Power grid intelligent operation and maintenance management method and system

Abstract

The invention provides an intelligent operation and maintenance management method and system for a power grid, and relates to the technical field of intelligent power grids. Performing feature screening on the state data, and determining health features related to life; performing fusion processing on the health features to generate a fusion feature sequence; determining the operation state classification of the key equipment based on the fused feature sequence; according to a running state classification result, constructing a random degradation model related to a load condition; predicting the residual life of the key equipment through a random degradation model, and generating an equipment health state vector; according to the equipment health state vector, constructing a load-dependent prediction maintenance collaborative optimization model, and generating a collaborative optimization decision; issuing the collaborative optimization decision to an execution unit, and monitoring the equipment state change and the decision execution effect in real time; and according to the equipment state change and the decision execution effect, dynamically updating and fusing the convolutional network and the operation and maintenance strategy to realize intelligent operation and maintenance of the power grid.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a method and system for intelligent operation and maintenance management of power grids. Background Technology

[0002] The significance of intelligent operation and maintenance management of power grids lies in ensuring that massive amounts of equipment data from power generation, transmission to distribution can be efficiently and accurately perceived, analyzed and utilized. This is crucial for the safe operation of large-scale power grids, distributed energy systems and new smart grids, and is a key technological link in realizing intelligent management and autonomous decision-making of power systems.

[0003] Currently, power grid operation and maintenance management mainly adopts a sequential management model based on fixed-cycle maintenance or threshold alarms, that is, maintenance and management are carried out in the order of "periodic inspection, abnormal alarm, and manual decision-making". This method is simple to implement, relying on traditional monitoring systems to complete equipment status collection and manual judgment, without the need for complex feature modeling and predictive analysis.

[0004] However, traditional methods based on fixed-period or threshold alarms have significant limitations when facing complex power grid operating environments. When the operating status of equipment is affected by load fluctuations, environmental disturbances, or the coupling of multiple factors, traditional methods cannot dynamically reflect the true health status of the equipment. With a large number of devices and high-dimensional monitoring data, data analysis efficiency is low and early warning is delayed, making it difficult to meet the comprehensive requirements of the power grid for real-time performance, reliability, and economy. This makes maintenance decisions still rely on experience-based judgments and makes it difficult to achieve coordinated optimization of operation and maintenance plans and system scheduling. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide a smart power grid operation and maintenance management method that can solve the obvious limitations of traditional methods based on fixed-cycle or threshold alarms when facing complex power grid operating environments. When the operating status of equipment is affected by load fluctuations, environmental disturbances, or the coupling of multiple factors, traditional methods cannot dynamically reflect the true health status of the equipment. With a large number of equipment and high-dimensional monitoring data, the data analysis efficiency is low and the early warning is delayed, making it difficult to meet the comprehensive requirements of the power grid for real-time performance, reliability, and economy. This makes maintenance decisions still rely on experience-based judgment and makes it difficult to achieve the technical problem of coordinated optimization of operation and maintenance plans and system scheduling.

[0006] A first aspect of this invention provides a method for intelligent operation and maintenance management of power grids, comprising:

[0007] S1: Obtain status data of key power grid equipment;

[0008] S2: By using a relational autoencoder model, feature filtering is performed on the state data to determine health characteristics related to lifespan;

[0009] S3: By fusing convolutional networks, health features are fused to generate a fused feature sequence;

[0010] S4: Based on the fused feature sequence, the operating status of key equipment is classified through a weighted voting mechanism;

[0011] S5: Based on the classification results of operating status and combined with historical data of key equipment, construct a stochastic degradation model related to the load conditions of key equipment;

[0012] S6: Predict the remaining lifespan of critical equipment using a stochastic degradation model and generate an equipment health status vector;

[0013] The equipment health status vector includes: lifetime distribution parameters, survival probability, and failure risk rate;

[0014] S7: Based on the equipment health status vector, construct a load-dependent predictive maintenance collaborative optimization model with predictive maintenance strategy as the core, and generate operation and maintenance strategy through the load-dependent predictive maintenance collaborative optimization model;

[0015] S8: Distribute the operation and maintenance strategy to the execution unit, and monitor the changes in device status and the effectiveness of decision execution in real time through the IoT platform;

[0016] S9: Based on changes in equipment status and the effectiveness of decision execution, dynamically update the fusion convolutional network and operation and maintenance strategies to achieve self-learning and closed-loop management of intelligent operation and maintenance of the power grid.

[0017] A second aspect of this invention provides a smart power grid operation and maintenance management system, comprising: a processor and a memory;

[0018] The memory stores programs or instructions that can run on a processor, and when the programs or instructions are executed by the processor, they implement the steps of the smart grid operation and maintenance management method as described in the first aspect.

[0019] A third aspect of the present invention provides a readable storage medium on which a program or instructions are stored, and when the program or instructions are executed by a processor, the steps of the power grid intelligent operation and maintenance management method of the first aspect are implemented.

[0020] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0021] In this embodiment of the invention, a relational autoencoder model is used to perform feature filtering and relational constraint modeling on multi-source state data, which can accurately extract health features related to lifespan and improve the relevance and interpretability of feature representation. By fusing convolutional networks and weighted voting mechanisms, high-precision identification of equipment operating status can be achieved in a multi-dimensional feature space. Through a load-dependent stochastic degradation model and a Bayesian dynamic update mechanism, the degradation process and remaining lifespan of the equipment can be accurately predicted, effectively addressing performance fluctuations under different load and environmental conditions. Furthermore, by constructing a load-dependent predictive maintenance collaborative optimization model, joint optimization of equipment-level health status and system-level scheduling decisions is achieved, significantly reducing system operation and maintenance costs and the risk of unplanned downtime. Attached Figure Description

[0022] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. Obviously, the drawings described below are merely some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0023] Figure 1 This is a flowchart illustrating a smart power grid operation and maintenance management method provided in an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of the structure of a smart power grid operation and maintenance management system provided in an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] The intelligent power grid operation and maintenance management method provided by the present invention will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0027] Reference manual attached Figure 1 The diagram illustrates a flowchart of a smart power grid operation and maintenance management method provided by an embodiment of the present invention.

