaptively fused according to a neural network, a multi-source fusion feature data set is obtained and input into a time series prediction model, a high-voltage cable state evaluation result and a high-voltage cable residual life prediction value are output, then the high-voltage cable residual life prediction value is corrected, and finally, a high-voltage cable maintenance decision is generated based on the high-voltage cable state evaluation result and the corrected high-voltage cable residual life prediction value.The application can improve the reliability of the maintenance decision.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance and condition management technology for high-voltage equipment in power systems, and particularly to a maintenance decision-making method and system based on high-voltage cable condition assessment and life prediction. Background Technology

[0002] High-voltage cables are the core infrastructure of power transmission systems, undertaking the crucial function of transmitting large-capacity power across regions. Their operational stability directly affects the safety of the power grid and the continuity of energy supply. With the continuous expansion of my country's power grid construction, a large number of high-voltage cables are entering the middle and late stages of their service life, and problems such as insulation aging, mechanical damage, and environmental corrosion are gradually becoming more prominent. Power outages caused by cable faults not only result in huge economic losses but also pose a severe challenge to the power supply for industrial production and residential use. How to accurately assess the condition of cables and predict their remaining lifespan has become a key issue that needs further research in the intelligent operation and maintenance system of the power grid.

[0003] Currently, existing technologies primarily rely on single electrical parameter monitoring or traditional machine learning modeling for cable condition assessment and lifespan prediction. Some solutions also attempt to improve assessment accuracy by incorporating multi-source data fusion and deep learning models. However, these technologies generally employ static indicator systems, failing to dynamically respond to sudden changes in the cable operating environment and abnormal parameter fluctuations. Furthermore, data-driven models lack physical mechanism support, resulting in poor model interpretability. Consequently, it is difficult to translate assessment and prediction results into actionable maintenance decisions, ultimately failing to provide reliable maintenance support for high-voltage cables. Therefore, existing technologies cannot dynamically respond to sudden changes in the cable operating environment or abnormal parameter fluctuations, posing a problem in providing reliable maintenance decisions for high-voltage cables. Summary of the Invention

[0004] This invention proposes a maintenance decision-making method and system based on high-voltage cable condition assessment and life prediction. It addresses the problem that existing technologies cannot dynamically respond to sudden changes in the cable operating environment or abnormal parameter fluctuations, thus failing to provide reliable maintenance decisions for high-voltage cables. This invention achieves this by constructing a dynamic evaluation index system and adaptively fusing the various indicators to adapt to sudden changes in the cable operating environment or abnormal parameter fluctuations, thereby improving the reliability of maintenance decisions.

[0005] To achieve the above objectives, embodiments of the present invention provide a maintenance decision-making method based on high-voltage cable condition assessment and lifespan prediction, comprising: constructing an initial assessment index system based on pre-collected real-time operating data, environmental parameters, and historical maintenance data of the high-voltage cable; dynamically adjusting the assessment index system whenever an index deviates from a preset baseline threshold or an abnormal operating condition is detected, to obtain a dynamic assessment index system; determining the weight of each index in the dynamic assessment index system based on a preset weighting algorithm, and adaptively fusing each index in the dynamic assessment index system according to a preset neural network to obtain a multi-source fusion feature dataset; inputting the multi-source fusion feature dataset into a pre-constructed time-series prediction model, outputting the high-voltage cable condition assessment result and the high-voltage cable remaining lifespan prediction value, and correcting the high-voltage cable remaining lifespan prediction value according to a preset physical mechanism model to obtain a corrected high-voltage cable remaining lifespan prediction value; and generating a high-voltage cable maintenance decision based on the high-voltage cable condition assessment result and the corrected high-voltage cable remaining lifespan prediction value.

[0006] This invention proposes a maintenance decision-making method based on high-voltage cable condition assessment and lifespan prediction. It constructs an initial assessment index system and dynamically adjusts it according to sudden environmental changes or abnormal parameter fluctuations. Multi-source fusion is then used to eliminate redundancy and interference from heterogeneous data. Finally, combining the advantages of a pre-constructed time-series model and a preset physical mechanism model, reliable high-voltage cable condition assessment results and remaining lifespan prediction results are output, thereby generating reliable high-voltage cable maintenance decisions. Thus, by constructing a dynamic assessment index system and adaptively fusing the indicators, the reliability of maintenance decisions is improved to adapt to sudden changes in the cable operating environment or abnormal parameter fluctuations.

[0007] Furthermore, based on the pre-collected real-time operation data, environmental parameters, and historical maintenance data of the high-voltage cable, an initial evaluation index system is constructed, including: preprocessing the collected real-time operation data, environmental parameters, and historical maintenance data of the high-voltage cable to obtain optimized real-time operation data, optimized environmental parameters, and optimized historical maintenance data of the high-voltage cable; constructing electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators based on the optimized real-time operation data of the high-voltage cable; constructing environmental factor indicators based on the optimized environmental parameters; constructing operational history indicators based on the optimized historical maintenance data; and integrating the electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operational history indicators to obtain the initial evaluation index system.

[0008] In the above scheme, the real-time operating data, environmental parameters, and operation and maintenance history data of high-voltage cables are preprocessed and optimized to construct a five-dimensional index system encompassing electrical performance, mechanical structure, insulation characteristics, environmental factors, and operating history. This system comprehensively covers the core dimensions affecting the cable's operating status, avoiding the one-sidedness of assessments caused by single indicators. The multi-dimensional indicators can also fully capture all the changes in the cable from its own performance to the external environment, providing reliable input data for subsequent high-voltage cable condition assessment and high-voltage cable remaining life prediction, which helps improve the reliability of maintenance decisions.

[0009] Furthermore, whenever an indicator in the initial evaluation indicator system deviates from a preset baseline threshold or an abnormal operating condition is detected, the evaluation indicator system is dynamically adjusted to obtain a dynamic evaluation indicator system. This includes: selecting target indicator parameters from the initial evaluation indicator system and obtaining real-time time-series data corresponding to the target indicator parameters; obtaining historical time-series data corresponding to the target indicator parameters through a preset time-series model, and comparing the real-time time-series data and the historical time-series data to obtain a comparison result; constructing a multi-dimensional feature vector based on the real-time time-series data, and performing density clustering on the multi-dimensional feature vector based on a preset clustering algorithm to classify abnormal operating conditions; whenever any target indicator parameter deviates from the preset baseline threshold or an abnormal operating condition is detected in the comparison result, adjusting the indicator topology of the initial evaluation indicator system to obtain an adjusted initial evaluation indicator system; and assigning importance ratings to the adjusted initial evaluation indicators based on a preset scoring model to obtain a dynamic evaluation indicator system.

[0010] In the above scheme, target indicator parameters are selected and corresponding real-time and historical time-series data are obtained. By comparing the real-time and historical time-series data, it can be determined whether the indicator deviates from the preset benchmark threshold. At the same time, multi-dimensional feature vectors are constructed using real-time time-series data and abnormal operating conditions are identified through density clustering. Then, when any target indicator parameter deviates from the preset baseline threshold or an abnormal operating condition occurs, the indicator topology of the initial evaluation indicator system is dynamically adjusted, and the indicator topology and weight ratio are optimized in real time. This avoids the shortcomings of static indicator systems that cannot adapt to environmental changes and parameter fluctuations. Furthermore, the reliability of the dynamic evaluation indicator system is further improved by combining importance scoring, which helps to improve the reliability of maintenance decisions.

[0011] Furthermore, the weights of each indicator in the dynamic evaluation indicator system are determined based on a preset weighting algorithm, and the indicators in the dynamic evaluation indicator system are adaptively fused according to a preset neural network to obtain a multi-source fused feature dataset. This includes: determining the individual weights of each indicator in the dynamic evaluation indicator system based on the preset weighting algorithm; constructing a weighted normalization matrix based on the individual weights of each indicator, and coordinating and optimizing the weighted normalization matrix to obtain optimized weights; constructing a weight vector matrix by combining the individual weights of each indicator and the optimized weights, and solving the weight vector matrix using a preset solution algorithm to obtain the combined indicator weights of each indicator in the dynamic evaluation indicator system; constructing a multi-source indicator association topology based on the combined indicator weights and the indicators in the dynamic evaluation indicator system, and calculating the attention weights of each indicator in the multi-source indicator association topology based on a preset attention weight algorithm; and adaptively fusion of each indicator in the dynamic evaluation indicator system based on the preset neural network and the attention weights of each indicator to obtain a multi-source fused feature dataset.

