A substation equipment condition maintenance auxiliary decision-making system and method

Abstract

The present invention provides a system and method for auxiliary decision-making for state maintenance of substation equipment, which belongs to the technical field of intelligent maintenance. The system includes a first feature module, a second feature module, a fault diagnosis module and a maintenance module. The first feature set is generated by collecting the best operation data and historical fault records of the substation equipment. Based on the parameter changes of the equipment state, relevant features are dynamically selected to form a second feature set. The two are used to build a fault diagnosis model, predict potential fault risks, and generate corresponding maintenance strategies and plans. Finally, intelligent maintenance and auxiliary decision-making are realized by simulating and optimizing maintenance operations, adapting to complex fault modes, and improving the safety and reliability of substation equipment operation.

Description

Technical Field

[0001] The present invention relates to the field of intelligent maintenance technology, and in particular to a state maintenance auxiliary decision-making system and method for substation equipment. Background Art

[0002] With the continuous development of power systems and the improvement of their intelligence level, the operational safety and reliability of substation equipment has become an important research topic. Traditional maintenance methods often rely on regular inspections and empirical judgments, and lack effective feature extraction and correlation analysis, making it difficult to adapt to rapidly changing equipment status and complex failure modes.

[0003] Therefore, the present invention provides a system and method for auxiliary decision-making for status maintenance of substation equipment. Summary of the invention

[0004] The present invention provides a substation equipment status maintenance decision-making support system and method, which generates a first feature set by collecting the best operating data and historical fault records of the substation equipment, and dynamically selects relevant features based on parameter changes of the equipment status to form a second feature set. The two are used to build a fault diagnosis model, predict potential fault risks, and generate corresponding maintenance strategies and plans. Finally, by simulating and optimizing maintenance operations, intelligent maintenance and decision-making support are achieved, which adapts to complex failure modes and improves the safety and reliability of substation equipment operation.

[0005] The present invention provides a substation equipment condition maintenance auxiliary decision-making system, comprising:

[0006] The first feature module collects the best operation data and historical fault records of the substation equipment, analyzes the historical fault records, defines the fault type and associates the best operation data to form an associated data set, extracts key features from the key data set, and generates a first feature set;

[0007] A second feature module: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set;

[0008] Fault diagnosis module: constructs a fault diagnosis model using the first feature set and the second feature set to perform fault prediction and obtain the potential fault risk corresponding to the current equipment state of the substation equipment;

[0009] Maintenance module: Generates maintenance strategies based on the potential failure risks, derives maintenance plans based on the current equipment status of the substation equipment, simulates the maintenance plans, outputs actual maintenance operations based on the simulation results, and assists decision-making on maintenance of the substation equipment status.

[0010] Preferably, the first feature module includes:

[0011] Associated data unit: defines the fault type and the corresponding annotation information according to the historical fault records of the substation equipment, and associates the collected optimal operation data of the substation equipment with the fault type according to the annotation information to form an associated data set;

[0012] Feature extraction unit: obtains equipment characteristics of the substation equipment, and selects a corresponding feature extraction method from a feature extraction table based on the equipment characteristics of the substation equipment and the fault type, and uses the feature extraction method to extract key features from the associated data set to obtain a first feature set.

[0013] Preferably, the second feature module includes:

[0014] Equipment status unit: Based on the first feature set, a sliding window is used to detect parameter changes of the substation equipment in real time, identify the time point of parameter changes, and a classification algorithm is used to distinguish the equipment status in real time based on the features in the first feature set to obtain the current equipment status;

[0015] Related feature unit: calculates the degree of influence of parameter changes on the degree of dependency between each feature in the first feature set and the current device state, screens the first feature set based on the influence degree, obtains related features, and then obtains the second feature set.

[0016] Preferably, the relevant characteristic units include:

[0017] Compute block:

[0018] ,in, Indicates the degree of dependency between feature X in the first feature set and the current device state Y; Indicates the degree to which the dependency is affected by parameter changes; X represents feature X in the first feature set; Y represents the current device state; x represents the value of feature X; y Indicates the value of the current device state Y; represents the marginal probability density function of feature X; Represents the marginal probability density function of the current device state Y; Represents the joint probability density function of feature X and device state Y; Represents parameter changes that affect the relationship between features and device states;

[0019] Related feature block: dynamically selects related features corresponding to the current device state according to the degree of influence, and then adjusts the first feature set to obtain the second feature set.

[0020] Preferably, the fault diagnosis module includes:

[0021] The first diagnostic model unit: determines a fault diagnosis algorithm based on the inspection of the state of the substation equipment, and uses the first feature set to train the fault diagnosis algorithm to generate a first-level diagnostic model;

[0022] The second diagnostic model unit: cross-validates the primary diagnostic model according to the evaluation index generated by the maintenance of the substation equipment status to obtain a cross-validation result, and makes a first adjustment to the primary diagnostic model based on the cross-validation result to obtain a secondary diagnostic model;

[0023] Hyperparameter adjustment unit: train the secondary diagnostic model using the second feature set, determine the hyperparameters that need to be tuned corresponding to the secondary diagnostic model and the value range of each hyperparameter based on the model-hyperparameter table, and then form a hyperparameter space, traverse the hyperparameter space using a grid search method, extract the best hyperparameter combination, and use the best hyperparameter combination to perform a second adjustment on the secondary diagnostic model to obtain a fault diagnosis model;

[0024] Fault output unit: performs fault prediction according to the fault diagnosis model, and then obtains the potential fault risk of the current equipment status.

