Relay protection device inspection method and device, computer device and storage medium

Abstract

The application relates to a relay protection device inspection method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: obtaining operation data of a target relay protection device, a plurality of inspection tasks and an inspection benchmark corresponding to each inspection task; for each inspection task, performing feature comparison on the operation data and the inspection benchmark corresponding to the inspection task to obtain a feature comparison result corresponding to the inspection task, determining the operation state of the target relay protection device under each inspection task based on the feature comparison result; determining an inspection task scheduling scheme through a dynamic scheduling model based on the operation state of the target relay protection device under each inspection task; and performing the plurality of inspection tasks on the target relay protection device according to the inspection task scheduling scheme to obtain an inspection result of the target relay protection device. The method can flexibly schedule the inspection tasks to inspect the relay protection device.

Description

Technical Field

[0001] This application relates to the field of substation operation and maintenance technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for testing relay protection equipment. Background Technology

[0002] Relay protection equipment is the core barrier for the safe and stable operation of substations and even the entire power system. Its operational reliability, measurement accuracy, and logical correctness directly determine the efficiency of power grid fault handling and the level of accident risk.

[0003] With the development trend of digital and intelligent power grids, traditional methods for inspecting relay protection equipment suffer from insufficient flexibility in scheduling multiple inspection tasks. Therefore, how to flexibly schedule inspection tasks to achieve the inspection of relay protection equipment is an urgent problem to be solved. Summary of the Invention

[0004] Therefore, it is necessary to provide a relay protection equipment testing method, device, computer equipment, computer-readable storage medium, and computer program product that can flexibly schedule testing tasks to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for testing relay protection equipment, including:

[0006] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0007] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0008] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0009] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0010] In one embodiment, feature comparison is performed between the running data and the inspection benchmark corresponding to the inspection task to obtain the feature comparison result corresponding to the inspection task, including:

[0011] For each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features.

[0012] The operational features and the benchmark features are sequentially aligned and compared in time series to obtain the feature comparison results corresponding to the testing task.

[0013] In one embodiment, determining the operating status of the target relay protection device under the inspection task based on the feature comparison results includes:

[0014] Based on the feature comparison results, the state evaluation score of the test task is obtained through a pre-built operational state evaluation model.

[0015] Based on the condition assessment score, the operating status of the target relay protection equipment under the inspection task is determined.

[0016] In one embodiment, based on the operating status of the target relay protection equipment under each inspection task, a dynamic scheduling model is used to determine the inspection task scheduling scheme, including:

[0017] Based on the operating status of the target relay protection equipment under each inspection task, the dynamic scheduling model is used to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment.

[0018] Multiple inspection tasks are sorted in descending order of their scheduling evaluation values ​​to determine the scheduling priority of each inspection task.

[0019] For each inspection task, based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task, the corresponding inspection resources are matched for the inspection task to obtain the inspection resource allocation result of the inspection task.

[0020] Based on the scheduling priority of each inspection task and the results of inspection resource allocation, an inspection task scheduling scheme is generated.

[0021] In one embodiment, based on the operating status of the target relay protection device under each inspection task, a dynamic scheduling model is used to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment, including:

[0022] For each inspection task, based on the operating state of the target relay protection equipment under the inspection task, the state space characteristics corresponding to the inspection task are determined, and the state space characteristics are vectorized to obtain the initial feature vector of the inspection task.

[0023] For each initial feature vector, the correlation between the initial feature vector and each test task is calculated using a self-attention approach, and the weight of the initial feature vector is determined based on the correlation.

[0024] Based on the weights of each initial feature vector, feature fusion is performed on each initial feature vector to obtain the target feature vector;

[0025] The target feature vector is input into the dynamic scheduling model, and the scheduling evaluation value of each inspection task in the current inspection environment is obtained by evaluating each inspection task.

[0026] In one embodiment, according to the inspection task scheduling scheme, multiple inspection tasks are performed on the target relay protection device to obtain the inspection results of the target relay protection device, including:

[0027] Based on the inspection task scheduling scheme, generate corresponding scheduling execution instructions for each inspection task;

[0028] Based on the current time occupancy status of inspection resources in the inspection environment, time conflict scheduling is performed on each scheduled execution instruction to determine the instruction time window for each scheduled execution instruction;

[0029] For each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, based on the scheduling priority indicated by the scheduling execution instruction and the test resource allocation result, the test task is performed on the target relay protection equipment, and the test result of the test task is obtained.

[0030] Based on the inspection results of each inspection task, the inspection results of the target relay protection equipment are generated.

[0031] Secondly, this application also provides a relay protection equipment testing device, comprising:

[0032] The acquisition module is used to acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task.

[0033] The comparison module is used to compare the running data with the inspection benchmark corresponding to each inspection task for each inspection task, obtain the feature comparison result corresponding to the inspection task, and determine the operating status of the target relay protection device under the inspection task based on the feature comparison result.

[0034] The determination module is used to determine the inspection task scheduling scheme based on the operating status of the target relay protection equipment under each inspection task through a dynamic scheduling model.

[0035] The execution module is used to perform multiple inspection tasks on the target relay protection equipment according to the inspection task scheduling plan, and obtain the inspection results of the target relay protection equipment.

[0036] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0037] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0038] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0039] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0040] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0041] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0042] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0043] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0044] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0045] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0046] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0047] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0048] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0049] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0050] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0051] The aforementioned relay protection equipment inspection method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task. For each inspection task, the operating data and the inspection benchmarks corresponding to the inspection task are compared to obtain the feature comparison results corresponding to the inspection task. Based on the feature comparison results, the operating status of the target relay protection equipment under each inspection task is determined. Based on the operating status of the target relay protection equipment under each inspection task, a dynamic scheduling model is used to determine the inspection task scheduling scheme. According to the inspection task scheduling scheme, multiple inspection tasks are executed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment. Through the above process, inspection resources can be dynamically allocated and the execution order of inspection tasks can be adjusted according to the actual operating status of the target relay protection equipment under different inspection tasks, improving the scheduling flexibility of multiple inspection tasks, effectively improving the efficiency and rationality of relay protection equipment inspection, ensuring that the inspection process can adapt to different field inspection environments and actual equipment states, providing reliable technical support for the status detection of relay protection equipment in the power grid operation and maintenance process, and realizing dynamic and adaptive control of the inspection process. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a diagram illustrating the application environment of a relay protection equipment testing method in one embodiment.