[0028] This invention provides a method for intelligent operation and maintenance management of power grids, which may include the following steps:

[0029] S1: Obtain status data of key power grid equipment.

[0030] Among them, the power grid refers to a complex energy network system composed of power generation, transmission, transformation, distribution and consumption links, and is the infrastructure for realizing the production and transmission of electrical energy in space.

[0031] Among them, key equipment are the core components in the power grid that have the greatest impact on the system's safety, stability, and reliability, such as transformers, circuit breakers, busbars, cables, surge arresters, and sensor terminals. Their changes in state often directly determine the operational safety of the power system.

[0032] Status data refers to multi-dimensional information data that can reflect the health status, performance parameters and environmental conditions of equipment, including electrical performance data (current, voltage, power factor, etc.), mechanical status data (vibration, temperature, noise), environmental data (humidity, air temperature, salt spray concentration), and insulation and leakage current data.

[0033] It should be noted that by fusing data from multiple sources of sensors and IoT terminals, multimodal information, including electrical, mechanical, and environmental data, can be acquired simultaneously, comprehensively reflecting the health status and operating conditions of the equipment. This step can collect key parameters such as temperature, current, vibration, and leakage current in real time, and achieve efficient transmission and preprocessing through edge computing and communication networks, thereby ensuring the timeliness and accuracy of the data.

[0034] In one possible implementation, the status data includes: sensor data, environmental data, and leakage current data.

[0035] Specifically, the sensor data includes electrical performance data, such as vibration, temperature, and current.

[0036] S2: By using a relational autoencoder model, feature filtering is performed on state data to determine health characteristics related to lifespan.

[0037] Among them, the relational autoencoder model is an improved deep learning structure consisting of an encoder and a decoder. It can not only compress and reconstruct high-dimensional features, but also introduce "relationship constraints between features" during training to maintain the correlation and structural features between different data modalities.

[0038] Among them, health characteristics refer to key characteristic variables that are closely related to equipment performance degradation, degradation rate or failure probability, such as temperature drift rate, leakage current growth rate, vibration spectrum energy change, etc.

[0039] It's worth noting that, compared to traditional principal component analysis or simple dimensionality reduction methods, this model not only automatically learns the latent structure of high-dimensional data but also maintains the physical correlation and temporal consistency between features during training, thus avoiding the loss of crucial information. This step effectively reduces redundant features, lowers model complexity, and improves the accuracy and stability of subsequent health assessments and lifespan predictions. Simultaneously, it enhances the model's generalization ability, enabling the system to maintain robust performance under different loads, environments, and device types.

[0040] In one possible implementation, S2 specifically includes:

[0041] S201: Using state data as input, construct an autoencoder model that includes the relationship between the encoder and decoder.

[0042] The encoder maps high-dimensional input features to a low-dimensional latent space, the decoder reconstructs the latent features to restore the original input features, and introduces a relational constraint term between features into the loss function.

[0043] S202: Calculate the relationship matrix between various data in the state data using a relational autoencoder model.

[0044] The relation matrix is ​​used to describe the similarity or dependency between multimodal features and is an important structural information reflecting the intrinsic relationship between different signal modes.

[0045] S203: Based on the encoder in the relational autoencoder model, and combined with the nonlinear mapping function, calculate the key feature matrix of the state data:

[0046]

[0047] Where Y represents the key feature matrix, f() represents the encoder output, i.e., the key feature matrix, and s f Let W represent the activation function, W represent the weight matrix, X represent the state data, and b represent the weight matrix. X This represents the bias vector.

[0048] S204: Based on the key feature matrix, the state data and the corresponding relation matrix are reconstructed using the decoder in the relational autoencoder model.

[0049] S205: Based on the reconstructed state data and relation matrix, determine the objective function based on relation preservation constraints:

[0050]

[0051]

[0052] in, This represents the parameters of the relational autoencoder model. Represents the weight parameters. This represents the loss from reconstructing the state data, i.e., the difference between the original state data and the reconstructed state data. This represents the loss from reconstructing the relation matrix, i.e., the state data relation matrix. Relationship matrix with reconstructed state data The gap between them This indicates the reconstruction of state data. T Indicates transpose. Represents the rectification function, r ij This represents the relationship matrix between state data i and j, where t represents the relationship threshold. This indicates that the solution is to minimize the model parameters θ. Indicates other.

[0053] S206: Generate health features by minimizing the objective function value.

[0054] It is important to note that, unlike traditional linear dimensionality reduction or independent feature extraction methods, this method considers both the similarity and dependencies between features during encoding and decoding, preserving the inherent coupling structure of different modal signals using a relation matrix. By introducing feature relationship preservation constraints into the loss function, the model not only focuses on the reconstruction error of the data itself but also maintains the relative geometric structure between the original features, thereby obtaining a more physically meaningful and interpretable latent feature representation. This process effectively eliminates redundant features, improves the representativeness and robustness of health features, and provides high-quality input for subsequent state classification and lifetime prediction. Ultimately, this method achieves an adaptive nonlinear mapping from raw data to health features, significantly improving the model's generalization performance and feature extraction efficiency, laying a solid foundation for accurate assessment and dynamic optimization of smart grid operation and maintenance.

[0055] S3: By fusing convolutional networks, health features are fused to generate a fused feature sequence.

