A power grid maintenance progress dynamic evaluation method, device, equipment and medium

Abstract

The application discloses a kind of power grid overhaul progress dynamic evaluation method, device, equipment and medium, belong to power grid overhaul progress dynamic evaluation field, the method is: the field working condition data and actual used overhaul duration of to-be-overhauled task are collected;According to the field working condition data, determine the comprehensive risk value, input the comprehensive risk value into the preset overhaul duration prediction model to obtain real-time predicted overhaul duration, combine the actual used overhaul duration to calculate the overhaul estimated progress value;When the overhaul estimated progress value is greater than the first threshold, the evaluation result of increasing personnel or adjusting process is generated, less than the second threshold, work specification prompt is generated, between the two is then according to preset interval continue to collect data.Therefore, by implementing the present application, the problem of low accuracy of power grid overhaul progress dynamic evaluation in the prior art can be solved.

Description

Technical Field

[0001] This application relates to the field of dynamic assessment of power grid maintenance progress, and in particular to a method, device, equipment and medium for dynamic assessment of power grid maintenance progress. Background Technology

[0002] As the power grid continues to expand, the requirements for power supply reliability are becoming increasingly stringent, leading to a growing workload for planned maintenance and emergency repairs. The speed of maintenance directly impacts power outages, operating costs, and on-site safety. In actual operations, factors such as sudden weather changes, varying equipment aging conditions, smooth coordination among multiple work teams, and the criticality of the load on the line under maintenance can all cause deviations from the original plan. Without real-time tracking and accurate assessment of progress, managers will find it difficult to promptly deploy additional personnel or adjust procedures when delays become apparent, and will also be unable to remind workers of safety regulations when progress is too rapid.

[0003] The current common practice is to statically compare actual progress with planned time. Dispatchers manually record the time elapsed on-site based on the pre-set planned duration on the task ticket, and then simply calculate the percentage of elapsed time to the planned time to roughly judge the progress. This approach is not very accurate. First, it doesn't consider changes happening on-site. For example, sudden heavy rain can slow down outdoor work, and proximity to live parts increases operational difficulty; these new risks all extend the remaining working time, but existing methods don't take these factors into account. Second, it only considers time, failing to incorporate risk factors such as the importance of related equipment and the number of users affected by power outages. Therefore, the calculated percentage does not accurately reflect the actual progress of maintenance work. These shortcomings make it difficult for dispatchers to make accurate and timely adjustment decisions based on existing data. Therefore, a dynamic assessment method for power grid maintenance progress is needed that can integrate real-time on-site data, dynamically predict progress, and proactively intervene. Summary of the Invention

[0004] This application provides a method, device, equipment, and medium for dynamic evaluation of power grid maintenance progress, which can solve the problem of low accuracy in dynamic evaluation of power grid maintenance progress in the prior art.

[0005] In a first aspect, embodiments of the present invention provide a method for dynamic evaluation of power grid maintenance progress, including: Collect on-site operating condition data and actual maintenance time for tasks to be repaired; The comprehensive risk value of the current maintenance is determined based on the on-site working condition data, and the comprehensive risk value is input into the preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained. Then, the estimated maintenance progress value is calculated based on the actual maintenance time used and the real-time predicted maintenance time. If the estimated maintenance progress value is greater than the first preset threshold, a first evaluation result is generated for dispatching additional workers or adjusting the work sequence. If the estimated maintenance progress value is less than the second preset threshold, a second evaluation result is generated to remind the workers of the work procedures. If the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, then the next round of on-site working condition data and actual maintenance time used will be collected according to the preset collection interval.

[0006] In this embodiment, by collecting on-site working condition data and actual maintenance time for the task to be repaired, the input information for progress assessment is transformed from static planned data to real-time dynamic on-site data, ensuring the accuracy and timeliness of the assessment basis from the source. Next, a comprehensive risk value is determined based on the on-site working condition data, quantifying multi-dimensional uncertainties such as weather, operational complexity, and personnel allocation into a calculable risk indicator, enabling subsequent predictions to truly reflect the actual difficulty level of the current working environment. Subsequently, this comprehensive risk value is input into a preset maintenance time prediction model to obtain the real-time predicted maintenance time. This dynamically corrected predicted value replaces the fixed planned time, overcoming the shortcomings of existing technologies that ignore changes in on-site risks, leading to large prediction deviations. Based on this, the estimated maintenance progress value is calculated based on the actual maintenance time and the real-time predicted maintenance time. This makes progress assessment no longer a simple percentage of time consumed, but a comprehensive progress indicator that integrates the remaining work difficulty and risk, significantly improving the accuracy of the progress value in representing the actual maintenance situation. Finally, by comparing the estimated maintenance progress value with a first preset threshold and a second preset threshold, a first assessment result is proactively generated when progress is lagging, to facilitate the deployment of additional personnel or adjustment of the work sequence; a second assessment result is generated when progress is ahead, to remind personnel of work procedures; and data is continuously updated at preset intervals when progress is normal. Therefore, this application can solve the problem of low accuracy in dynamic assessment of power grid maintenance progress in existing technologies.