[0054] Figure 2 This is a flowchart illustrating a relay protection device inspection method in one embodiment;

[0055] Figure 3 This is a flowchart illustrating the relay protection equipment inspection method in another embodiment;

[0056] Figure 4 This is a structural block diagram of a relay protection equipment testing device in one embodiment;

[0057] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0059] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0060] The relay protection device testing method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, the target relay protection device 102 communicates with the server 104 via a network. A data storage system can store the data that the server 104 needs to process. The data storage system can be integrated onto the server 104 or placed on a cloud or other network server. Specifically, the server 104 acquires the operating data of the target relay protection device 102, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task. For each inspection task, the server 104 performs feature comparison between the operating data and the inspection benchmarks corresponding to the inspection task to obtain the feature comparison results for the inspection task. Based on the feature comparison results, the server 104 determines the operating status of the target relay protection device 102 under each inspection task. Based on the operating status of the target relay protection device 102 under each inspection task, the server 104 determines the inspection task scheduling scheme through a dynamic scheduling model. The server 104 executes multiple inspection tasks on the target relay protection device 102 according to the inspection task scheduling scheme to obtain the inspection results of the target relay protection device 102. The server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0061] In one exemplary embodiment, such as Figure 2 As shown, a method for testing relay protection equipment is provided, which is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 202 to 208. Wherein:

[0062] Step 202: Obtain the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task.

[0063] The target relay protection equipment's operational data includes body operation status data, circuit action and communication status data, waveform data, and localized equipment-specific data. Body operation status data includes inverter power supply operating voltage, function board status, setting control word parameters, time synchronization anomaly signals, operation and maintenance board status, and remote signaling and telemetry data. Circuit action and communication status data includes the target relay protection equipment's input status, output status, maintenance quality level, failure start signal, associated circuit signal, and output trip reverse calibration message. Waveform data includes steady-state analog quantity RMS value, transient fault waveform data, zero-drift analog quantity, and switch quantity change information. Localized equipment-specific data includes insulation performance parameters, AC circuit power consumption data, optical port optical power value, device power-on self-test status, and hard contact on / off status.

[0064] For example, the system acquires the physical operating status data of the target relay protection device; it acquires data from the target relay protection device through the SV (sampled value) and GOOSE (substation event for general objects) network message acquisition unit to obtain circuit operation and communication status data; it acquires data from the operation process of the target relay protection device through the intelligent waveform recorder system to obtain waveform data of the target relay protection device; and it acquires data from the target relay protection device through a dedicated detection module to obtain localized device-specific data.

[0065] Optionally, multiple inspection tasks and corresponding inspection benchmarks for each inspection task can be stored in a data storage system on a server or in a cloud database. The inspection standards are standardized and adapted for different types and manufacturers of relay protection equipment. These include state benchmark templates such as pressure plate status benchmark templates, setting and control word benchmark templates, software version and download time CRC code benchmark templates, and GOOSE input status benchmark templates. They also include error thresholds such as high and low operating voltage limits, zero-drift analog quantity thresholds, steady-state and transient accuracy error thresholds, frequency thresholds, and time synchronization thresholds. Furthermore, they include complete action sequence specifications for protection start-up, protection action, circuit breaker tripping, failure start-up, reclosing action, and intelligent terminal back-calibration. The inspection benchmarks for each inspection task ensure that each inspection task has a clear and unified reference basis, avoiding judgment deviations caused by inconsistent benchmarks.

[0066] The inspection tasks are performed on the target relay protection equipment using external inspection equipment. These include basic inspection tasks, electrical performance testing tasks, and functional and overall logic testing tasks. External inspection equipment refers to the equipment used to perform these inspection tasks, such as relay protection testers and online monitoring platforms. Basic inspection tasks involve performing basic checks on the target relay protection equipment, such as external line checks, wiring checks, terminal tightness checks, secondary circuit verification, setting sheet checks, software version checks, CRC verification, pressure plate checks, control word checks, and time synchronization checks. Electrical performance testing tasks involve performing basic checks on the target relay protection equipment. The tasks of testing the performance of protection equipment include, for example, insulation resistance testing, withstand voltage testing, current or voltage action value testing, return value testing, return coefficient testing, sampling accuracy testing, time setting value testing, action time error testing, impedance characteristic testing, directional characteristic testing, and harmonic restraint testing. Functional and overall logic testing tasks are tasks that test the function of the target relay protection equipment and the overall logic of the entire equipment. Examples include testing the logic of each protection stage (instantaneous overcurrent, zero-sequence, distance, differential, etc.), interlocking, starting, reclosing, and post-acceleration logic, output matrix testing, trip / signal circuit testing, GOOSE / SV communication testing, and time synchronization accuracy testing.

[0067] Step 204: For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0068] The feature comparison result is a time-series deviation feature matrix, which includes deviation information between the running data and the corresponding testing benchmark at each time point. This deviation information can be data such as deviation value, absolute deviation, relative deviation, or deviation change rate. The time-series deviation feature matrix is ​​a feature matrix with the same dimension as the multi-dimensional time-series feature sequence. Each element in the matrix corresponds to the deviation information of the feature parameters at a given time point, used to completely record the deviation changes of the testing task over continuous time. The operating states include normal state, critical degradation state, and abnormal state.

[0069] For example, before comparing the features of the running data and the inspection benchmark corresponding to the inspection task, the running data needs to be preprocessed. For example, the Kalman filter algorithm is used to filter out invalid interference data in the running data; outliers in the running data are removed; the running data is time-stamped using the IRIG-B time synchronization protocol to ensure time consistency of different acquisition channels and different types of data (analog quantities, switch quantities, message data); and the format of the running data is uniformly converted into the XML standard format, etc.

[0070] For example, feature extraction is performed on the operational data to obtain operational features. For each inspection task, the target operational features and the inspection benchmark features are compared. The deviation information between each target operational feature and the corresponding inspection benchmark feature at each time point is calculated. The deviation information at each time point is integrated to construct a time-series deviation feature matrix, which is used as the feature comparison result corresponding to the inspection task. Based on the feature comparison results, the operational status of the target relay protection equipment under the inspection task is determined through a pre-constructed operational status evaluation model.

[0071] Step 206: Based on the operating status of the target relay protection equipment under each inspection task, determine the inspection task scheduling scheme through a dynamic scheduling model.

[0072] The dynamic scheduling model, trained using reinforcement learning algorithms, employs multiple objective optimization functions, including verification efficiency, verification coverage, and power grid safety operation coefficient. It outputs the optimal inspection task scheduling scheme based on different operating states. The inspection task scheduling scheme includes the priority of each inspection task and the matching inspection resources.

[0073] For example, based on the operating status of the target relay protection equipment under each inspection task, a scheduling evaluation value for each inspection task is determined through a dynamic scheduling model; based on the scheduling evaluation value of each inspection task, the priority of each inspection task and the matching inspection resources are determined; based on the priority of each inspection task and the matching inspection resources, an inspection task scheduling scheme is determined.

[0074] Specifically, for each inspection task, if the target relay protection device is in an abnormal operating state under the inspection task, the inspection task is identified as having the highest priority; if the target relay protection device is in a critical degradation state under the inspection task, the inspection task is processed with the second highest priority; if the target relay protection device is in a normal operating state under the inspection task, the inspection task is processed sequentially according to the cycle. Based on the processing method corresponding to the inspection task, corresponding inspection resources are matched for the inspection task. For inspection tasks processed with the highest priority, high-precision testing equipment, experienced maintenance personnel, and critical operation periods are allocated first, while routine resources are reasonably allocated for inspection tasks processed with the second highest priority, so as to maximize resource utilization.