[0056] Among them, the fusion convolutional network is a deep neural network model that combines multi-layer convolutional structure with feature fusion mechanism. It can simultaneously process health feature data from different sensors, different modalities or different time scales. It extracts local patterns through convolution and then achieves the collaborative expression of multi-dimensional information through feature fusion layer.

[0057] Among them, the fusion feature sequence refers to the temporal high-dimensional feature set generated after the feature fusion of multiple layers of convolutional network. It can comprehensively reflect the health evolution law of equipment under different working conditions and time periods, and provide a unified input expression for subsequent classification and prediction.

[0058] It should be noted that the deep fusion and dynamic feature extraction of multimodal health features are achieved through the fusion convolutional network, significantly improving the completeness of data representation and the model's discriminative ability. Traditional feature fusion methods often employ simple weighting or concatenation, making it difficult to capture the spatial correlations and temporal dependencies between different features. This step, however, utilizes the local perception and parameter sharing characteristics of convolutional neural networks to automatically extract high-order correlation patterns of different health features across multiple convolutional layers, achieving structured fusion at the feature level. Simultaneously, the fusion convolutional network can adaptively learn from the input time-series health data, generating a fusion feature sequence reflecting degradation trends, allowing device state changes to exhibit continuous evolution in the feature space. This method effectively suppresses noise interference and data redundancy, enhancing feature stability and discriminability, and providing more representative input features for subsequent operational status classification and lifetime prediction.

[0059] S4: Based on the fused feature sequence, the operating status of key equipment is classified through a weighted voting mechanism.

[0060] Among them, the weighted voting mechanism is an integrated decision-making method. It integrates the discrimination results of multiple classification models or different feature subspaces, and assigns weights according to the importance of each classifier or feature. It then performs weighted summation or probability fusion on the final output categories to obtain more robust classification results.

[0061] Among them, the operation status classification refers to dividing the current health status of key power grid equipment into different categories, such as "normal operation", "minor degradation", "severe degradation" or "potential fault", in order to guide subsequent predictive maintenance and scheduling decisions.

[0062] It should be noted that the weighted voting mechanism achieves the synergistic fusion of multi-source feature information and multi-model judgment results, significantly improving the accuracy and stability of operational status identification. By fusing multi-layered information from feature sequences, the weighted voting mechanism can comprehensively integrate temporal, spatial, and modal features at the global level, enabling multi-faceted analysis of equipment operational status. This step not only improves the accuracy and anti-interference capability of status identification but also provides a reliable status classification basis for subsequent life prediction and maintenance optimization, promoting the transformation of power grid operation and maintenance decisions from experience-based judgment to intelligent identification.

[0063] S5: Based on the classification results of operating status and combined with historical data of key equipment, construct a stochastic degradation model related to the load conditions of key equipment.

[0064] Historical data refers to the condition monitoring records accumulated by the equipment during long-term operation, including temperature, current, vibration, partial discharge intensity and environmental conditions, which can reflect the time characteristics of equipment degradation and evolution.

[0065] Among them, load conditions refer to the level of electrical or mechanical load that the equipment bears under different operating conditions, such as voltage fluctuations, power output, environmental stress, etc., which are important external factors affecting the degradation rate and lifespan.

[0066] Among them, the stochastic degradation model is a mathematical model that uses probability statistics and stochastic process theory to describe the degradation law of equipment performance over time. It can take into account the uncertainty of degradation rate and the impact of load on degradation process, and is used to predict remaining life and failure rate.

[0067] It should be noted that by combining the results of operational status classification with historical load data, a stochastic degradation model that dynamically reflects the true aging patterns of equipment was constructed, thereby achieving a quantitative description of equipment lifespan changes. Traditional lifespan models are mostly based on the assumption of a fixed degradation rate, which is difficult to adapt to the variable load conditions and environmental conditions in the power grid. This step, by embedding the load effect into the drift and diffusion terms of the degradation model, can capture the dynamic changes in the degradation rate under different operating conditions, making the model closer to actual operating conditions. This step provides a precise mathematical foundation for subsequent remaining lifespan prediction, realizing a leap from qualitative health assessment to quantitative lifespan modeling, and significantly improving the scientific nature and prediction accuracy of power grid equipment health management.

[0068] In one possible implementation, S5 specifically includes:

[0069] S50A: Based on the classification results of operating status, the set of key power grid equipment under different operating conditions is stratified, and the key power grid equipment is sliced ​​according to the load level to form a homogeneous sample set.

[0070] Among them, the homogeneous sample set refers to the set of device samples under similar load levels, environmental conditions and state categories, which is used to reduce statistical bias caused by data heterogeneity.

[0071] S50B: In a homogeneous sample set, a discrete increment is defined for the degradation signal of each critical device, and the average increment under the reference load is used as the normalization factor.

[0072]

[0073] in, This represents the discrete degradation increment of device i during the j-th time interval. This indicates that device i at time t j ,load The cumulative degradation level below, Indicates that device i is at time t j-1 ,load The cumulative degradation level below.

[0074] Among them, degradation signals represent the process of equipment performance deterioration over time, such as changes in temperature rise, partial discharge intensity, or leakage current, and are key indicators reflecting the aging state of equipment.

[0075] Among them, discrete increments represent the changes in the amount of degradation of the device within adjacent time intervals, and are used to quantify the degradation rate.

[0076] S50C: Normalizes discrete increments based on a normalization factor.

[0077] S50D: Based on the normalized discrete increment, the load influence function is determined by fitting the multiplicative relationship between the load level and degradation rate of key equipment through nonlinear regression.

[0078] Among them, the load effect function is a nonlinear function that characterizes the degree of influence of load level on the rate of equipment degradation.