[0025] Preferably, the maintenance module comprises:

[0026] Priority unit: matches the failure mode corresponding to the potential failure risk from the risk-mode table, constructs a risk matrix according to the failure mode, and calculates the risk priority number of each failure mode based on the risk matrix;

[0027] Mode determination unit: divides the risk priority numbers based on the risk priority values ​​to obtain the first level risk failure mode, the second level risk failure mode and the third level risk failure mode;

[0028] An association unit: associates the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode with specific components corresponding to the power transformation equipment, and identifies the affected components;

[0029] Maintenance planning unit: determining corresponding maintenance modes for the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode, formulating maintenance strategies based on the affected components, the current equipment status, and the maintenance modes, and determining maintenance plans based on the maintenance strategies;

[0030] Practical operation unit: simulates the maintenance plan, evaluates the feasibility of the maintenance plan based on the simulation result, and performs actual maintenance operation on the maintenance plan with a feasibility higher than the preset feasibility.

[0031] Preferably, the practical operation unit includes:

[0032] Simulation block: construct a simulation scenario according to the maintenance plan, perform operations according to the steps of the maintenance plan, and record the execution status and corresponding execution time of each step;

[0033] Judgment block: Evaluate the feasibility of the maintenance plan according to the execution status and execution time, compare the feasibility of the maintenance plan with the preset feasibility, if the feasibility corresponding to the maintenance plan is higher than the preset feasibility, perform actual maintenance operations, if the feasibility corresponding to the maintenance plan is equal to or lower than the preset feasibility, identify potential risk points, adjust the maintenance plan based on the potential risk points, and perform secondary simulation on the adjusted maintenance plan.

[0034] The present invention provides a method for auxiliary decision-making of state maintenance of substation equipment, comprising:

[0035] Step 1: Collect the best operation data and historical fault records of the substation equipment, analyze the historical fault records, define the fault type and associate it with the best operation data to form an associated data set, extract key features from the key data set, and generate a first feature set;

[0036] Step 2: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set;

[0037] Step 3: Use the first feature set and the second feature set to build a fault diagnosis model, perform fault prediction, and obtain the potential fault risk corresponding to the current device state of the substation equipment;

[0038] Step 4: Generate a maintenance strategy based on the potential failure risk, derive a maintenance plan based on the current equipment status of the substation equipment, simulate the maintenance plan, output actual maintenance operations based on the simulation results, and assist in decision-making on maintenance of the substation equipment status.

[0039] Compared with the prior art, the beneficial effects of the present application are as follows: by collecting the best operating data and historical fault records of substation equipment, a first feature set is generated; based on the parameter changes of the equipment status, relevant features are dynamically selected to form a second feature set; these two are used to build a fault diagnosis model, predict potential fault risks, and generate corresponding maintenance strategies and plans; ultimately, through simulation and optimization of maintenance operations, intelligent maintenance and auxiliary decision-making are achieved, complex failure modes are adapted, and the safety and reliability of substation equipment operation are improved.

[0040] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0041] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0043] Figure 1 It is a structural diagram of a substation equipment condition maintenance auxiliary decision-making system provided by an embodiment of the present invention;

[0044] Figure 2 It is a flow chart of a method for auxiliary decision-making for status maintenance of substation equipment provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0046] Embodiment 1:

[0047] The embodiment of the present invention provides a substation equipment condition maintenance auxiliary decision-making system, such as Figure 1 As shown, including:

[0048] The first feature module collects the best operation data and historical fault records of the substation equipment, analyzes the historical fault records, defines the fault type and associates the best operation data to form an associated data set, extracts key features from the key data set, and generates a first feature set;

[0049] A second feature module: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set;

[0050] Fault diagnosis module: constructs a fault diagnosis model using the first feature set and the second feature set to perform fault prediction and obtain the potential fault risk corresponding to the current equipment state of the substation equipment;

[0051] Maintenance module: Generates maintenance strategies based on the potential failure risks, derives maintenance plans based on the current equipment status of the substation equipment, simulates the maintenance plans, outputs actual maintenance operations based on the simulation results, and assists decision-making on maintenance of the substation equipment status.

[0052] In this embodiment, the historical fault record is a record of fault events that occurred in the substation equipment in the past, including the time, type, cause and processing result of the fault. For example, a transformer had three faults in the past year, namely overheating, insulation aging and short circuit.

[0053] In this embodiment, the fault type is a classification of different faults in the historical fault records, such as mechanical fault, electrical fault, etc. For example, the fault types of the substation equipment include "overload", "short circuit" and "insulation failure".