[0075] For example, the inspection task scheduling scheme includes: initiating local closed-loop testing and special investigation for inspection tasks processed with the highest priority; conducting trend tracking, accuracy verification and uninterrupted power-off reinforcement verification for inspection tasks processed with the second highest priority; and maintaining online monitoring and automatically extending the verification cycle for inspection tasks corresponding to normal status.

[0076] Step 208: According to the inspection task scheduling plan, perform multiple inspection tasks on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0077] The test results include the execution results of multiple test tasks.

[0078] For example, according to the inspection task scheduling scheme, scheduling execution instructions for each inspection task are generated; the scheduling execution instructions for each inspection task are time-sequentially scheduled to obtain the time window for each scheduling execution instruction; the corresponding scheduling execution instructions are issued within the time window of each scheduling execution instruction; in response to each scheduling execution instruction, multiple inspection tasks are performed on the target relay protection device to obtain the inspection results of the target relay device.

[0079] The aforementioned relay protection equipment inspection method involves acquiring the target relay protection equipment's operating data, multiple inspection tasks, and the inspection benchmark corresponding to each inspection task. For each inspection task, the operating data and the corresponding inspection benchmark are compared using features to obtain the feature comparison result. Based on the feature comparison result, the operating status of the target relay protection equipment under each inspection task is determined. Based on the operating status of the target relay protection equipment under each inspection task, a dynamic scheduling model is used to determine the inspection task scheduling scheme. According to the inspection task scheduling scheme, multiple inspection tasks are executed on the target relay protection equipment to obtain the inspection result. Through this process, inspection resources can be dynamically allocated and the execution order of inspection tasks can be adjusted according to the actual operating status of the target relay protection equipment under different inspection tasks. This improves the scheduling flexibility of multiple inspection tasks, effectively enhances the efficiency and rationality of relay protection equipment inspection, and ensures that the inspection process can adapt to different field inspection environments and actual equipment conditions. This provides reliable technical support for the status detection of relay protection equipment in power grid operation and maintenance, realizing dynamic and adaptive control of the inspection process.

[0080] In an exemplary embodiment, feature comparison is performed between the running data and the inspection benchmark corresponding to the inspection task to obtain the feature comparison result corresponding to the inspection task. This includes: for each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features; the running features and inspection benchmark features are sequentially aligned and compared with time series features to obtain the feature comparison result corresponding to the inspection task.

[0081] For example, hierarchical feature extraction is performed on the operational data along the time dimension to construct operational features corresponding to each inspection task. Specifically, the preprocessed operational data is time-series sliced ​​using a preset time window (set according to the type of inspection task; for example, the time window for basic inspection tasks and electrical performance testing tasks (steady-state tasks) is set to 10-30 seconds, and the time window for functional and overall logic inspection tasks (transient tasks) is set to 1-5 seconds). For each inspection task, features are extracted from the operational data within the corresponding time window to obtain the operational features corresponding to that inspection task. The operational features are arranged in chronological order to construct the time-series operational features corresponding to each inspection task. The operational characteristics include electrical quantity timing characteristics (time-sequence changes in the effective values, phases, fundamental frequency and harmonic amplitudes of current and voltage), status timing characteristics (time-sequence switching records of pressure plate status and setpoint control words), communication timing characteristics (time-sequence changes in transmission delay and packet loss rate of SV / GOOSE messages), and self-test timing characteristics (time-sequence records of the results of device power-on self-test and periodic self-test). The timing operational characteristics are used to fully reflect the operational status change patterns of the target relay protection equipment over a continuous period of time.

[0082] For example, for each inspection task, feature extraction is performed on the inspection benchmark corresponding to the inspection task to obtain inspection benchmark features. For instance, for the pressure plate / fixed value / CRC / GOOSE status in the status benchmark template, discrete feature vectors (0 / 1, matching degree) can be extracted as inspection benchmark features; for the zero drift, accuracy, frequency and time synchronization error in the error threshold value, numerical feature vectors (threshold, upper and lower bounds) can be extracted as inspection benchmark features; for the action timing and time window in the entire time series specification, time series template features (standard time series sequence) can be extracted as inspection benchmark features.

[0083] Optionally, using the Dynamic Time Warping (DTW) algorithm, for each inspection task, the corresponding operational features and benchmark features are sequentially aligned and compared in time series. The deviation information of each feature parameter at each time node is calculated from the corresponding benchmark feature. The deviation information of each time node is integrated to construct the time series deviation feature matrix of the inspection task. The time series deviation matrix is ​​used as the feature comparison result.

[0084] Optionally, for non-time-series inspection benchmark features (such as fixed value control word benchmark, CRC code benchmark), the deviation information between the running features and the inspection benchmark features can be directly calculated.

[0085] In this embodiment, by performing feature extraction, timing alignment, and timing feature comparison on the operating data and the inspection benchmark corresponding to the inspection task, respectively, the feature comparison results corresponding to the inspection task that accurately reflect the difference between the actual operation of the equipment and the inspection benchmark can be obtained, thus reflecting the state deviation of the target relay protection equipment under different time periods and different inspection tasks.

[0086] In one embodiment, determining the operating status of the target relay protection device under the inspection task based on the feature comparison results includes: obtaining the status assessment score of the inspection task through a pre-built operating status assessment model based on the feature comparison results; and determining the operating status of the target relay protection device under the inspection task based on the status assessment score.

[0087] The pre-built runtime evaluation model includes a temporal coding layer, an attention mechanism layer, and a regression output layer. The temporal coding layer comprises an input layer, a hidden layer, and an output layer, while the attention mechanism layer is located between the temporal coding layer and the regression output layer. The dimension of the input layer is consistent with that of the temporal bias feature matrix (determined according to the complexity of the testing task; the input dimension can be set to 64-128 dimensions to adaptively match the number of features for different testing tasks). One to two hidden layers are set, with 128-256 hidden units per layer, using the tanh activation function to achieve non-linear transformation of the input features and uncover deep hidden features. The output layer adopts a fully connected structure, with the output dimension consistent with the number of hidden units, ensuring that the corresponding hidden state features are output at each time step.

[0088] Furthermore, the temporal coding layer serves as the feature extraction center for the operational status assessment model. It performs in-depth processing on the input temporal deviation feature matrix, uncovering long-term dependencies, short-term fluctuation patterns, and feature correlations within the time series, providing high-quality latent state features for subsequent attention weighting. The temporal coding layer introduces a dropout regularization mechanism (dropout probability set to 0.2-0.3) to prevent overfitting of the operational status assessment model, improve its generalization ability, and adapt to inspection scenarios for different substations and different types of relay protection equipment. Specifically, the temporal coding layer is constructed using a bidirectional long short-term memory network. This network avoids the gradient vanishing or exploding problems of traditional recurrent neural networks and effectively captures the temporal correlation information of the temporal deviation features.

