Method and system for automatic testing of substation relay protection functions

Abstract

The application discloses a kind of substation relay protection function automatic test method and system, it is related to substation relay protection function automatic test field, comprising: the knowledge base of relay protection is constructed, and standard relay protection function detection contrast template is formulated;Setting simulation multi-type pulse self-checking device and relay protection action logic signal acquisition device, construct adaptive automatic detection mechanism;Using professional detection technology to detect relay protection function, construct static detection mechanism;Based on LSTM neural network, the internal action logic output result of relay protection device under the influence of test signal is optimized;Based on static detection mechanism, adaptive automatic detection mechanism and the stability of the function of adaptive automatic detection mechanism is comprehensively evaluated based on optimized historical data.The application has the advantages that: adaptive automatic detection mechanism is set, the online automatic test of substation relay protection function is realized, and the automatic detection efficiency and detection accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of automatic testing of substation relay protection functions, specifically to a method and system for automatic testing of substation relay protection functions. Background Technology

[0002] In modern power systems, substations serve as crucial hubs for power transmission and distribution, making the reliability and accuracy of their relay protection devices paramount. These devices must act quickly to isolate faulty components and ensure stable operation of the power system during power system failures. However, with the continuous development of power systems, substations are expanding in scale, and equipment types and operating conditions are becoming increasingly complex. On the one hand, the large number of nonlinear loads, new energy power sources connected to the grid, and the influence of the substation environment cause distortions in the voltage and current waveforms of the grid, interfering with the normal operation of relay protection devices and increasing the risk of maloperation and failure to operate. On the other hand, traditional relay protection detection methods are mostly periodic offline detections, which are insufficient to meet real-time requirements and cannot promptly detect potential faults that may occur during the operation of relay protection devices.

[0003] Therefore, realizing online automatic detection of substation relay protection functions has become a key requirement for ensuring the safe and stable operation of the power system. At the same time, using advanced algorithms and technologies to optimize and correct the detection results can further improve the reliability and performance of relay protection devices.

[0004] Traditional relay protection testing relies heavily on manual periodic inspections and offline testing, which cannot monitor the operating status of relay protection devices in real time. If a fault occurs during the interval between two tests, it may not be detected and dealt with in time, leading to the expansion of the fault range and affecting the stable operation of the power system. Secondly, existing online automatic detection technology for relay protection functions mainly relies on electrical quantity monitoring to provide feedback on the operating status of relay protection devices. It lacks comprehensive consideration of the impact of the environment and equipment load on relay protection devices, resulting in inaccurate monitoring results, which in turn leads to system misjudgments and waste of resources. Summary of the Invention

[0005] To address the aforementioned technical problems, this paper provides a method and system for automatic testing of substation relay protection functions. This technical solution solves the problems mentioned in the background technology, which rely heavily on manual periodic inspections and offline testing, failing to monitor the operating status of relay protection devices in real time. If a fault occurs in the device during the interval between two tests, it may not be detected and dealt with in time, leading to the expansion of the fault range and affecting the stable operation of the power system. Furthermore, the lack of comprehensive consideration of the impact of the environment and equipment load on relay protection devices results in inaccurate monitoring results, leading to system misjudgments and wasting resources.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for automatically testing the relay protection function of a substation includes:

[0008] Based on the testing requirements of substation relay protection functions, a knowledge base for relay protection is constructed, and a standard test template for relay protection functions is developed.

[0009] Set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and dynamically adjust parameters according to the substation equipment load to build an adaptive automatic detection mechanism;

[0010] Based on the feedback of the adaptive automatic detection mechanism, abnormal signals are extracted, and professional detection technology is used to specifically detect the relay protection function to build a static detection mechanism.

[0011] Based on the LSTM neural network algorithm, the output results of the internal action logic of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals are corrected and optimized.

[0012] The stability of the adaptive automatic detection mechanism is comprehensively evaluated based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data.