[0054] In this embodiment, the optimal operating data is the equipment performance data collected under normal and efficient operating conditions, such as current, voltage, temperature, etc. For example, under normal operating conditions, the current of the transformer is 500A and the temperature is maintained at 70°C.

[0055] In this embodiment, the association process is to match the optimal operating data with the fault type to identify the device status when the fault occurs. For example, the current (600A) recorded when the "overload" fault occurs is compared with the optimal operating data (500A) to form an association.

[0056] In this embodiment, the associated data set is a data set generated by combining the optimal operating data and the fault type, which is used for analysis and feature extraction. For example, it contains records of the current, temperature and corresponding optimal operating data when the "overload" fault occurs. More specifically, equipment type: transformer, rated voltage (kV): 110, rated power (MVA): 50, fault type: short circuit, operating status: abnormal, current (A) 600, voltage (V) 110000, temperature (°C): 70, annotation information: short circuit causes the equipment to stop operating and requires maintenance.

[0057] In this embodiment, the key features are features extracted from the data set that are of great significance to fault prediction and equipment status assessment, for example, the temperature mean and current standard deviation of the transformer are used as key features.

[0058] In this embodiment, the first feature set is a data set composed of key features, which is used for subsequent fault diagnosis and decision support, for example, a feature set including the transformer temperature mean, load current mean and historical fault frequency.

[0059] In this embodiment, the parameter changes of the substation equipment mainly refer to the changes in various technical parameters related to the operating status and performance of the equipment, including electrical parameter changes, temperature changes, mechanical parameter changes, operating time changes, environmental conditions changes, and maintenance record changes. For example, the area where the substation is located has experienced continuous humid weather, and the ambient humidity has increased from 50% to 80%, which may affect the insulation performance of the equipment.

[0060] In this embodiment, the relevant features are features with significant dependencies and influences during the screening process, and can effectively reflect the device status. For example, temperature and current are screened out as relevant features because they are highly correlated with the device status.

[0061] In this embodiment, the second feature set is a set of related features selected from the first feature set, focusing on features that have a significant impact on the current device state. For example, the second feature set may only include temperature and current, while the first feature set may include multiple features such as temperature, current, and vibration.

[0062] In this embodiment, the difference between the second feature set and the first feature set is that the first feature set includes all extracted features, which may include redundant or irrelevant features, and the second feature set includes features screened out based on dependencies and impact levels, focuses on features that are highly relevant to the device status, and is usually smaller and more targeted.

[0063] In this embodiment, the current device state is the device operating status obtained based on real-time data and classification algorithms, such as "normal", "overload", "fault", etc. For example, the transformer is judged to be in the "overload" state in real time.

[0064] In this embodiment, the potential failure risk is the probability or risk level of a device failure predicted by the fault diagnosis model. For example, the model predicts that the risk of a transformer failure within the next 30 days is 10%.

[0065] In this embodiment, the maintenance mode is a maintenance method or strategy formulated for a specific failure mode. For example, for "insulation damage", the maintenance mode may be "replace insulation material" or "repair insulation layer"; the maintenance strategy is a specific maintenance plan based on the status of affected components and equipment. For example, after the cable insulation is found to be damaged, the maintenance strategy may include "regularly monitor the insulation status" and "perform insulation inspection every quarter"; the maintenance plan is the operational steps and time schedule that concretizes the maintenance strategy. For example, all aging cables are scheduled to be inspected next month and replaced at the end of the quarter.

[0066] In this embodiment, the actual maintenance operation is a specific maintenance task performed after the feasibility assessment is passed, such as actually replacing or repairing the cable insulation layer.

[0067] In this embodiment, the simulation result refers to the result obtained after simulating the generated maintenance plan in the maintenance module.

[0068] The working principle and beneficial effects of the above technical solution are: by collecting the best operating data and historical fault records of substation equipment, a first feature set is generated, and based on the parameter changes of the equipment status, relevant features are dynamically selected to form a second feature set. These two are used to build a fault diagnosis model, predict potential fault risks, and generate corresponding maintenance strategies and plans. Finally, through simulation and optimization of maintenance operations, intelligent maintenance and auxiliary decision-making are achieved, which adapts to complex failure modes and improves the safety and reliability of substation equipment operation.

[0069] Embodiment 2:

[0070] Based on Example 1, the first feature module includes:

[0071] Associated data unit: defines the fault type and the corresponding annotation information according to the historical fault records of the substation equipment, and associates the collected optimal operation data of the substation equipment with the fault type according to the annotation information to form an associated data set;

[0072] Feature extraction unit: obtains equipment characteristics of the substation equipment, and selects a corresponding feature extraction method from a feature extraction table based on the equipment characteristics of the substation equipment and the fault type, and uses the feature extraction method to extract key features from the associated data set to obtain a first feature set.

[0073] In this embodiment, the annotation information is a detailed description of the fault type and related information, which is used to identify the characteristics and impact of the fault. For example, the annotation information is "overload fault, causing the device temperature to exceed the normal range, and immediate maintenance is required."