[0007] As a preferred example of the first aspect, the on-site operating condition data includes meteorological data, operational data of the task to be maintained, personnel arrangement data of the task to be maintained, topology association data, and user area association data. The determination of the comprehensive risk value based on the on-site operating condition data includes: Based on the meteorological data, a weather risk value is calculated; based on the operational data, an operational risk value is calculated; based on the personnel arrangement data, a personnel risk value is calculated; based on the topology association data, a topology risk value is calculated; and based on the user area association data, a user area risk value is calculated. The comprehensive risk value is obtained by weighting the weather risk value, the operational risk value, the personnel risk value, the topology risk value, and the user area risk value according to preset weighting coefficients.

[0008] In this preferred example, by quantifying five types of risks—weather, operation, personnel, topology, and user area—a multi-dimensional decoupled assessment of power grid maintenance risks is achieved, overcoming the limitations of existing technologies that only assess single factors. Weather risk values ​​reflect uncontrollable environmental factors; operation and personnel risk values ​​characterize the standardization and capability level of the work performers; topology risk values ​​reflect the vulnerability of equipment and grid structure; and user area risk values ​​incorporate the social impact of public opinion and power supply reliability. Finally, by using preset weighting coefficients for weighted synthesis, the discrete, heterogeneous risk data are integrated into a unified quantitative indicator, providing a comprehensive, objective, and comparable risk assessment basis for maintenance decisions, thereby effectively supporting subsequent efficiency planning and dynamic control.

[0009] As a preferred example of the first aspect, the process of constructing the maintenance duration prediction model includes: Obtain information on equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual maintenance duration for several historical maintenance tasks. Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed. Each of the historical maintenance samples is input into a preset initial regression model based on gradient boosting decision tree for iterative training until the loss function value of the current iteration satisfies the first preset condition. The regression model of the current iteration is then output as the maintenance duration prediction model.

[0010] In this preferred example, by constructing a multi-dimensional historical sample that includes equipment attributes, shift configuration, and comprehensive risk values, risk factors are explicitly incorporated into the input feature space of the maintenance duration prediction model, thus overcoming the deficiency of existing prediction methods that ignore the impact of risks. Iterative training using a regression model based on gradient boosting decision trees automatically learns the complex nonlinear mapping relationship between risk variables and maintenance duration, and uses the gradient of the loss function to guide the model to gradually approach the optimal prediction accuracy. The maintenance duration prediction model obtained through this training can dynamically output accurate predicted durations based on real-time risk levels, providing crucial data-driven decision-making support for the quantitative assessment of the linkage between risk status and maintenance efficiency, as well as subsequent dynamic control.

[0011] As a preferred example of the first aspect, the construction of several historical maintenance samples is based on the equipment type, equipment operating years, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, including: For each historical maintenance task, the equipment type, equipment operating years, voltage level, number of construction team members, and comprehensive risk value corresponding to the historical maintenance task are combined into a feature vector, and the actual historical maintenance duration corresponding to the historical maintenance task is used as the label value corresponding to the feature vector. Then, a historical maintenance sample corresponding to the historical maintenance task is constructed based on the feature vector and the label value.

[0012] In this preferred example, a standardized supervised learning sample structure is established by uniformly constructing a feature vector from equipment type, years of operation, voltage level, number of shift workers, and comprehensive risk value, and using historical actual maintenance time as the labeled value. Equipment attributes and shift configuration reflect the inherent complexity of the maintenance task, while the comprehensive risk value quantifies the dynamic uncertainties of the external environment and personnel behavior; the labeled value provides a true efficiency benchmark for the model. This construction method ensures that the training data can fully map the intrinsic relationship between "risk level" and "maintenance time," laying a data foundation for the subsequent gradient boosting decision tree model to fully learn the multi-factor nonlinear coupling effects and achieve high-precision time prediction.

[0013] As a preferred example of the first aspect, the step of calculating the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time includes: Divide the actual maintenance time by the real-time predicted maintenance time to obtain the time ratio, and multiply the time ratio by a preset percentage value to obtain the estimated maintenance progress value.

[0014] As a preferred example of the first aspect, the evaluation method further includes: The initial environmental data of the task to be repaired is obtained, and the pre-existing comprehensive risk value is determined based on the initial environmental data. Then, the pre-existing comprehensive risk value is input into the maintenance time prediction model to obtain the initial predicted maintenance time of the task to be repaired. The risk deviation rate is determined based on the pre-existing comprehensive risk value and the comprehensive risk value.

[0015] In this preferred example, by acquiring initial environmental data and calculating the pre-maintenance comprehensive risk value before maintenance, and then using a maintenance duration prediction model to output the initial predicted maintenance duration, an objective progress benchmark is provided for subsequent efficiency monitoring. Based on this, the risk deviation rate between the pre-maintenance comprehensive risk value and the dynamically updated comprehensive risk value during maintenance is further calculated, quantifying the risk prediction error into an assessable indicator. This process, on the one hand, enables scientific planning of efficiency targets before maintenance, making scheduling decisions based on evidence.

[0016] As a preferred example of the first aspect, the evaluation method further includes: Obtain the actual total maintenance time of the task to be maintained, and calculate the maintenance time deviation rate based on the actual total maintenance time and the real-time predicted maintenance time. Based on the maintenance time deviation rate and the comprehensive risk value, a correlation index is calculated, and the maintenance time prediction model is adjusted according to the correlation index.