[0089] Furthermore, the attention mechanism layer performs weighted fusion of the latent state features output by the temporal coding layer to obtain the target temporal features. These target temporal features highlight features that contribute significantly to the state evaluation of the testing task (such as time step features with significant deviation changes and high correlation with anomalies) and suppress redundant features (such as features with stable deviations and no practical significance). The attention mechanism layer uses a scaled dot-product attention structure to compute the weighted results of different attention heads in parallel to obtain the target temporal features. The scaled dot-product attention structure includes three trainable parameter matrices: a query vector (Q), a key vector (K), and a value vector (V). The dimensions of these three matrices are consistent with the dimensions of the latent state features output by the temporal coding layer, and the number of attention heads is set to 4-8 (using a multi-head attention structure), which further improves the comprehensiveness of feature fusion and avoids feature omissions caused by a single attention head. In addition, the attention mechanism layer incorporates a softmax normalization function to normalize the attention weights to the 0-1 range, ensuring that the sum of the weights is 1, thus achieving a reasonable weighted allocation of the latent state features.

[0090] Furthermore, the regression output layer performs high-dimensional dimensionality reduction and nonlinear mapping on the target time-series features output by the attention mechanism layer, converting the high-dimensional features into continuous state evaluation scores within the 0-100 range, thus obtaining the state evaluation score. The regression output layer is used to quantitatively evaluate the operating status of the target relay protection equipment under the inspection task. The higher the state evaluation score, the more stable the operating status of the target relay protection equipment under the inspection task and the higher the fit with the inspection benchmark; the lower the state evaluation score, the greater the probability of anomalies in the operating status of the target relay protection equipment under the inspection task and the greater the deviation from the inspection benchmark. The regression output layer employs a structure combining two fully connected layers and regression activation. Specifically, the first fully connected layer uses the ReLU activation function to reduce the dimensionality of the target temporal features (with the same dimension as the hidden state features) output from the attention mechanism layer to 32 dimensions. The ReLU activation function filters out invalid features, reduces model complexity, and avoids gradient vanishing. The second fully connected layer uses the sigmoid activation function to further reduce the 32-dimensional features to 1 dimension, mapping the output to the 0-1 range. Then, a linear scaling (multiplied by 100) is applied to obtain a state evaluation score of 0-100. Simultaneously, the regression output layer introduces an L2 regularization mechanism. This L2 regularization constrains the model parameters, further improving the model's generalization ability and ensuring that the output state evaluation score has good stability and reliability, accurately matching the inspection requirements of the target relay protection equipment.

[0091] For example, the feature comparison result is a time-series deviation feature matrix (the dimension is "time step × feature dimension", where the time step is set according to the type of verification project; 10-30 time steps for steady-state projects and 5-15 time steps for transient projects). The time-series deviation feature matrix is ​​sequentially input into the time-series encoding layer of a pre-built operational status assessment model in batches to extract the latent state features at each time step. The latent state features at each time step are then input into the attention mechanism layer of the operational status assessment model for weighted fusion to obtain the target time-series features. Finally, the target time-series features are input into the regression output layer for dimensionality reduction mapping to obtain the status assessment score for the verification task. The latent state features include the correlation information between the deviation features at the current time step and the deviation features at previous and subsequent time steps. The target time-series features are used to centrally characterize the overall operational status of the target relay protection equipment under the verification task.

[0092] Specifically, the temporal coding layer processes the temporal bias feature matrix step-by-step through the forward propagation of a bidirectional LSTM. The forward LSTM branch starts from the first time step and processes the bias features of each time step sequentially, uncovering positive dependencies in the time series (e.g., the correlation between the bias at the current time step and the biases of multiple previous time steps). The backward LSTM branch starts from the last time step and processes the bias features of each time step in reverse, uncovering negative dependencies in the time series (e.g., the correlation between the bias at the current time step and the biases of multiple subsequent time steps). After processing by the bidirectional LSTM, each time step outputs a corresponding hidden state feature. Based on these hidden state features, the temporal coding layer performs temporal alignment to obtain a sequence of hidden state features.

[0093] Furthermore, the hidden state feature sequence output from the temporal coding layer is fed into the attention mechanism layer as a whole. Through adaptive weighted fusion by the attention mechanism layer, the target temporal feature is obtained. Specifically, the hidden state feature sequence is mapped to a query vector (Q), a key vector (K), and a value vector (V), respectively. This mapping process is achieved through matrix multiplication. , , ,in The hidden state feature sequence, , and The trainable parameter matrices for the query vector, key vector, and value vector are obtained during the pre-training process of the pre-built operational status assessment model. The dot product of the query vector (Q) and key vector (K) is calculated to obtain the attention score matrix. To avoid excessively large dot product results due to high feature dimensions and overly concentrated weight distribution after softmax normalization, the attention score matrix is ​​scaled using a scaling factor (the square root of the key vector dimension). The scaled attention score matrix is ​​then normalized using the softmax function to obtain the attention weights for each time step and each feature dimension. The normalized attention weights are then weighted and summed with the value vector (V) to obtain the target temporal feature, which is the product of the attention weights and the value vector. This target temporal feature highlights the impact of key deviation features (e.g., time step features of deviation mutations and deviation exceeding limits) and suppresses the interference of redundant deviation features, thus more accurately representing the overall operational status of the assessment project. A larger attention weight indicates a higher contribution of the implicit state features of that time step to the state assessment of the assessment task.

[0094] Furthermore, the target temporal features output from the attention mechanism layer are input into the regression output layer. Through dimensionality reduction and nonlinear mapping in this layer, the state evaluation scores for each testing task are finally output. Specifically, the target temporal features are input into the first fully connected layer of the regression output layer. The first fully connected layer reduces the dimensionality of the high-dimensional target temporal features (the dimension is the same as the number of hidden units) to 32 dimensions using a trainable weight matrix. At the same time, the ReLU activation function is used to filter out invalid feature components, reducing the computational complexity of the running state evaluation model and avoiding overfitting. The 32-dimensional features after dimensionality reduction are input into the second fully connected layer. The second fully connected layer further reduces the dimensionality of the 32-dimensional features to 1 dimension using a weight matrix. Then, the sigmoid activation function is used to map the output result to a continuous interval of 0-1. Through a linear scaling operation (multiplying the output result by 100), it is converted into a state evaluation score of 0-100, completing the state quantification evaluation of a single testing task.

[0095] For example, since the pre-built operational status assessment model adopts a multi-branch parallel architecture, with each branch corresponding to an independent inspection task, the above-mentioned time-series encoding, attention weighting, and regression mapping processes can be executed synchronously and in parallel on the time-series deviation feature matrix of each inspection task, and synchronously output the status assessment scores of all inspection tasks, ensuring the efficiency and synchronicity of status assessment, and providing accurate and quantitative input basis for subsequent status classification and dynamic scheduling.