[0079] S50E: Based on the load influence function, the discrete increments for each observation interval are standardized to obtain the standardized degradation increments:

[0080]

[0081] in, This represents the standardized degradation increment of device i in the k-th observation interval. Indicates device i in t k Time, load The cumulative degradation level below, Indicates device i in t k Time, load The cumulative degradation level below, This indicates the load condition of the device at time s. This represents the load effect function.

[0082] S50F: Calculate the degradation rate parameters of critical power grid equipment based on the expected properties of the standardized degradation increment. :

[0083]

[0084] in, The parameter represents the estimated degradation rate parameter of device i, k represents the observation interval index, and h represents the value of the parameter. i This indicates the number of observation intervals for device i. This represents the time length of the k-th observation interval.

[0085] S50G: Calculates degradation fluctuation intensity parameters based on residual statistics after degradation rate parameter estimation.

[0086]

[0087] in, This represents the estimated value of the degradation fluctuation intensity parameter for device i. This represents the time length of the k-th interval.

[0088] S50H: Calculate the initial degradation level based on the historical degradation trajectory of power grid equipment.

[0089] S50I: Based on the state category and load level, statistical analysis is performed on samples of initial degradation level, degradation rate parameters and degradation fluctuation intensity parameters of different key power grid equipment to obtain a family of prior parameters.

[0090] S50J: In continuous time, by combining the family prior parameter set and stochastic differential equations, the loading effect is embedded into the drift and diffusion terms of the degradation model to establish a stochastic degradation model related to the loading conditions.

[0091]

[0092]

[0093]

[0094] in, This indicates that device i is affected by external operating conditions at time t. The cumulative level of degradation caused by the impact r represents the initial degradation parameter or fixed effect term of device i. i ( ) represents the degradation rate function of device i. This represents the load condition of the device at time s, where d represents the differential symbol, and v... i ( ) represents the degradation fluctuation intensity function of device i, and W(s) represents the unpredictable random disturbance or noise source of the degradation signal at time s. The parameter representing the inherent degradation rate of device i. This represents the degradation fluctuation variance when there is no load effect.

[0095] Among them, stochastic differential equations are used to describe the mathematical model of the degradation process over time, combining drift terms (average degradation trend) and diffusion terms (fluctuation intensity) to characterize the randomness of degradation.

[0096] It should be noted that, firstly, by stratifying the equipment sets under different operating conditions and forming a homogeneous sample set, the structural differences between samples are effectively eliminated, making the model training more representative and stable. Secondly, by calculating the discrete degradation increment and performing normalization and standardization, a nonlinear mapping relationship between load level and degradation rate is established, quantifying the impact of load on equipment aging. Furthermore, by introducing degradation rate parameters and fluctuation intensity parameters, the model can describe both the average degradation trend and the fluctuation behavior under random disturbances, thus achieving higher fitting accuracy and prediction robustness. Finally, by embedding the load effect into the drift and diffusion terms through stochastic differential equations, dynamic modeling and uncertainty quantification of the degradation process are achieved. This method overcomes the limitations of traditional fixed degradation assumptions, making equipment life prediction more consistent with actual operating conditions and providing a scientific mathematical foundation for subsequent health assessment and maintenance optimization.

[0097] In one possible implementation, the process after S5 and before S6 includes:

[0098] Based on the state data, the stochastic degradation model is updated using a Bayesian framework.

[0099] It's important to note that by introducing a Bayesian framework to dynamically update the stochastic degradation model, the model transitions from static estimation to adaptive learning. Traditional degradation models are typically built upon fixed historical data, and their parameters cannot be adjusted in a timely manner to adapt to environmental changes during equipment operation, leading to a gradual deviation of lifespan predictions from reality. However, through the Bayesian inference mechanism, the system can use real-time collected state data as new evidence, fusing it with prior model parameters to obtain a posterior distribution reflecting the current health status, thereby achieving parameter self-updating and uncertainty quantification. This method enables the model to maintain high prediction accuracy and robustness even when facing load fluctuations, environmental changes, or abnormal disturbances.

[0100] S6: Predict the remaining lifespan of critical equipment using a stochastic degradation model and generate an equipment health status vector.

[0101] Among them, remaining lifetime prediction refers to the calculation of the time distribution between the current state of the equipment and the failure threshold based on the degradation model, which is used to assess how long the equipment can still operate safely.

[0102] Among them, the equipment health status vector is a quantitative expression of the equipment's operational health status, which includes multi-dimensional information such as lifespan distribution parameters, survival probability, and failure risk rate, and is used to comprehensively assess the equipment's health level and risk level.

[0103] It is important to note that quantitatively predicting the remaining lifespan of critical equipment using a stochastic degradation model enables a shift from passive monitoring to proactive prediction, significantly improving the intelligence level of power grid operation and maintenance. Traditional methods often rely on experience or thresholds to judge equipment health, making it difficult to accurately reflect the uncertainty and randomness of degradation. This step, however, utilizes the updated stochastic degradation model from the previous step to comprehensively consider the degradation rate, fluctuation intensity, and load impact of the equipment, obtaining a probability distribution of lifespan. This allows for the prediction of equipment failure probabilities and survival functions at different time scales. By generating equipment health state vectors, the system can express the health status of multiple devices under a unified data structure, enabling horizontal comparisons and vertical trend analysis. This process not only improves the accuracy and interpretability of lifespan prediction but also provides a quantitative decision-making basis for subsequent maintenance optimization and resource scheduling.

[0104] In one possible implementation, S6 specifically includes:

[0105] S601: Calculate the equivalent degradation time of key power grid equipment based on the updated stochastic degradation model.

[0106]

[0107] in, This represents the equivalent degradation time at time t. This represents the load effect function, where s represents the time variable.