[0017] In this preferred example, the maintenance time deviation rate is obtained by comparing the actual total maintenance time with the real-time predicted maintenance time. This quantifies the accuracy of efficiency prediction into an assessable indicator, intuitively reflecting the impact of dynamic risk changes on the project schedule. Subsequently, a correlation index is calculated based on this deviation rate and the comprehensive risk value, quantitatively characterizing the inherent coupling relationship between risk level and efficiency deviation. Finally, the maintenance time prediction model is adjusted based on this correlation index, enabling the model to adaptively correct the weighting of risk factors in time prediction. This achieves closed-loop linkage optimization between risk assessment and efficiency prediction, thereby continuously improving the accuracy of efficiency monitoring and the reliability of control decisions in subsequent maintenance tasks.

[0018] Secondly, the present invention provides a dynamic evaluation device for power grid maintenance progress, comprising: a data acquisition module, a first monitoring module, a second monitoring module, a third monitoring module, and a fourth monitoring module; The data acquisition module is used to collect on-site operating data and actual maintenance time of the task to be repaired. The first monitoring module is used to determine the comprehensive risk value based on the on-site working condition data, and input the comprehensive risk value into a preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained, and then calculate the maintenance estimated progress value based on the actual maintenance time used and the real-time predicted maintenance time. The second monitoring module is used to output control instructions for increasing the number of workers or adjusting the work sequence if the estimated maintenance progress value is greater than the first preset threshold. The third monitoring module is used to output a safety compliance reminder message to remind workers of the work procedures if the estimated maintenance progress value is less than the second preset threshold. The fourth monitoring module is used to collect the next round of on-site working condition data and actual maintenance time according to the preset collection interval if the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold.

[0019] As a preferred example of the second aspect, the on-site operating condition data includes meteorological data, operational data of the task to be maintained, personnel arrangement data of the task to be maintained, topology association data, and user area association data. The determination of the comprehensive risk value based on the on-site operating condition data includes: Based on the meteorological data, a weather risk value is calculated; based on the operational data, an operational risk value is calculated; based on the personnel arrangement data, a personnel risk value is calculated; based on the topology association data, a topology risk value is calculated; and based on the user area association data, a user area risk value is calculated. The comprehensive risk value is obtained by weighting the weather risk value, the operational risk value, the personnel risk value, the topology risk value, and the user area risk value according to preset weighting coefficients.

[0020] As a preferred example of the second aspect, the process of constructing the maintenance duration prediction model includes: Obtain information on equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual maintenance duration for several historical maintenance tasks. Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed. Each of the historical maintenance samples is input into a preset initial regression model based on gradient boosting decision tree for iterative training until the loss function value of the current iteration satisfies the first preset condition. The regression model of the current iteration is then output as the maintenance duration prediction model.

[0021] As a preferred example of the second aspect, the construction of several historical maintenance samples is based on the equipment type, equipment operating years, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, including: For each historical maintenance task, the equipment type, equipment operating years, voltage level, number of construction team members, and comprehensive risk value corresponding to the historical maintenance task are combined into a feature vector, and the actual historical maintenance duration corresponding to the historical maintenance task is used as the label value corresponding to the feature vector. Then, a historical maintenance sample corresponding to the historical maintenance task is constructed based on the feature vector and the label value.

[0022] As a preferred example of the second aspect, the calculation of the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time includes: Divide the actual maintenance time by the real-time predicted maintenance time to obtain the time ratio, and multiply the time ratio by a preset percentage value to obtain the estimated maintenance progress value.

[0023] As a preferred example of the second aspect, the evaluation method further includes: The initial environmental data of the task to be repaired is obtained, and the pre-existing comprehensive risk value is determined based on the initial environmental data. Then, the pre-existing comprehensive risk value is input into the maintenance time prediction model to obtain the initial predicted maintenance time of the task to be repaired. The risk deviation rate is determined based on the pre-existing comprehensive risk value and the comprehensive risk value.

[0024] As a preferred example of the second aspect, the evaluation method further includes: Obtain the actual total maintenance time of the task to be maintained, and calculate the maintenance time deviation rate based on the actual total maintenance time and the real-time predicted maintenance time. Based on the maintenance time deviation rate and the comprehensive risk value, a correlation index is calculated, and the maintenance time prediction model is adjusted according to the correlation index.

[0025] In summary, this embodiment of the application, by collecting on-site working condition data and actual maintenance time for the task to be repaired, transforms the input information for progress assessment from static planned data to real-time dynamic on-site data, ensuring the accuracy and timeliness of the assessment basis from the source. Next, a comprehensive risk value is determined based on the on-site working condition data, quantifying multi-dimensional uncertainties such as weather, operational complexity, and personnel allocation into a calculable risk indicator, enabling subsequent predictions to truly reflect the actual difficulty level of the current working environment. Subsequently, this comprehensive risk value is input into a preset maintenance time prediction model to obtain the real-time predicted maintenance time. The dynamically corrected predicted value replaces the fixed planned time, overcoming the shortcomings of existing technologies that ignore changes in on-site risks, leading to large prediction deviations. Based on this, the estimated maintenance progress value is calculated based on the actual maintenance time and the real-time predicted maintenance time. This makes progress assessment no longer a simple percentage of time consumed, but a comprehensive progress indicator that integrates the remaining work difficulty and risk, significantly improving the accuracy of the progress value in representing the actual maintenance situation. Finally, by comparing the estimated maintenance progress value with a first preset threshold and a second preset threshold, a first assessment result is proactively generated when progress is lagging, to facilitate the deployment of additional personnel or adjustment of the work sequence; a second assessment result is generated when progress is ahead, to remind personnel of work procedures; and data is continuously updated at preset intervals when progress is normal. Therefore, this application can solve the problem of low accuracy in dynamic assessment of power grid maintenance progress in existing technologies.

