Methods, devices, equipment, media, and procedures for predicting construction risks of power transmission lines.

Abstract

This invention discloses a method, device, equipment, medium, and program product for predicting construction risks of transmission lines. The method includes: analyzing historical construction samples of the construction stages of transmission lines to obtain construction characteristic information of each historical construction sample and the risk level of multiple construction hazard types of each historical construction sample; constructing a historical construction hazard information matrix of transmission lines based on the risk level and construction characteristic information of each construction hazard type; inputting the historical construction hazard information matrix into a risk prediction model to obtain the risk prediction results of the transmission lines; achieving objective and comprehensive risk prediction of the transmission line construction plan stages from multiple perspectives; improving the accuracy of risk prediction; timely detection of potential hazards throughout the entire life cycle of the construction stage; and achieving advance risk detection to ensure early avoidance of risks during the construction stage of transmission lines.

Description

Technical Field

[0001] This invention relates to the field of transmission line risk prediction, and in particular to a method, apparatus, equipment, medium, and program product for predicting construction risks of transmission lines. Background Technology

[0002] As a crucial component of the power grid, the safe construction of transmission lines directly impacts the stable operation of the power system. With the continuous growth of electricity demand, transmission networks are increasingly expanding towards higher voltage, larger capacity, and longer distances, leading to more extensive spatial distribution and more complex engineering scenarios for transmission line construction projects. Furthermore, the complex coupling relationship between environmental conditions, construction methods, and construction risks during the construction process results in an increasingly diverse range of potential engineering hazards. Against this backdrop, effectively assessing the potential risks of various optional construction schemes and making reasonable selections becomes a critical step in ensuring the smooth progress of projects.

[0003] Compared to real-time risk management during construction, risk identification in the planning phase is more forward-looking: on the one hand, early intervention can provide guidance for scheme optimization, resource allocation, and schedule planning, significantly improving the safety and economy of the project; on the other hand, early identification of potential high-risk hazards helps avoid cost waste and schedule delays caused by later design changes and emergency interventions. However, current risk prediction in the transmission line planning phase lacks effective objective evidence, making it impossible to deeply explore different types of characteristics and hazards from complex multi-dimensional information, leading to prediction errors. Summary of the Invention

[0004] The main objective of this invention is to provide a method, device, equipment, medium, and program product for predicting construction risks of transmission lines. This invention aims to solve the technical problem that existing technologies for risk prediction in the planning stage of transmission lines lack effective objective evidence and cannot deeply explore different types of characteristics and hidden dangers from complex information in multiple dimensions, leading to prediction errors.

[0005] To achieve the above objectives, the present invention provides a method for predicting construction risks of power transmission lines, the method comprising the following steps:

[0006] Historical construction samples from the construction phase of transmission lines are analyzed to obtain construction characteristic information of each historical construction sample and risk levels of multiple construction hazard types for each historical construction sample. The construction hazard types are the construction hazards included in each key process during the construction phase of transmission lines.

[0007] Based on the risk level of each type of construction hazard and the construction feature information, a historical construction hazard information matrix for transmission lines is constructed. The historical construction hazard information matrix includes a target information matrix corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples.

[0008] The historical construction hazard information matrix is ​​input into a pre-built risk prediction model to obtain the risk prediction result of the transmission line. The risk prediction model is built based on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model has a dual-channel weighted architecture and includes a discrete feature mining module and a continuous feature mining module.

[0009] Optionally, the discrete feature mining module is configured to calculate the target BPA distribution parameters of construction discrete features and generate candidate discrete features based on the target BPA distribution parameters;

[0010] The continuous feature mining module is configured to calculate the target BPA distribution parameters of each interval of the construction continuous feature by means of equal distance dispersion, form a weight distribution curve by Gaussian kernel regression, and generate candidate continuous features based on the weight distribution curve.

[0011] The risk prediction model also includes a feature selection module, which is configured to perform dynamic feature selection on the candidate discrete features and the candidate continuous features based on an uncertainty threshold to obtain target discrete features and target continuous features.

[0012] The risk prediction model also includes a conflict-guided modular evidence fusion module, which is configured to perform a hazard risk assessment based on the discrete features and continuous features of the target, and output the risk prediction results of the transmission line.

[0013] Optionally, the discrete feature mining module is a single-factor evidence mining module, which is configured to calculate the initial basic probability distribution parameters of the construction discrete features, referring to the following formula:

[0014]

[0015] in, Representing discrete features Corresponding risk level The initial basic probability distribution parameters, The scoring function representing the joint confidence and lift is used to measure the strength of the association between discrete features and risk levels. Representing discrete features Pointing to risk level confidence level Representing discrete features Pointing to risk level The degree of improvement An adjustment coefficient representing the degree of lift;

[0016] The single-factor evidence mining module is also configured to impose uncertainty constraints on the initial basic probability distribution parameters and calculate the target BPA distribution parameters, referring to the following formula:

[0017]

[0018]

[0019] in, Representing discrete features Corresponding risk level Target BPA distribution parameters, This represents the summation of the initial basic probability distribution parameters for the three risk levels in the initial basic probability distribution parameters. This represents the sum of risk level weights for discrete features. Representing discrete features The basic probability distribution parameters of the uncertainty term;

[0020] The weights of the uncertainty terms of the discrete features are assigned according to the following formula:

[0021]

[0022] in, The retention factor represents the uncertainty term. The uncertainty term represents the set of all possible risk levels of the discrete feature.

[0023] Optionally, the continuous feature mining module is a weight distribution curve generation module, which is configured to divide the continuous construction features into multiple equal-width intervals, as shown in the following formula:

[0024]

[0025]

[0026] in, Indicates the first A series of equal-width intervals, This represents the total number of equal-width intervals divided by the continuous construction characteristics. Indicates the interval length. and These represent the lower and upper limits of the interval, respectively.

[0027] The weight distribution curve generation module is also configured to calculate the target BPA distribution parameters for each equal-width interval, and calculate the initial response weights of the construction continuity characteristics based on the target BPA distribution parameters for each equal-width interval, referring to the following formula:

[0028]

[0029]

[0030] in, Representing continuous features Values Risk level The initial response weights, Indicates bandwidth is The Gaussian kernel function is used to measure... and the center of the interval similarity, Indicates the first BPA distribution parameters for risk levels across discrete intervals;

[0031] The weight distribution curve generation module is further configured to apply uncertainty constraints to the initial response weights, calculate the target response weights, and generate a weight distribution curve based on the target response weights, referring to the following formula:

[0032]

[0033]

[0034] in, Representing continuous features Values Risk level The target response weight, This represents the sum of weights for all risk levels corresponding to a continuous feature;

[0035] The weights of the uncertainty terms of the continuous features are assigned according to the following formula:

[0036]

[0037] in, The retention factor represents the uncertainty term. The basic probability distribution parameters represent the uncertainty term of continuous features.