[0108] Among them, the equivalent degradation time is the "degradation equivalent time" obtained by weighting the operating time of the equipment under different load conditions, and is used to uniformly measure the cumulative aging effect under different operating conditions.

[0109] S602: Calculate the scale and shape parameters of the inverse Gaussian distribution based on the updated stochastic degradation model.

[0110] Among them, the Inverse Gaussian Distribution (IG) is a probability distribution commonly used for lifetime data modeling, which can accurately describe the time distribution characteristics of equipment reaching the failure threshold during random degradation.

[0111] Among them, the scale parameter and the shape parameter are two core parameters of the inverse Gaussian distribution, which respectively control the average level and dispersion of the lifetime distribution, reflecting the expected and uncertain aspects of the equipment lifetime.

[0112] S603: Substitute the shape parameters, scale parameters, and equivalent degradation time into the cumulative distribution function of the inverse Gaussian distribution to determine the probability of equipment failure, i.e., the lifetime distribution parameters:

[0113]

[0114] Where P() represents the probability of failure. Let represent the remaining lifespan of the i-th critical device at time t, where t represents the time index. This represents the drift parameter of the random degradation model. This represents the load condition of device i at time t. Represents the time transformation function. A function representing the load severity of device i at time s. This represents an inverse Gaussian distribution, i.e., This represents the cumulative load impact time of the i-th critical device at time t, i.e., the transition time. This represents the scale parameter of device i. This represents the shape parameters of device i. Let CDF represent the standard normal distribution, and exp() represent the exponential function.

[0115] It should be noted that this method first maps operating conditions to equivalent time τ(t) using the load influence function Ψ(γ), achieving a unified measure of degradation rate under different load conditions. Subsequently, based on the updated stochastic degradation model, the inverse Gaussian distribution parameters are estimated, enabling the model to reflect both the average degradation trend and the fluctuations and uncertainties, thus more closely approximating the true lifespan distribution of the equipment. By substituting these parameters into the cumulative distribution function, the failure probability at any given time can be quantitatively calculated, providing a mathematical basis for risk prediction. This method achieves a closed-loop connection from degradation process modeling to lifespan probability assessment, not only improving the accuracy and reliability of remaining life prediction but also providing interpretable quantitative indicators for risk classification and maintenance decisions of power grid equipment.

[0116] S7: Based on the equipment health status vector, construct a load-dependent predictive maintenance collaborative optimization model with predictive maintenance strategy as the core, and generate operation and maintenance strategies through the load-dependent predictive maintenance collaborative optimization model.

[0117] Among them, predictive maintenance strategy is an intelligent operation and maintenance strategy that plans maintenance time and methods in advance based on the health status and life prediction results of equipment. It aims to perform maintenance before equipment failure, thereby reducing the rate of sudden failures and operation and maintenance costs.

[0118] Among them, the load-dependent predictive maintenance collaborative optimization model is an optimization model that integrates equipment health information and system operating conditions. It takes into account the timing of equipment maintenance, load allocation and system operating economy in a unified manner, and achieves globally optimal maintenance and scheduling decisions through optimization algorithms.

[0119] Among them, collaborative optimization decision refers to the comprehensive decision results obtained through model solving, including the optimal maintenance time, maintenance sequence, load allocation scheme and system resource coordination strategy of the equipment, so as to realize the collaboration between the operation and maintenance layer and the scheduling layer.

[0120] It should be noted that, using the equipment health state vector as the core input, and leveraging a load-dependent predictive maintenance collaborative optimization model, this method integrates equipment remaining lifespan, maintenance costs, system power balance, and operational economy into a single optimization framework, generating optimal collaborative decisions through multi-objective solutions. This approach can achieve optimal allocation of maintenance resources and minimize operation and maintenance costs while ensuring grid security. More importantly, it supports rolling updates and dynamic adjustments, enabling the system to possess adaptive optimization capabilities. This significantly improves the intelligence and economy of grid operation and maintenance, providing key technical support for realizing a new grid management model of "state-driven, collaborative optimization."

[0121] In one possible implementation, S7 specifically includes:

[0122] S701: Calculate the dynamic maintenance cost function based on the remaining lifetime distribution parameters for maintenance at different future times:

[0123]

[0124] in, Indicates the time i of device i at time t and the equivalent degradation time. The expected maintenance cost rate, i.e., the dynamic maintenance cost function, This represents the planned maintenance cost of device i. This represents the failure and maintenance cost of device i. This represents the survival probability, i.e., the remaining lifetime of device i. Greater than the equivalent degradation time The probability, This represents the probability of failure, i.e., the remaining lifetime of device i. Less than or equal to the equivalent degradation time The probability, z represents the time from 0 to the equivalent degradation time. A time coordinate for traversal.

[0125] S702: Based on the dynamic maintenance cost function, construct the objective function of the load-dependent predictive maintenance collaborative optimization model with the goal of minimizing the total system cost.

[0126] Specifically, based on the dynamic maintenance cost function, the predictive maintenance costs of each key piece of equipment in different time periods are summed. Secondly, the output loss caused by equipment maintenance downtime, the power generation dispatch cost of the system (including voltage regulation, power up / down adjustment and reserve capacity), and the DC and AC loss costs in the transmission network are all included in the overall optimization objective. Under the effect of weighting coefficients, the economic efficiency of maintenance and the economic efficiency of system operation are balanced to minimize the total cost of the entire system.

[0127] It should be noted that it organically integrates the health status of the equipment layer with the operation and scheduling of the system layer, realizing multi-layer optimization of "state-driven and cost-constrained": it avoids the resource waste caused by traditional fixed-cycle maintenance, and takes into account both operational safety and economy; through the optimization solution of a unified objective function, it can dynamically determine when to maintain, which equipment to maintain, and how to schedule the system, thereby significantly reducing the total operation and maintenance cost while ensuring the reliability of the power grid, and realizing truly intelligent and collaborative power grid operation and maintenance management.