[0074] In this embodiment, the device characteristics describe the physical and functional characteristics of the substation equipment, such as rated power, operating temperature range, etc. For example, the rated power of a transformer is 2000 kVA, and the operating temperature range is -10°C to +40°C.

[0075] In this embodiment, the feature extraction table lists various feature extraction methods and their applicable conditions, which are used to extract key features from the data set. For example, the feature extraction table includes statistical feature extraction methods such as "mean", "variance", and "kurtosis".

[0076] In this embodiment, the feature extraction method is a specific algorithm or technology used to extract key features from a related data set, for example, using "mean" and "standard deviation" to extract key features of transformer temperature data.

[0077] The working principle and beneficial effect of the above technical solution are: by analyzing the historical fault records of the substation equipment, defining the fault type and associating the best operating data to form a related data set. Then, according to the equipment characteristics and fault type, a suitable method is selected from the feature extraction table to extract key features from the related data set to generate a first feature set, which provides basic data for subsequent fault diagnosis and maintenance decisions, improves the accuracy of fault identification, and enhances the flexibility of the system.

[0078] Embodiment 3:

[0079] Based on the first embodiment, the second characteristic module includes:

[0080] Equipment status unit: Based on the first feature set, a sliding window is used to detect parameter changes of the substation equipment in real time, identify the time point of parameter changes, and a classification algorithm is used to distinguish the equipment status in real time based on the features in the first feature set to obtain the current equipment status;

[0081] Related feature unit: calculates the degree of influence of parameter changes on the degree of dependency between each feature in the first feature set and the current device state, screens the first feature set based on the influence degree, obtains related features, and then obtains the second feature set.

[0082] In this embodiment, the sliding window is a data processing technology that analyzes data changes in real time by moving a fixed-size window on time series data. For example, the temperature and current data of the substation equipment are collected every 5 minutes to form a 5-minute sliding window for analysis.

[0083] In this embodiment, the classification algorithm is applied to classify the device status using a machine learning algorithm, such as a decision tree, a support vector machine, etc. For example, a support vector machine is used to classify the operating data of the transformer to determine whether its status is "normal" or "faulty".

[0084] In this embodiment, the degree of dependency describes the correlation between the feature and the device state, which is usually calculated by statistical methods. For example, the correlation coefficient between temperature change and the "overload" state is 0.85, indicating a high correlation.

[0085] In this embodiment, the influence degree is the influence degree of the characteristic change on the device state, which can be obtained by regression analysis and other methods. For example, the influence degree of the current change on the device state is 0.7, indicating that the current change has a certain influence on the state.

[0086] The working principle and beneficial effects of the above technical solution are: real-time monitoring of parameter changes of substation equipment through a sliding window, identifying the time point of change, and applying a classification algorithm in combination with the first feature set to determine the current equipment status, calculating the dependency relationship between the features in the first feature set and the current equipment status, screening out features that are significantly affected by parameter changes, and forming a second feature set to provide a basis for subsequent fault warnings and maintenance decisions.

[0087] Embodiment 4:

[0088] Based on Example 3, the relevant characteristic units include:

[0089] Compute block:

[0090] ,in, Indicates the degree of dependency between feature X in the first feature set and the current device state Y; Indicates the degree to which the dependency is affected by parameter changes; X represents feature X in the first feature set; Y represents the current device state; x represents the value of feature X; y Indicates the value of the current device state Y; represents the marginal probability density function of feature X; Represents the marginal probability density function of the current device state Y; Represents the joint probability density function of feature X and device state Y; Represents parameter changes that affect the relationship between features and device states;

[0091] Related feature block: dynamically selects related features corresponding to the current device state according to the degree of influence, and then adjusts the first feature set to obtain the second feature set.

[0092] In this embodiment, the preset influence degree is compared with the influence degree, and the features with significant correlation with the current device state are screened to obtain relevant features. Assuming that the following features are included in the first feature set: temperature X1, vibration X2, current X3, voltage X4, after calculation, the following influence degrees are obtained: I(X1;Y)=0.6, I(X2;Y)=0.2, I(X3;Y)=0.8, I(X4;Y)=0.4. Assuming that the preset threshold is 0.5, the screening result is: retain features X1 (temperature) and X3 (current) because their influence degrees are greater than 0.5, discard features X2 (vibration) and X4 (voltage) because their influence degrees are less than or equal to 0.5. Finally, the second feature set screened out is: relevant feature set = {temperature X1, current X3}.

[0093] The working principle and beneficial effects of the above technical solution are: by calculating the degree of dependency between features and equipment status, combining the impact of parameter changes, dynamically selecting relevant features, optimizing the feature set to improve the accuracy of the classification algorithm, achieving real-time monitoring and discrimination of the status of substation equipment, achieving more efficient feature extraction and association analysis, and enhancing the safety and reliability of equipment operation.