[0012] In the above scheme, after determining the individual weights of each indicator using a preset weighting algorithm, the optimized weights are obtained through coordination optimization using a weighted standardization matrix. Then, the individual weights and optimized weights are combined to construct a weight vector matrix and solve for the combined indicator weights. Subsequently, a multi-source indicator correlation topology is constructed and attention weights are calculated. Finally, adaptive fusion is achieved through a preset neural network. This eliminates human bias and data bias caused by single weighting and suppresses the interference of abnormal data through the attention mechanism. It provides reliable input data for subsequent high-voltage cable condition assessment and high-voltage cable remaining life prediction, which helps to improve the reliability of maintenance decisions.

[0013] Furthermore, the step of inputting the multi-source fusion feature dataset into a pre-constructed time-series prediction model, outputting high-voltage cable status assessment results and high-voltage cable remaining life prediction values, and correcting the high-voltage cable remaining life prediction values ​​according to a preset physical mechanism model to obtain corrected high-voltage cable remaining life prediction values ​​includes: training an initial time-series prediction model based on the multi-source fusion feature dataset to obtain a target time-series prediction model; inputting the multi-source fusion feature dataset into the target time-series prediction model to output high-voltage cable operation status assessment results and initial remaining life prediction values; inputting the initial remaining life prediction values ​​into a preset physical mechanism model to obtain adjusted remaining life prediction values; and calibrating the adjusted remaining life prediction values ​​based on the operation and maintenance historical data to obtain corrected high-voltage cable remaining life prediction values.

[0014] In the above scheme, the trained target time-series prediction model outputs the high-voltage cable operation status assessment results and initial remaining life prediction values. Then, the remaining life prediction values ​​are corrected by the physical mechanism model. Furthermore, the remaining life prediction values ​​are calibrated a second time by combining historical operation and maintenance data. This approach leverages the time-series prediction model's ability to capture time-series characteristics and compensates for the lack of physical rationality in pure data-driven models through the physical mechanism model. At the same time, the calibration of historical operation and maintenance data further adapts to the actual service conditions of the cable. The final output of remaining life results is more accurate and reasonable, accurately reflecting the true degree of cable degradation and improving the reliability of maintenance decisions.

[0015] Furthermore, the step of training the initial time-series prediction model based on the multi-source fusion feature dataset to obtain the target time-series prediction model includes: performing stratified sampling of the multi-source fusion feature dataset to obtain a model training set, a model validation set, and a model test set; constructing the initial time-series prediction model, training the initial time-series prediction model using the model training set, and optimizing the hyperparameters of the initial time-series prediction model based on the model validation set and the model test set to obtain the target time-series prediction model.

[0016] In the above scheme, stratified sampling is used to divide the multi-source fusion dataset. Stratified sampling can ensure the consistency of the distribution of each dataset and avoid the model overfitting problem caused by data skew. Then, the model training set is used to complete the model training, and the model validation set and model test set are used to achieve multi-dataset iterative optimization of the model hyperparameters. This can improve the generalization ability and prediction accuracy of the time series prediction model, and at the same time ensure that the model can output stable and reliable prediction results in cable samples in different working conditions and different regions, thereby improving the reliability of maintenance decisions.

[0017] Furthermore, the step of generating a high-voltage cable maintenance decision based on the high-voltage cable condition assessment results and the corrected high-voltage cable remaining life prediction value includes: classifying the high-voltage cable condition level from the high-voltage cable condition assessment results, and classifying the remaining life level from the corrected high-voltage cable remaining life prediction value; determining the high-voltage cable degradation type and degradation location based on the high-voltage cable condition level and the classified remaining life level; and determining the corresponding maintenance method, maintenance cycle, and maintenance scope based on the high-voltage cable degradation type, high-voltage cable degradation location, and the operation and maintenance historical data, thereby generating a high-voltage cable maintenance decision.

[0018] In the above scheme, the cable condition assessment results are used to classify the cable into condition levels, and the revised remaining life prediction values ​​are used to classify the cable life levels. The cable degradation type and location are determined by comprehensively analyzing both levels. Then, historical maintenance data is combined to determine the corresponding maintenance methods, maintenance cycles, and maintenance scope, ultimately generating a high-voltage cable maintenance decision. Therefore, by jointly analyzing the condition level and life level, the degradation type and location can be accurately identified, and differentiated maintenance strategies can be formulated accordingly to prevent resource waste or insufficient maintenance, thereby improving the reliability of maintenance decisions.

[0019] This invention also provides a maintenance decision-making system based on high-voltage cable condition assessment and lifespan prediction, including an initial indicator system construction module, a dynamic evaluation indicator system construction module, a data fusion module, a high-voltage cable condition assessment and lifespan prediction module, and a maintenance decision output module; wherein: the initial indicator system construction module is used to construct an initial evaluation indicator system based on pre-collected real-time operating data, environmental parameters, and historical operation and maintenance data of the high-voltage cable; the dynamic evaluation indicator system construction module is used to dynamically adjust the evaluation indicator system whenever an indicator in the initial evaluation indicator system deviates from a preset baseline threshold or an abnormal operating condition is detected, thereby obtaining a dynamic evaluation indicator system; the data fusion module is used to... The weights of each indicator in the dynamic evaluation index system are determined by a preset weighting algorithm, and the indicators in the dynamic evaluation index system are adaptively fused according to a preset neural network to obtain a multi-source fused feature dataset. The high-voltage cable condition assessment and life prediction module is used to input the multi-source fused feature dataset into a pre-constructed time-series prediction model, output the high-voltage cable condition assessment result and the high-voltage cable remaining life prediction value, and correct the high-voltage cable remaining life prediction value according to a preset physical mechanism model to obtain the corrected high-voltage cable remaining life prediction value. The maintenance decision output module is used to generate a high-voltage cable maintenance decision based on the high-voltage cable condition assessment result and the corrected high-voltage cable remaining life prediction value.

[0020] This invention proposes a maintenance decision-making system based on high-voltage cable condition assessment and lifespan prediction. It constructs an initial assessment index system and dynamically adjusts it according to sudden environmental changes or abnormal parameter fluctuations. Multi-source fusion is then used to eliminate redundancy and interference from heterogeneous data. Finally, combining the advantages of a pre-constructed time-series model and a preset physical mechanism model, reliable high-voltage cable condition assessment results and remaining lifespan prediction results are output, thereby generating reliable high-voltage cable maintenance decisions. Thus, by constructing a dynamic assessment index system and adaptively fusing the various indicators, the system can adapt to sudden changes in the cable operating environment or abnormal parameter fluctuations, improving the reliability of maintenance decisions.

[0021] Furthermore, the initial indicator system construction module includes a data preprocessing unit, a first indicator construction unit, a second indicator construction unit, a third indicator construction unit, and an indicator integration unit; wherein: the data preprocessing unit is used to preprocess the collected real-time operation data, environmental parameters, and historical operation and maintenance data of the high-voltage cable to obtain optimized real-time operation data, optimized environmental parameters, and optimized historical operation and maintenance data of the high-voltage cable; the first indicator construction unit is used to construct electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators based on the optimized real-time operation data of the high-voltage cable; the second indicator construction unit is used to construct environmental factor indicators based on the optimized environmental parameters; the third indicator construction unit is used to construct operation history indicators based on the optimized historical operation and maintenance data; and the indicator integration unit is used to integrate the electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operation history indicators to obtain the initial evaluation indicator system.

[0022] Furthermore, the dynamic evaluation index system construction module includes a real-time data acquisition unit, a data comparison unit, an abnormal operating condition classification unit, an index system adjustment unit, and an importance rating unit; wherein: the real-time data acquisition unit is used to select target index parameters from the initial evaluation index system and acquire the real-time time series data corresponding to the target index parameters; the data comparison unit is used to acquire the historical time series data corresponding to the target index parameters through a preset time series model, and compare the real-time time series data and the historical time series data to obtain a comparison result; the abnormal operating condition classification unit is used to construct a multi-dimensional feature vector based on the real-time time series data, and perform density clustering on the multi-dimensional feature vector based on a preset clustering algorithm to classify abnormal operating conditions; the index system adjustment unit is used to adjust the index topology of the initial evaluation index system whenever any target index parameter deviates from the preset baseline threshold or an abnormal operating condition occurs in the comparison result, to obtain an adjusted initial evaluation index system; the importance rating unit is used to perform importance rating on the adjusted initial evaluation index based on a preset scoring model to obtain a dynamic evaluation index system. Attached Figure Description

[0023] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 A flowchart illustrating the steps of a maintenance decision-making method based on high-voltage cable condition assessment and life prediction, provided in a certain embodiment of the present invention; Figure 2This is a schematic diagram of the module structure of a maintenance decision system based on high-voltage cable condition assessment and life prediction, provided in one embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0027] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0028] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0029] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0030] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0031] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in the embodiments of this application according to the specific circumstances.