[0128] S703: Determine the constraint set of the load-dependent predictive maintenance collaborative optimization model based on the power grid system operation constraints and maintenance logic.

[0129] S704: Under the constraints of the constraint set, with the objective function as the goal, the load dependency prediction and maintenance collaborative optimization model is solved by the rolling time domain method to obtain the operation and maintenance strategy.

[0130] In this embodiment of the invention, firstly, a dynamic maintenance cost rate is established using remaining lifetime distribution parameters. This quantifies the economic risks of maintenance at different times, shifting maintenance decisions from static experience to data-driven, refined optimization. Secondly, by constructing a load-dependent predictive maintenance collaborative optimization model, equipment health, load changes, system energy consumption, and economic constraints are comprehensively incorporated into a unified objective function, achieving the optimization goal of "state-driven, cost minimization." This model can automatically balance preventative maintenance and system operating economy under different operating conditions, avoiding resource waste and failure risks caused by over-maintenance and delayed maintenance. Furthermore, by combining the rolling time-domain method for solving, the system possesses dynamic response capabilities, continuously adjusting maintenance and scheduling strategies based on real-time status. Overall, this step achieves global optimization and dynamic collaboration in power grid operation and maintenance, significantly improving the system's economy, reliability, and adaptive intelligence level.

[0131] In one possible implementation, the constraint set includes: maintenance operation constraints, unit scheduling constraints, and load dependency degradation constraints.

[0132] The specific constraints on maintenance operations include: ensuring that each critical power generation equipment is only subjected to maintenance operation once during the entire dispatch cycle, and that the maintenance time is reasonably selected within the remaining maintenance cycle to avoid resource waste caused by repeated maintenance or premature maintenance.

[0133] Ensure that the number of maintenance tasks performed simultaneously by the system does not exceed the upper limit of available labor or maintenance resources within any given time period, so as to achieve dynamic matching between operation and maintenance tasks and human resources.

[0134] The specific constraints of unit scheduling include: ensuring that when the equipment is under maintenance, its start-stop variables remain closed within the corresponding period to avoid timing conflicts between maintenance and operation, so as to ensure the safety of maintenance operations and the consistency of operation scheduling.

[0135] Ensure that the unit can resume operation according to the start-stop logic after maintenance, and meet the minimum start-stop time, ramp rate and energy balance requirements, thereby maintaining the overall stable operation and power balance of the system.

[0136] The load-dependent degradation constraints include ensuring that maintenance time is selected within the load-weighted transformation time domain and dynamically corresponds to the operating load level of the equipment, thereby achieving adaptive matching of maintenance timing with degradation rate.

[0137] Ensure that when the equipment is under maintenance, its operating load automatically drops to zero, and that different load level variables maintain a monotonic relationship to avoid inconsistencies between load status and maintenance status.

[0138] In one possible implementation, S704 specifically includes:

[0139] S704A: Initializes the integrated health status vector and maintenance parameters of key power grid equipment.

[0140] The comprehensive health status vector includes: age after time transformation, current degradation, dynamic maintenance cost, and remaining maintenance time. The maintenance parameters include: total system maintenance cost, unit operating cost, predictive maintenance count, and fault count.

[0141] S704B: Sets the total number of iterations, window length, and scroll step size for the rolling time domain.

[0142] S704C: Solve the load dependency prediction and maintenance collaborative optimization model to obtain the optimal maintenance and load allocation strategy.

[0143] S704D: Based on the optimal maintenance and load distribution strategy, within the window length, the degradation of key power grid equipment is updated based on the load level, and random drift and diffusion terms are calculated.

[0144] S704E: Based on the updated degradation amount, random drift term, and diffusion term, determine whether the critical equipment of the power grid has reached the maintenance threshold; if yes, proceed to step S704F; otherwise, continue to operate the critical equipment of the power grid and proceed to step S704G.

[0145] S704F: Update the availability status and maintenance costs of critical power grid equipment.

[0146] The availability status is down.

[0147] S704G: Based on the current load level, update the time transformation variable and degradation signal, and determine whether the updated degradation signal is greater than or equal to the failure threshold. If so, determine that the critical power grid equipment has entered a failure state and proceed to step S704H. Otherwise, repeat steps S704D-S704F.

[0148] S704H: Triggers unplanned maintenance operations, records the failure maintenance costs of critical power grid equipment, and updates equipment health parameters and availability status.

[0149] S704I: Solve the unit operation scheduling optimization model based on the updated equipment health parameters and availability status.

[0150] S704J: Updates system operating parameters and the rolling time domain window based on the solution results.

[0151] S704K: Within the updated rolling time window, repeat steps 704C–S704H until the set total number of iterations for the rolling time is reached, and output the collaborative optimization decision.

[0152] It should be noted that, firstly, the initialization of the integrated health state vector and maintenance parameters enables the system to achieve global observability, allowing for a unified assessment of equipment operating status across time, economy, and health dimensions. Secondly, the rolling time-domain method, through phased optimization and real-time updates, enables maintenance and scheduling strategies to dynamically adjust based on equipment degradation changes, load fluctuations, and system constraints, thus maintaining the timeliness and flexibility of decision-making. By updating drift and diffusion terms within a window, the model can correct degradation trends and uncertainties in real time, achieving accurate tracking of equipment status. Furthermore, through a dual judgment mechanism of maintenance and failure thresholds, the system can proactively trigger predictive maintenance or emergency repairs, significantly reducing the risk of unplanned downtime. Finally, this process achieves collaborative linkage between the operation and maintenance layer and the scheduling layer, enabling the entire power grid to achieve optimal economy, highest reliability, and continuous self-learning optimization of maintenance strategies in long-term operation, possessing truly intelligent closed-loop management capabilities.