[0038] Optionally, the feature filtering module is further configured to superimpose the target BPA distribution parameters of each feature to obtain the total BPA distribution parameters, as shown in the following formula:

[0039]

[0040] in, This represents the total BPA distribution parameter. Indicate the feature pair of the first Target BPA distribution parameters for each risk level The target BPA distribution parameters represent the uncertainty term of the feature;

[0041] The feature selection module is further configured to calculate the uncertainty weight ratio of candidate discrete features and the uncertainty weight ratio of candidate continuous features based on the total BPA distribution parameters, referring to the following formula:

[0042]

[0043]

[0044] in, Representing discrete features Uncertainty weighting ratio Representing discrete features The corresponding uncertainty is assigned a basic probability value. Representing continuous features Uncertainty weighting ratio Representing continuous features Throughout the entire value range The integral of the basic probability assignment of internal uncertainty. Representing continuous features The length of the range of values;

[0045] The feature filtering module is further configured to filter candidate discrete features and candidate continuous features based on an uncertainty threshold, and remove features whose uncertainty weight ratio is higher than the uncertainty threshold to obtain target discrete features and target continuous features.

[0046] Optionally, the conflict-guided modular evidence fusion module includes a multi-factor evidence mining module, and the conflict-guided modular evidence fusion module is further configured to calculate the conflict degree between each feature, referring to the following formula:

[0047]

[0048] in, Representation of features With features The degree of conflict between them Representation of features Corresponding risk level Target BPA distribution parameters, Representation of features Corresponding risk level Target BPA distribution parameters, A set of uncertainties representing risk levels;

[0049] The conflict-guided modular evidence fusion module is further configured to construct a feature conflict degree matrix based on the conflict degree, and to construct a graph structure based on the feature conflict degree matrix, and to generate multiple high-conflict feature pairs and low-conflict feature pairs based on the graph structure. The high-conflict feature pairs are feature pairs whose conflict degree exceeds the conflict threshold, and the low-conflict feature pairs are feature pairs whose conflict degree does not exceed the conflict threshold.

[0050] The multi-factor evidence mining module is configured to perform evidence mining based on the high-conflict feature pairs to obtain the BPA distribution parameters of the high-conflict feature pairs, referring to the following formula:

[0051]

[0052]

[0053]

[0054] in, Indicates by features With features The high-conflict characteristic pair formed by the combination, Representation of features With features Risk level after joint action The BPA distribution parameters, The scoring function representing the joint confidence and lift. High conflict characteristics indicate the level of risk. confidence level High conflict characteristics indicate the level of risk. The degree of improvement The adjustment coefficient representing the degree of lift. This indicates that the features are satisfied simultaneously. With features And the risk level is The number of samples, This indicates that the features are satisfied simultaneously. With features The number of samples, This represents the total number of samples. Indicates a risk level of The number of samples;

[0055] The multi-factor evidence mining module is further configured to perform joint evidence mining on the high-conflict feature pairs based on the BPA distribution parameters of the high-conflict feature pairs, and generate joint BPA parameters for the high-conflict features.

[0056] The conflict-guided modular evidence fusion module is further configured to fuse the joint BPA parameters of high-conflict features with the BPA distribution parameters of each low-conflict feature pair to obtain fused BPA parameters, perform a hazard risk assessment based on the fused BPA parameters, and output the risk prediction result of the transmission line, referring to the following formula:

[0057]

[0058] in, Indicates the risk prediction results. Indicates the risk level The fusion of BPA parameters.

[0059] Furthermore, to achieve the above objectives, the present invention also proposes a transmission line construction risk prediction device, the transmission line construction risk prediction device comprising:

[0060] The construction sample analysis module is used to analyze historical construction samples of the construction phase of transmission lines to obtain construction characteristic information of each historical construction sample and risk level of multiple construction hazard types of each historical construction sample. The construction hazard types are the construction hazards contained in each key process of the construction phase of the transmission line.

[0061] The information matrix construction module is used to construct a historical construction hazard information matrix for transmission lines based on the risk level of each type of construction hazard and the construction feature information. The historical construction hazard information matrix includes a target information matrix corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples.

[0062] The construction risk prediction module is used to input the historical construction hazard information matrix into a pre-built risk prediction model to obtain the risk prediction result of the transmission line. The risk prediction model is built based on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model is a dual-channel weighted architecture and includes a discrete feature mining module and a continuous feature mining module.

[0063] In addition, to achieve the above objectives, this application also proposes a transmission line construction risk prediction device, the device comprising: a memory, a processor, and a transmission line construction risk prediction program stored in the memory, the processor being used to run the transmission line construction risk prediction program, the computer program being configured to implement the steps of the transmission line construction risk prediction method as described above.

[0064] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the transmission line construction risk prediction method described above.

[0065] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the transmission line construction risk prediction method described above.

[0066] This invention analyzes historical construction samples from the construction phases of transmission lines to obtain construction characteristic information and risk levels for multiple construction hazard types in each historical construction sample. These construction hazard types are those included in key processes during the construction phases of the transmission line. Based on the risk levels of each construction hazard type and the construction characteristic information, a historical construction hazard information matrix for the transmission line is constructed. This matrix includes target information matrices corresponding to each construction hazard type. The target information matrix contains input record numbers, discrete construction characteristics, continuous construction characteristics, and hazard risk levels for multiple historical construction samples. The historical construction hazard information matrix is ​​then input into a pre-constructed risk prediction model to obtain the risk prediction results for the transmission line. This risk prediction model is data-driven. The invention constructs a dynamic conflict perception parallel evidence reasoning system. The risk prediction model is a dual-channel weighted architecture, comprising a discrete feature mining module and a continuous feature mining module. By capturing construction features and potential hazards from historical construction samples during the construction phase of transmission lines, and constructing a historical construction hazard information matrix based on multi-dimensional information, the invention achieves objective, multi-faceted, comprehensive risk analysis of the transmission line construction plan phase. The dual-channel weighted risk prediction model separately mines discrete and continuous features during the transmission line construction phase, improving the accuracy of risk prediction, timely identifying potential hazards throughout the entire lifecycle of the construction phase, enabling advance risk detection, ensuring early avoidance of risks during the transmission line construction phase, and effectively preventing cost waste and project delays caused by later design changes and emergency interventions. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a schematic diagram of the structure of the transmission line construction risk prediction device in the hardware operating environment involved in the embodiments of the present invention;

[0069] Figure 2 This is a flowchart illustrating the first embodiment of the transmission line construction risk prediction method of the present invention;

[0070] Figure 3(a) is a schematic diagram of the discrete interval calculation of continuous features in one embodiment of the transmission line construction risk prediction method of the present invention;

[0071] Figure 3(b) is a schematic diagram of the WDC fitting results of continuous features in one embodiment of the transmission line construction risk prediction method of the present invention;

[0072] Figure 4 This is a schematic diagram of the processing flow of the conflict-guided modular evidence fusion module in one embodiment of the transmission line construction risk prediction method of the present invention;

[0073] Figure 5 This is a schematic diagram of the risk prediction model in one embodiment of the transmission line construction risk prediction method of the present invention;

[0074] Figure 6 This is a structural block diagram of the first embodiment of the transmission line construction risk prediction device of the present invention.