[0032] To address the problem that existing technologies cannot dynamically respond to sudden changes in cable operating environments or abnormal parameter fluctuations, thus failing to provide reliable maintenance decisions for high-voltage cables, see [reference needed]. Figure 1 , Figure 1 This is a flowchart illustrating the steps of a maintenance decision-making method based on high-voltage cable condition assessment and lifespan prediction, provided in one embodiment of the present invention. Figure 1 As shown in the figure, this invention proposes a maintenance decision-making method based on high-voltage cable condition assessment and life prediction, including steps 101 to 105, each step of which is as follows: Step 101: Based on the pre-collected real-time operation data, environmental parameters, and historical operation and maintenance data of the high-voltage cable, construct an initial evaluation index system; Step 102: Whenever an index in the initial evaluation index system deviates from the preset baseline threshold or an abnormal working condition is detected, the evaluation index system is dynamically adjusted to obtain a dynamic evaluation index system. Step 103: Determine the weight of each indicator in the dynamic evaluation index system based on the preset weighting algorithm, and adaptively fuse each indicator in the dynamic evaluation index system according to the preset neural network to obtain a multi-source fusion feature dataset. Step 104: Input the multi-source fusion feature dataset into the pre-constructed time series prediction model, output the high-voltage cable condition assessment result and the high-voltage cable remaining life prediction value, and correct the high-voltage cable remaining life prediction value according to the preset physical mechanism model to obtain the corrected high-voltage cable remaining life prediction value. Step 105: Based on the high-voltage cable condition assessment results and the corrected high-voltage cable remaining life prediction value, generate a high-voltage cable maintenance decision.

[0033] One specific implementation method involves acquiring real-time operating data, environmental parameters, and maintenance history data of high-voltage cables at key locations such as the high-voltage cable body, intermediate joints, and terminal heads using a combination of online real-time acquisition and offline supplementary acquisition. After performing preprocessing operations such as outlier removal, missing value completion, noise removal, dimensional standardization, and key feature extraction on the acquired multi-source heterogeneous data, an initial evaluation index system covering five dimensions—electrical performance, mechanical structure, insulation characteristics, environmental factors, and operating history—is integrated and constructed. This initial evaluation index system comprehensively covers the core influencing factors affecting the operating status and lifespan of high-voltage cables. Then, the operating status of each indicator within the system is continuously monitored in real time. For continuously changing indicator parameters such as temperature, partial discharge, and dielectric loss, a time series model is used to establish historical time series data for each indicator. The currently collected time series data is compared with the historical time series data in real time. At the same time, the real-time operating data and environmental parameters are constructed into a multi-dimensional feature vector. The normal operating conditions and abnormal operating conditions are divided by density clustering algorithm. When any indicator parameter is detected to deviate from the preset baseline threshold, or when an abnormal operating condition that is not classified into the normal clustering area is identified, the adjustment process of the evaluation indicator system is automatically triggered. Combined with a preset scoring model that can simulate the experience of domain experts, the importance rating of the adjusted indicators is evaluated to complete the optimization of the indicator hierarchy and weight ratio, resulting in a dynamic evaluation indicator system that adapts to the current operating conditions, ensuring that the indicator system can respond in real time to sudden changes in the cable operating environment and abnormal fluctuations in parameters. After obtaining the dynamic evaluation index system, a preset combined weighting algorithm is used to calculate the weight of each index. In this embodiment, the preset combined weighting algorithm includes subjective weighting, objective weighting, dimensionality reduction weight calculation, and multi-weight coordination optimization process. Then, the optimal combined weight of each index is constructed by combining the objective laws of the data itself, that is, the combined index weight of each index. Subsequently, a multi-source index association topology is constructed with the index data of each dimension as nodes and the correlation strength between the indicators as edges. The contribution weight of each node to state assessment and lifetime prediction is calculated through a self-attention mechanism. Combined with a preset graph neural network, the adaptive fusion of multi-source heterogeneous data is completed to obtain a multi-source fusion feature dataset for subsequent model training and prediction.

[0034] The obtained multi-source fusion feature dataset is input into a pre-built and trained time-series prediction model. The model outputs the high-voltage cable condition assessment results in the 0-1 interval and the initial high-voltage cable remaining life prediction value. Subsequently, a preset physical mechanism model is introduced to correct the initial remaining life prediction value. The activation energy of the cable insulation material and the actual operating temperature are combined to complete the physical calibration of the life value. Then, the corrected life value is calibrated again by combining the cable's historical fault records and maintenance records, finally obtaining the corrected high-voltage cable remaining life prediction value. Finally, the cable operating status is classified according to the preset condition classification standard. At the same time, combined with the interval classification results of the remaining life, the degree of cable degradation, degradation type and potential degradation location are located. The historical operation and maintenance records of the cable are matched with the current operation and maintenance resource configuration of the power grid to determine the appropriate maintenance method, maintenance cycle and maintenance scope, and finally generate a high-voltage cable maintenance decision that can be directly implemented.

[0035] This invention proposes a maintenance decision-making method based on high-voltage cable condition assessment and lifespan prediction. It constructs an initial assessment index system and dynamically adjusts it according to sudden environmental changes or abnormal parameter fluctuations. Multi-source fusion is then used to eliminate redundancy and interference from heterogeneous data. Finally, combining the advantages of a pre-constructed time-series model and a preset physical mechanism model, reliable high-voltage cable condition assessment results and remaining lifespan prediction results are output, thereby generating reliable high-voltage cable maintenance decisions. Thus, by constructing a dynamic assessment index system and adaptively fusing the indicators, the reliability of maintenance decisions is improved to adapt to sudden changes in the cable operating environment or abnormal parameter fluctuations.

[0036] In a preferred embodiment, the initial evaluation index system is constructed based on pre-collected real-time operation data, environmental parameters, and historical maintenance data of the high-voltage cable. This includes: preprocessing the collected real-time operation data, environmental parameters, and historical maintenance data of the high-voltage cable to obtain optimized real-time operation data, optimized environmental parameters, and optimized historical maintenance data; constructing electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators based on the optimized real-time operation data of the high-voltage cable; constructing environmental factor indicators based on the optimized environmental parameters; constructing historical operation indicators based on the optimized historical maintenance data; and integrating the electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and historical operation indicators to obtain the initial evaluation index system.

[0037] For example, preprocessing operations are performed on the collected real-time operation data, environmental parameters, and historical maintenance data of high-voltage cables. These preprocessing operations include using the 3σ criterion to remove outliers exceeding the mean ± 3 standard deviations; and for missing values ​​generated during data collection, K-nearest neighbor interpolation is used to fill in missing values ​​based on weights assigned according to the Euclidean distance between adjacent samples. Specific calculation formulas include: ; In the formula, To complete the missing data in the j-th dimension for the i-th sample. For the k-th nearest neighbor sample, the j-th dimension is the valid data. The weight of the k-th nearest neighbor sample is calculated as follows: , Let be the Euclidean distance between the i-th sample and the k-th nearest neighbor sample.

[0038] To address electromagnetic interference in partial discharge signals and random noise in temperature data, a db4 wavelet is selected as the base wavelet for wavelet threshold denoising with 5-level decomposition. This preserves the effective features of the data while eliminating noise interference. Specifically, this includes: ; In the formula, λ is the wavelet threshold, σ is the noise standard deviation, n is the data length, and the estimation formula for σ is: ; In the formula, x represents the original data before noise reduction; After noise reduction, the effective features of the data are preserved while noise interference is eliminated to ensure the authenticity of partial discharge signals and temperature data. Finally, considering the dimensional differences between different types of data, the min-max normalization method is used to map all data to the [0,1] interval to eliminate the influence of dimensions on subsequent calculations. The calculation process includes: ; In the formula, For the standardized data of the i-th sample in the j-th dimension, The original data before standardization. The minimum value of all data in the j-th dimension. The maximum value of all data in the j-th dimension is given. Finally, optimized real-time operation data of the high-voltage cable, optimized environmental parameters, and optimized historical operation and maintenance data are obtained after cleaning, noise reduction, completion, and standardization.