[0026] Another embodiment of this application also provides a terminal device, including: 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 the steps of the power grid maintenance progress dynamic evaluation method of this application.

[0027] Another embodiment of this application also provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform the steps of the power grid maintenance progress dynamic assessment method of this application. Attached Figure Description

[0028] 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.

[0029] Figure 1 This is a flowchart illustrating an embodiment of a method for dynamic evaluation of power grid maintenance progress provided by the present invention. Figure 2This is a module structure diagram of one embodiment of a power grid maintenance progress dynamic evaluation device provided by the present invention. Detailed Implementation

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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.

[0035] 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).

[0036] 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. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0037] Example 1 Please refer to Figure 1 To address the issue of low accuracy in dynamic assessment of power grid maintenance progress in existing technologies, this application provides a method for dynamic assessment of power grid maintenance progress, comprising: S1. Collect on-site operating data and actual maintenance time of the task to be repaired; Specifically, real-time data is obtained from various sensing devices deployed at the maintenance site and external data interfaces. The on-site operating data includes at least four dimensions: First, real-time meteorological data for the substation or line area, collected through data interfaces with meteorological departments or small on-site weather stations, covering sudden thunderstorms, sudden increases in rainfall, and sudden changes in wind speed; second, personnel behavior data, utilizing on-site cameras for real-time image recognition of the work area, automatically capturing violations such as not wearing safety helmets, not following procedures, and crossing safety fences, and recording the time and type of occurrence; third, equipment status sensing data, including real-time status data transmitted from temperature sensors and opening / closing contact points on primary equipment such as circuit breakers and disconnectors, used to determine whether the equipment is performing its intended displacement or exhibiting abnormal overheating; and fourth, user feedback data, obtained by calling social media platform APIs to obtain the frequency of mentions of relevant power outage complaint keywords and news exposures in the maintenance area, while simultaneously extracting incremental call complaints by region and user type from the customer service system in real time. The collection of "actual maintenance time" is automatically accumulated through timestamps recorded by the mobile terminal APP used by maintenance personnel. The start time of maintenance is issued by the dispatch system to the mobile terminal with a time tag. The start and end of each key operation item are confirmed by the on-site personnel or automatically associated with the time point through the equipment status change signal. In this embodiment, the time span from the start time to the current time is continuously accumulated as the current maintenance time.

[0038] S2. Determine the comprehensive risk value of the current maintenance based on the on-site working condition data, and input the comprehensive risk value into the preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained. Then, calculate the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time. In some embodiments, the field operating condition data includes meteorological data, operational data of the task to be maintained, personnel arrangement data of the task to be maintained, topology association data, and user area association data. Determining the comprehensive risk value based on the field operating condition data includes: Based on the meteorological data, a weather risk value is calculated; based on the operational data, an operational risk value is calculated; based on the personnel arrangement data, a personnel risk value is calculated; based on the topology association data, a topology risk value is calculated; and based on the user area association data, a user area risk value is calculated. The comprehensive risk value is obtained by weighting the weather risk value, the operational risk value, the personnel risk value, the topology risk value, and the user area risk value according to preset weighting coefficients.

[0039] Specifically, the calculation process for the comprehensive risk value can be as follows: in, This is the weather risk value, used to measure weather risk. It is quantified using an exponential function mapping method, with the specific formula being: Where k is the fitting coefficient. This is the rainfall intensity index. Wind force index The lightning probability index is obtained through standardization of meteorological data.

[0040] This is the operational risk value, used to measure operational risk. It is quantified using the job condition hazard assessment method, and the specific formula is as follows: Where L represents the probability of an accident occurring, E represents the frequency of exposure, and C represents the severity of the accident's consequences.

[0041] This is the personnel risk value, used to measure personnel risk. It is quantified through a nonlinear coupling model, and the specific formula is as follows: , where a and b are fitting coefficients, N is the number of violations, and G is the skill level.

[0042] This is the topology risk value, used to measure topology risk. It is quantified by fitting a Weibull curve, and the specific formula is as follows: ,in, The initial failure rate is given by t, which represents the equipment's operating years. For environmental factors, (load factor) β represents the characteristic lifespan, and β is a device characteristic parameter.

[0043] This is the user area risk value, used to measure user area risk. It is quantified through a nonlinear coupling model, and the specific formula is as follows: Where P is the regional influence coefficient, Q is the user importance coefficient, and Y is the delay sensitivity coefficient.

[0044] The weighting coefficients can be objectively calculated using the entropy weighting method.