[0153] S8: Distribute operation and maintenance strategies to the execution unit and monitor device status changes and decision execution effects in real time through the IoT platform.

[0154] The execution unit refers to the physical unit that specifically executes maintenance, repair, and dispatch commands, including intelligent control terminals, automated actuators, dispatch center control systems, and field maintenance equipment.

[0155] Among them, the Internet of Things (IoT) platform is a communication and data management platform that enables device interconnection, data sharing and remote control. Through the collaboration of sensors, edge nodes and cloud systems, it enables real-time monitoring of device status and closed-loop feedback of commands.

[0156] Among them, the decision execution effect refers to the actual operational results generated after the collaborative optimization decision is issued, including evaluation indicators such as equipment operation reliability, maintenance completion rate and system economy.

[0157] It should be noted that by distributing collaborative optimization decisions to the execution units, the system can automatically trigger corresponding maintenance and scheduling operations, greatly improving the timeliness and efficiency of operation and maintenance response. Leveraging the IoT platform, equipment operating status and maintenance processes can be monitored in real time, ensuring that the execution results of decisions are visualized and traceable. This process forms a closed-loop mechanism of "perception—decision—execution—feedback," enabling the optimization model to continuously correct and learn based on real-time data. Compared to traditional manual scheduling, this step significantly reduces human intervention and communication delays, improving the system's automation and intelligence levels.

[0158] S9: Based on changes in equipment status and the effectiveness of decision execution, dynamically update the fusion convolutional network and operation and maintenance strategies to achieve self-learning and closed-loop management of intelligent operation and maintenance of the power grid.

[0159] It should be noted that by acquiring real-time data on equipment status changes and decision execution effects, the fused convolutional network undergoes parameter retraining and feature relearning. This enables the model to adapt to different operating conditions, environments, and degradation features, maintaining high accuracy and generalization ability. Simultaneously, the system dynamically adjusts its operation and maintenance strategies based on execution feedback, achieving iterative optimization of maintenance plans and scheduling strategies. This mechanism tightly couples the perception layer, decision layer, and execution layer, forming a data-driven closed-loop management system. Through continuous learning and optimization, the power grid can maintain optimal operation and maintenance status under varying operating environments, significantly improving equipment lifespan utilization, operational reliability, and system economy, ultimately achieving truly intelligent and autonomous operation and maintenance management.

[0160] In this embodiment of the invention, a relational autoencoder model is used to perform feature filtering and relational constraint modeling on multi-source state data, which can accurately extract health features related to lifespan and improve the relevance and interpretability of feature representation. By fusing convolutional networks and weighted voting mechanisms, high-precision identification of equipment operating status can be achieved in a multi-dimensional feature space. Through a load-dependent stochastic degradation model and a Bayesian dynamic update mechanism, the degradation process and remaining lifespan of the equipment can be accurately predicted, effectively addressing performance fluctuations under different load and environmental conditions. Furthermore, by constructing a load-dependent predictive maintenance collaborative optimization model, joint optimization of equipment-level health status and system-level scheduling decisions is achieved, significantly reducing system operation and maintenance costs and the risk of unplanned downtime.

[0161] Reference manual attached Figure 2 The diagram shows a structural schematic of a smart power grid operation and maintenance management system provided in an embodiment of the present invention.

[0162] This invention provides a smart power grid operation and maintenance management system 20, including: a processor 201 and a memory 202;

[0163] The memory 202 stores programs or instructions that can run on the processor 201. When the program or instructions are executed by the processor 201, they implement the steps of the above-mentioned smart grid operation and maintenance management method and achieve the same technical effect. To avoid repetition, the present invention will not elaborate further.

[0164] It should be understood that the processor 201 in this embodiment of the invention may be a central processing unit (CPU), or it may be 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. The general-purpose processor may be a microprocessor or any conventional processor.

[0165] It should also be understood that the memory 202 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM).

[0166] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0167] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0168] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0169] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0170] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0171] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0172] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0173] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0174] This invention provides a readable storage medium comprising: storing a program or instructions on the readable storage medium, wherein when the program or instructions are executed by a processor, the program or instructions implement the steps of the above-described intelligent power grid operation and maintenance management method and achieve the same technical effect. To avoid repetition, this invention will not elaborate further.

[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent operation and maintenance management of power grids, characterized in that, include: S1: Obtain status data of key power grid equipment; S2: By using a relational autoencoder model, feature filtering is performed on the state data to determine health characteristics related to lifespan; S3: The health features are fused using a fusion convolutional network to generate a fused feature sequence; S4: Based on the fused feature sequence, the key equipment is classified according to its operating status through a weighted voting mechanism; S5: Based on the classification results of operating status and combined with historical data of key equipment, construct a stochastic degradation model related to the load conditions of key equipment; S6: Using the stochastic degradation model, the remaining lifespan of the key equipment is predicted, and an equipment health status vector is generated; The device health status vector includes: lifetime distribution parameters, survival probability, and failure risk rate; S7: Based on the device health status vector, construct a load-dependent predictive maintenance collaborative optimization model with predictive maintenance strategy as the core, and generate operation and maintenance strategy through the load-dependent predictive maintenance collaborative optimization model; S8: The operation and maintenance strategy is sent to the execution unit, and the device status changes and decision execution effects are monitored in real time through the Internet of Things platform; S9: Based on the changes in the device status and the effect of the decision execution, dynamically update the fused convolutional network and the operation and maintenance strategy to realize self-learning and closed-loop management of intelligent operation and maintenance of the power grid.

