Control method and device of relay protection device, equipment and medium

Abstract

The invention relates to the technical field of power system relay protection, in particular to a control method and device of a relay protection device, equipment and a medium, and the method comprises the steps: obtaining power grid operation data when the relay protection device triggers a protection action; the operation data are input into an action recognition model, the action recognition model outputs the confidence level of the protection action, the action recognition model adopts an artificial intelligence technology to model behavior modes of the relay protection device, and the behavior modes comprise a normal protection action mode and a maloperation mode; and controlling the relay protection device according to the confidence level and the operation data. Therefore, the problems that the fault is judged and the protection action is triggered by detecting the change of the power grid based on the fixed threshold criterion and the preset logic rule in the related technology, so that misjudgment is easily caused, and the use of the power grid is influenced are solved.

Description

Technical Field

[0001] The present application relates to the technical field of relay protection for power systems, and in particular to a control method, device, equipment, and medium for a relay protection device. Background Art

[0002] With the increasing complexity of power system structures and the increasing proportion of renewable energy integration, relay protection devices face more complex fault identification challenges during actual operation. For example, when faced with short-term disturbances such as instantaneous grounding, lightning disturbances, and arc flashovers, traditional protection devices may mistakenly identify them as permanent faults and trip, resulting in "false tripping."

[0003] In related technologies, traditional protection devices lack coordination with backup protection and dispatching systems, which can easily create protection blind spots or amplify the impact of faults. Relay protection systems primarily rely on fixed threshold criteria and preset logic rules, detecting changes in the amplitude or phase of electrical quantities such as current and voltage to identify faults and trigger protective actions. For example, overcurrent protection uses set current thresholds to identify short-circuit faults, and differential protection determines the fault location by comparing the current difference between the two ends of the line. While these methods have a certain degree of reliability in conventional power grids, their limitations are becoming increasingly apparent with the expansion of power systems and the high proportion of renewable energy (such as photovoltaic and wind power). Specifically, traditional methods have difficulty distinguishing true faults from transient disturbances (such as lightning strikes and capacitor switching) or complex harmonic interference. For example, harmonics generated by renewable energy inverters can cause current distortion, which can be misjudged as a short-circuit fault and cause false tripping. Parameter setting of existing protection devices relies on manual experience and cannot dynamically adapt to changes in grid topology (such as renewable energy integration and load fluctuations) or equipment aging (such as decreased relay sensitivity). Traditional systems are mostly open-loop control systems, lacking real-time feedback and self-optimization mechanisms, and are unable to optimize strategies based on historical action data. Therefore, detecting changes in the power grid to determine faults and trigger protection actions can easily lead to misjudgments and affect the use of the power grid; most of them rely on fixed setting values ​​and protection logic, lack the ability to learn historical data and dynamic operating conditions, and lack adaptive adjustment and fault-tolerant mechanisms for false action scenarios. Summary of the Invention

[0004] The present application provides a control method, device, equipment and medium for a relay protection device to solve the problems in the related art of judging faults and triggering protection actions by detecting changes in the power grid based on fixed threshold criteria and preset logical rules, which can easily lead to misjudgment and affect the use of the power grid.

[0005] The first aspect of the present application provides a control method for a relay protection device, comprising the following steps: obtaining grid operation data when the relay protection device triggers a protection action; inputting the operation data into an action recognition model, and the action recognition model outputs a confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false operation mode; and controlling the relay protection device according to the confidence level and the operation data.

[0006] Optionally, before inputting the operating data into the action recognition model, it includes: obtaining training data carrying real labels of protection actions, wherein the training data includes: historical protection action records, power grid operation data when the relay protection device triggers protection actions, and protection action recognition results; training and optimizing the action recognition model according to the training data until the action recognition model converges.

[0007] Optionally, the action recognition model includes: an input layer, a preprocessing layer, a feature extraction layer, a deep learning layer, a fusion layer and an output layer, wherein the input layer is used to input the operating data when the relay protection device triggers the protection action; the preprocessing layer is used to filter, normalize, align events and perform sliding window processing on the operating data; the feature extraction layer is used to extract the time domain features and frequency domain features of the preprocessed operating data; the deep learning layer includes a convolutional neural network and a long short-term memory network, wherein the convolutional neural network is used to process the recorded waveform data to generate spatial features, and the long short-term memory network is used to process the operating data to generate temporal features; the fusion layer is used to fuse the spatial features and the temporal features to generate a joint feature vector, and dynamically allocate the weight of the joint feature vector using the attention weighting mechanism; the output layer is used to output the action classification confidence based on the joint feature vector to generate a confidence level of the protection action.