[0094] Embodiment 5:

[0095] Based on the first embodiment, the fault diagnosis module includes:

[0096] The first diagnostic model unit: determines a fault diagnosis algorithm based on the inspection of the state of the substation equipment, and uses the first feature set to train the fault diagnosis algorithm to generate a first-level diagnostic model;

[0097] The second diagnostic model unit: cross-validates the primary diagnostic model according to the evaluation index generated by the maintenance of the substation equipment status to obtain a cross-validation result, and makes a first adjustment to the primary diagnostic model based on the cross-validation result to obtain a secondary diagnostic model;

[0098] Hyperparameter adjustment unit: train the secondary diagnostic model using the second feature set, determine the hyperparameters that need to be tuned corresponding to the secondary diagnostic model and the value range of each hyperparameter based on the model-hyperparameter table, and then form a hyperparameter space, traverse the hyperparameter space using a grid search method, extract the best hyperparameter combination, and use the best hyperparameter combination to perform a second adjustment on the secondary diagnostic model to obtain a fault diagnosis model;

[0099] Fault output unit: performs fault prediction according to the fault diagnosis model, and then obtains the potential fault risk of the current equipment status.

[0100] In this embodiment, the inspection and maintenance of the status of the power transformation equipment is to perform regular or irregular inspection and maintenance on the power transformation equipment to ensure its normal operation, for example, regularly checking the quality and temperature of the insulating oil of the transformer.

[0101] In this embodiment, the fault diagnosis algorithm is an algorithm for analyzing device status data, identifying and locating faults, for example, using a decision tree algorithm to determine whether a transformer has an overheating fault.

[0102] In this embodiment, the primary diagnosis model is a model trained based on a preliminary feature set and is used for preliminary fault judgment, for example, a simple random forest model that performs fault identification through temperature and vibration data.

[0103] In this embodiment, the evaluation index is a standard for measuring model performance, such as accuracy, recall rate, etc. For example, the F1-score is used to evaluate the performance of the substation equipment fault diagnosis model.

[0104] In this embodiment, cross-validation is a model validation method that reduces overfitting by dividing a data set into multiple subsets for training and validation. For example, the status data of the substation equipment is divided into five parts and a 5-fold cross-validation is performed. The cross-validation result is a model performance evaluation result obtained through cross-validation. For example, after cross-validation, the average accuracy of the model is 85%.

[0105] In this embodiment, the secondary diagnosis model is a more complex model that is adjusted based on the primary model and cross-validation results, for example, using a support vector machine (SVM) model that integrates more features and optimized parameters.

[0106] In this embodiment, the first adjustment process is a step of optimizing the primary diagnostic model, adjusting the model parameters based on the evaluation index, for example, adjusting the feature selection and weight in the model according to the cross-validation results.

[0107] In this embodiment, the model-hyperparameter table is a table that lists all hyperparameters in the model and their adjustable ranges. For example, for an SVM model, the hyperparameter table may include C (penalty parameter) and kernel function type.

[0108] In this embodiment, the value range is the range of values ​​that each hyperparameter can take, for example, the value range of C is [0.1, 10], and the kernel function type is {linear, RBF, polynomial}.

[0109] In this embodiment, the hyperparameter space is a space composed of combinations of all hyperparameter values. For example, for C and kernel function, the hyperparameter space may contain multiple combinations, such as (0.1, linear), (1, RBF), etc.

[0110] In this embodiment, the best hyperparameter combination is a hyperparameter combination that can achieve the best model performance after optimization. For example, the combination of C=1 and kernel function=RBF performs best on the validation set.

[0111] In this embodiment, the grid search method is a systematic hyperparameter optimization method, which finds the best combination by traversing all possible hyperparameter combinations. For example, grid search is used to traverse all combinations of C and kernel functions to evaluate the model performance of each combination.

[0112] In this embodiment, the second adjustment is a process of further optimizing the secondary diagnosis model after obtaining the best hyperparameter combination, for example, retraining the SVM model according to the best hyperparameter combination to improve accuracy.

[0113] In this embodiment, the fault diagnosis model is a final model after adjustment and optimization, and is used for real-time fault prediction. For example, the final SVM model can identify potential faults of substation equipment in real time.

[0114] The working principle and beneficial effects of the above technical solution are: by constructing a multi-level fault diagnosis model, using feature sets and cross-validation to continuously optimize model performance, and ultimately achieving accurate fault prediction and risk assessment of the status of substation equipment, improving the operating safety and reliability of substation equipment, reducing the limitations of traditional maintenance methods, improving the timeliness and accuracy of fault detection, and enhancing the real-time monitoring and analysis capabilities of equipment status.

[0115] Embodiment 6:

[0116] Based on the first embodiment, the maintenance module includes:

[0117] Priority unit: matches the failure mode corresponding to the potential failure risk from the risk-mode table, constructs a risk matrix according to the failure mode, and calculates the risk priority number of each failure mode based on the risk matrix;

[0118] Mode determination unit: divides the risk priority numbers based on the risk priority values ​​to obtain the first level risk failure mode, the second level risk failure mode and the third level risk failure mode;

[0119] An association unit: associates the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode with specific components corresponding to the power transformation equipment, and identifies the affected components;

[0120] Maintenance planning unit: determining corresponding maintenance modes for the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode, formulating maintenance strategies based on the affected components, the current equipment status, and the maintenance modes, and determining maintenance plans based on the maintenance strategies;

[0121] Practical operation unit: simulates the maintenance plan, evaluates the feasibility of the maintenance plan based on the simulation result, and performs actual maintenance operation on the maintenance plan with a feasibility higher than the preset feasibility.