2. The intelligent power grid operation and maintenance management method according to claim 1, characterized in that, The status data includes: sensor data, environmental data, and leakage current data.

3. The intelligent power grid operation and maintenance management method according to claim 1, characterized in that, S2 specifically includes: S201: Using the state data as input, construct an autoencoder model that includes the relationship between the encoder and the decoder; S202: Calculate the relationship matrix between the data in the state data using a relational autoencoder model; S203: Calculate the key feature matrix of the state data based on the encoder in the relational autoencoder model and the nonlinear mapping function; S204: Based on the key feature matrix, the state data and the corresponding relation matrix are reconstructed using the decoder in the relation autoencoder model; S205: Based on the reconstructed state data and relation matrix, determine the objective function based on relation preservation constraints; S206: The health feature is generated by minimizing the objective function value.

4. The intelligent power grid operation and maintenance management method according to claim 1, characterized in that, S5 specifically includes: S50A: Based on the classification results of the operating status, the set of key power grid equipment under different operating conditions is layered, and the key power grid equipment is sliced ​​according to the load level to form a homogeneous sample set. S50B: In the homogeneous sample set, a discrete increment is defined for the degradation signal of each critical device, and the average increment under the reference load is used as the normalization factor. S50C: Normalize the discrete increment according to the normalization factor; S50D: Based on the normalized discrete increment, the load influence function is determined by fitting the multiplicative relationship between the load level and degradation rate of key equipment through nonlinear regression. S50E: Based on the load influence function, the discrete increment of each observation interval is standardized to obtain the standardized degradation increment; S50F: Calculate the degradation rate parameters of the key power grid equipment based on the expected properties of the standardized degradation increment; S50G: Calculate the degradation fluctuation intensity parameter based on the residual statistics after estimating the degradation rate parameter; S50H: Calculate the initial degradation level based on the historical degradation trajectory of the power grid equipment; S50I: Based on the state category and load level, perform statistical analysis on the initial degradation level, degradation rate parameter and degradation fluctuation intensity parameter samples of different key power grid equipment to obtain a family prior parameter set; S50J: In continuous time, by combining the family of prior parameters and stochastic differential equations, the load effect is embedded into the drift and diffusion terms of the degradation model to establish a stochastic degradation model related to the load conditions.

5. The intelligent power grid operation and maintenance management method according to claim 1, characterized in that, The process includes the following steps after S5 and before S6: Based on the state data, the random degradation model is updated using a Bayesian framework.

6. The intelligent power grid operation and maintenance management method according to claim 5, characterized in that, S6 specifically includes: S601: Calculate the equivalent degradation time of key power grid equipment based on the updated stochastic degradation model; S602: Based on the updated stochastic degradation model, calculate the scale and shape parameters of the inverse Gaussian distribution: S603: Substitute the shape parameter, the scale parameter, and the equivalent degradation time into the cumulative distribution function of the inverse Gaussian distribution to determine the probability of equipment failure, i.e., the lifetime distribution parameter.

7. The intelligent power grid operation and maintenance management method according to claim 1, characterized in that, Specifically, S7 includes: S701: Calculate the dynamic maintenance cost function for maintenance at different future times based on the remaining lifetime distribution parameters; S702: Based on the dynamic maintenance cost function, construct the objective function of the load-dependent predictive maintenance collaborative optimization model with the goal of minimizing the total system cost; S703: Determine the constraint set of the load-dependent predictive maintenance collaborative optimization model based on the power grid system operation constraints and maintenance logic; S704: Under the constraints of the constraints in the constraint set, with the objective function as the goal, the load dependency prediction and maintenance collaborative optimization model is solved by the rolling time domain method to obtain the operation and maintenance strategy.

8. The intelligent power grid operation and maintenance management method according to claim 7, characterized in that, The constraint set includes: maintenance operation constraints, unit scheduling constraints, and load dependency degradation constraints.

9. The intelligent power grid operation and maintenance management method according to claim 7, characterized in that, Specifically, S704 includes: S704A: Initialize the comprehensive health status vector and maintenance parameters of the key power grid equipment; S704B: Sets the total number of iterations, window length, and scroll step size for the rolling time domain; S704C: Solve the load dependency prediction and maintenance collaborative optimization model to obtain the optimal maintenance and load allocation strategy; S704D: Based on the optimal maintenance and the load allocation strategy, within the window length, update the degradation of the key power grid equipment based on the load level, and calculate the random drift term and diffusion term; S704E: Based on the updated degradation amount, the random drift term, and the diffusion term, determine whether the key power grid equipment has reached the maintenance threshold; if yes, proceed to step S704F; otherwise, continue operating the key power grid equipment and proceed to step S704G. S704F: Update the availability status and maintenance costs of the aforementioned critical power grid equipment; S704G: Based on the current load level, update the time transformation variable and degradation signal, and determine whether the updated degradation signal is greater than or equal to the failure threshold; if so, determine that the key power grid equipment has entered a failure state and proceed to step S704H; otherwise, repeat steps S704D-S704F. S704H: Triggers unplanned maintenance operations, records the failure maintenance costs of critical power grid equipment, and updates equipment health parameters and availability status; S704I: Solve the unit operation scheduling optimization model based on the updated equipment health parameters and availability status; S704J: Update system operating parameters and rolling time domain window based on solution results; S704K: Within the updated rolling time-domain window, repeat steps 704C–S704H until the set total number of iterations for the rolling time-domain is reached, and output the collaborative optimization decision.

10. A smart power grid operation and maintenance management system, characterized in that, include: Processor and memory; The memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of the smart grid operation and maintenance management method as described in any one of claims 1 to 9.

Images