[0096] For example, if the status assessment score is less than a first threshold, the operating status of the target relay protection device under the inspection task is determined to be normal; if the status assessment score is greater than or equal to the first threshold and less than a second threshold, the operating status of the target relay protection device under the inspection task is determined to be critically degraded; if the status assessment score is greater than or equal to the second threshold, the operating status of the target relay protection device under the inspection task is determined to be abnormal; wherein, the first threshold is less than the second threshold.

[0097] Specifically, the first and second thresholds are pre-set based on the operational importance of the relay protection equipment, the inspection accuracy requirements, and historical operation and maintenance data. If the state assessment score is less than the first threshold, all deviation characteristics of the inspection task are within the allowable range, the timing deviation feature matrix closely matches the verification benchmark, and the target relay protection equipment operates stably and reliably. Therefore, the operating state corresponding to this inspection task is determined as the normal state. If the state assessment score is greater than or equal to the first threshold but less than the second threshold, the inspection task has shown phenomena such as increased local deviations and decreased feature consistency. The operating performance of the target relay protection equipment has degraded to a certain extent, but has not yet exceeded the safe operating boundary, and is in the transition range between performance decline and abnormal occurrence. Therefore, the operating state corresponding to this inspection task is determined as the degradation critical state. If the state assessment score is greater than or equal to the second threshold, the timing deviation of the inspection task has significantly exceeded the allowable range, exhibiting obvious abnormal characteristics or severely deviating from the verification benchmark. Continued operation may lead to problems such as protection maloperation, failure to operate, or functional failure of the target relay protection equipment, posing a safety hazard to the power grid operation. Therefore, the operating state corresponding to this inspection task is determined as the abnormal state.

[0098] In this embodiment, the hierarchical processing and weighted fusion of features are completed through the operation status assessment model, realizing the transformation from original deviation features to quantitative status scores. This can accurately identify minor performance degradation and early abnormal signs of relay protection equipment, solving the problem of traditional manual inspection relying on experience judgment and having strong subjectivity in the results. It provides an objective and quantitative decision-making basis for condition-based maintenance of relay protection equipment.

[0099] In one embodiment, based on the operating status of the target relay protection device under each inspection task, a dynamic scheduling model is used to determine the inspection task scheduling scheme, including: evaluating each inspection task based on the operating status of the target relay protection device under each inspection task using a dynamic scheduling model to obtain the scheduling evaluation value of each inspection task in the current inspection environment; sorting multiple inspection tasks in descending order of scheduling evaluation values ​​to determine the scheduling priority of each inspection task; for each inspection task, matching corresponding inspection resources to the inspection task based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task to obtain the inspection resource allocation result of the inspection task; and generating an inspection task scheduling scheme based on the scheduling priority of each inspection task and the inspection resource allocation result.

[0100] The availability of inspection resources includes the qualifications and number of maintenance personnel, the type and number of relay protection testing instruments, the time window for on-site operations, spare parts and communication channels, etc.

[0101] For example, multiple inspection tasks are sorted in descending order of scheduling evaluation values ​​to obtain a sorting result. The sorting result is then mapped to the priority of the multiple inspection tasks. For example, the first inspection task in the sorting result is mapped to the first execution priority. Among them, multiple inspection tasks with similar scheduling evaluation values ​​and related coupling relationships are clustered and merged, and set to the same execution priority to avoid duplicate operations and multiple power outages.

[0102] For example, for each inspection task, based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task, corresponding inspection resources are matched to the inspection task to obtain the inspection resource allocation result. Specifically, high-priority inspection tasks are given priority in configuring high-precision testing equipment, experienced operators, and maintenance windows during off-peak hours; medium- and low-priority inspection tasks are reasonably allocated conventional resources and general testing methods. At the same time, inspection resources in the current inspection environment are balanced and scheduled to avoid resource overload or idleness, ultimately forming a resource allocation scheme that takes into account safety, efficiency, and economy.

[0103] In this embodiment, by allocating inspection resources based on scheduling priority, the resource supply for high-risk and high-priority inspection tasks can be prioritized, avoiding the waste of resources on low-risk routine inspection items. Under the condition of limited inspection resources, the detection efficiency of hidden dangers in relay protection equipment can be maximized, which is more suitable for the actual scenario of limited on-site inspection resources and urgent needs, and ensures that the power system can always maintain a safe and reliable operating state.

[0104] In one embodiment, based on the operating status of the target relay protection device under each inspection task, a dynamic scheduling model is used to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment. This includes: for each inspection task, determining the state space characteristics corresponding to the inspection task based on the operating status of the target relay protection device under the inspection task; vectorizing the state space characteristics to obtain the initial feature vector of the inspection task; for each initial feature vector, calculating the correlation degree between the initial feature vector and each inspection task using a self-attention method; determining the weight of the initial feature vector based on the correlation degree; performing feature fusion on the initial feature vectors based on their weights to obtain the target feature vector; and inputting the target feature vector into the dynamic scheduling model to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment.

[0105] The state-space features include information such as the operating status, importance level, operating conditions of the power grid, historical anomaly records, and time window constraints of the target relay protection equipment corresponding to the inspection task. The dynamic scheduling model includes a built-in feature encoding network and a global correlation reasoning module. The built-in feature encoding network is used to uniformly embed and normalize the target feature vector to eliminate dimensional differences and feature redundancy. The global correlation reasoning module is used to mine the loop correlations, logical correlations, and risk transmission relationships between different inspection tasks, identifying the chain reactions that abnormal inspection tasks may trigger. The target feature vector is used to reflect the overall verification environment situation, so that subsequent scheduling decisions can take into account both local risks and global security. The scheduling evaluation value is a quantitative value reflecting the severity of the anomaly of the inspection task, its impact on power grid security, functional criticality, maintenance timeliness requirements, and resource input benefits. It is used to characterize the urgency and importance of the inspection task; the higher the scheduling evaluation value, the higher the urgency and importance of the inspection task.

[0106] For example, the operating state of the target relay protection equipment under the inspection task is quantitatively encoded to obtain the initial state space characteristics corresponding to the inspection task. Constraints such as the importance level of the target relay protection equipment, the current operation mode of the power grid, the load level, the number of available maintenance personnel, the number of testing equipment, and the operation time window are integrated into the initial state space characteristics to obtain the state space characteristics.

[0107] Optionally, each initial feature vector is linearly mapped to obtain the query vector, key vector, and value vector corresponding to each inspection task. For each query vector, the dot product similarity between the query vector and each key vector is calculated to obtain the correlation between the inspection task corresponding to the query vector and other inspection tasks. The correlation between the inspection task corresponding to each initial feature vector and other inspection tasks is converted into a correlation weight matrix with values ​​between 0 and 1 and a sum of 1 for each row through the Softmax function. This correlation weight matrix is ​​used to characterize the risk transmission relationship, logical dependency relationship, and maintenance coupling degree between different inspection tasks. Based on the correlation weight matrix, feature fusion is performed on the value vectors corresponding to all inspection tasks to obtain the target feature vector.