[0122] In this embodiment, the risk-mode table is a table that records potential fault risks and their corresponding fault modes. For example, the potential fault risks include cable aging, transformer overheating, and grounding fault, and the corresponding fault modes include insulation damage, overload, cooling failure, and grounding wire breakage.

[0123] In this embodiment, the failure mode is a specific failure manifestation that may occur in the device under a specific risk. For example, for the risk of "cable aging", the failure mode may include aging of the insulation layer and current leakage.

[0124] In this embodiment, the risk matrix is ​​a tool for quantifying the fault modes, their probability of occurrence and the degree of impact, and is usually presented in the form of a two-dimensional table. For example, the fault modes include insulation damage, overload, and ground wire breakage, with high, medium, and low probability of occurrence, severe, high, and medium impact, and risk priority numbers of 15, 10, and 4.

[0125] In this embodiment, the risk priority number is a value calculated by the risk matrix, which indicates the priority of the failure mode and is usually the product of the probability of occurrence and the degree of impact. For example, for "insulation damage", if the probability of occurrence is "high" (value is 5) and the degree of impact is "serious" (value is 3), then the risk priority number is 5×3=15.

[0126] In this embodiment, the first level risk failure mode, the second level risk failure mode, and the third level risk failure mode are failure mode levels divided according to risk priority values. For example, the first level risk failure mode: insulation damage (risk priority number 15), the second level risk failure mode: overload (risk priority number 10), and the third level risk failure mode: ground wire breakage (risk priority number 4).

[0127] In this embodiment, the process of associating specific components is to link the identified risk failure mode with the specific components inside the equipment to identify the affected components. For example, the "insulation damage" mode is associated with the "cable insulation layer" to identify the affected component as the "cable".

[0128] In this embodiment, the affected components are specific equipment components that are affected by potential faults in fault mode identification. For example, for "insulation damage", the affected components include "cable insulation layer", "transformer insulation material" and the like.

[0129] In this embodiment, feasibility is an evaluation criterion for whether the maintenance plan can be successfully implemented in actual operation, which is usually expressed as a percentage. For example, the simulation results show that the feasibility of the maintenance plan for cable replacement is 90%, which is higher than the set threshold of 70%.

[0130] The working principle and beneficial effects of the above technical solution are: through priority assessment and risk analysis, potential failure risks are associated with failure modes to form a risk matrix, and then the affected components are identified, targeted maintenance plans are formulated, and the feasibility of the maintenance plan is evaluated through simulation to optimize maintenance decisions, effectively allocate maintenance resources, and reduce unnecessary maintenance.

[0131] Embodiment 7:

[0132] Based on Example 6, the practical operation unit includes:

[0133] Simulation block: construct a simulation scenario according to the maintenance plan, perform operations according to the steps of the maintenance plan, and record the execution status and corresponding execution time of each step;

[0134] Judgment block: Evaluate the feasibility of the maintenance plan according to the execution status and execution time, compare the feasibility of the maintenance plan with the preset feasibility, if the feasibility corresponding to the maintenance plan is higher than the preset feasibility, perform actual maintenance operations, if the feasibility corresponding to the maintenance plan is equal to or lower than the preset feasibility, identify potential risk points, adjust the maintenance plan based on the potential risk points, and perform secondary simulation on the adjusted maintenance plan.

[0135] In this embodiment, the simulation scenario is a virtual environment created according to the maintenance plan, simulating the various steps and conditions in the maintenance process. For example, in transformer maintenance, the simulation scenario may include equipment power outage, maintenance tool preparation, personnel allocation, and maintenance steps (such as replacing insulating oil, checking the grounding system, etc.).

[0136] In this embodiment, the execution status refers to the actual execution status of each step in the simulation process, including success or failure, abnormal conditions, etc. For example, in the step of replacing the insulating oil of the transformer, it is recorded whether it is successfully completed, whether leakage occurs during the replacement process, etc.

[0137] In this embodiment, the execution time is the actual time required to complete each step, for example, the time for replacing the insulating oil is recorded as 30 minutes, while the time for checking the grounding system is recorded as 15 minutes.

[0138] In this embodiment, the potential risk point is a link in the maintenance process that may cause failure or safety hazards. For example, in transformer maintenance, the potential risk point may be failure to properly cut off the power during the maintenance process, resulting in an electrical accident.

[0139] In this embodiment, the maintenance plan adjustment process is to modify the original maintenance plan to eliminate or reduce risks based on the identified potential risk points. For example, if it is found in the simulation that the power-off step is not strictly implemented, the maintenance plan is adjusted to add power-off confirmation and safety inspection steps.