[0039] The optimized real-time operation data of the high-voltage cable is used to construct three categories of core operation indicators. The electrical performance indicators cover partial discharge, insulation resistance, dielectric loss tangent, metal sheath grounding current, conductor temperature, and voltage fluctuation, comprehensively reflecting the cable's electrical operational stability. The mechanical structure indicators cover mechanical stress and strain, vibration parameters, sheath deformation and breakage rate, axial stress, and bending radius deviation, characterizing the physical integrity of the cable body and its auxiliary structures. The insulation characteristic indicators cover polarization index, space charge density, crosslinking degree, oxidation induction period, and carbonyl index, quantifying the aging degree of the cable insulation material. Environmental factor indicators are constructed based on optimized environmental parameters, covering soil corrosion index, temperature and humidity gradient, air pressure, stray current intensity, and electromagnetic interference intensity, fully incorporating the corrosive effects of the external environment on the cable. Finally, operational history indicators are constructed based on optimized maintenance history data, covering service life, overload cycles, fault trip rate, maintenance frequency, and joint replacement records, incorporating the cumulative effects of the cable's entire lifecycle on its condition.

[0040] After completing the construction of indicators for each dimension, electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operating history indicators are used as primary indicators, and the quantifiable parameters corresponding to each primary indicator are used as secondary indicators, thus forming a hierarchical and comprehensive initial evaluation indicator system.

[0041] In the above scheme, the real-time operating data, environmental parameters, and operation and maintenance history data of high-voltage cables are preprocessed and optimized to construct a five-dimensional index system encompassing electrical performance, mechanical structure, insulation characteristics, environmental factors, and operating history. This system comprehensively covers the core dimensions affecting the cable's operating status, avoiding the one-sidedness of assessments caused by single indicators. The multi-dimensional indicators can also fully capture all the changes in the cable from its own performance to the external environment, providing reliable input data for subsequent high-voltage cable condition assessment and high-voltage cable remaining life prediction, which helps improve the reliability of maintenance decisions.

[0042] In a preferred embodiment, whenever an indicator in the initial evaluation indicator system deviates from a preset baseline threshold or an abnormal operating condition is detected, the evaluation indicator system is dynamically adjusted to obtain a dynamic evaluation indicator system. This includes: selecting target indicator parameters from the initial evaluation indicator system and obtaining real-time time-series data corresponding to the target indicator parameters; obtaining historical time-series data corresponding to the target indicator parameters through a preset time-series model, and comparing the real-time time-series data and the historical time-series data to obtain a comparison result; constructing a multi-dimensional feature vector based on the real-time time-series data, and performing density clustering on the multi-dimensional feature vector based on a preset clustering algorithm to classify abnormal operating conditions; whenever any target indicator parameter deviates from the preset baseline threshold or an abnormal operating condition is detected in the comparison result, adjusting the indicator topology of the initial evaluation indicator system to obtain an adjusted initial evaluation indicator system; and assigning importance ratings to the adjusted initial evaluation indicators based on a preset scoring model to obtain a dynamic evaluation indicator system.

[0043] For example, core parameters that are highly sensitive to changes in cable condition are selected from the established initial evaluation index system as target index parameters. In this embodiment, these include key continuously changing parameters such as electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators. Then, principal component analysis is used for dimensionality reduction, retaining only principal components with a cumulative contribution rate ≥ 85%, specifically including: ; In the formula, Y is the feature matrix after dimensionality reduction, X is the original feature matrix after extraction, and W is the principal component projection matrix, which is composed of the eigenvectors of the covariance matrix of the feature matrix.

[0044] Then, a distributed sensor network is used to continuously collect and acquire real-time time-series data corresponding to the dimensionality-reduced target index parameters. A preset time series model is then used, in this embodiment, the ARIMA time series model is used. Based on the historical normal operation data of the target index parameters, the historical change trend and baseline threshold of each index are obtained, and the historical time-series data of the corresponding standard are acquired. The real-time time-series data and the historical time-series data are compared point by point, and the deviation value, parameter change slope and fluctuation amplitude of the real-time data and the baseline threshold are calculated to obtain a complete comparison result. This is used to determine whether each target index parameter has a continuous offset or abnormal fluctuation. In this embodiment, the preset baseline threshold can be set according to the actual application situation, and is not specifically limited in this embodiment.

[0045] Simultaneously, real-time operating data and environmental factor indicators within the same time window are integrated to construct a multi-dimensional feature vector. A preset clustering algorithm is used. In this embodiment, the DBSCAN clustering algorithm is used to perform density clustering operations on the multi-dimensional feature vector. Samples that conform to normal operating rules are identified as stable clustering regions, and outlier samples that cannot be classified into normal clustering regions and have sparse density are classified as abnormal operating conditions. Abnormal operating conditions specifically include abnormal cable operating states such as sudden increase in dielectric loss, excessive electromagnetic interference, abnormal enhancement of partial discharge, and sudden change in sheath stress. When the comparison results show that any target indicator parameter deviates from the preset baseline threshold, or when clustering operation identifies an abnormal working condition, the adjustment process of the evaluation indicator system is automatically triggered. This includes adjusting the indicator topology of the initial evaluation indicator system according to the current anomaly type and parameter offset. In this embodiment, the adjustment process can improve the level of indicators strongly correlated with the current anomaly, add corresponding subdivided monitoring indicators, remove redundant indicators unrelated to the current working condition, and adjust the monitoring frequency of core indicators, etc., and finally obtain the adjusted initial evaluation indicator system. Finally, a pre-set scoring model is introduced. This model can simulate professionals in the field of high-voltage cable operation and maintenance and research. It can rate the importance of each adjusted indicator from multiple dimensions, such as the importance of physical mechanism, the relevance of operation and maintenance decision-making, and the sensitivity to abnormal states. Based on the rating results, the hierarchical structure and weight ratio of the indicators are further dynamically optimized, invalid indicators with too low an importance rating are eliminated, and the weight ratio of core indicators is strengthened. Finally, a dynamic evaluation indicator system that can adapt to the current cable operating conditions and can dynamically respond to sudden changes in the operating environment and abnormal fluctuations in parameters is obtained.

[0046] In the above scheme, target indicator parameters are selected and corresponding real-time and historical time-series data are obtained. By comparing the real-time and historical time-series data, it can be determined whether the indicator deviates from the preset benchmark threshold. At the same time, multi-dimensional feature vectors are constructed using real-time time-series data and abnormal operating conditions are identified through density clustering. Then, when any target indicator parameter deviates from the preset baseline threshold or an abnormal operating condition occurs, the indicator topology of the initial evaluation indicator system is dynamically adjusted, and the indicator topology and weight ratio are optimized in real time. This avoids the shortcomings of static indicator systems that cannot adapt to environmental changes and parameter fluctuations. Furthermore, the reliability of the dynamic evaluation indicator system is further improved by combining importance scoring, which helps to improve the reliability of maintenance decisions.

[0047] A preferred embodiment involves determining the weights of each indicator in the dynamic evaluation indicator system based on a preset weighting algorithm, and adaptively fusing the indicators in the dynamic evaluation indicator system according to a preset neural network to obtain a multi-source fused feature dataset. The embodiment includes: determining the individual weights of each indicator in the dynamic evaluation indicator system based on the preset weighting algorithm; constructing a weighted normalization matrix based on the individual weights of each indicator, and performing coordinated optimization on the weighted normalization matrix to obtain optimized weights; constructing a weight vector matrix by combining the individual weights of each indicator and the optimized weights, and solving the weight vector matrix using a preset solution algorithm to obtain the combined indicator weights of each indicator in the dynamic evaluation indicator system; constructing a multi-source indicator association topology based on the combined indicator weights and the indicators in the dynamic evaluation indicator system, and calculating the attention weights of each indicator in the multi-source indicator association topology based on a preset attention weight algorithm; and adaptively fusing the indicators in the dynamic evaluation indicator system based on the preset neural network and the attention weights of each indicator to obtain a multi-source fused feature dataset.