[0045] In some embodiments, the process of constructing the maintenance duration prediction model includes: Obtain information on equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual maintenance duration for several historical maintenance tasks. Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed. Each of the historical maintenance samples is input into a preset initial regression model based on gradient boosting decision tree for iterative training until the loss function value of the current iteration satisfies the first preset condition. The regression model of the current iteration is then output as the maintenance duration prediction model.

[0046] In some embodiments, the construction of several historical maintenance samples is based on the equipment type, equipment operating years, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, including: For each historical maintenance task, the equipment type, equipment operating years, voltage level, number of construction team members, and comprehensive risk value corresponding to the historical maintenance task are combined into a feature vector, and the actual historical maintenance duration corresponding to the historical maintenance task is used as the label value corresponding to the feature vector. Then, a historical maintenance sample corresponding to the historical maintenance task is constructed based on the feature vector and the label value.

[0047] Specifically, regarding the acquisition of historical data and the construction of samples, it is necessary to retrieve relevant records of each past maintenance task from the power grid's production management system, dispatch logs, and historical maintenance work orders, and extract several categories of key information. The first category is the static attributes of the equipment itself, including whether the equipment to be repaired is a transformer, circuit breaker, or transmission line, i.e., the equipment type; the second category is work resource information, i.e., how many people were assigned to that maintenance task, i.e., the number of construction team members; the third category is the comprehensive risk value corresponding to that task, which is a comprehensive score calculated using the multi-dimensional risk quantification model mentioned earlier, taking into account data such as the weather conditions at the time, personnel violation records, equipment topology location, and user area impact. In addition to these feature data, it is also necessary to record the actual time taken to complete that task, i.e., the historical actual maintenance time.

[0048] Next, training samples are constructed based on the aforementioned data. For each completed historical maintenance task, the equipment type, equipment age, voltage level, number of construction team members, and overall risk value are grouped together to form a feature vector. This vector represents all the known conditions of that maintenance task. Then, the actual historical maintenance time corresponding to that task is used as the label value of this feature vector, which tells the model: under these conditions, this is the actual time taken. In this way, a feature vector plus a label value constitutes a complete historical maintenance sample. By processing all historical maintenance tasks in this way, a sample set is accumulated for subsequent model training.

[0049] Finally, let's look at the specific model training process. The model chosen is based on a gradient boosting decision tree framework, specifically LightGBM. The previously constructed historical maintenance samples are fed into this initial regression model in batches, allowing it to begin iterative learning. In each iteration, the model first outputs a predicted maintenance duration for each sample based on the current parameters. Then, it calculates a loss function value using this predicted value and the actual maintenance duration recorded in the sample. Next, based on this error, the model calculates the first and second gradients. The first gradient reflects the direction of error change, and the second gradient reflects the curvature of the error change. Using this gradient information, the model searches for the optimal split point on each feature, determining which feature and what value to use to split the data for more accurate predictions in the next iteration. After the split points are found, a new decision tree grows. Each leaf node in the tree calculates an output value based on the sum of the gradients of the samples falling within it, serving as a correction for the prediction error of that batch of samples. This process continues, adding up the outputs of all previous trees in each round to obtain the final prediction. This process is repeated until the mean absolute error (MAE) or root mean square error (RMSE) calculated on the validation set reaches a preset threshold, such as MAE less than or equal to 10 or RMSE less than or equal to 10, or the error stops decreasing for several consecutive rounds. At this point, the model training is considered successful. The iteration stops, and the current set of model parameters is archived and solidified as a maintenance time prediction model that can be directly used later.

[0050] For example, the following is a detailed description of the training process for the maintenance time prediction model: ① Training sample construction: Sample set Where N is the number of historical maintenance samples. For feature vectors, This refers to the actual historical maintenance time.

[0051] ② Model training methods: The formula for calculating the loss function is shown below: in, This refers to the actual maintenance time. These are the model's predicted values.

[0052] The formula for calculating the first-order gradient in the direction of reaction error is as follows: The formula for calculating the second gradient of the response error curvature is shown below: For the b-th interval of each feature (the default number of intervals is 255), calculate the sum of the first-order gradients and the sum of the second-order gradients of all samples within that interval, as shown in the following formula: Next, the splitting gain is calculated, and the optimal splitting point is selected. The specific formula is shown below: in, This represents the sum of the gradients of the left subtree. This represents the sum of the gradients of the right subtree. This is the regularization coefficient (default setting is 0.1). Sets the leaf split gain threshold (default is 0.01) to filter out meaningless splits.

[0053] The optimal output of a leaf node (i.e., the correction value for a single leaf) is calculated using the following formula: in, This represents the j-th leaf node.

[0054] The formula for calculating the predicted value is shown below: Where M is the total number of ensemble trees, Represents the m-th tree pair of samples The predicted value (i.e., the sample falling into a leaf node) ).

[0055] In some embodiments, calculating the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time includes: Divide the actual maintenance time by the real-time predicted maintenance time to obtain the time ratio, and multiply the time ratio by a preset percentage value to obtain the estimated maintenance progress value.

[0056] S3. If the estimated maintenance progress value is greater than the first preset threshold, a first evaluation result is generated for dispatching more workers or adjusting the work sequence. S4. If the estimated maintenance progress value is less than the second preset threshold, a second evaluation result is generated to remind the operators of the work procedures. S5. If the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, then collect the next round of on-site working condition data and the actual maintenance time used according to the preset collection interval.