[0075] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0076] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0077] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a power transmission line construction risk prediction device in the hardware operating environment of an embodiment of the present invention.

[0078] like Figure 1As shown, the transmission line construction risk prediction device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen or an input unit such as a keyboard; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0079] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the transmission line construction risk prediction equipment, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0080] like Figure 1 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a power transmission line construction risk prediction program.

[0081] exist Figure 1 In the transmission line construction risk prediction device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the transmission line construction risk prediction device of the present invention can be set in the transmission line construction risk prediction device, and the transmission line construction risk prediction device calls the transmission line construction risk prediction program stored in the memory 1005 through the processor 1001 and executes the transmission line construction risk prediction method provided in the embodiment of the present invention.

[0082] This invention provides a method for predicting construction risks of power transmission lines, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the transmission line construction risk prediction method of the present invention.

[0083] In this embodiment, the method for predicting construction risks of transmission lines includes the following steps:

[0084] Step S10: Analyze historical construction samples of the transmission line construction phase to obtain construction characteristic information of each historical construction sample and risk level of multiple construction hazard types of each historical construction sample.

[0085] It should be noted that this embodiment is applied to risk prediction in the planning stage of transmission line construction schemes. It integrates multi-source data in the planning stage to conduct a comprehensive, objective, and interpretable assessment of the potential risks of each alternative scheme, thereby providing engineering construction personnel with a scientific and feasible basis for selecting the best scheme.

[0086] It should be noted that with the development of sensing and monitoring technologies and engineering digitization technologies, data-driven methods based on machine learning have been widely used in engineering risk prediction. Compared with traditional assessment methods that rely on expert experience, machine learning can uncover potential patterns in historical data, establish complex nonlinear mapping relationships, and possess automated modeling and multi-source feature fusion capabilities. However, machine learning methods have the following problems:

[0087] First, some methods lack the flexibility for comprehensive prediction:

[0088] Similar to mechanism-based modeling, some methods are designed specifically for a typical failure cause event and require certain input features or specific mathematical models. Although these methods have high accuracy and interpretability under specific mechanisms, their generality is limited and they are difficult to apply directly to complex engineering environments with multiple coexisting risks or unknown failure mechanisms.

[0089] Second, the problem of model interpretability:

[0090] The "black box" problem is one of the major obstacles affecting the practical application of machine learning. While most mainstream machine learning algorithms can handle complex nonlinear relationships, they often struggle to clearly explain the logical mechanisms behind model decisions. Although tree models (such as random forests and XGBoost) are more intuitive than neural networks in interpreting feature importance, their output feature importance still falls short of the engineering field's requirements for causal transparency and decision traceability. Firstly, different tree models calculate feature importance significantly differently. For example, random forests typically evaluate features based on the Gini index or average change in information gain during node splits, while XGBoost uses weighted calculations based on metrics such as the improvement in the objective function, sample coverage, or split frequency for each split. These methods may produce completely different feature ranking results when faced with the same dataset. Secondly, this type of feature importance only represents the variable's contribution to improving model predictive performance, rather than its causal role in real engineering mechanisms.

[0091] The DS evidence fusion method constructs a basic probability assignment (BPA) for each input feature to a risk level, transforming raw data into "support" for different risk states, and integrates multi-source judgment information based on fusion rules. The BPA assignment mechanism is highly flexible, allowing for assignment based on feature confidence, relevance, or discriminative power. This enhances the model's ability to represent "critical minority" risks without relying on large-scale samples, effectively mitigating sample imbalance. Furthermore, the DS fusion process possesses explicit evidence combination and conflict management logic, enabling the decomposition of feature contribution paths class by class, thus exhibiting a degree of interpretability.

[0092] Currently, the Dempster-Shafer (DS) evidence theory has made significant progress in risk prediction research, but the following problems still exist:

[0093] First, the Basic Probability Assignment (BPA) generation method lacks objectivity.

[0094] The Basic Probability Assignment (BPA) serves as the foundation for fusion, and its generation method directly affects the final result. In practical engineering, BPA often relies on expert experience, fuzzy mapping, or model output probabilities for conversion, lacking objective evaluation standards.

[0095] Second, the fusion of highly conflicting evidence leads to prediction errors.

[0096] When there is inconsistency among multiple sources of evidence, the standard Dempster synthesis rule may incorrectly transfer conflicting quality to non-conflicting assumptions, thus leading to a bias in judgment.

[0097] It should be understood that the executing entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a terminal electronic device capable of performing the above functions. The following description uses a power transmission line construction risk prediction device as an example to illustrate this embodiment and the subsequent embodiments.

[0098] It should be noted that the types of construction hazards mentioned refer to the construction hazards included in each key process during the construction phase of transmission lines. Key processes can include foundation excavation, concrete pouring, road construction, tower material transportation, tower erection, and line erection during the construction phase of transmission lines. Each stage of a key process faces specific safety risks.

[0099] For example, during the foundation construction phase, a pit collapse may occur, burying personnel; during tower erection, personal injury accidents may occur due to pole instability or equipment overturning. The direct causes of such safety incidents are usually human factors such as operational errors and improper maintenance, making accurate prediction difficult in the initial project planning stage. Therefore, this embodiment focuses on objective hazards that can be identified in the early stages and uses them as risk prediction targets. The selection of construction hazard types should meet the following three principles:

[0100] First, it is representative.

[0101] The construction of power transmission lines is a long and complex process, with risk events occurring at multiple stages of the project. Therefore, the predictive targets of risk assessments should cover the key phases or core processes of the project and demonstrate sufficient representativeness.

[0102] Second, it is predictable.

[0103] The essence of data-driven models relies on the regular relationship between feature variables and target variables. Therefore, the selected prediction target should be predictable, meaning that its occurrence is driven by environmental factors, design parameters, or construction plans, rather than dominated by human factors or chance.

[0104] Third, it has clear classification standards.

[0105] In order to achieve model training and risk classification, the prediction target should have clear classification criteria to facilitate label definition and data annotation.

[0106] Based on the above analysis, the following six types of hidden dangers are selected as research objects, as shown in Table 1. Table 1 is a table of hidden danger categories for power transmission line construction:

[0107] Table 1. Categories of Hidden Dangers During Transmission Line Construction

[0108]

[0109] It should be noted that this embodiment analyzes and extracts response feature variables for each historical construction sample, covering external environmental factors and construction method parameters, thereby screening out the construction features in the historical construction samples. The specific feature composition is shown in Table 2, which is the transmission line construction feature table.

[0110] Table 2, Construction Characteristics of Transmission Lines

[0111]

[0112] Step S20: Construct a historical construction hazard information matrix for transmission lines based on the risk level of each type of construction hazard and the construction characteristic information.

[0113] It should be noted that the historical construction hazard information matrix includes target information matrices corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples.