[0008] Optionally, controlling the relay protection device according to the confidence level and the operating data includes: determining the risk of malfunction of the relay protection device according to the confidence level; if the risk of malfunction of the relay protection device meets the error correction condition, correcting the protection action according to the operating data.

[0009] Optionally, determining the risk of malfunction of the relay protection device based on the confidence level includes: if the confidence level is higher than a first preset threshold, the risk of malfunction of the relay protection device is low risk; if the confidence level is higher than a second preset threshold and lower than the first threshold, the risk of malfunction of the relay protection device is medium risk; if the confidence level is lower than the second preset threshold, the risk of malfunction of the relay protection device is high risk, wherein the first preset threshold is higher than the second preset threshold.

[0010] Optionally, the correction of the protection action includes one or more of: delayed confirmation, multi-source verification, backup protection triggering, over-tripping, and manual confirmation intervention.

[0011] Optionally, the error correction condition includes: the risk of malfunction of the relay protection device is medium or high risk.

[0012] The second aspect of the present application provides a control device for a relay protection device, including: an acquisition module for acquiring grid operation data when the relay protection device triggers a protection action; a processing module for inputting the operation data into an action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false operation mode; a control module for controlling the relay protection device according to the confidence level and the operation data.

[0013] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to execute the control method of the relay protection device as described in the above embodiment.

[0014] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to perform the control method of the relay protection device as described in the above embodiment.

[0015] Therefore, this application has at least the following beneficial effects:

[0016] The embodiment of the present application can obtain the power grid operation data when the relay protection device triggers a protection action; the operation data is input into the action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false operation mode; the relay protection device is controlled according to the confidence level and the operation data, and the artificial intelligence model can efficiently distinguish between normal protection action and false operation mode. The system can accurately identify the false operation risk caused by non-fault factors such as harmonic interference and transient disturbances, improve the fault identification accuracy, and dynamically optimize the model and closed-loop control in real time to improve the stability of the smart grid.

[0017] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0019] Figure 1 This is a flow chart of a control method for a relay protection device provided according to an embodiment of the present application;

[0020] Figure 2 A flowchart of a work flow according to an embodiment of the present application is provided;

[0021] Figure 3 This is an internal logic diagram of a behavior recognition model module provided according to an embodiment of the present application;

[0022] Figure 4 This is an example diagram of a control device of a relay protection device provided according to an embodiment of the present application;

[0023] Figure 5 This is a diagram of the overall system architecture provided according to an embodiment of the present application;

[0024] Figure 6 A schematic diagram of the structure of a system according to an embodiment of the present application;

[0025] Figure 7 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0026] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0027] The following describes the control method, device, equipment and medium of the relay protection device according to the embodiment of the present application with reference to the accompanying drawings.

[0028] Specifically, Figure 1 A flow chart of a control method for a relay protection device provided in an embodiment of the present application.

[0029] like Figure 1 As shown, the control method of the relay protection device includes the following steps:

[0030] In step S101, the grid operation data when the relay protection device triggers the protection action is obtained.

[0031] It is understandable that the embodiments of the present application can obtain the power grid operation data when the relay protection device triggers the protection action, so as to facilitate the subsequent identification of the confidence level of the protection action.

[0032] It should be noted that the power grid operation data of this application includes electrical quantity data, recorded data and protection action information in the power system. The electrical quantity data includes current, voltage, frequency, active power, reactive power and phase angle. The recorded data includes waveform data, duration and frequency domain characteristics. The protection action information includes device action signals, etc.

[0033] Specifically, current and voltage: measure the current and voltage levels of different nodes or lines to analyze whether there is overload or short circuit; frequency: monitor the frequency stability of the power system, abnormal frequency changes may be an indication of a fault; active power and reactive power: understand the energy flow of the power grid, which helps to judge the load status and potential imbalance problems; phase angle: monitor the angle difference between each phase, which is very important for detecting asymmetric faults; device action signal: record the action of the relay protection device, such as tripping commands, switch status, etc., to help determine the specific protection behavior; waveform data: capture the voltage and current waveforms before and after the fault occurs, and provide detailed time series data for subsequent analysis; duration: record the duration of the fault event, distinguish between transient disturbances and permanent faults; frequency domain features: extract harmonic components and other frequency domain characteristics in the waveform for more in-depth fault analysis, etc.