[0048] For example, a preset combined weighting algorithm is used to sequentially calculate the individual weights of each indicator in the dynamic evaluation index system. In this embodiment, the preset combined weighting algorithm includes subjective weighting, objective weighting, dimensionality reduction weight calculation, and multi-weight coordination optimization processes. First, subjective weights are calculated using the analytic hierarchy process (AHP). Five to ten experts in the high-voltage cable field are invited to conduct pairwise ratings of the relative importance of the primary and secondary indicators to construct a judgment matrix. The eigenvector method is then used to calculate the weights, and the consistency ratio (CR) is controlled within 0.1 to ensure rating consistency. Specifically, the judgment matrix... A The construction process includes: ; In the formula: Indicates the first The first indicator is relative to the first The importance comparison values ​​of each indicator; The consistency verification process includes: ; ; In the formula, To determine the matrix The largest eigenvalue, It is a random consistency indicator; when At that time, it is assumed that the judgment matrix satisfies the consistency requirement, and the results of each indicator are obtained accordingly. The subjective weight vector is denoted as: ; Then, the objective weights are calculated using the entropy weight method. Information entropy and difference coefficients are calculated on the standardized indicator data to fully extract the distribution information inherent in the data and obtain the objective weights. Specifically, after the indicator data has been standardized, a standardized data matrix is ​​set up. X for: ; In the formula, Indicates the first The sample at the th Standardized values ​​for each indicator Represents the total number of samples. Indicates the total number of indicators; Among them, the The first indicator in the The proportion in each sample for: ; Then calculate the first... The information entropy of each indicator is calculated as follows: ; In the formula: ; No. Coefficient of difference of each indicator for: ; Then, the corresponding objective weights of the entropy weight method are calculated using the following formula: ; The final objective weight vector obtained by the entropy weight method is denoted as: ; Finally, principal component analysis was used to calculate the dimensionality reduction weights. Principal components with a cumulative contribution rate of no less than 85% were extracted from the high-dimensional index data. The dimensionality reduction weights were calculated based on the loading coefficients of each index in the principal components. Specifically, the eigenvalues ​​corresponding to each principal component were set as follows: Before selection Each principal component satisfies a cumulative contribution rate that is not less than a preset contribution threshold, i.e.: ; In the formula, The number of principal components selected; Then set Represented as the first The first indicator in the The loading coefficients on the principal components, then the... The PCA dimensionality reduction weights corresponding to each indicator are expressed as follows: ; This yields the PCA dimensionality reduction weight vector, expressed as: ; This completes the calculation of the weights for the three categories of individual items.

[0049] The obtained three types of individual weights are used as input to construct a weighted standardized matrix. The TOPSIS method is used to coordinate and optimize the weights. The positive and negative ideal solutions corresponding to each index are determined respectively. The Euclidean distance and relative proximity between each index and the two types of ideal solutions are calculated. After normalizing the relative proximity, the TOPSIS optimized weights are obtained. Specifically, let the evaluation matrix formed by different weighting methods be represented as: ; In the formula, This indicates the h-th weighting method. The weight values ​​of each indicator, Indicates the number of weighting methods involved in the optimization; Then for the first Each indicator is used to determine the positive ideal solution. With negative ideal solution Then its distances to the positive ideal solution and the negative ideal solution are respectively: ; ; Calculate the next The relative similarity of the indicators is expressed as: ; Then to After normalization, the TOPSIS optimization weights are obtained, which are expressed as: ; The final optimized weights are represented as follows: ; The three types of individual weights and the TOPSIS optimized weights are combined to construct a 4xm weight vector matrix, where m is the total number of indicators in the dynamic evaluation index system. The sum of the Euclidean distances between each individual weight and the final combined weight is used as the deviation function. The optimal combination coefficient that minimizes the deviation function is obtained through game theory equilibrium solution. Then, the combined index weights of each indicator in the dynamic evaluation index system are calculated. Specifically, let the final combined weight vector be: ; In the formula, the combination coefficient The following relationship must be satisfied: ; Then, to minimize the overall deviation between the final combined weights and the individual weighting results, a deviation function is constructed, expressed as: ; In the formula, Indicates the first The weight vector obtained by a single weighting method is solved to make The combination coefficient corresponding to the minimum value The final combined weight vector is obtained. .

[0050] Then, using the indicator data of each dimension in the dynamic evaluation indicator system as graph nodes, and the Pearson correlation coefficient between indicators as the correlation strength of the edges between nodes, a multi-source indicator correlation topology is constructed, represented as: ; In the formula, Let be the correlation strength between the i-th node and the j-th node, i.e., the correlation coefficient, with a value range of [-1, 1]. and These are the data from the i-th node and the j-th node in the k-th sample, respectively. and These are the mean values ​​of all data at the i-th node and the mean values ​​of all data at the j-th node, respectively, where n is the number of samples. In this embodiment, the larger the absolute value of the correlation strength, the stronger the correlation between the two indicators, and the higher the weight of the corresponding edge.

[0051] Subsequently, the attention weights of each node in the multi-source index association topology are calculated based on a preset self-attention weight algorithm. The score value of each node is obtained by calculating the node features and weight matrix, and the corresponding attention weight is obtained after normalization by the softmax function. The calculation process is as follows: ; In the formula, Let be the attention weight of the i-th node. Let m be the rating value of the i-th node, and m be the total number of nodes, i.e., the total number of data sources; rating value It is calculated from the node features and weight matrix, i.e. ,in and This is the weight matrix. and Let be the feature vectors of the i-th node and the j-th node. In this embodiment, the weight directly reflects the contribution of the index to cable condition assessment and life prediction. At the same time, a dynamic adjustment mechanism is set up to recalculate the node association strength and attention weight at fixed intervals. The weight allocation is adaptively adjusted according to the changes in cable operating status. When a data source has continuous anomalies or is missing, its attention weight is automatically reduced to below 0.05.

[0052] Finally, based on the pre-defined graph neural network, and combined with the attention weights of each node, adaptive fusion of multi-source heterogeneous data is achieved. First, the graph neural network completes information propagation and feature aggregation between adjacent nodes. Then, the aggregated features of each node are weighted and fused using attention weights. The specific calculation process is as follows: ; In the formula, This is the final fused feature vector. Here is the weight matrix of the graph neural network. For bias terms, For the first Attention weights for each node, Let be the feature vector of the i-th node. In this embodiment, the proportion of high contribution indicators in the fusion result is increased, and the influence of abnormal or low contribution indicators is effectively suppressed. Finally, a multi-source fusion feature dataset is obtained that retains the key features of each data source and eliminates data redundancy and interference, providing high-quality input data for subsequent time series prediction model training and state evaluation.

[0053] In the above scheme, after determining the individual weights of each indicator using a preset weighting algorithm, the optimized weights are obtained through coordination optimization using a weighted standardization matrix. Then, the individual weights and optimized weights are combined to construct a weight vector matrix and solve for the combined indicator weights. Subsequently, a multi-source indicator correlation topology is constructed and attention weights are calculated. Finally, adaptive fusion is achieved through a preset neural network. This eliminates human bias and data bias caused by single weighting and suppresses the interference of abnormal data through the attention mechanism. It provides reliable input data for subsequent high-voltage cable condition assessment and high-voltage cable remaining life prediction, which helps to improve the reliability of maintenance decisions.

[0054] In a preferred embodiment, the step of inputting the multi-source fusion feature dataset into a pre-constructed time-series prediction model, outputting a high-voltage cable status assessment result and a high-voltage cable remaining life prediction value, and correcting the high-voltage cable remaining life prediction value according to a preset physical mechanism model to obtain a corrected high-voltage cable remaining life prediction value includes: training an initial time-series prediction model based on the multi-source fusion feature dataset to obtain a target time-series prediction model; inputting the multi-source fusion feature dataset into the target time-series prediction model to output a high-voltage cable operation status assessment result and an initial remaining life prediction value; inputting the initial remaining life prediction value into a preset physical mechanism model to obtain an adjusted remaining life prediction value; and calibrating the adjusted remaining life prediction value based on the historical operation and maintenance data to obtain the corrected high-voltage cable remaining life prediction value.

[0055] For example, the multi-source fusion feature dataset that has completed dynamic fusion is used as the training data for the initial time series prediction model. After the dataset is divided into normalized parts, it is used for parameter training, hyperparameter validation and performance testing of the model. By constructing an initial time series prediction model with a long short-term memory network as the core architecture, the model parameters are updated and optimized through multiple rounds of iteration. At the same time, the core configuration of the model is adjusted in combination with the validation set results, and an early stopping strategy is adopted to avoid overfitting of the model. Finally, the target time series prediction model that has completed training and whose performance meets the requirements of operation and maintenance is obtained.

[0056] After obtaining the target time series prediction model, the multi-source fusion feature dataset obtained through real-time acquisition, standardized preprocessing, and dynamic fusion is input into the target time series prediction model. The model outputs two types of core prediction results. The first type is the evaluation result of the high-voltage cable operation status with values ​​ranging from 0 to 1. The closer the value is to 1, the better the overall operation status and insulation performance of the cable. The second type is the initial remaining life prediction value in years. At the same time, the target time series prediction model will also output the confidence level of this status evaluation result. When the confidence level is greater than or equal to 90%, the evaluation and prediction results are considered valid. If the confidence level is less than 90%, the full-dimensional data of the corresponding cable is re-acquired, and the above process is repeated.