[0114] It should be understood that, considering the various environmental characteristics and construction methods involved in potential hazards during power transmission line construction, in order to fully analyze the input information, it is necessary to aggregate the raw data into a unified data processing space to address the specific hazards. Construct a matrix of historical construction hazards for power transmission lines. The mathematical expression for the historical construction hazard information matrix is ​​as follows:

[0115]

[0116] in, , indicating the number of the input record; This indicates that each record contains 20 environmental characteristics and construction methods; Indicates the first The first record The values ​​of each feature; This indicates the risk level of the potential hazard in each record.

[0117] Furthermore, the features Structurally divided into Discrete features and Continuous features ,matrix It can be rewritten in the following form:

[0118]

[0119] Step S30: Input the historical construction hazard information matrix into the pre-built risk prediction model to obtain the risk prediction result of the transmission line.

[0120] It should be noted that the risk prediction model is built on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model is a dual-channel evidence weighting architecture (DC-EWA), which includes a discrete feature mining module and a continuous feature mining module.

[0121] It should be noted that the risk prediction model can be a neural network model built on the Data-Driven Conflict-Informed Parallel Evidence Reasoning (D-CIPER) system.

[0122] Understandably, the discrete feature mining module is used to mine discrete construction features input into the risk prediction model, while the continuous feature mining module is used to mine continuous construction features input into the risk prediction model.

[0123] In its implementation, the risk prediction model fuses the target discrete features output by the discrete feature mining module with the target continuous features output by the continuous feature mining module, and then performs risk prediction based on the fused features, outputting the risk prediction results for the transmission line.

[0124] Furthermore, in order to accurately mine the discrete and continuous features of the transmission line construction stage, in one embodiment, the discrete feature mining module is configured to calculate the target BPA distribution parameters of the construction discrete features and generate candidate discrete features based on the target BPA distribution parameters.

[0125] The continuous feature mining module is configured to calculate the target BPA distribution parameters of each interval of the construction continuous feature by means of equal distance dispersion, form a weight distribution curve by Gaussian kernel regression, and generate candidate continuous features based on the weight distribution curve.

[0126] The risk prediction model also includes a feature selection module, which is configured to perform dynamic feature selection on the candidate discrete features and the candidate continuous features based on an uncertainty threshold to obtain target discrete features and target continuous features.

[0127] The risk prediction model also includes a conflict-guided modular evidence fusion module, which is configured to perform a hazard risk assessment based on the discrete features and continuous features of the target, and output the risk prediction results of the transmission line.

[0128] It should be noted that in the dual-channel weighted architecture, the historical construction hazard information matrix is ​​used as input. The risk prediction model calculates the BPA distribution of each discrete factor through the discrete feature mining module. For continuous features, the BPA of each interval is calculated using an equidistant dispersion method, and an importance distribution curve is formed through Gaussian kernel regression. Then, the feature filtering module performs dynamic feature filtering based on the uncertainty ratio. Finally, the filtered feature information is input into the Conflict-Guided Modular Evidential Fusion (CGMEF) module for prediction, thereby obtaining the risk prediction result of transmission line hazards.

[0129] Furthermore, to accurately measure the correlation strength between discrete features and various risk levels, improve the mining accuracy of discrete features, and ensure prediction reliability, in one embodiment, the discrete feature mining module is a Single-Factor Evidential Mining (SFEM) module. The SFEM module is configured to calculate the initial basic probability distribution parameters of the construction discrete features, referring to the following formula:

[0130]

[0131] in, Representing discrete features Corresponding risk level The initial basic probability distribution parameters, The scoring function representing the joint confidence and lift is used to measure the strength of the association between discrete features and risk levels. Representing discrete features Pointing to risk level confidence level Representing discrete features Pointing to risk level The degree of improvement An adjustment coefficient representing the degree of lift;

[0132] The single-factor evidence mining module is also configured to impose uncertainty constraints on the initial basic probability distribution parameters and calculate the target BPA distribution parameters, referring to the following formula:

[0133]

[0134]

[0135] in, Representing discrete features Corresponding risk level Target BPA distribution parameters, This represents the summation of the initial basic probability distribution parameters for the three risk levels in the initial basic probability distribution parameters. This represents the sum of risk level weights for discrete features. Representing discrete features The basic probability distribution parameters of the uncertainty term;

[0136] The weights of the uncertainty terms of the discrete features are assigned according to the following formula:

[0137]

[0138] in, The retention factor represents the uncertainty term. The uncertainty term represents the set of all possible risk levels of the discrete feature.

[0139] It should be noted that we define a set of propositions (i.e., a set of risk levels). , Corresponding to the three risk levels in Table 1 above, It represents the complete set of all possible risk levels, describing the uncertainty of the judgment.

[0140] Assuming discrete characteristics It contains several types of elements, namely In DS evidence theory, based on a certain discrete characteristic element... Risk level The information relationships between them are used to construct the Basic Probability Assignment (BPA), which is represented as follows:

[0141]

[0142] In solving for the weight of evidence, a system is constructed from... The evidence weight distribution for each risk level, i.e., the BPA value vector. ,satisfy:

[0143]

[0144] in This indicates that the current evidence is unclear regarding the specific level of uncertainty, and therefore requires retention in the overall set of uncertainty weights. In this embodiment, we have:

[0145] .

[0146] Association rules are a method for uncovering potential dependencies between items in a dataset, and can be used to extract discrete feature values. Risk level The correspondence between these characteristics is used to measure the strength and credibility of the support for a certain risk level, providing a basis for subsequent allocation of BPA in the DS evidence theory. Traditional association rule evaluation methods mainly rely on support (…). ), confidence level ( ) and improvement ( Three indicators:

[0147] Support is expressed across all samples The conditions are met simultaneously The result is The proportion of the sample reflects the frequency of the rule's occurrence in the sample.

[0148]

[0149] Confidence level indicates that when the condition is met... What percentage of the sample belongs to This reflects the accuracy or credibility of the rule.

[0150]

[0151] Lift degree indicates the extent to which the premise is considered. In this case, The probability of an occurrence is a multiple of its natural probability of occurrence, reflecting the strength of the association between rules.

[0152]

[0153] Support only reflects the frequency of a rule in a sample. However, in engineering risk prediction scenarios, low-frequency but highly correlated features often have extremely high discriminative value. Introducing support as a weighting criterion may ignore such key low-frequency rules. Confidence reflects the accuracy of a conclusion under a given condition and is the core indicator for measuring the reliability of a prediction. Lift is used to assess the magnitude of the influence of the premise on the probability of the consequent and can effectively distinguish between causality and spurious associations caused by codependent variables.