[0057] In some embodiments, the evaluation method further includes: The initial environmental data of the task to be repaired is obtained, and the pre-existing comprehensive risk value is determined based on the initial environmental data. Then, the pre-existing comprehensive risk value is input into the maintenance time prediction model to obtain the initial predicted maintenance time of the task to be repaired. The risk deviation rate is determined based on the pre-existing comprehensive risk value and the comprehensive risk value.

[0058] Specifically, the risk deviation rate can be obtained by subtracting the comprehensive risk value from the comprehensive risk beforehand, and then dividing by the comprehensive risk value.

[0059] In some embodiments, the evaluation method further includes: Obtain the actual total maintenance time of the task to be maintained, and calculate the maintenance time deviation rate based on the actual total maintenance time and the real-time predicted maintenance time. Based on the maintenance time deviation rate and the comprehensive risk value, a correlation index is calculated, and the maintenance time prediction model is adjusted according to the correlation index.

[0060] Specifically, after all maintenance work is completed and archived, this embodiment automatically retrieves the actual total maintenance time of this maintenance task from the historical database of the data storage module. This time is accumulated and uploaded by the on-site mobile terminal according to the start and end times of the operation items and confirmed after time sequence alignment verification. This embodiment compares the actual total maintenance time with the real-time predicted maintenance time obtained by dynamic correction through a lightweight gradient boosting regression model during the last stage, and calculates the maintenance time deviation rate. The calculation formula is the difference between the actual total maintenance time and the real-time predicted maintenance time divided by the actual total maintenance time, and then multiplied by 100%. This indicator reflects the overall prediction accuracy of the maintenance time prediction model. Subsequently, this embodiment calls a single nonlinear multivariate coupled correlation model to conduct in-depth quantitative analysis of the intrinsic correlation between the risk prediction results and the efficiency deviation results. This model uses the pre-event comprehensive risk value and the in-event comprehensive risk value as input variables, and performs a coupling operation with the absolute value of the maintenance time deviation rate through an exponential nonlinear increasing function to obtain a comprehensive correlation index.

[0061] Specifically, the expression for the single nonlinear multi-element coupling correlation model is as follows: Among them, in the model The terms show a non-linear increasing trend, indicating that , The larger the risk level, the stronger its impact on the efficiency deviation rate, and the growth is non-uniform. This represents the deviation rate of total maintenance time.

[0062] The design of this exponential nonlinear increasing function follows the patterns observed in actual maintenance: when the risk value is in a low range, for every zero-unit increase in the risk value, the correlation index increases by only three to five percentage points, indicating that the impact of risk on efficiency deviation is still weak and can be effectively offset by conventional control measures; however, when the risk value climbs to a high range, the same zero-unit increase in risk leads to a sharp increase in the correlation index at a rate of twelve to fifteen percentage points. This accurately reflects the actual operational characteristics of high-risk conditions, which easily lead to uncontrolled maintenance time and a significant increase in deviation rate. The calculation results of the correlation index are used, on the one hand, in the post-maintenance review and analysis report to reveal the quantitative linkage between risk variables and efficiency variables; on the other hand, they directly trigger the closed-loop optimization submodule of the power grid maintenance efficiency and risk control device to adjust the maintenance time prediction model. The specific adjustment methods include: First, the feature vector of this maintenance task and the corresponding actual total maintenance time are used as new samples and added to the training sample set of the lightweight gradient boosting regression model. Second, in this embodiment, the model is retrained based on the updated complete sample set on a quarterly basis or under manual triggering authorized by management. During the retraining process, the model's learning rate parameter is adaptively and iteratively adjusted according to the changes in the prediction error of the old and new sample sets. The adjustment method is to multiply the original learning rate by the ratio of the average absolute error of the old sample set to the estimated average absolute error after adding the new samples, so as to ensure that the model can still maintain convergence stability and prediction accuracy after incorporating the latest maintenance data. In addition, if the correlation index continues to show an abnormal amplification effect of certain specific risk dimensions on the maintenance time deviation, this embodiment also supports maintenance personnel to manually adjust the fitting coefficient in the corresponding risk quantification formula through the visual interaction module. For example, the index fitting coefficient used in weather risk calculation is increased from 0.8 to 1.0, so that the mapping relationship between the risk prediction value and the efficiency prediction value is more in line with the actual situation on site. Through the aforementioned incremental learning of samples driven by correlation indicators, periodic retraining of the model, and adaptive parameter adjustment mechanism, the maintenance duration prediction model can be continuously optimized after each maintenance repair, gradually reducing the duration prediction error rate of subsequent maintenance tasks, and achieving a closed-loop iterative improvement in maintenance efficiency control capabilities.