[0034] In step S102, the operation data is input into the action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false action mode.

[0035] Among them, such as Figure 3 As shown, the action recognition model includes: an input layer, a preprocessing layer, a feature extraction layer, a deep learning layer, a fusion layer and an output layer, wherein the input layer is used to input the operating data when the relay protection device triggers the protection action; the preprocessing layer is used to filter, normalize, align events and perform sliding window processing on the operating data; the feature extraction layer is used to extract the time domain features and frequency domain features of the preprocessed operating data; the deep learning layer includes a convolutional neural network and a long short-term memory network, wherein the convolutional neural network is used to process the recorded waveform data to generate spatial features, and the long short-term memory network is used to process the operating data to generate temporal features; the fusion layer is used to fuse the spatial features and temporal features to generate a joint feature vector, and dynamically allocate the weight of the joint feature vector using the attention weighting mechanism; the output layer is used to output the action classification confidence according to the joint feature vector to generate the confidence level of the protection action.

[0036] It can be understood that the embodiments of the present application can more accurately identify complex fault modes and distinguish normal operations from malfunctions by adopting action recognition models, significantly reduce the false alarm rate, complete analysis and provide results in a very short time, and ensure the ability to respond quickly to abnormal conditions in the power grid.

[0037] It should be noted that the normal protection action mode refers to the expected response behavior of the relay protection device when facing a real fault event (such as short circuit, grounding, overload, etc.); the false operation mode refers to the situation where the relay protection device erroneously performs tripping or other protection actions when there is no actual fault or the fault does not reach the protection setting value.

[0038] The action recognition model is based on machine learning models (such as random forest, convolutional neural network CNN, long short-term memory network LSTM, etc.), which is used to analyze the action behavior of relay protection devices in real time and determine whether there is a risk of false operation.

[0039] The model captures the differences between normal protection operation and false operation modes by analyzing multi-dimensional data such as time domain (current / voltage change rate, trip point) and frequency domain (harmonic energy distribution, transient characteristics). For example, the current surge slope of a short-circuit fault and the transient fluctuation caused by electromagnetic interference have significantly different waveform characteristics.

[0040] In an embodiment of the present application, before inputting the operating data into the action recognition model, it includes: obtaining training data carrying real labels of protection actions, wherein the training data includes: historical protection action records, power grid operation data when the relay protection device triggers the protection action, and protection action recognition results; training and optimizing the action recognition model according to the training data until the action recognition model converges.

[0041] It can be understood that the embodiments of the present application can obtain training data carrying real labels of protective actions, and train and optimize the action recognition model based on the training data until the action recognition model converges, so as to improve the recognition accuracy of the model and improve the robustness of the model.

[0042] Specifically, training data includes historical protection action records, grid operation data, and protection action recognition results. Historical protection action records include detailed logs of normal actions (such as proper tripping and blocking) and malfunction cases (such as false tripping and delayed action), annotated with the action type, timestamp, and fault cause. Grid operation data includes multi-dimensional data from when relay protection devices trigger actions, including: electrical quantities: three-phase current, voltage, frequency, phase angle, and zero-sequence component; recorded waveforms: high-precision time-domain waveforms before and after the fault; protection signals: trip commands, switch status, and device self-test status; and protection action recognition results: action effectiveness assessments from expert systems or historical operation and maintenance reports. Data preprocessing includes filtering and denoising, normalization, sliding window and event alignment, and more. Model training: Training data is input into the action recognition model, which outputs a confidence level for the protection action. A loss function is calculated based on the true labels of the training data and the output confidence level. If the loss function value exceeds the target threshold, the parameters of the action recognition model are optimized until the action recognition model converges. The behavioral model parameters are dynamically updated based on new protection action data and fault analysis results in actual applications, so that the identification and error correction strategies can continuously evolve and the training data can be continuously updated to achieve online continuous training and optimization of the model.