[0108] For example, the training and optimization process of the dynamic scheduling model is similar to that of the operation status assessment model. It relies on historical scheduling cases and decision results to complete the model parameter iteration, adapt to the testing task scheduling needs of substations of different sizes and under different operating modes, and ensure the accuracy and timeliness of the scheduling results.

[0109] In this embodiment, the correlation between different inspection tasks is explored through a self-attention mechanism, avoiding global decision-making bias caused by scheduling based solely on the risk of a single task. This enables the generation of a more scientific and reasonable inspection scheduling scheme from a global perspective of power grid operation, ensuring that high-risk anomalies are prioritized for handling and improving the overall safety of regional power grid operation.

[0110] In one embodiment, according to the inspection task scheduling scheme, multiple inspection tasks are performed on the target relay protection device to obtain the inspection results of the target relay protection device. This includes: generating corresponding scheduling execution instructions for each inspection task according to the inspection task scheduling scheme; performing timing conflict scheduling on each scheduling execution instruction according to the time occupancy status of inspection resources in the current inspection environment to determine the instruction time window of each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, performing the inspection task on the target relay protection device based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result to obtain the inspection result of the inspection task; and generating the inspection result of the target relay protection device based on the inspection results of each inspection task.

[0111] For example, according to the inspection task scheduling scheme, corresponding scheduling execution instructions are generated for each inspection task. The scheduling execution instructions adopt a standardized structured format to indicate the scheduling priority of the inspection task, the result of the inspection resource allocation, and the content corresponding to the inspection task.

[0112] For example, the real-time time occupancy status of all current inspection resources (maintenance personnel, testing equipment, and work sites) is collected to establish a resource time occupancy ledger, clarifying the occupied time periods, idle time periods, and available time periods for each resource. Based on the scheduling priority corresponding to the inspection task scheduling scheme, timing conflict detection is performed on each scheduling execution instruction, focusing on identifying issues such as multiple instructions occupying the same resource in the same time period, and timing conflicts between scheduling execution instructions corresponding to high-priority inspection tasks and those corresponding to low-priority inspection tasks. In the event of a conflict, priority is given to ensuring the resource occupancy of scheduling execution instructions corresponding to high-priority inspection tasks, while adjusting the execution time of scheduling execution instructions corresponding to low-priority inspection tasks, or allocating resources to resolve the conflict. Simultaneously, combined with the power grid operating conditions (such as off-peak load and heavy load operation), the optimal verification time window is allocated to each scheduling execution instruction. High-priority abnormal inspection tasks are prioritized for allocation to emergency time windows such as off-peak load to ensure that the safe operation of the power grid is not affected, while medium- and low-priority inspection tasks reasonably avoid the heavy load period of the power grid to improve verification efficiency. By using time-series scheduling, the specific start and end times of each scheduled execution instruction are clearly defined, thereby optimizing resource time allocation and avoiding resource idleness or conflicts.

[0113] For example, the inspection results adopt a standardized document format, including: basic information of each inspection task (name, device, voltage level), scheduling priority, and the operating status of the corresponding target relay equipment; the matching scheduling time window (start time, end time, duration); inspection resource details (configuration and usage requirements of maintenance personnel, testing equipment, and spare parts); inspection task scheduling plan (process, testing items, judgment criteria, and safety measures); time-series scheduling (conflict handling methods and resource allocation logic); and closed-loop archiving requirements (execution records, result feedback, and data upload specifications).

[0114] In this embodiment, timing conflict scheduling avoids resource contention and execution conflicts between different inspection tasks, ensuring the orderly progress of the entire inspection process, further improving the execution efficiency of inspection tasks, and avoiding inspection interruptions or resource misallocation due to unreasonable scheduling.

[0115] In one of the most specific embodiments, such as Figure 3 As shown, the most specific steps of the relay protection device inspection method of this application are as follows:

[0116] Step 1: Obtain the inspection data of the relay protection equipment.

[0117] The relay protection equipment inspection data includes the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task.

[0118] Step 2: Perform feature comparison between the running data and the verification benchmark.

[0119] For each inspection task, features are extracted from the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features. The running features and inspection benchmark features are then sequentially aligned and compared in time series to obtain the feature comparison results corresponding to the inspection task.

[0120] Step 3: Determine the operating status of the target relay protection equipment under each inspection task.

[0121] Based on the feature comparison results, the status assessment score of the inspection task is obtained through a pre-built operation status assessment model; based on the status assessment score, the operation status of the target relay protection equipment under the inspection task is determined.

[0122] Step 4: Determine the inspection task scheduling plan.

[0123] For each inspection task, based on the operating state of the target relay protection equipment under the inspection task, the state space characteristics corresponding to the inspection task are determined. These state space characteristics are then vectorized to obtain the initial feature vector of the inspection task. For each initial feature vector, the correlation degree between the initial feature vector and each inspection task is calculated using a self-attention approach. Based on this correlation degree, the weight of the initial feature vector is determined. Based on the weights of each initial feature vector, feature fusion is performed to obtain the target feature vector. The target feature vector is input into the dynamic scheduling model. By evaluating each inspection task, the scheduling evaluation value of each inspection task in the current inspection environment is obtained. Multiple inspection tasks are sorted according to their scheduling evaluation values ​​from largest to smallest to determine their scheduling priority. For each inspection task, based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task, corresponding inspection resources are matched to the inspection task to obtain the inspection resource allocation result. Based on the scheduling priority and inspection resource allocation result of each inspection task, an inspection task scheduling scheme is generated.

[0124] Step 5: Perform multiple inspection tasks on the target relay protection equipment.

[0125] According to the inspection task scheduling scheme, corresponding scheduling execution instructions are generated for each inspection task; based on the time occupancy status of inspection resources in the current inspection environment, timing conflict scheduling is performed on each scheduling execution instruction to determine the instruction time window for each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window, based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result, the inspection task of the target relay protection equipment is executed to obtain the inspection result of the inspection task; based on the inspection results of each inspection task, the inspection result of the target relay protection equipment is generated.

[0126] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0127] Based on the same inventive concept, this application also provides a relay protection device testing apparatus for implementing the relay protection device testing method described above. The solution provided by this apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the relay protection device testing apparatus provided below can be found in the limitations of the relay protection device testing method described above, and will not be repeated here.

[0128] In one exemplary embodiment, such as Figure 4 As shown, a relay protection equipment testing device 400 is provided, including: an acquisition module 402, a comparison module 404, a determination module 406, and an execution module 408, wherein:

[0129] The acquisition module 402 is used to acquire the operating data of the target relay protection device, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task.

[0130] The comparison module 404 is used to compare the running data and the inspection benchmark corresponding to the inspection task for each inspection task, obtain the feature comparison result corresponding to the inspection task, and determine the operating status of the target relay protection device under the inspection task based on the feature comparison result.

[0131] The determination module 406 is used to determine the inspection task scheduling scheme based on the operating status of the target relay protection equipment under each inspection task through a dynamic scheduling model.