[0140] In this embodiment, the secondary simulation is re-simulated based on the adjusted maintenance plan to verify the effectiveness of the adjustment. For example, in the adjusted maintenance plan, the transformer maintenance process is simulated again to ensure that the newly added safety inspection steps can effectively prevent risks.

[0141] In this embodiment, the preset feasibility is the minimum feasibility standard for the successful execution of the maintenance plan set based on historical data and experience. For example, the preset feasibility is 80%, which means that the maintenance plan must have at least an 80% success probability before actual maintenance can be carried out.

[0142] The working principle and beneficial effects of the above technical solution are: by constructing a simulation scenario, the maintenance plan is executed step by step and the execution status and time are recorded, the feasibility of the maintenance plan is evaluated, and if the feasibility is higher than the preset value, actual maintenance is carried out; if it is lower than or equal to the preset value, potential risks are identified, the maintenance plan is adjusted and a secondary simulation is carried out to ensure the effectiveness and safety of the maintenance operation, and through effective maintenance management and risk control, the operating reliability of the substation equipment is improved and the service life of the equipment is extended.

[0143] Embodiment 8:

[0144] The embodiment of the present invention provides a method for auxiliary decision making of state maintenance of substation equipment, such as Figure 2 As shown, including:

[0145] Step 1: Collect the best operation data and historical fault records of the substation equipment, analyze the historical fault records, define the fault type and associate it with the best operation data to form an associated data set, extract key features from the key data set, and generate a first feature set;

[0146] Step 2: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set;

[0147] Step 3: Use the first feature set and the second feature set to build a fault diagnosis model, perform fault prediction, and obtain the potential fault risk corresponding to the current device state of the substation equipment;

[0148] Step 4: Generate a maintenance strategy based on the potential failure risk, derive a maintenance plan based on the current equipment status of the substation equipment, simulate the maintenance plan, output actual maintenance operations based on the simulation results, and assist in decision-making on maintenance of the substation equipment status.

[0149] The working principle and beneficial effects of the above technical solution are: by collecting the best operating data and historical fault records of substation equipment, a first feature set is generated, and based on the parameter changes of the equipment status, relevant features are dynamically selected to form a second feature set. These two are used to build a fault diagnosis model, predict potential fault risks, and generate corresponding maintenance strategies and plans. Finally, through simulation and optimization of maintenance operations, intelligent maintenance and auxiliary decision-making are achieved, which adapts to complex failure modes and improves the safety and reliability of substation equipment operation.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A substation equipment condition maintenance auxiliary decision-making system, characterized in that: include: The first feature module collects the best operation data and historical fault records of the substation equipment, analyzes the historical fault records, defines the fault type and associates the best operation data to form an associated data set, extracts key features from the key data set, and generates a first feature set; A second feature module: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set; Fault diagnosis module: constructs a fault diagnosis model using the first feature set and the second feature set to perform fault prediction and obtain the potential fault risk corresponding to the current equipment state of the substation equipment; Maintenance module: Generates maintenance strategies based on the potential failure risks, derives maintenance plans based on the current equipment status of the substation equipment, simulates the maintenance plans, outputs actual maintenance operations based on the simulation results, and assists decision-making on maintenance of the substation equipment status; The first feature module includes: Associated data unit: defines the fault type and the corresponding annotation information according to the historical fault records of the substation equipment, and associates the collected optimal operation data of the substation equipment with the fault type according to the annotation information to form an associated data set; A feature extraction unit: obtaining device characteristics of the substation equipment, and selecting a corresponding feature extraction method from a feature extraction table based on the device characteristics of the substation equipment and the fault type, and using the feature extraction method to extract key features from the associated data set to obtain a first feature set; The second feature module includes: Equipment status unit: Based on the first feature set, a sliding window is used to detect parameter changes of the substation equipment in real time, identify the time point of parameter changes, and a classification algorithm is used to distinguish the equipment status in real time based on the features in the first feature set to obtain the current equipment status; Related feature unit: calculates the degree of influence of parameter change on the degree of dependency between each feature in the first feature set and the current device state, screens the first feature set based on the influence degree, obtains related features, and further obtains the second feature set; The relevant feature units include: Calculation block: ,in, Indicates the degree of dependency between feature X in the first feature set and the current device state Y; Indicates the degree to which the dependency is affected by parameter changes; X represents feature X in the first feature set; Y represents the current device state; x represents the value of feature X; y Indicates the value of the current device state Y; represents the marginal probability density function of feature X; Represents the marginal probability density function of the current device state Y; Represents the joint probability density function of feature X and device state Y; Represents parameter changes that affect the relationship between features and device states; Related feature block: dynamically selects related features corresponding to the current device state according to the degree of influence, and then adjusts the first feature set to obtain the second feature set.