[0043] In step S103 , the relay protection device is controlled according to the confidence level and the operating data.

[0044] It is understood that the embodiments of the present application can control the relay protection device according to the confidence level and operating data to ensure that the power system can make accurate and timely responses when facing complex faults or disturbances.

[0045] In an embodiment of the present application, the relay protection device is controlled according to the confidence level and operating data, including: determining the risk of malfunction of the relay protection device according to the confidence level; if the risk of malfunction of the relay protection device meets the error correction condition, correcting the protection action according to the operating data.

[0046] Corrective protection actions include one or more of: delayed confirmation, multi-source verification, backup protection triggering, over-tripping, and manual confirmation intervention. The risk of relay protection device malfunction is medium to high.

[0047] It can be understood that the embodiments of the present application dynamically distinguish between false operations and real faults through confidence level control and multi-source data-driven error correction strategies, reduce the misjudgment rate, and quickly respond to closed-loop control to ensure power grid safety.

[0048] For example, the confidence level is mapped to the risk of false activation:

[0049] (1) High confidence (≥0.8): Risk determination: Clearly identify a false action (e.g., false tripping, over-tripping) or a correct action (reasonable tripping). Correction conditions: If it is a false action, immediately block the tripping; if it is a correct action, execute normally. For example: If the model determines "false tripping" (confidence 0.9), the system directly blocks the tripping command.

[0050] (2) Medium confidence (0.5-0.8): Risk assessment: suspected malfunction or fuzzy action logic (such as transient disturbances or partial data anomalies). Error correction conditions: Trigger multi-source verification (PMU (Phasor Measurement Unit) phase, SCADA (Supervisory Control and Data Acquisition) topology) + delayed confirmation (1-2 power frequency cycles). Example: A lightning transient fault is judged as a "false trip" (confidence level 0.7). The system delays 40ms and cancels the trip after rechecking the PMU data.

[0051] (3) Low confidence (<0.5): Risk assessment: Poor data quality (e.g., communication interruption) or new, unknown fault. Error correction conditions: Mandatory manual intervention + device self-test (e.g., sensor health check). Example: Data anomaly results in a confidence level of 0.3, an alarm is pushed to the dispatch center, and maintenance is performed after manual confirmation.

[0052] In an embodiment of the present application, the risk of malfunction of the relay protection device is determined according to the confidence level, including: if the confidence level is higher than the first preset threshold, the risk of malfunction of the relay protection device is low risk; if the confidence level is higher than the second preset threshold and lower than the first threshold, the risk of malfunction of the relay protection device is medium risk; if the confidence level is lower than the second preset threshold, the risk of malfunction of the relay protection device is high risk, wherein the first preset threshold is higher than the second preset threshold.

[0053] The first preset threshold and the second preset threshold can be set according to actual needs and are not specifically limited.

[0054] It can be understood that the embodiments of the present application can determine the risk of false operation of the relay protection device based on the confidence level to control the subsequent protection strategy, dynamically distinguish between false operation and real faults, reduce the misjudgment rate, and quickly respond to closed-loop control to ensure the safety of the power grid.

[0055] For example, the first-level strategy (high-risk false operation): immediate blocking tripping: sending a locking instruction to the protection device and recording the event; over-tripping: if the main protection fails and the fault may spread, triggering the upper-level circuit breaker to operate (master station authorization is required). The second-level strategy (medium-risk false operation): backup protection linkage: calling differential protection or distance protection to review fault characteristics; delayed re-evaluation: waiting for 1 to 2 power frequency cycles and then re-analyzing the data to confirm whether the fault persists. The third-level strategy (low-risk uncertainty): early warning and manual intervention: pushing detailed data (recorded waveform, phase information) to the operation and maintenance interface; equipment self-test: automatically performing circuit breaker mechanical characteristic test or sensor calibration, without specific restrictions.

[0056] According to the control method of the relay protection device proposed in the embodiment of the present application, the grid operation data when the relay protection device triggers the protection action is obtained; the operation data is input into the action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false operation mode; the relay protection device is controlled according to the confidence level and the operation data, and the artificial intelligence model can efficiently distinguish between normal protection action and false operation mode. The system can accurately identify the false operation risk caused by non-fault factors such as harmonic interference and transient disturbances, improve the fault identification accuracy, and dynamically optimize the model and closed-loop control in real time to improve the stability of the smart grid.