[0063] Example 2 like Figure 2 As shown, based on the above method embodiments, corresponding device embodiments are provided; An embodiment of the present invention provides a dynamic evaluation device for power grid maintenance progress, comprising: a data acquisition module 21, a first monitoring module 22, a second monitoring module 23, a third monitoring module 24, and a fourth monitoring module 25; Data acquisition module 21 is used to collect on-site operating condition data and actual maintenance time of the task to be repaired; The first monitoring module 22 is used to determine the comprehensive risk value based on the on-site working condition data, and input the comprehensive risk value into a preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained, and then calculate the maintenance estimated progress value based on the actual maintenance time used and the real-time predicted maintenance time. The second monitoring module 23 is used to output control instructions for increasing the number of workers or adjusting the work sequence if the estimated maintenance progress value is greater than the first preset threshold. The third monitoring module 24 is used to output a safety compliance reminder message to remind the workers of the work procedures if the estimated maintenance progress value is less than the second preset threshold. The fourth monitoring module 25 is used to collect the next round of on-site working condition data and the actual maintenance time used according to the preset collection interval if the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold.

[0064] In some embodiments, the field operating condition data includes meteorological data, operational data of the task to be maintained, personnel arrangement data of the task to be maintained, topology association data, and user area association data. Determining the comprehensive risk value based on the field operating condition data includes: Based on the meteorological data, a weather risk value is calculated; based on the operational data, an operational risk value is calculated; based on the personnel arrangement data, a personnel risk value is calculated; based on the topology association data, a topology risk value is calculated; and based on the user area association data, a user area risk value is calculated. The comprehensive risk value is obtained by weighting the weather risk value, the operational risk value, the personnel risk value, the topology risk value, and the user area risk value according to preset weighting coefficients.

[0065] In some embodiments, the process of constructing the maintenance duration prediction model includes: Obtain information on equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual maintenance duration for several historical maintenance tasks. Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed. Each of the historical maintenance samples is input into a preset initial regression model based on gradient boosting decision tree for iterative training until the loss function value of the current iteration satisfies the first preset condition. The regression model of the current iteration is then output as the maintenance duration prediction model.

[0066] In some embodiments, the construction of several historical maintenance samples is based on the equipment type, equipment operating years, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, including: For each historical maintenance task, the equipment type, equipment operating years, voltage level, number of construction team members, and comprehensive risk value corresponding to the historical maintenance task are combined into a feature vector, and the actual historical maintenance duration corresponding to the historical maintenance task is used as the label value corresponding to the feature vector. Then, a historical maintenance sample corresponding to the historical maintenance task is constructed based on the feature vector and the label value.

[0067] In some embodiments, calculating the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time includes: Divide the actual maintenance time by the real-time predicted maintenance time to obtain the time ratio, and multiply the time ratio by a preset percentage value to obtain the estimated maintenance progress value.

[0068] In some embodiments, the evaluation method further includes: The initial environmental data of the task to be repaired is obtained, and the pre-existing comprehensive risk value is determined based on the initial environmental data. Then, the pre-existing comprehensive risk value is input into the maintenance time prediction model to obtain the initial predicted maintenance time of the task to be repaired. The risk deviation rate is determined based on the pre-existing comprehensive risk value and the comprehensive risk value.

[0069] In some embodiments, the evaluation method further includes: Obtain the actual total maintenance time of the task to be maintained, and calculate the maintenance time deviation rate based on the actual total maintenance time and the real-time predicted maintenance time. Based on the maintenance time deviation rate and the comprehensive risk value, a correlation index is calculated, and the maintenance time prediction model is adjusted according to the correlation index.

[0070] For more detailed steps and working principles of this embodiment, please refer to the relevant description in Embodiment 1, but not limited to these descriptions.

[0071] In summary, this embodiment of the application, by collecting on-site working condition data and actual maintenance time for the task to be repaired, transforms the input information for progress assessment from static planned data to real-time dynamic on-site data, ensuring the accuracy and timeliness of the assessment basis from the source. Next, a comprehensive risk value is determined based on the on-site working condition data, quantifying multi-dimensional uncertainties such as weather, operational complexity, and personnel allocation into a calculable risk indicator, enabling subsequent predictions to truly reflect the actual difficulty level of the current working environment. Subsequently, this comprehensive risk value is input into a preset maintenance time prediction model to obtain the real-time predicted maintenance time. The dynamically corrected predicted value replaces the fixed planned time, overcoming the shortcomings of existing technologies that ignore changes in on-site risks, leading to large prediction deviations. Based on this, the estimated maintenance progress value is calculated based on the actual maintenance time and the real-time predicted maintenance time. This makes progress assessment no longer a simple percentage of time consumed, but a comprehensive progress indicator that integrates the remaining work difficulty and risk, significantly improving the accuracy of the progress value in representing the actual maintenance situation. Finally, by comparing the estimated maintenance progress value with a first preset threshold and a second preset threshold, a first assessment result is proactively generated when progress is lagging, to facilitate the deployment of additional personnel or adjustment of the work sequence; a second assessment result is generated when progress is ahead, to remind personnel of work procedures; and data is continuously updated at preset intervals when progress is normal. Therefore, this application can solve the problem of low accuracy in dynamic assessment of power grid maintenance progress in existing technologies.

[0072] 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 in this application, 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.

[0073] Example 3 Based on the above embodiments of the dynamic assessment method for power grid maintenance progress, another embodiment of this application 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 the dynamic assessment method for power grid maintenance progress of any embodiment of this application.

[0074] 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 this application. The one or more module units 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.

[0075] 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.

[0076] 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.