[0154] Therefore, single-factor evidence mining (SFEM) uses a scoring function that jointly considers confidence and lift. This method measures the correlation strength between discrete feature values ​​and various risk levels, and assigns a Basic Probability Assignment (BPA) in the DS theory accordingly. While maintaining prediction reliability, this method also considers sensitivity to rare, highly correlated factors, making it suitable for predicting engineering hazards in scenarios with unevenly distributed data. Its calculation expression is as follows:

[0155]

[0156] In the case of imbalanced data distribution An abnormally high value could lead to an overemphasis on the importance of factors, resulting in an unreasonable BPA weighting. To mitigate this situation, a coefficient is introduced. Adjustments are made to the feature elements. Total risk level weights With constraints applied, the BPA calculation formula is as follows:

[0157]

[0158] in For uncertain terms The retention coefficients and the weights assigned to the uncertain terms are:

[0159]

[0160] Furthermore, to improve the stability of BPA calculation for continuous features, avoid the loss of continuous feature information, and improve prediction accuracy, in one embodiment, the continuous feature mining module is a weight distribution curve generation module. This module is configured to divide the continuous construction features into multiple equal-width intervals, as shown in the following formula:

[0161]

[0162]

[0163] in, Indicates the first A series of equal-width intervals, This represents the total number of equal-width intervals divided by the continuous construction characteristics. Indicates the interval length. and These represent the lower and upper limits of the interval, respectively.

[0164] The weight distribution curve generation module is also configured to calculate the target BPA distribution parameters for each equal-width interval, and calculate the initial response weights of the construction continuity characteristics based on the target BPA distribution parameters for each equal-width interval, referring to the following formula:

[0165]

[0166]

[0167] in, Representing continuous features Values Risk level The initial response weights, Indicates bandwidth is The Gaussian kernel function is used to measure... and the center of the interval similarity, Indicates the first BPA distribution parameters for risk levels across discrete intervals;

[0168] The weight distribution curve generation module is further configured to apply uncertainty constraints to the initial response weights, calculate the target response weights, and generate a weight distribution curve based on the target response weights, referring to the following formula:

[0169]

[0170]

[0171] in, Representing continuous features Values Risk level The target response weight, This represents the sum of weights for all risk levels corresponding to a continuous feature;

[0172] The weights of the uncertainty terms of the continuous features are assigned according to the following formula:

[0173]

[0174] in, The retention factor represents the uncertainty term. The basic probability distribution parameters represent the uncertainty term of continuous features.

[0175] It should be noted that in multi-source information risk assessment, some features are continuous variables and cannot be directly calculated using SFEM (Single Factor Evidence Mining Module) to calculate the BPA distribution. One approach is to discretize the continuous features and calculate the BPA for each discrete interval. However, this method has the following limitations:

[0176] 1. All samples falling into the same interval will be assigned the same risk expression, thus ignoring the numerical differences within the interval and resulting in the loss of original information.

[0177] 2. Even small numerical changes can lead to abrupt changes in risk level assessment if they cross the boundary. For example, the BPA of samples with slopes of 9.9° and 10.1° may differ significantly due to the different intervals they belong to.

[0178] 3. In actual engineering data, continuous variables often exhibit non-uniform distribution. For intervals with sparse data samples, BPA calculation is prone to instability, resulting in local misleading situations and affecting the overall fusion accuracy.

[0179] 4. Most potential risks exhibit a smooth response characteristic as the characteristic values ​​increase. Traditional discretization methods can only output step-like level classifications, failing to reflect the dynamic evolution trajectory of risks during continuous changes, thus limiting the interpretability of the model.

[0180] To address the aforementioned issues, this paper implements a continuous representation of BPA by constructing a WDC (Wide Distributed Data Structure). Firstly, for continuous features... Let the range of values ​​be . Divide it into A series of equal-width intervals:

[0181]

[0182] in The interval length is [length], and each interval [length] is [length]. Treated as a discrete factor, its BPA distribution is calculated using SFEM, ultimately yielding the weights in each interval. , The value is the center value of the interval.

[0183] Then, Nadaraya-Watson kernel regression is introduced to fit the discrete points and construct a continuous risk level response function:

[0184]

[0185] in Indicates continuous eigenvalues Risk level Response weights; Indicates bandwidth is The kernel function used in this embodiment is the Gaussian kernel function:

[0186]

[0187] Continuous features For risk range WDC function As shown in Figures 3(a) and 3(b), these figures illustrate the construction of the WDC for continuous features. For continuous features, assuming the feature value range is 0-1, it is divided into several discrete intervals, and the BPA of each interval is calculated, as shown in Figure 3(a). Then, the WDC is fitted using a Gaussian kernel function, as shown in Figure 3(b), with the top layer representing the uncertainty ratio.

[0188] Similar to SFEM, coefficients are introduced when constructing WDC. For each point The sum of risk level weights Apply uncertainty constraints.

[0189] Eigenvalues Risk level The response weights are:

[0190]

[0191] The weights assigned to the uncertain terms are:

[0192]

[0193] Furthermore, to improve fusion efficiency and prediction accuracy, and to effectively evaluate the identification ability of each feature in risk level classification, thereby identifying and eliminating redundant or inefficient features, in one embodiment, the feature filtering module is further configured to superimpose the target BPA distribution parameters of each feature to obtain the total BPA distribution parameters, as shown in the following formula:

[0194]

[0195] in, This represents the total BPA distribution parameter. Indicate the feature pair of the first Target BPA distribution parameters for each risk level The target BPA distribution parameters represent the uncertainty term of the feature;

[0196] The feature selection module is further configured to calculate the uncertainty weight ratio of candidate discrete features and the uncertainty weight ratio of candidate continuous features based on the total BPA distribution parameters, referring to the following formula:

[0197]

[0198]

[0199] in, Representing discrete features Uncertainty weighting ratio Representing discrete features The corresponding uncertainty is assigned a basic probability value. Representing continuous features Uncertainty weighting ratio Representing continuous features Throughout the entire value range The integral of the basic probability assignment of internal uncertainty. Representing continuous features The length of the range of values;

[0200] The feature filtering module is further configured to filter candidate discrete features and candidate continuous features based on an uncertainty threshold, and remove features whose uncertainty weight ratio is higher than the uncertainty threshold to obtain target discrete features and target continuous features.

[0201] It should be noted that, to further improve fusion efficiency and prediction accuracy, it is necessary to evaluate the distinguishing ability of each feature in risk level classification, and identify and eliminate redundant or inefficient features. This section proposes a feature selection method based on the proportion of uncertainty weights to measure the ability of features to distinguish specific risk levels under different conditions. First, the BPA distributions of each feature are superimposed:

[0202]

[0203] Uncertainty Located at the top level, it indicates the ambiguity or uncertainty of a feature when it fails to clearly support a certain risk level.

[0204] To quantitatively analyze the discriminative power of features, the Uncertainty Ratio (UR) is used as a discriminant metric to measure the degree of uncertainty introduced by a feature within its value range. A higher UR indicates that the feature cannot effectively support a specific risk level under most conditions, potentially introducing redundant information or amplifying conflicts during the fusion process, thereby reducing the credibility of the inference results. Discrete feature factors are defined. of for:

[0205]

[0206] For continuous features Calculate it using the following formula :

[0207]

[0208] For example, setting an uncertainty threshold. When the features If the value exceeds this threshold, it is considered a redundant or irrelevant feature and should be removed.