[0057] The following will be combined Figure 2 The control method of the relay protection device of the present application is described in detail with specific embodiments, and the specific steps are as follows:

[0058] Step 1: Real-time data collection

[0059] The grid operation data when the relay protection device triggers the protection action includes: electrical quantity data: current, voltage, frequency, phase angle, zero-sequence component; recorded data: high-precision waveforms before and after the fault; protection signals: trip command, switch status, and device self-test information.

[0060] Step 2: Behavior recognition (based on behavior recognition model)

[0061] Grid operation data is input into the behavior recognition model, which extracts time-domain and frequency-domain features. The convolutional neural network in the deep learning model is used to process the recorded waveform and capture spatial features. The long short-term memory network analyzes time series data and models dynamic evolution. The CNN and LSTM features are dynamically weighted and fused to generate action classification probabilities (normal action, malfunction type) and confidence levels (0-1).

[0062] Step 3: Misoperation judgment

[0063] High-risk false trip: Confidence level ≥ 0.8 and features matching historical false trip patterns (e.g., harmonic interference, electromagnetic transients);

[0064] Medium-risk malfunction: Confidence level 0.5-0.8, requiring secondary verification based on multi-source data (PMU phase, SCADA topology);

[0065] Low-risk uncertainty: Confidence level < 0.5, triggering manual review and device self-check.

[0066] Step 4, error correction control module

[0067] Delayed tripping: Applicable scenarios: transient faults (such as lightning strikes) and non-permanent disturbances; Implementation: Suspend the trip command for 1 to 2 power frequency cycles (20 to 40ms) and wait for data re-evaluation.

[0068] Backup protection linkage: Applicable scenarios: medium-risk malfunction or main protection failure; Implementation: Call differential protection and directional protection to cross-verify fault characteristics.

[0069] Alarm / manual confirmation: Applicable scenarios: low confidence or data anomalies (such as communication interruption); Implementation: push alarms to the dispatch console, synchronize recorded waveforms and PMU data for manual decision-making.

[0070] Step 5: Feedback optimization behavior recognition model

[0071] Using historical malfunction samples, operational feedback, and accident reports as training sets, we continuously optimize the behavior recognition model to achieve self-learning and self-evolution.

[0072] Step 6: Interact with the dispatching system, recording system, PMU, and other protection devices to achieve fault information sharing and action coordination.

[0073] Example 1: When a line experiences a sudden current change due to a lightning strike and a brief grounding, protection device A detects that the current exceeds a set threshold and is about to trip. The behavior recognition model analyzes the short duration of the short circuit, the sudden drop in the zero-sequence component, and the subsequent stable recovery indicated by the waveform recording. This identifies the fault as a "non-permanent transient fault" and indicates the risk of a false trip. The system initiates a "delayed confirmation + backup check" strategy, temporarily suspending the primary protection trip until the line recovers, effectively preventing false trips.

[0074] Example 2: Lightning Disturbance Scenario of a 220kV Transmission Line

[0075] 1. The system detects that the line current rises instantaneously and exceeds the setting threshold, and protection device A is ready to trip;

[0076] 2. The behavior recognition model analyzed the recorded data and found that the fault lasted only 100ms, and the voltage and current subsequently returned to normal, with no signs of a persistent short circuit;

[0077] 3. The model outputs a "non-permanent fault + high risk of malfunction" label, and the malfunction identification module confirms that the behavior has deviated;

[0078] 4. The error correction control module starts with a delay + backup protection confirmation strategy to prevent the main protection from tripping directly;

[0079] 5. Automatic fault recovery without power outage, realizing "self-rescue" protection;

[0080] 6. The events are stored in the system learning module to further optimize the false operation judgment model.

[0081] Next, the control device of the relay protection device proposed in accordance with the embodiment of the present application will be described with reference to the accompanying drawings.

[0082] Figure 4 It is a block diagram of a control device of a relay protection device according to an embodiment of the present application.

[0083] like Figure 4 As shown, the control device 10 of the relay protection device includes: an acquisition module 100 , a processing module 200 and a control module 300 .

[0084] Among them, the acquisition module 100 is used to obtain the power grid operation data when the relay protection device triggers the protection action; the processing module 200 is used to input the operation data into the action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes normal protection action mode and false operation mode; the control module 300 is used to control the relay protection device according to the confidence level and operation data.