2. A substation equipment condition maintenance auxiliary decision system according to claim 1, characterized in that: Fault diagnosis module, including: The first diagnostic model unit: determines a fault diagnosis algorithm based on the inspection of the state of the substation equipment, and uses the first feature set to train the fault diagnosis algorithm to generate a first-level diagnostic model; The second diagnostic model unit: cross-validates the primary diagnostic model according to the evaluation index generated by the maintenance of the substation equipment status to obtain a cross-validation result, and makes a first adjustment to the primary diagnostic model based on the cross-validation result to obtain a secondary diagnostic model; Hyperparameter adjustment unit: train the secondary diagnostic model using the second feature set, determine the hyperparameters that need to be tuned corresponding to the secondary diagnostic model and the value range of each hyperparameter based on the model-hyperparameter table, and then form a hyperparameter space, traverse the hyperparameter space using a grid search method, extract the best hyperparameter combination, and use the best hyperparameter combination to perform a second adjustment on the secondary diagnostic model to obtain a fault diagnosis model; Fault output unit: performs fault prediction according to the fault diagnosis model, and then obtains the potential fault risk of the current equipment status.

3. The substation equipment condition maintenance auxiliary decision-making system according to claim 1 is characterized in that: Maintenance module, including: Priority unit: matches the failure mode corresponding to the potential failure risk from the risk-mode table, constructs a risk matrix according to the failure mode, and calculates the risk priority number of each failure mode based on the risk matrix; Mode determination unit: divides the risk priority numbers based on the risk priority values ​​to obtain the first level risk failure mode, the second level risk failure mode and the third level risk failure mode; An association unit: associates the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode with specific components corresponding to the power transformation equipment, and identifies the affected components; Maintenance planning unit: determining corresponding maintenance modes for the first-level risk failure mode, the second-level risk failure mode, and the third-level risk failure mode, formulating maintenance strategies based on the affected components, the current equipment status, and the maintenance modes, and determining maintenance plans based on the maintenance strategies; Practical operation unit: simulates the maintenance plan, evaluates the feasibility of the maintenance plan based on the simulation result, and performs actual maintenance operation on the maintenance plan with a feasibility higher than the preset feasibility.

4. A substation equipment condition maintenance auxiliary decision system according to claim 3, characterized in that: Practical modules include: Simulation block: construct a simulation scenario according to the maintenance plan, perform operations according to the steps of the maintenance plan, and record the execution status and corresponding execution time of each step; Judgment block: Evaluate the feasibility of the maintenance plan according to the execution status and execution time, compare the feasibility of the maintenance plan with the preset feasibility, if the feasibility corresponding to the maintenance plan is higher than the preset feasibility, perform actual maintenance operations, if the feasibility corresponding to the maintenance plan is equal to or lower than the preset feasibility, identify potential risk points, adjust the maintenance plan based on the potential risk points, and perform secondary simulation on the adjusted maintenance plan.

5. A method for auxiliary decision-making for condition maintenance of substation equipment, characterized in that: include: Step 1: Collect the best operation data and historical fault records of the substation equipment, analyze the historical fault records, define the fault type and associate it with the best operation data to form an associated data set, extract key features from the key data set, and generate a first feature set; Step 2: determining parameter changes of the substation equipment based on the first feature set, dynamically selecting relevant features corresponding to the current equipment state according to the parameter changes, and adjusting the first feature set according to the relevant features to obtain a second feature set; Step 3: Use the first feature set and the second feature set to build a fault diagnosis model, perform fault prediction, and obtain the potential fault risk corresponding to the current device state of the substation equipment; Step 4: Generate a maintenance strategy based on the potential fault risk, derive a maintenance plan based on the current equipment status of the substation equipment, simulate the maintenance plan, output the actual maintenance operation according to the simulation result, and assist in decision-making on the maintenance of the substation equipment status; Regarding step 1, it also includes: Step 11: According to the historical fault records of the substation equipment, the fault type and the corresponding annotation information are defined, and the collected optimal operation data of the substation equipment are associated with the fault type according to the annotation information to form an associated data set; Step 12: obtaining equipment characteristics of the substation equipment, and selecting a corresponding feature extraction method from a feature extraction table based on the equipment characteristics of the substation equipment and the fault type, and using the feature extraction method to extract key features from the associated data set to obtain a first feature set; Regarding step 2, it also includes: Step 21: Based on the first feature set, a sliding window is used to detect parameter changes of the substation equipment in real time, identify the time point of the parameter change, and a classification algorithm is used to distinguish the equipment state in real time in combination with the features in the first feature set to obtain the current equipment state; Step 22: Calculate the degree of influence of the parameter change on the degree of dependency between each feature in the first feature set and the current device state, and screen the first feature set based on the influence degree to obtain relevant features, thereby obtaining the second feature set; Regarding step 22, it also includes: Step 221: ,in, Indicates the degree of dependency between feature X in the first feature set and the current device state Y; Indicates the degree to which the dependency is affected by parameter changes; X represents feature X in the first feature set; Y represents the current device state; x represents the value of feature X; y Indicates the value of the current device state Y; represents the marginal probability density function of feature X; Represents the marginal probability density function of the current device state Y; Represents the joint probability density function of feature X and device state Y; Represents parameter changes that affect the relationship between features and device states; Step 222: Dynamically select relevant features corresponding to the current device state according to the impact degree, and then adjust the first feature set to obtain a second feature set.

Images