[0077] Example 4 Based on the above-described method embodiments, another embodiment of this application 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 power grid maintenance progress dynamic assessment method described in any of the above-described method embodiments of this application.

[0078] 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 methods of the above embodiments 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, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

Claims

1. A method for dynamic assessment of power grid maintenance progress, characterized in that, include: Collect on-site operating data and actual maintenance time for tasks to be repaired; The comprehensive risk value of the current maintenance is determined based on the on-site working condition data, and the comprehensive risk value is input into the preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained. Then, the estimated maintenance progress value is calculated based on the actual maintenance time used and the real-time predicted maintenance time. If the estimated maintenance progress value is greater than the first preset threshold, a first evaluation result is generated for dispatching additional workers or adjusting the work sequence. If the estimated maintenance progress value is less than the second preset threshold, a second evaluation result is generated to remind the workers of the work procedures. If the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, then the next round of on-site working condition data and actual maintenance time used will be collected according to the preset collection interval.

2. The method of claim 1, wherein, The on-site operating data includes meteorological data, operational data of the tasks to be maintained, personnel arrangement data for the tasks to be maintained, topology association data, and user area association data. Determining the comprehensive risk value based on the on-site operating data includes: Based on the meteorological data, a weather risk value is calculated; based on the operational data, an operational risk value is calculated; based on the personnel arrangement data, a personnel risk value is calculated; based on the topology association data, a topology risk value is calculated; and based on the user area association data, a user area risk value is calculated. The comprehensive risk value is obtained by weighting the weather risk value, the operational risk value, the personnel risk value, the topology risk value, and the user area risk value according to preset weighting coefficients.

3. The method for dynamic evaluation of power grid maintenance progress as described in claim 1, characterized in that, The process of constructing the maintenance duration prediction model includes: Obtain information on equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual maintenance duration for several historical maintenance tasks. Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed. Each of the historical maintenance samples is input into a preset initial regression model based on gradient boosting decision tree for iterative training until the loss function value of the current iteration satisfies the first preset condition. The regression model of the current iteration is then output as the maintenance duration prediction model.

4. The method for dynamic evaluation of power grid maintenance progress as described in claim 3, characterized in that, Based on the equipment type, equipment service life, voltage level, number of construction team members, comprehensive risk value, and actual historical maintenance duration corresponding to each historical maintenance task, several historical maintenance samples are constructed, including: For each historical maintenance task, the equipment type, equipment operating years, voltage level, number of construction team members, and comprehensive risk value corresponding to the historical maintenance task are combined into a feature vector, and the actual historical maintenance duration corresponding to the historical maintenance task is used as the label value corresponding to the feature vector. Then, a historical maintenance sample corresponding to the historical maintenance task is constructed based on the feature vector and the label value.

5. The method for dynamic evaluation of power grid maintenance progress as described in claim 1, characterized in that, The step of calculating the estimated maintenance progress value based on the actual maintenance time used and the real-time predicted maintenance time includes: Divide the actual maintenance time by the real-time predicted maintenance time to obtain the time ratio, and multiply the time ratio by a preset percentage value to obtain the estimated maintenance progress value.

6. The method for dynamic evaluation of power grid maintenance progress as described in claim 1, characterized in that, The evaluation method further includes: The initial environmental data of the task to be repaired is obtained, and the pre-existing comprehensive risk value is determined based on the initial environmental data. Then, the pre-existing comprehensive risk value is input into the maintenance time prediction model to obtain the initial predicted maintenance time of the task to be repaired. The risk deviation rate is determined based on the pre-existing comprehensive risk value and the comprehensive risk value.

7. The method for dynamic evaluation of power grid maintenance progress as described in claim 1, characterized in that, The evaluation method further includes: Obtain the actual total maintenance time of the task to be maintained, and calculate the maintenance time deviation rate based on the actual total maintenance time and the real-time predicted maintenance time. Based on the maintenance time deviation rate and the comprehensive risk value, a correlation index is calculated, and the maintenance time prediction model is adjusted according to the correlation index.

8. A dynamic evaluation device for power grid maintenance progress, characterized in that, include: The system includes a data acquisition module, a first monitoring module, a second monitoring module, a third monitoring module, and a fourth monitoring module. The data acquisition module is used to collect on-site operating condition data and actual maintenance time of the task to be repaired. The first monitoring module is used to determine the comprehensive risk value based on the on-site working condition data, and input the comprehensive risk value into a preset maintenance time prediction model to obtain the real-time predicted maintenance time of the task to be maintained, and then calculate the maintenance estimated progress value based on the actual maintenance time used and the real-time predicted maintenance time. The second monitoring module is used to output control instructions for increasing the number of workers or adjusting the work sequence if the estimated maintenance progress value is greater than the first preset threshold. The third monitoring module is used to output a safety compliance reminder message to remind workers of the work procedures if the estimated maintenance progress value is less than the second preset threshold. The fourth monitoring module is used to collect the next round of on-site working condition data and actual maintenance time according to the preset collection interval if the estimated maintenance progress value is less than or equal to the first preset threshold and greater than or equal to the second preset threshold.

9. A terminal device, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements a method for dynamic assessment of power grid maintenance progress as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform a dynamic assessment method for power grid maintenance progress as described in any one of claims 1 to 7.

Images