[0085] It should be noted that the above explanation of the embodiment of the control method of the relay protection device is also applicable to the control device of the relay protection device of this embodiment, and will not be repeated here.

[0086] According to the control device of the relay protection device proposed in the embodiment of the present application, the grid operation data when the relay protection device triggers the protection action is obtained; the operation data is input into the action recognition model, and the action recognition model outputs the confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model the behavior mode of the relay protection device, and the behavior mode includes a normal protection action mode and a false operation mode; the relay protection device is controlled according to the confidence level and the operation data, and the abnormal action behavior of the relay protection device under a special operating state can be identified, and the protection action mode can be dynamically adjusted according to the predefined strategy to prevent the occurrence of problems such as false tripping, false switching, and refusal to operate, thereby improving the intelligence and fault tolerance of the protection system.

[0087] The following will be combined Figure 5-6The control system of the relay protection device of this application is described in detail as follows:

[0088] The control system of the relay protection device includes a data acquisition module, a behavior recognition model module, a malfunction judgment module, an error correction control module, an online learning module and a communication interface module.

[0089] Data acquisition module: used to collect key operating parameters of the power grid in real time, including current, voltage, frequency, active power, reactive power, phase angle, recorded data, device action signals, etc. by integrating multi-source data such as electrical quantities, action records, and recorded data.

[0090] Behavior Recognition Model Module: This module, based on machine learning models (such as random forests, convolutional neural networks (CNNs), and long-short-term memory (LSTMs), analyzes the behavior of relay protection devices in real time to determine whether there is a risk of misoperation. The model is trained using historical action data to form a library of protection behavior labels and then calculates the matching degree based on real-time data.

[0091] Misoperation identification module: This module receives the output of the behavior recognition model and makes intelligent judgments on protection actions. If the system identifies that the current protection behavior does not match the actual fault characteristics, it will be judged as having a "high risk of misoperation."

[0092] Error correction control module: When a high risk of malfunction is detected, the system initiates an error correction control strategy based on the built-in rule base, including but not limited to the following methods:

[0093] (1) Delayed confirmation action;

[0094] (2) Cross-check with backup protection devices;

[0095] (3) Strengthen the location of the fault section and prevent tripping after confirming the fault area;

[0096] (3) Prompt for manual review intervention (optional);

[0097] (4) Issue a tripping command when necessary to prevent the system from expanding the scope of the fault.

[0098] Online learning module: supports continuous online model training and optimization, dynamically updates behavioral model parameters based on new protection action data and fault analysis results, and continuously evolves identification and error correction strategies.

[0099] Communication interface module: used to realize real-time data interaction with the master station dispatching system, fault recording system, PMU device, SCADA system, etc. to ensure the coordinated operation of the system.

[0100] The system leverages artificial intelligence technologies such as deep learning to identify in real time whether relay protection devices are at risk of false tripping. If a false trip is detected, it automatically implements error correction strategies such as delayed confirmation, backup verification, and over-tripping to avoid unnecessary tripping and improve the system's fault tolerance and stability. The system also features online learning capabilities, continuously optimizing protection strategies to adapt to different scenarios and operating environments. This system can effectively reduce problems such as false relay protection tripping, enhance the intelligence and safety of power grid operations, and is suitable for various transmission and distribution systems, as well as power supply systems for key users.

[0101] In summary, this application uses a behavior recognition model to conduct an in-depth analysis of the rationality of relay protection actions. This allows for early identification of actions with the risk of false operation and dynamic adjustment of action strategies, thereby:

[0102] (1) Reduce the false tripping rate and reduce unnecessary power outages in non-fault sections;

[0103] (2) Avoid the risk of accidents caused by protection refusal to operate;

[0104] (3) Redundant protection is formed through leapfrogging control and backup strategies;

[0105] (4) Enhance the adaptability of the protection system to special fault scenarios such as new energy and power electronic disturbances;

[0106] (5) It realizes the transformation of relay protection system from “static setting” to “dynamic intelligence”, which has extensive engineering application value.

[0107] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0108] A memory 701 , a processor 702 , and a computer program stored in the memory 701 and executable on the processor 702 .