[0209] Furthermore, in order to effectively avoid feature conflicts during multi-source information fusion and improve prediction reliability, in one embodiment, the conflict-guided modular evidence fusion module includes a multi-factor evidence mining module. The conflict-guided modular evidence fusion module is also configured to calculate the conflict degree between each feature, referring to the following formula:

[0210]

[0211] in, Representation of features With features The degree of conflict between them Representation of features Corresponding risk level Target BPA distribution parameters, Representation of features Corresponding risk level Target BPA distribution parameters, A set of uncertainties representing risk levels;

[0212] The conflict-guided modular evidence fusion module is further configured to construct a feature conflict degree matrix based on the conflict degree, and to construct a graph structure based on the feature conflict degree matrix, and to generate multiple high-conflict feature pairs and low-conflict feature pairs based on the graph structure. The high-conflict feature pairs are feature pairs whose conflict degree exceeds the conflict threshold, and the low-conflict feature pairs are feature pairs whose conflict degree does not exceed the conflict threshold.

[0213] The multi-factor evidence mining module is configured to perform evidence mining based on the high-conflict feature pairs to obtain the BPA distribution parameters of the high-conflict feature pairs, referring to the following formula:

[0214]

[0215]

[0216]

[0217] in, Indicates by features With features The high-conflict characteristic pair formed by the combination, Representation of features With features Risk level after joint action The BPA distribution parameters, The scoring function representing the joint confidence and lift. High conflict characteristics indicate the level of risk. confidence level High conflict characteristics indicate the level of risk. The degree of improvement The adjustment coefficient representing the degree of lift. This indicates that the features are satisfied simultaneously. With features And the risk level is The number of samples, This indicates that the features are satisfied simultaneously. With features The number of samples, This represents the total number of samples. Indicates a risk level of The number of samples;

[0218] The multi-factor evidence mining module is further configured to perform joint evidence mining on the high-conflict feature pairs based on the BPA distribution parameters of the high-conflict feature pairs, and generate joint BPA parameters for the high-conflict features.

[0219] The conflict-guided modular evidence fusion module is further configured to fuse the joint BPA parameters of high-conflict features with the BPA distribution parameters of each low-conflict feature pair to obtain fused BPA parameters, perform a hazard risk assessment based on the fused BPA parameters, and output the risk prediction result of the transmission line, referring to the following formula:

[0220]

[0221] in, Indicates the risk prediction results. Indicates the risk level The fusion of BPA parameters.

[0222] It should be noted that DS evidence theory is often used for multi-source information fusion in engineering risk prediction. However, when the number of features is large and the information conflicts between features are significant, the classic DS fusion rule may fall into the Zadeh paradox, that is, the forced fusion of highly conflicting evidence leads to misclassification and reduces prediction reliability. Therefore, this embodiment proposes a Conflict-Guided Modular Evidential Fusion (CGMEF) module. (Refer to...) Figure 4 , Figure 4 The diagram illustrates the processing flow of the conflict-guided modular evidence fusion module. The specific process of this module is as follows:

[0223] First, construct the feature conflict degree matrix:

[0224] To address potential conflicts between different features, this embodiment introduces a Conflict Matrix (CM) before fusion to identify the degree of inconsistency between features, serving as the basis for feature grouping and module partitioning. For the selected features... To quantify the degree of conflict between the evidence provided by any two features, a matrix is ​​constructed. Its elements Representation of features and The degree of conflict is calculated using the following formula:

[0225]

[0226] when The two features are completely identical. The closer a value is to 1, the more severe the conflict between pieces of evidence. A conflict threshold is set. When the conflict between features exceeds a threshold, direct fusion is not advisable. For example, the conflict threshold can be set to... .

[0227] (i.e., the feature conflict matrix) can be represented as:

[0228]

[0229] because Therefore, the feature conflict degree matrix It is a symmetric matrix, and its main diagonal is zero.

[0230] Second, conflict pair identification and feature grouping fusion:

[0231] First, based on Constructing feature map structure , where the set of nodes Represents all features, any feature pair Form an edge whose weight corresponds to the degree of conflict. ,like Then these two nodes (features) form a high-conflict edge. ;like Then it is defined as a low-conflict edge. ,satisfy ,and .

[0232] Then, in The algorithm employs a greedy matching strategy, selecting non-shared nodes sequentially according to their edge weights (conflict degrees) from highest to lowest. ,structure High-conflict feature pairs:

[0233]

[0234] Pair each high-conflict feature Treating it as a new consortium, mining its joint evidence distribution in historical samples. The calculation formula for Joint-Factor Evidential Mining (JFEM) is as follows:

[0235]

[0236]

[0237]

[0238] Among them, the uncertainty term The processing and BPA normalization calculation are consistent with the calculation formula for the target BPA distribution parameters in the single-factor evidence mining module. This yields the joint BPA distribution. .

[0239] Remove nodes from high-conflict feature pairs from the graph, and define the remaining graph structure as follows: ,satisfy ,set up Each feature is considered a low-conflict evidence source and fused using the DS rule. In this embodiment, the set... For two different features, it targets Fusion results Defined as:

[0240]

[0241] For all the above evidence sources, the fusion result can be calculated iteratively:

[0242]

[0243] in, This indicates a DS synthesis operation, ultimately predicting the risk level. Qualitative analysis is conducted based on the principle of maximum confidence:

[0244]

[0245] In some embodiments, the workflow of the risk prediction model is as follows: Figure 5 As shown, Figure 5This is a schematic diagram of the risk prediction model. The risk prediction model receives an input historical construction hazard information matrix, extracts discrete and continuous construction features from the matrix, processes the discrete features using the SFEM module (single-factor evidence mining module), processes the continuous features using the WDC module (weight distribution curve generation module), and then filters features based on an uncertainty threshold using a feature filtering module. The filtered features are then input into the CGMEF module (conflict-guided modular evidence fusion module). The CGMEF module constructs a feature conflict degree matrix based on the target discrete and continuous features, builds a graph structure based on the feature conflict degree matrix, and obtains low-conflict and high-conflict feature pairs based on the graph structure. The JFEM module (multi-factor evidence mining module) processes the high-conflict feature pairs to generate joint BPA parameters for high-conflict features. Then, the joint BPA parameters for high-conflict features are fused with the BPA parameters for low-conflict feature pairs to obtain fused BPA parameters. Risk prediction is then performed based on the fused BPA parameters, and the risk prediction result is output.