[0057] Then, the initial remaining lifetime prediction value output by the target time series prediction model is input into the preset Arrhenius physical mechanism model for correction, specifically: ; In the formula, L represents the corrected remaining cable life, in years. To optimize the original remaining life prediction values ​​output by the model, in years, is the activation energy of the insulating material, k is the Boltzmann constant, and T is the average operating temperature of the cable, in °C. In this embodiment, the physical mechanism model is a classic physical model in the field of cable life prediction, which is used to describe the influence of operating temperature on the aging rate of cable insulation material. Combining the activation energy of cable insulation material, Boltzmann constant and historical average operating temperature of the cable section, the initial life prediction value obtained by pure data-driven model is adapted and adjusted at the physical level to make up for the defects of pure data-driven model lacking physical mechanism support and insufficient generalization, thereby obtaining the remaining life prediction value after physical mechanism adjustment.

[0058] Finally, based on the collected historical operation and maintenance data of the entire life cycle of this high-voltage cable, the adjusted remaining life prediction value is calibrated a second time. Combining the cable's actual service life, historical fault trip records, historical maintenance frequency, intermediate joint and terminal head replacement records, and other historical operation and maintenance information, a suitable calibration coefficient is determined by matching the standard maintenance frequency under the corresponding service life. The adjusted life value is then calibrated specifically. Specifically: ; In the formula, The final remaining cable life, in years. The calibration factor, determined in this embodiment, is based on the cable's service life. This represents the number of historical maintenance visits to the cable. The standard number of maintenance checks corresponding to the total service life of the cable; Therefore, by eliminating the impact of operation and maintenance interventions and fault repairs on the actual remaining life of the cable, a corrected prediction value for the remaining life of the high-voltage cable is obtained, which has both data-driven prediction accuracy and physical mechanism rationality, and is also adapted to the actual operation and maintenance of the cable. When outputting the life value, one decimal place is retained and the corresponding prediction error range is given simultaneously.

[0059] In the above scheme, the trained target time-series prediction model outputs the high-voltage cable operation status assessment results and initial remaining life prediction values. Then, the remaining life prediction values ​​are corrected by the physical mechanism model. Furthermore, the remaining life prediction values ​​are calibrated a second time by combining historical operation and maintenance data. This approach leverages the time-series prediction model's ability to capture time-series characteristics and compensates for the lack of physical rationality in pure data-driven models through the physical mechanism model. At the same time, the calibration of historical operation and maintenance data further adapts to the actual service conditions of the cable. The final output of remaining life results is more accurate and reasonable, accurately reflecting the true degree of cable degradation and improving the reliability of maintenance decisions.

[0060] In a preferred embodiment, training an initial time-series prediction model based on the multi-source fusion feature dataset to obtain a target time-series prediction model includes: performing stratified sampling on the multi-source fusion feature dataset to obtain a model training set, a model validation set, and a model test set; constructing an initial time-series prediction model; training the initial time-series prediction model using the model training set; and optimizing the hyperparameters of the initial time-series prediction model based on the model validation set and the model test set to obtain the target time-series prediction model.

[0061] For example, the multi-source fusion feature dataset that has completed dynamic fusion is first divided into a model training set, a model validation set, and a model test set in a ratio of 7:2:1 using a stratified sampling method. During the stratified sampling process, the service life, operating condition level, and insulation aging degree of the high-voltage cable are used as the stratification criteria to ensure that the sample distribution of the three datasets is consistent and to avoid model overfitting caused by data skew. In this embodiment, the model training set is used for iterative training of the model's weight parameters, the model validation set is used for model performance verification during the hyperparameter optimization process, and the model test set is used for the final generalization performance test of the model after training.

[0062] Then, an initial time-series prediction model with a Long Short-Term Memory (LSTM) network as its core architecture is constructed. The initial time-series prediction model is set as a three-layer LSTM hidden layer structure. The dimension of the input layer matches the dimension of the multi-source fusion feature vector. In this embodiment, the dimension of the multi-source fusion feature vector is determined by the number of principal components after dimensionality reduction by PCA. Then, the first hidden layer is set with 64 hidden units, the second layer with 32 hidden units, and the third layer with 16 hidden units. The hidden layers use the ReLU activation function to avoid the gradient vanishing problem during training. The output layer dimension is set to 2, corresponding to the high-voltage cable operating status evaluation value and the remaining life prediction value, respectively. At the same time, Dropout regularization and L2 regularization are configured for the model. The Dropout probability is set to 0.2 and the weight decay coefficient is set to 0.001 to constrain the model parameter size and further reduce the risk of overfitting.

[0063] In this embodiment, the calculation process of the LSTM model includes: ; ; ; ; ; ; In the formula, , and These are the output values ​​of the input gate, forget gate, and output gate, respectively. Candidate cell state, The current cell state, Output for the current hidden layer. For the current input feature vector, and Equal to the weight matrix, and Equal to the bias term, It is the sigmoid activation function. Dot product of elements; During model training, mean squared error is used as the loss function, and the Adam optimizer is selected to complete the parameter iterative update. The learning rate is set to decay adaptively with the number of iterations according to a decay coefficient of 0.95. Every 10 iterations, the model validation set is used to verify the prediction performance of the current model. If the loss function value of the validation set does not decrease for 20 consecutive iterations, the early stopping strategy is triggered to terminate the training and avoid overfitting of the model.

[0064] During training, a Bayesian optimization algorithm is simultaneously employed to adaptively optimize the key hyperparameters of the initial time-series prediction model. The optimization scope covers the learning rate, number of hidden layer units, Dropout probability, training batch size, and number of iterations. A Gaussian process probabilistic model is constructed with the root mean square error of the model validation set as the objective function. The probabilistic model is continuously updated through iterative sampling to find the optimal combination of hyperparameters that minimizes the objective function. Specifically: ; In the formula, The validation set contains true values, which include cable condition assessment values ​​and remaining lifetime values. Here, n represents the predicted value of the model, and n is the number of samples in the validation set; when the change in the objective function value over 10 consecutive iterations is less than 1... The optimization process terminates when the optimal hyperparameter configuration is obtained.

[0065] After completing hyperparameter optimization, the model is retrained using the optimal hyperparameter configuration. After training, the model's generalization performance and prediction accuracy are tested using a model test set. If the test results do not meet the preset accuracy requirements, the hyperparameter optimization range is readjusted and the training and optimization process is repeated until the model performance meets the preset accuracy requirements for high-voltage cable condition assessment and life prediction. Finally, the target time series prediction model is obtained after training.

[0066] In the above scheme, stratified sampling is used to divide the multi-source fusion dataset. Stratified sampling can ensure the consistency of the distribution of each dataset and avoid the model overfitting problem caused by data skew. Then, the model training set is used to complete the model training, and the model validation set and model test set are used to achieve multi-dataset iterative optimization of the model hyperparameters. This can improve the generalization ability and prediction accuracy of the time series prediction model, and at the same time ensure that the model can output stable and reliable prediction results in cable samples in different working conditions and different regions, thereby improving the reliability of maintenance decisions.

[0067] In a preferred embodiment, generating a high-voltage cable maintenance decision based on the high-voltage cable condition assessment results and the corrected predicted remaining life of the high-voltage cable includes: classifying the high-voltage cable condition level from the high-voltage cable condition assessment results, and classifying the remaining life level from the corrected predicted remaining life of the high-voltage cable; determining the high-voltage cable degradation type and degradation location based on the high-voltage cable condition level and the classified remaining life level; and determining the corresponding maintenance method, maintenance cycle, and maintenance scope based on the high-voltage cable degradation type, degradation location, and historical maintenance data to generate a high-voltage cable maintenance decision.

[0068] For example, the output high-voltage cable condition assessment results are classified into condition levels according to preset classification standards, as shown in Table 1: Table 1 Classification Criteria In Table 1, an evaluation value between 0.8 and 1.0 is classified as excellent, corresponding to slight aging of the cable insulation and good overall performance; between 0.6 and 0.8 is classified as good, corresponding to slight aging of the cable and insulation performance basically meeting operational requirements; between 0.4 and 0.6 is classified as fair, corresponding to moderate aging of the cable and a significant decline in insulation performance; and between 0.0 and 0.4 is classified as poor, corresponding to severe aging of the cable and the presence of clear potential faults. At the same time, the remaining life prediction values ​​of the revised high-voltage cables were also classified into remaining life levels. Those with a remaining life greater than 15 years were classified as long life level, those with a remaining life of 5 to 15 years were classified as medium life level, and those with a remaining life of less than 5 years were classified as short life level, thus completing the classification and matching of the two levels.

[0069] After completing the classification and matching of the two levels, and combining the high-voltage cable condition level and remaining life level obtained from the classification, the abnormalities of each indicator in the dynamic evaluation index system are traced back to locate the type and location of cable degradation. When electrical performance indicators show significant abnormalities, it is judged as electrical insulation degradation; when mechanical structure indicators deviate from the normal range, it is judged as mechanical structure damage degradation; when environmental factor indicators continuously exceed the standard, it is judged as external environmental corrosion degradation; when operating history indicators show multiple overloads and faults, it is judged as cumulative effect degradation. At the same time, combined with the monitoring point information of each indicator, the degradation location is accurately located to the specific location of the high-voltage cable body, intermediate joints or terminal heads, etc., to clarify the core causes and scope of influence of degradation.