[0132] The execution module 408 is used to perform multiple inspection tasks on the target relay protection equipment according to the inspection task scheduling plan, and obtain the inspection results of the target relay protection equipment.

[0133] In one embodiment, the comparison module 404 is further configured to extract features from the running data and the inspection benchmark corresponding to the inspection task for each inspection task, so as to obtain the corresponding running features and inspection benchmark features; and to perform time-series alignment and time-series feature comparison on the running features and inspection benchmark features in sequence, so as to obtain the feature comparison result corresponding to the inspection task.

[0134] In one embodiment, the comparison module 404 is further configured to obtain the status assessment score of the inspection task based on the feature comparison results and through a pre-built operating status assessment model; and to determine the operating status of the target relay protection device under the inspection task based on the status assessment score.

[0135] In one embodiment, the determining module 406 is further configured to evaluate each inspection task based on the operating status of the target relay protection device under each inspection task using a dynamic scheduling model, and obtain the scheduling evaluation value of each inspection task in the current inspection environment; sort the multiple inspection tasks in descending order of scheduling evaluation values ​​to determine the scheduling priority of each inspection task; for each inspection task, match the corresponding inspection resources for the inspection task based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task, and obtain the inspection resource allocation result of the inspection task; and generate an inspection task scheduling scheme based on the scheduling priority of each inspection task and the inspection resource allocation result.

[0136] In one embodiment, the determining module 406 is further configured to, for each inspection task, determine the state space features corresponding to the inspection task based on the operating state of the target relay protection device under the inspection task, vectorize the state space features to obtain the initial feature vector of the inspection task; for each initial feature vector, calculate the correlation degree between the initial feature vector and each inspection task through self-attention, and determine the weight of the initial feature vector based on the correlation degree; perform feature fusion on each initial feature vector based on the weight of each initial feature vector to obtain the target feature vector; input the target feature vector into the dynamic scheduling model, and obtain the scheduling evaluation value of each inspection task in the current inspection environment by evaluating each inspection task.

[0137] In one embodiment, the execution module 408 is further configured to generate corresponding scheduling execution instructions for each inspection task according to the inspection task scheduling scheme; perform timing conflict scheduling for each scheduling execution instruction according to the time occupancy status of inspection resources in the current inspection environment, and determine the instruction time window for each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, perform the inspection task on the target relay protection device based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result, and obtain the inspection result of the inspection task; and generate the inspection result of the target relay protection device based on the inspection results of each inspection task.

[0138] Each module in the aforementioned relay protection equipment testing device 400 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0139] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores the operating data of the target relay protection device. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a relay protection device testing method.

[0140] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0141] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0142] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0143] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0144] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0145] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0146] In one embodiment, when the processor executes the computer program, it further performs the following steps: for each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features; the running features and inspection benchmark features are sequentially aligned and compared in time to obtain the feature comparison result corresponding to the inspection task.

[0147] In one embodiment, when the processor executes the computer program, it also performs the following steps: based on the feature comparison results, obtains the status assessment score of the inspection task through a pre-built operating status assessment model; and determines the operating status of the target relay protection device under the inspection task based on the status assessment score.

[0148] In one embodiment, when the processor executes the computer program, it further performs the following steps: based on the operating status of the target relay protection device under each inspection task, evaluate each inspection task through a dynamic scheduling model to obtain the scheduling evaluation value of each inspection task in the current inspection environment; sort the multiple inspection tasks in descending order of scheduling evaluation values ​​to determine the scheduling priority of each inspection task; for each inspection task, match the corresponding inspection resources for the inspection task based on the availability status of inspection resources in the current inspection environment and the scheduling priority of the inspection task to obtain the inspection resource allocation result of the inspection task; and generate an inspection task scheduling scheme based on the scheduling priority of each inspection task and the inspection resource allocation result.

[0149] In one embodiment, when the processor executes the computer program, it further implements the following steps: for each inspection task, based on the operating state of the target relay protection device under the inspection task, the state space features corresponding to the inspection task are determined, and the state space features are vectorized to obtain the initial feature vector of the inspection task; for each initial feature vector, based on each initial feature vector, the correlation degree between the initial feature vector and each inspection task is calculated using a self-attention method, and the weight of the initial feature vector is determined based on the correlation degree; based on the weight of each initial feature vector, feature fusion is performed on each initial feature vector to obtain the target feature vector; the target feature vector is input into the dynamic scheduling model, and the scheduling evaluation value of each inspection task in the current inspection environment is obtained by evaluating each inspection task.

[0150] In one embodiment, when the processor executes the computer program, it further performs the following steps: generating corresponding scheduling execution instructions for each inspection task according to the inspection task scheduling scheme; performing timing conflict scheduling for each scheduling execution instruction based on the time occupancy status of inspection resources in the current inspection environment, and determining the instruction time window for each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, executing the inspection task on the target relay protection device based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result, and obtaining the inspection result of the inspection task; and generating the inspection result of the target relay protection device based on the inspection results of each inspection task.

[0151] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0152] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0153] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0154] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0155] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0156] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: for each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features; the running features and inspection benchmark features are sequentially aligned and compared in time to obtain the feature comparison results corresponding to the inspection task.

[0157] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: based on the feature comparison results, obtains the status assessment score of the inspection task through a pre-built operating status assessment model; and determines the operating status of the target relay protection device under the inspection task based on the status assessment score.

[0158] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: based on the operating status of the target relay protection device under each inspection task, evaluate each inspection task through a dynamic scheduling model to obtain the scheduling evaluation value of each inspection task in the current inspection environment; sort the multiple inspection tasks in descending order of scheduling evaluation values ​​to determine the scheduling priority of each inspection task; for each inspection task, match the corresponding inspection resources for the inspection task based on the availability status of inspection resources in the current inspection environment and the scheduling priority of the inspection task to obtain the inspection resource allocation result of the inspection task; and generate an inspection task scheduling scheme based on the scheduling priority of each inspection task and the inspection resource allocation result.

[0159] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: for each inspection task, based on the operating state of the target relay protection device under the inspection task, the state space features corresponding to the inspection task are determined, and the state space features are vectorized to obtain the initial feature vector of the inspection task; for each initial feature vector, based on each initial feature vector, the correlation degree between the initial feature vector and each inspection task is calculated using a self-attention method, and the weight of the initial feature vector is determined based on the correlation degree; based on the weight of each initial feature vector, feature fusion is performed on each initial feature vector to obtain the target feature vector; the target feature vector is input into the dynamic scheduling model, and the scheduling evaluation value of each inspection task in the current inspection environment is obtained by evaluating each inspection task.

[0160] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: generating corresponding scheduling execution instructions for each inspection task according to the inspection task scheduling scheme; performing timing conflict scheduling for each scheduling execution instruction based on the time occupancy status of inspection resources in the current inspection environment, and determining the instruction time window for each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, performing the inspection task on the target relay protection device based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result, and obtaining the inspection result of the inspection task; and generating the inspection result of the target relay protection device based on the inspection results of each inspection task.