[0246] This embodiment analyzes historical construction samples from the construction phases of transmission lines to obtain construction characteristic information and risk levels for multiple construction hazard types in each historical construction sample. These hazard types are construction hazards included in each key process during the transmission line construction phase. Based on the risk levels of each hazard type and the construction characteristic information, a historical construction hazard information matrix for the transmission line is constructed. This matrix includes target information matrices corresponding to each hazard type, and each target information matrix contains input record numbers, discrete construction characteristics, continuous construction characteristics, and hazard risk levels for multiple historical construction samples. The historical construction hazard information matrix is ​​then input into a pre-constructed risk prediction model to obtain the risk prediction results for the transmission line. This risk prediction model is data-driven. A dynamic conflict-aware parallel evidence reasoning system is constructed. The risk prediction model adopts a dual-channel weighted architecture, which includes a discrete feature mining module and a continuous feature mining module. This embodiment captures construction features and potential hazards from historical construction samples during the construction phase of transmission lines. Based on multi-dimensional information, a historical construction hazard information matrix is ​​constructed to objectively conduct comprehensive risk analysis of the transmission line construction plan from multiple perspectives. The dual-channel weighted risk prediction model separately mines discrete and continuous features during the transmission line construction phase, improving the accuracy of risk prediction, timely identifying potential hazards throughout the entire lifecycle of the construction phase, achieving advance risk detection, ensuring early avoidance of risks during the transmission line construction phase, and effectively preventing cost waste and project delays caused by later design changes and emergency interventions.

[0247] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium storing a transmission line construction risk prediction program, which, when executed by a processor, implements the steps of the transmission line construction risk prediction method described above.

[0248] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0249] The aforementioned computer-readable storage medium may be included in the transmission line construction risk prediction equipment; or it may exist independently and not be assembled into the transmission line construction risk prediction equipment.

[0250] Furthermore, this invention also proposes a computer program product, including a transmission line construction risk prediction program, which, when executed by a processor, implements the steps of the transmission line construction risk prediction method as described above.

[0251] The specific implementation of the computer program product of the present invention is basically the same as the embodiments of the above-mentioned transmission line construction risk prediction method, and will not be repeated here.

[0252] Reference Figure 6 , Figure 6 This is a structural block diagram of the first embodiment of the transmission line construction risk prediction device of the present invention.

[0253] like Figure 6 As shown, the transmission line construction risk prediction device proposed in this embodiment of the invention includes:

[0254] The construction sample analysis module 10 is used to analyze historical construction samples of the construction phase of the transmission line to obtain construction characteristic information of each historical construction sample and risk level of multiple construction hazard types of each historical construction sample. The construction hazard types are the construction hazards contained in each key process of the construction phase of the transmission line.

[0255] The information matrix construction module 20 is used to construct a historical construction hazard information matrix for transmission lines based on the risk level of each type of construction hazard and the construction feature information. The historical construction hazard information matrix includes a target information matrix corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples.

[0256] The construction risk prediction module 30 is used to input the historical construction hazard information matrix into the pre-built risk prediction model to obtain the risk prediction result of the transmission line. The risk prediction model is built based on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model is a dual-channel weighted architecture and includes a discrete feature mining module and a continuous feature mining module.

[0257] This embodiment analyzes historical construction samples from the construction phases of transmission lines to obtain construction characteristic information and risk levels for multiple construction hazard types in each historical construction sample. These hazard types are construction hazards included in each key process during the transmission line construction phase. Based on the risk levels of each hazard type and the construction characteristic information, a historical construction hazard information matrix for the transmission line is constructed. This matrix includes target information matrices corresponding to each hazard type, and each target information matrix contains input record numbers, discrete construction characteristics, continuous construction characteristics, and hazard risk levels for multiple historical construction samples. The historical construction hazard information matrix is ​​then input into a pre-constructed risk prediction model to obtain the risk prediction results for the transmission line. This risk prediction model is data-driven. A dynamic conflict-aware parallel evidence reasoning system is constructed. The risk prediction model adopts a dual-channel weighted architecture, which includes a discrete feature mining module and a continuous feature mining module. This embodiment captures construction features and potential hazards from historical construction samples during the construction phase of transmission lines. Based on multi-dimensional information, a historical construction hazard information matrix is ​​constructed to objectively conduct comprehensive risk analysis of the transmission line construction plan from multiple perspectives. The dual-channel weighted risk prediction model separately mines discrete and continuous features during the transmission line construction phase, improving the accuracy of risk prediction, timely identifying potential hazards throughout the entire lifecycle of the construction phase, achieving advance risk detection, ensuring early avoidance of risks during the transmission line construction phase, and effectively preventing cost waste and project delays caused by later design changes and emergency interventions.

[0258] The transmission line construction risk prediction device provided in this application, employing the transmission line construction risk prediction method in the above embodiments, can solve the technical problem of transmission line construction risk prediction. Compared with the prior art, the beneficial effects of the transmission line construction risk prediction device provided in this application are the same as those of the transmission line construction risk prediction method provided in the above embodiments, and other technical features in the transmission line construction risk prediction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0259] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0260] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0261] In addition, for technical details not described in detail in this embodiment, please refer to the transmission line construction risk prediction method provided in any embodiment of the present invention, which will not be repeated here.

[0262] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0263] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0264] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0265] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for predicting construction risks of power transmission lines, characterized in that, The method for predicting construction risks of transmission lines includes: Historical construction samples from the construction phase of transmission lines are analyzed to obtain construction characteristic information of each historical construction sample and risk levels of multiple construction hazard types for each historical construction sample. The construction hazard types are the construction hazards included in each key process during the construction phase of transmission lines. Based on the risk level of each type of construction hazard and the construction feature information, a historical construction hazard information matrix for transmission lines is constructed. The historical construction hazard information matrix includes a target information matrix corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples. The historical construction hazard information matrix is ​​input into a pre-built risk prediction model to obtain the risk prediction result of the transmission line. The risk prediction model is built based on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model is a dual-channel weighted architecture and includes a discrete feature mining module and a continuous feature mining module. The discrete feature mining module is configured to calculate the target BPA distribution parameters of construction discrete features and generate candidate discrete features based on the target BPA distribution parameters. The continuous feature mining module is configured to calculate the target BPA distribution parameters of each interval of the construction continuous feature by means of equal distance dispersion, form a weight distribution curve by Gaussian kernel regression, and generate candidate continuous features based on the weight distribution curve. The risk prediction model also includes a feature selection module, which is configured to perform dynamic feature selection on the candidate discrete features and the candidate continuous features based on an uncertainty threshold to obtain target discrete features and target continuous features. The risk prediction model also includes a conflict-guided modular evidence fusion module, which is configured to perform a hidden danger risk assessment based on the discrete features and continuous features of the target, and output the risk prediction result of the transmission line. The conflict-guided modular evidence fusion module includes a multi-factor evidence mining module. This module is further configured to calculate the conflict degree between each feature, using the following formula: in, Representation of features With features The degree of conflict between them Representation of features Corresponding risk level Target BPA distribution parameters Representation of features Corresponding risk level Target BPA distribution parameters A set of uncertainties representing risk levels; The conflict-guided modular evidence fusion module is further configured to construct a feature conflict degree matrix based on the conflict degree, and to construct a graph structure based on the feature conflict degree matrix, and to generate multiple high-conflict feature pairs and low-conflict feature pairs based on the graph structure. The high-conflict feature pairs are feature pairs whose conflict degree exceeds the conflict threshold, and the low-conflict feature pairs are feature pairs whose conflict degree does not exceed the conflict threshold. The multi-factor evidence mining module is configured to perform evidence mining based on the high-conflict feature pairs to obtain the BPA distribution parameters of the high-conflict feature pairs, referring to the following formula: in, Indicates by features With features The high-conflict characteristic pair formed by the combination, Representation of features With features Risk level after joint action The BPA distribution parameters, The scoring function representing the joint confidence and lift. High conflict characteristics indicate the level of risk. Confidence level, High conflict characteristics indicate the level of risk. The degree of improvement The adjustment coefficient representing the degree of lift. This indicates that the features are satisfied simultaneously. With features And the risk level is The number of samples, This indicates that the features are satisfied simultaneously. With features The number of samples, This represents the total number of samples. Indicates a risk level of The number of samples; The multi-factor evidence mining module is further configured to perform joint evidence mining on the high-conflict feature pairs based on the BPA distribution parameters of the high-conflict feature pairs, and generate joint BPA parameters for the high-conflict features. The conflict-guided modular evidence fusion module is further configured to fuse the joint BPA parameters of high-conflict features with the BPA distribution parameters of each low-conflict feature pair to obtain fused BPA parameters, perform a hazard risk assessment based on the fused BPA parameters, and output the risk prediction result of the transmission line, referring to the following formula: in, Indicates the risk prediction results. Indicates the risk level The fusion of BPA parameters.