[0109] When the processor 702 executes the program, the control method of the relay protection device provided in the above embodiment is implemented.

[0110] Furthermore, the electronic device further includes:

[0111] The communication interface 703 is used for communication between the memory 701 and the processor 702 .

[0112] The memory 701 is used to store computer programs that can be run on the processor 702 .

[0113] The memory 701 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0114] If the memory 701, processor 702, and communication interface 703 are implemented independently, the communication interface 703, memory 701, and processor 702 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0115] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can communicate with each other through an internal interface.

[0116] The processor 702 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0117] An embodiment of the present application further provides a computer-readable storage medium having a computer program or instruction stored thereon. When the computer program or instruction is executed by a processor, the control method of the relay protection device as described above is implemented.

[0118] An embodiment of the present application also provides a computer program product, including a computer program or instructions, which, when executed, implements the control method of the relay protection device as described above.

[0119] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0120] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0121] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0122] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using any one or a combination of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0123] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

Claims

1. A control method for a relay protection device, characterized in that: The following steps are involved: Obtaining grid operation data when the relay protection device triggers protection action; Inputting the operating data into an action recognition model, the action recognition model outputting a confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model a behavior mode of the relay protection device, the behavior mode including a normal protection action mode and a malfunction mode; The relay protection device is controlled according to the confidence level and the operating data.

2. The control method of the relay protection device according to claim 1, characterized in that: Before inputting the running data into the action recognition model, the method includes: Obtain training data with real labels of protection actions, where the training data includes: historical protection action records, grid operation data when the relay protection device triggers the protection action, and protection action recognition results; The action recognition model is trained and optimized according to the training data until the action recognition model converges.

3. The control method of the relay protection device according to claim 1 or 2, characterized in that: The action recognition model includes: an input layer, a preprocessing layer, a feature extraction layer, a deep learning layer, a fusion layer and an output layer, wherein: The input layer is used to input the operating data when the relay protection device triggers the protection action; The preprocessing layer is used to perform filtering, normalization, event alignment and sliding window processing on the operating data; The feature extraction layer is used to extract the time domain features and frequency domain features of the pre-processed operating data; The deep learning layer includes a convolutional neural network and a long short-term memory network, wherein the convolutional neural network is used to process the recorded waveform data to generate spatial features, and the long short-term memory network is used to process the running data to generate temporal features; The fusion layer is used to fuse the spatial features and the temporal features to generate a joint feature vector, and dynamically assign the weight of the joint feature vector using an attention weighting mechanism; The output layer is used to generate a confidence level of the protection action according to the action classification confidence output by the joint feature vector.

4. The control method of the relay protection device according to claim 1, characterized in that: The controlling the relay protection device according to the confidence level and the operating data includes: determining the risk of malfunction of the relay protection device according to the confidence level; If the malfunction risk of the relay protection device meets the error correction condition, the protection action is corrected according to the operation data.

5. The control method of the relay protection device according to claim 4, characterized in that: Determining the malfunction risk of the relay protection device according to the confidence level includes: If the confidence level is higher than the first preset threshold, the risk of malfunction of the relay protection device is low; If the confidence level is higher than the second preset threshold and lower than the first threshold, the risk of malfunction of the relay protection device is medium risk; If the confidence level is lower than a second preset threshold, the risk of malfunction of the relay protection device is high, wherein the first preset threshold is higher than the second preset threshold.

6. The control method of the relay protection device according to claim 4, characterized in that: The correction of the protection action includes one or more of: delayed confirmation, multi-source verification, backup protection triggering, over-tripping, and manual confirmation intervention.

7. The control method of the relay protection device according to claim 4, characterized in that: The error correction condition includes: the risk of malfunction of the relay protection device is medium or high risk.

8. A control device for a relay protection device, characterized in that: include: An acquisition module is used to obtain the power grid operation data when the relay protection device triggers the protection action; a processing module, configured to input the operating data into an action recognition model, wherein the action recognition model outputs a confidence level of the protection action, wherein the action recognition model uses artificial intelligence technology to model a behavior mode of the relay protection device, wherein the behavior mode includes a normal protection action mode and a malfunction mode; A control module is used to control the relay protection device according to the confidence level and the operating data.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the control method for the relay protection device according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the control method of the relay protection device according to any one of claims 1 to 7.

Images