[0070] Then, based on the determined degradation type and location, and combined with the complete maintenance history data of the cable section, including historical maintenance records, fault trip records, overload counts, and joint replacement records, a suitable maintenance method, maintenance cycle, and maintenance scope are determined. For cables with an excellent condition level and a long remaining life, daily inspections are adopted, with a quarterly inspection cycle, and the maintenance scope covers routine parameter monitoring of the entire cable section. For cables with a good condition level and a medium remaining life, live-line testing is added to the daily inspection, and the inspection cycle is adjusted to once every six months, focusing on degradation trends. Specialized monitoring will be conducted on the corresponding parts of the indicators. For cables with a general condition level, specialized insulation diagnosis and structural testing will be carried out, shortening the testing cycle to 3 months and covering all sections with abnormal indicators. For cables with a poor condition level and short remaining life, a work plan for power outage maintenance or complete replacement will be immediately formulated and given priority in the power grid maintenance scheduling. The maintenance scope will cover all deteriorated parts and adjacent related sections. At the same time, based on the cable's historical maintenance situation, the maintenance priority of sections with multiple faults and multiple maintenance will be increased, and the frequency and items of testing will be appropriately increased to avoid the recurrence of faults.

[0071] Ultimately, all information is integrated to form a complete high-voltage cable maintenance decision, clarifying the priority of maintenance work, specific work content, work standards, personnel configuration requirements and spare parts preparation suggestions, and at the same time, providing subsequent operation monitoring requirements, transforming abstract condition assessment and life prediction results into operation and maintenance decisions that can be directly implemented.

[0072] In the above scheme, the cable condition assessment results are used to classify the cable into condition levels, and the revised remaining life prediction values ​​are used to classify the cable life levels. The cable degradation type and location are determined by comprehensively analyzing both levels. Then, historical maintenance data is combined to determine the corresponding maintenance methods, maintenance cycles, and maintenance scope, ultimately generating a high-voltage cable maintenance decision. Therefore, by jointly analyzing the condition level and life level, the degradation type and location can be accurately identified, and differentiated maintenance strategies can be formulated accordingly to prevent resource waste or insufficient maintenance, thereby improving the reliability of maintenance decisions.

[0073] Based on the above method embodiments, corresponding apparatus embodiments are provided; see [link to apparatus embodiments]. Figure 2 , Figure 2 This is a schematic diagram of the module structure of a maintenance decision-making system based on high-voltage cable condition assessment and life prediction, provided in one embodiment of the present invention. Figure 2 As shown, this embodiment of the invention also provides a maintenance decision system based on high-voltage cable condition assessment and life prediction, including an initial indicator system construction module 201, a dynamic evaluation indicator system construction module 202, a data fusion module 203, a high-voltage cable condition assessment and life prediction module 204, and a maintenance decision output module 205; wherein: the initial indicator system construction module 201 is used to construct an initial evaluation indicator system based on pre-collected real-time operating data, environmental parameters, and historical operation and maintenance data of the high-voltage cable; the dynamic evaluation indicator system construction module 202 is used to dynamically adjust the evaluation indicator system whenever an indicator in the initial evaluation indicator system deviates from a preset baseline threshold or an abnormal operating condition is detected, thereby obtaining a dynamic evaluation indicator system; the data The fusion module 203 is used to determine the weight of each indicator in the dynamic evaluation index system based on a preset weighting algorithm, and to adaptively fuse each indicator in the dynamic evaluation index system according to a preset neural network to obtain a multi-source fusion feature dataset; the high-voltage cable condition assessment and life prediction module 204 is used to input the multi-source fusion feature dataset into a pre-constructed time series prediction model, output the high-voltage cable condition assessment result and the high-voltage cable remaining life prediction value, and correct the high-voltage cable remaining life prediction value according to a preset physical mechanism model to obtain the corrected high-voltage cable remaining life prediction value; the maintenance decision output module 205 is used to generate a high-voltage cable maintenance decision based on the high-voltage cable condition assessment result and the corrected high-voltage cable remaining life prediction value.

[0074] This invention proposes a maintenance decision-making system based on high-voltage cable condition assessment and lifespan prediction. It constructs an initial assessment index system and dynamically adjusts it according to sudden environmental changes or abnormal parameter fluctuations. Multi-source fusion is then used to eliminate redundancy and interference from heterogeneous data. Finally, combining the advantages of a pre-constructed time-series model and a preset physical mechanism model, reliable high-voltage cable condition assessment results and remaining lifespan prediction results are output, thereby generating reliable high-voltage cable maintenance decisions. Thus, by constructing a dynamic assessment index system and adaptively fusing the various indicators, the system can adapt to sudden changes in the cable operating environment or abnormal parameter fluctuations, improving the reliability of maintenance decisions.

[0075] Furthermore, the initial indicator system construction module 201 includes a data preprocessing unit 301, a first indicator construction unit 302, a second indicator construction unit 303, a third indicator construction unit 304, and an indicator integration unit 305; wherein: the data preprocessing unit 301 is used to preprocess the collected real-time operation data, environmental parameters, and historical operation and maintenance data of the high-voltage cable to obtain optimized real-time operation data, optimized environmental parameters, and optimized historical operation and maintenance data of the high-voltage cable; the first indicator construction unit 302 is used to construct electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators based on the optimized real-time operation data of the high-voltage cable; the second indicator construction unit 303 is used to construct environmental factor indicators based on the optimized environmental parameters; the third indicator construction unit 304 is used to construct operation history indicators based on the optimized historical operation and maintenance data; and the indicator integration unit 305 is used to integrate the electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operation history indicators to obtain the initial evaluation indicator system.

[0076] Furthermore, the dynamic evaluation index system construction module 202 includes a real-time data acquisition unit 401, a data comparison unit 402, an abnormal working condition classification unit 403, an index system adjustment unit 404, and an importance rating unit 405; wherein: the real-time data acquisition unit 401 is used to select target index parameters from the initial evaluation index system and acquire the real-time time series data corresponding to the target index parameters; the data comparison unit 402 is used to acquire the historical time series data corresponding to the target index parameters through a preset time series model, and compare the real-time time series data and the historical time series data to obtain a comparison result. As a result, the abnormal operating condition classification unit 403 is used to construct a multi-dimensional feature vector based on the real-time time series data, and perform density clustering on the multi-dimensional feature vector based on a preset clustering algorithm to classify abnormal operating conditions; the indicator system adjustment unit 404 is used to adjust the indicator topology of the initial evaluation indicator system whenever any target indicator parameter deviates from the preset baseline threshold or an abnormal operating condition occurs in the comparison result, to obtain an adjusted initial evaluation indicator system; the importance rating unit 405 is used to perform importance rating on the adjusted initial evaluation indicators based on a preset scoring model to obtain a dynamic evaluation indicator system.

[0077] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can realize the maintenance decision-making method based on high-voltage cable condition assessment and life prediction provided by any of the above-described method embodiments of the present invention.

[0078] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0079] Based on the above embodiment of a maintenance decision-making method based on high-voltage cable condition assessment and life prediction, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a maintenance decision-making method based on high-voltage cable condition assessment and life prediction according to any embodiment of the present invention.

[0080] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0081] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0082] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0083] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium, including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in any of the above-described method embodiments of the present invention.

[0084] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0085] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A maintenance decision-making method based on high-voltage cable condition assessment and life prediction, characterized in that, include: Based on the pre-collected real-time operation data, environmental parameters, and historical operation and maintenance data of high-voltage cables, an initial evaluation index system is constructed. Whenever an indicator in the initial evaluation indicator system deviates from the preset baseline threshold or an abnormal operating condition is detected, the evaluation indicator system is dynamically adjusted to obtain a dynamic evaluation indicator system. The weights of each indicator in the dynamic evaluation index system are determined based on a preset weighting algorithm, and the indicators in the dynamic evaluation index system are adaptively fused according to a preset neural network to obtain a multi-source fusion feature dataset. The multi-source fusion feature dataset is input into a pre-built time-series prediction model, which outputs the high-voltage cable condition assessment result and the high-voltage cable remaining life prediction value. The high-voltage cable remaining life prediction value is then corrected according to a preset physical mechanism model to obtain the corrected high-voltage cable remaining life prediction value. Based on the high-voltage cable condition assessment results and the corrected high-voltage cable remaining life prediction value, a high-voltage cable maintenance decision is generated.

2. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 1, characterized in that, The initial evaluation index system is constructed based on pre-collected real-time operation data of high-voltage cables, environmental parameters, and historical operation and maintenance data, including: The preprocessing of the collected real-time operating data, environmental parameters, and historical maintenance data of the high-voltage cable yields optimized real-time operating data, optimized environmental parameters, and optimized historical maintenance data of the high-voltage cable. Based on the optimized real-time operation data of high-voltage cables, electrical performance indicators, mechanical structure indicators, and insulation characteristic indicators are constructed. Based on the optimized environmental parameters, environmental factor indicators are constructed. Based on the optimized historical operation and maintenance data, construct historical operation indicators; By integrating electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operating history indicators, an initial evaluation indicator system is obtained.

3. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 2, characterized in that, Whenever an index in the initial evaluation index system deviates from a preset baseline threshold or an abnormal operating condition is detected, the evaluation index system is dynamically adjusted to obtain a dynamic evaluation index system, including: Select target indicator parameters from the initial evaluation indicator system and obtain the real-time time series data corresponding to the target indicator parameters; Historical time series data corresponding to the target indicator parameters are obtained by using a preset time series model, and the real-time time series data and the historical time series data are compared to obtain the comparison results. Based on the real-time time series data, a multi-dimensional feature vector is constructed, and the multi-dimensional feature vector is subjected to density clustering based on a preset clustering algorithm to divide abnormal working conditions. Whenever any target indicator parameter deviates from the preset baseline threshold or an abnormal working condition occurs in the comparison result, the indicator topology of the initial evaluation indicator system is adjusted to obtain the adjusted initial evaluation indicator system. Based on the preset scoring model, the importance rating of the adjusted initial evaluation indicators is obtained to obtain a dynamic evaluation indicator system.

4. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 1, characterized in that, The weights of each indicator in the dynamic evaluation indicator system are determined based on a preset weighting algorithm, and the indicators in the dynamic evaluation indicator system are adaptively fused according to a preset neural network to obtain a multi-source fusion feature dataset, including: The individual weights of each indicator in the dynamic evaluation index system are determined based on a preset weighting algorithm. A weighted standardized matrix is ​​constructed based on the individual weights of each indicator, and the weighted standardized matrix is ​​coordinated and optimized to obtain the optimized weights. The individual weights of each indicator and the optimized weights are used to construct a weight vector matrix, and the weight vector matrix is ​​solved by a preset solution algorithm to obtain the combined indicator weights of each indicator in the dynamic evaluation indicator system. Based on the combined index weights of each index in the dynamic evaluation index system and the indexes in the dynamic evaluation index system, a multi-source index association topology is constructed, and the attention weights of each index in the multi-source index association topology are calculated based on a preset attention weight algorithm. Based on the preset neural network and the attention weights of each indicator, the indicators in the dynamic evaluation indicator system are adaptively fused to obtain a multi-source fusion feature dataset.

5. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 1, characterized in that, The process of inputting the multi-source fusion feature dataset into a pre-constructed time-series prediction model, outputting high-voltage cable condition assessment results and high-voltage cable remaining life prediction values, and correcting the high-voltage cable remaining life prediction values ​​according to a preset physical mechanism model to obtain corrected high-voltage cable remaining life prediction values ​​includes: The initial time-series prediction model is trained based on the multi-source fusion feature dataset to obtain the target time-series prediction model; The multi-source fusion feature dataset is input into the target time series prediction model, and the high-voltage cable operation status assessment result and initial remaining life prediction value are output. The initial remaining lifetime prediction value is input into a preset physical mechanism model to obtain the adjusted remaining lifetime prediction value; Based on the historical maintenance data, the adjusted remaining life prediction value is calibrated to obtain the corrected remaining life prediction value for the high-voltage cable.

6. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 5, characterized in that, The step of training the initial time-series prediction model based on the multi-source fusion feature dataset to obtain the target time-series prediction model includes: The multi-source fusion feature dataset is hierarchically sampled and divided to obtain a model training set, a model validation set, and a model test set. An initial time series prediction model is constructed, trained using the model training set, and its hyperparameters are optimized based on the model validation set and the model test set to obtain the target time series prediction model.

7. The maintenance decision-making method based on high-voltage cable condition assessment and life prediction as described in claim 6, characterized in that, The process of generating a high-voltage cable maintenance decision based on the high-voltage cable condition assessment results and the corrected predicted remaining life of the high-voltage cable includes: The high-voltage cable condition level is determined from the high-voltage cable condition assessment results, and the remaining life level is determined from the corrected high-voltage cable remaining life prediction value. The type and location of high-voltage cable degradation are determined based on the condition level of the high-voltage cable and the classification of remaining life levels. Based on the type of high-voltage cable degradation, the location of the degradation, and the historical maintenance data, the corresponding maintenance method, maintenance cycle, and maintenance scope are determined to generate a high-voltage cable maintenance decision.

8. A maintenance decision-making system based on high-voltage cable condition assessment and life prediction, characterized in that, It includes an initial indicator system construction module, a dynamic evaluation indicator system construction module, a data fusion module, a high-voltage cable condition assessment and life prediction module, and a maintenance decision output module; among which: The initial indicator system construction module is used to construct an initial evaluation indicator system based on the pre-collected real-time operation data, environmental parameters, and historical operation and maintenance data of high-voltage cables. The dynamic evaluation index system construction module is used to dynamically adjust the evaluation index system whenever an index in the initial evaluation index system deviates from the preset baseline threshold or an abnormal working condition occurs, thereby obtaining a dynamic evaluation index system. The data fusion module is used to determine the weight of each indicator in the dynamic evaluation indicator system based on a preset weighting algorithm, and to adaptively fuse each indicator in the dynamic evaluation indicator system according to a preset neural network to obtain a multi-source fusion feature dataset. The high-voltage cable condition assessment and life prediction module is used to input the multi-source fusion feature dataset into a pre-built time-series prediction model, output the high-voltage cable condition assessment result and the high-voltage cable remaining life prediction value, and correct the high-voltage cable remaining life prediction value according to the preset physical mechanism model to obtain the corrected high-voltage cable remaining life prediction value. The maintenance decision output module is used to generate a high-voltage cable maintenance decision based on the high-voltage cable condition assessment results and the corrected high-voltage cable remaining life prediction value.

9. The maintenance decision-making system based on high-voltage cable condition assessment and life prediction as described in claim 8, characterized in that, The initial indicator system construction module includes a data preprocessing unit, a first indicator construction unit, a second indicator construction unit, a third indicator construction unit, and an indicator integration unit; wherein: The data preprocessing unit is used to preprocess the collected real-time operating data, environmental parameters, and historical operation and maintenance data of the high-voltage cable to obtain optimized real-time operating data, optimized environmental parameters, and optimized historical operation and maintenance data of the high-voltage cable. The first index construction unit is used to construct electrical performance indexes, mechanical structure indexes, and insulation characteristic indexes based on the optimized real-time operation data of the high-voltage cable; The second indicator construction unit is used to construct environmental factor indicators based on the optimized environmental parameters; The third indicator construction unit is used to construct historical operation indicators based on the optimized historical operation and maintenance data; The indicator integration unit is used to integrate electrical performance indicators, mechanical structure indicators, insulation characteristic indicators, environmental factor indicators, and operating history indicators to obtain an initial evaluation indicator system.

10. The maintenance decision-making system based on high-voltage cable condition assessment and life prediction as described in claim 9, characterized in that, The dynamic evaluation indicator system construction module includes a real-time data acquisition unit, a data comparison unit, an abnormal operating condition classification unit, an indicator system adjustment unit, and an importance rating unit; wherein: The real-time data acquisition unit is used to select target indicator parameters from the initial evaluation indicator system and acquire the real-time time series data corresponding to the target indicator parameters; The data comparison unit is used to obtain historical time series data corresponding to the target indicator parameters through a preset time series model, and compare the real-time time series data with the historical time series data to obtain the comparison result; The abnormal operating condition division unit is used to construct a multi-dimensional feature vector based on the real-time time series data, and perform density clustering on the multi-dimensional feature vector based on a preset clustering algorithm to divide the abnormal operating conditions. The indicator system adjustment unit is used to adjust the indicator topology of the initial evaluation indicator system whenever any target indicator parameter deviates from the preset baseline threshold or an abnormal working condition occurs in the comparison result, so as to obtain the adjusted initial evaluation indicator system. The importance rating unit is used to rate the importance of the adjusted initial evaluation indicators based on a preset scoring model, thereby obtaining a dynamic evaluation indicator system.

Images