[0161] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0162] Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task;

[0163] For each inspection task, the running data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under each inspection task is determined.

[0164] Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model.

[0165] According to the inspection task scheduling plan, multiple inspection tasks are performed on the target relay protection equipment to obtain the inspection results of the target relay protection equipment.

[0166] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: for each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain the corresponding running features and inspection benchmark features; the running features and inspection benchmark features are sequentially aligned and compared in time to obtain the feature comparison results corresponding to the inspection task.

[0167] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: based on the feature comparison results, obtains the status assessment score of the inspection task through a pre-built operating status assessment model; and determines the operating status of the target relay protection device under the inspection task based on the status assessment score.

[0168] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: based on the operating status of the target relay protection device under each inspection task, evaluate each inspection task through a dynamic scheduling model to obtain the scheduling evaluation value of each inspection task in the current inspection environment; sort the multiple inspection tasks in descending order of scheduling evaluation values ​​to determine the scheduling priority of each inspection task; for each inspection task, match the corresponding inspection resources for the inspection task based on the availability status of inspection resources in the current inspection environment and the scheduling priority of the inspection task to obtain the inspection resource allocation result of the inspection task; and generate an inspection task scheduling scheme based on the scheduling priority of each inspection task and the inspection resource allocation result.

[0169] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: for each inspection task, based on the operating state of the target relay protection device under the inspection task, the state space features corresponding to the inspection task are determined, and the state space features are vectorized to obtain the initial feature vector of the inspection task; for each initial feature vector, based on each initial feature vector, the correlation degree between the initial feature vector and each inspection task is calculated using a self-attention method, and the weight of the initial feature vector is determined based on the correlation degree; based on the weight of each initial feature vector, feature fusion is performed on each initial feature vector to obtain the target feature vector; the target feature vector is input into the dynamic scheduling model, and the scheduling evaluation value of each inspection task in the current inspection environment is obtained by evaluating each inspection task.

[0170] In one embodiment, when the computer program is executed by the processor, it further performs the following steps: generating corresponding scheduling execution instructions for each inspection task according to the inspection task scheduling scheme; performing timing conflict scheduling for each scheduling execution instruction based on the time occupancy status of inspection resources in the current inspection environment, and determining the instruction time window for each scheduling execution instruction; for each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, performing the inspection task on the target relay protection device based on the scheduling priority indicated by the scheduling execution instruction and the inspection resource allocation result, and obtaining the inspection result of the inspection task; and generating the inspection result of the target relay protection device based on the inspection results of each inspection task.

[0171] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0172] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0173] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0174] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for inspecting relay protection equipment, characterized in that, The method includes: Acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task; For each inspection task, the operating data and the inspection benchmark corresponding to the inspection task are compared by feature comparison to obtain the feature comparison result corresponding to the inspection task. Based on the feature comparison result, the operating status of the target relay protection device under the inspection task is determined. Based on the operating status of the target relay protection equipment under each inspection task, the inspection task scheduling scheme is determined through a dynamic scheduling model. According to the inspection task scheduling scheme, multiple inspection tasks are performed on the target relay protection device to obtain the inspection results of the target relay protection device.

2. The method according to claim 1, characterized in that, The step of performing feature comparison between the running data and the inspection benchmark corresponding to the inspection task to obtain the feature comparison result corresponding to the inspection task includes: For each inspection task, feature extraction is performed on the running data and the inspection benchmark corresponding to the inspection task to obtain their respective running features and inspection benchmark features. The operational features and the benchmark features are sequentially aligned and compared in time to obtain the feature comparison results corresponding to the testing task.

3. The method according to claim 1, characterized in that, The step of determining the operating status of the target relay protection device under the test task based on the feature comparison results includes: Based on the feature comparison results, the status evaluation score of the inspection task is obtained through a pre-built operational status evaluation model. Based on the status assessment score, the operating status of the target relay protection device under the inspection task is determined.

4. The method according to claim 1, characterized in that, Based on the operating status of the target relay protection equipment under each inspection task, a dynamic scheduling model is used to determine the inspection task scheduling scheme, including: Based on the operating status of the target relay protection equipment under each inspection task, the dynamic scheduling model is used to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment. Multiple inspection tasks are sorted in descending order of their scheduling evaluation values ​​to determine the scheduling priority of each inspection task. For each inspection task, based on the availability of inspection resources in the current inspection environment and the scheduling priority of the inspection task, corresponding inspection resources are matched for the inspection task to obtain the inspection resource allocation result of the inspection task. Based on the scheduling priority of each inspection task and the results of inspection resource allocation, an inspection task scheduling scheme is generated.

5. The method according to claim 4, characterized in that, Based on the operating status of the target relay protection equipment under each inspection task, a dynamic scheduling model is used to evaluate each inspection task and obtain the scheduling evaluation value of each inspection task in the current inspection environment, including: For each inspection task, based on the operating state of the target relay protection device under the inspection task, the state space features corresponding to the inspection task are determined, and the state space features are vectorized to obtain the initial feature vector of the inspection task. For each initial feature vector, the correlation between the initial feature vector and each test task is calculated using a self-attention method, and the weight of the initial feature vector is determined based on the correlation. Based on the weights of each initial feature vector, feature fusion is performed on each initial feature vector to obtain the target feature vector; The target feature vector is input into the dynamic scheduling model, and the scheduling evaluation value of each inspection task in the current inspection environment is obtained by evaluating each inspection task.

6. The method according to claim 1, characterized in that, The step of performing multiple inspection tasks on the target relay protection device according to the inspection task scheduling scheme to obtain the inspection results of the target relay protection device includes: Based on the inspection task scheduling scheme, generate corresponding scheduling execution instructions for each inspection task; Based on the current time occupancy status of inspection resources in the inspection environment, time conflict scheduling is performed on each scheduled execution instruction to determine the instruction time window for each scheduled execution instruction; For each scheduling execution instruction, within the instruction time window of the scheduling execution instruction, based on the scheduling priority indicated by the scheduling execution instruction and the test resource allocation result, the test task is executed on the target relay protection device to obtain the test result of the test task; Based on the inspection results of each inspection task, the inspection results of the target relay protection device are generated.

7. A relay protection equipment testing device, characterized in that, The device includes: The acquisition module is used to acquire the operating data of the target relay protection equipment, multiple inspection tasks, and the inspection benchmarks corresponding to each inspection task. The comparison module is used to perform feature comparison between the running data and the inspection benchmark corresponding to the inspection task for each inspection task, obtain the feature comparison result corresponding to the inspection task, and determine the operating status of the target relay protection device under each inspection task based on the feature comparison result. The determination module is used to determine the inspection task scheduling scheme based on the operating status of the target relay protection equipment under each inspection task through a dynamic scheduling model. The execution module is used to perform multiple inspection tasks on the target relay protection device according to the inspection task scheduling scheme, and obtain the inspection results of the target relay protection device.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images