2. The method for predicting construction risks of transmission lines as described in claim 1, characterized in that, The discrete feature mining module is a single-factor evidence mining module, which is configured to calculate the initial basic probability distribution parameters of the construction discrete features, referring to the following formula: in, Representing discrete features Corresponding risk level The initial basic probability distribution parameters, The scoring function representing the joint confidence and lift is used to measure the strength of the association between discrete features and risk levels. Representing discrete features Pointing to risk level confidence level Representing discrete features Pointing to risk level The degree of improvement An adjustment coefficient representing the degree of lift; The single-factor evidence mining module is also configured to impose uncertainty constraints on the initial basic probability distribution parameters and calculate the target BPA distribution parameters, referring to the following formula: in, Representing discrete features Corresponding risk level Target BPA distribution parameters, This represents the summation of the initial basic probability distribution parameters for the three risk levels in the initial basic probability distribution parameters. This represents the sum of risk level weights for discrete features. Representing discrete features The basic probability distribution parameters of the uncertainty term; The weights of the uncertainty terms of the discrete features are assigned according to the following formula: in, The retention factor represents the uncertainty term. The uncertainty term represents the set of all possible risk levels of the discrete feature.

3. The method for predicting construction risks of transmission lines as described in claim 2, characterized in that, The continuous feature mining module is a weight distribution curve generation module, which is configured to divide the continuous construction features into multiple equal-width intervals, as shown in the following formula: in, Indicates the first A series of equal-width intervals, This represents the total number of equal-width intervals divided by the continuous construction characteristics. Indicates the interval length. and These represent the lower and upper limits of the interval, respectively. The weight distribution curve generation module is also configured to calculate the target BPA distribution parameters for each equal-width interval, and calculate the initial response weights of the construction continuity characteristics based on the target BPA distribution parameters for each equal-width interval, referring to the following formula: in, Representing continuous features Values Risk level The initial response weights, Indicates bandwidth is The Gaussian kernel function is used to measure... and the center of the interval similarity, Indicates the first BPA distribution parameters for risk levels across discrete intervals; The weight distribution curve generation module is further configured to apply uncertainty constraints to the initial response weights, calculate the target response weights, and generate a weight distribution curve based on the target response weights, referring to the following formula: in, Representing continuous features Values Risk level The target response weight, This represents the sum of weights for all risk levels corresponding to a continuous feature; The weights of the uncertainty terms of the continuous features are assigned according to the following formula: in, The retention factor represents the uncertainty term. The basic probability distribution parameters represent the uncertainty term of continuous features.

4. The method for predicting construction risks of transmission lines as described in claim 3, characterized in that, The feature filtering module is also configured to superimpose the target BPA distribution parameters of each feature to obtain the total BPA distribution parameters, as shown in the following formula: in, This represents the total BPA distribution parameter. Indicate the feature pair of the first Target BPA distribution parameters for each risk level The target BPA distribution parameters represent the uncertainty term of the feature; The feature selection module is further configured to calculate the uncertainty weight ratio of candidate discrete features and the uncertainty weight ratio of candidate continuous features based on the total BPA distribution parameters, referring to the following formula: in, Representing discrete features The proportion of uncertainty weight, Representing discrete features The corresponding uncertainty is assigned a basic probability value. Representing continuous features Uncertainty weighting ratio Representing continuous features Throughout the entire value range The integral of the basic probability assignment of internal uncertainty. Representing continuous features The length of the range of values; The feature filtering module is further configured to filter candidate discrete features and candidate continuous features based on an uncertainty threshold, and remove features whose uncertainty weight ratio is higher than the uncertainty threshold to obtain target discrete features and target continuous features.

5. A transmission line construction risk prediction device, characterized in that, The device applies the transmission line construction risk prediction method according to any one of claims 1 to 4, and the transmission line construction risk prediction device comprises: The construction sample analysis module is used to analyze historical construction samples of the construction phase of transmission lines to obtain construction characteristic information of each historical construction sample and risk level of multiple construction hazard types of each historical construction sample. The construction hazard types are the construction hazards contained in each key process of the construction phase of the transmission line. The information matrix construction module is used to construct a historical construction hazard information matrix for transmission lines based on the risk level of each type of construction hazard and the construction feature information. The historical construction hazard information matrix includes a target information matrix corresponding to each type of construction hazard. The target information matrix includes the input record number, construction discrete features, construction continuous features, and hazard risk level of multiple historical construction samples. The construction risk prediction module is used to input the historical construction hazard information matrix into a pre-built risk prediction model to obtain the risk prediction result of the transmission line. The risk prediction model is built based on a data-driven conflict-aware parallel evidence reasoning system. The risk prediction model is a dual-channel weighted architecture and includes a discrete feature mining module and a continuous feature mining module.

6. A transmission line construction risk prediction device, characterized in that, The transmission line construction risk prediction device includes: a memory, a processor, and a transmission line construction risk prediction program stored in the memory. The processor is used to run the transmission line construction risk prediction program, and the transmission line construction risk prediction program is configured to implement the transmission line construction risk prediction method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a transmission line construction risk prediction program, which, when executed by a processor, implements the transmission line construction risk prediction method as described in any one of claims 1 to 4.

8. A computer program product, characterized in that, The computer program product includes a transmission line construction risk prediction program, which, when executed by a processor, implements the steps of the transmission line construction risk prediction method as described in any one of claims 1 to 4.

Images