[0013] Preferably, the self-testing device simulating multiple types of pulses and the relay protection action logic signal acquisition device, which dynamically adjust parameters based on the substation equipment load to construct an adaptive automatic detection mechanism, specifically includes:

[0014] In the secondary circuit of the relay protection device, a self-testing device simulating multiple types of pulses is set up, and different types of pulse test signals are superimposed on the signal under test through isolation devices;

[0015] Set up a relay protection action logic signal acquisition device to collect the internal action logic output of the relay protection device under the influence of test signals, and present it through indicator lights and data groups;

[0016] The system acquires dynamic load adjustment parameter data of substation equipment and electromagnetic interference data around relay protection devices, and then filters and normalizes the data using Kalman filtering algorithm and normalization formula.

[0017] Based on the dynamic adjustment of substation equipment load and electromagnetic interference around relay protection devices, the load fluctuation characteristics and electromagnetic interference characteristics affecting the detection of pulse test signals are extracted respectively.

[0018] The load fluctuation characteristics include: voltage, current, active / reactive power, load rate, and harmonic components; the electromagnetic interference characteristics include: interference amplitude, frequency distribution, pulse width, and frequency of occurrence.

[0019] The load fluctuation characteristics and electromagnetic interference characteristics are used as inputs, and the pulse test signal trigger time step is used as the output.

[0020] With the goal of minimizing pulse test signal fluctuations, a loss function for stabilizing pulse test signals is constructed based on the mean square error formula.

[0021] Based on historical data of relay protection function detection and / or by conducting test experiments, training sets, test sets, and validation sets are constructed for training neural network algorithms, respectively.

[0022] Based on the collected dynamic adjustment parameters of substation equipment load and electromagnetic interference data around the relay protection device, combined with the trained neural network model, the pulse test signal trigger time step is output.

[0023] Based on the frequency requirements of automatic testing of substation relay protection functions, a pulse test signal trigger time reference value is set, and a time step is superimposed to determine the adaptive trigger time interval of the pulse test signal.

[0024] Preferably, the step of correcting and optimizing the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals based on the LSTM neural network algorithm specifically includes:

[0025] Historical data from adaptive automatic detection mechanisms, static detection mechanisms, and environmental monitoring are acquired, and the data are filtered and normalized using filtering algorithms and normalization formulas.

[0026] Based on historical data from the adaptive automatic detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the theoretical relay protection function detection and the comparison template is calculated and denoted as . ;

[0027] Based on historical data from the static detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the actual relay protection function detection and the comparison template is calculated and denoted as . ;

[0028] by With the goal of minimizing the error, a loss function for the detection distortion rate of relay protection function is constructed based on the mean square error formula.

[0029] Based on environmental monitoring data, an environmental loss function for relay protection function testing is constructed with the goal of minimizing the error of environmental parameters between theoretical and actual relay protection function testing.

[0030] Based on the loss function of the distortion rate of relay protection function detection and the environmental loss function during relay protection function detection, a comprehensive loss function for relay protection function detection is constructed.

[0031] In the hidden state, constraints are set to store historical environmental disturbance features and standard data as the cell state.

[0032] Based on the LSTM neural network algorithm and combined with the detection data of the adaptive automatic detection mechanism, the output results of the internal action logic of the relay protection device under the influence of the test signal in the adaptive automatic detection mechanism are corrected and optimized.

[0033] Preferably, the comprehensive evaluation of the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data specifically includes:

[0034] The static detection mechanism, the adaptive automatic detection mechanism, and the optimized historical data are obtained, and the data are normalized using a normalization formula.

[0035] The heterogeneity rates of static detection mechanism, adaptive automatic detection mechanism and optimized historical data and control templates were obtained respectively;

[0036] Based on the static detection mechanism, the adaptive automatic detection mechanism, and the optimized alienation rate, a stability model of the adaptive automatic detection mechanism is established.

[0037] Based on the stability model of the adaptive automatic detection mechanism, the stability of the adaptive automatic detection mechanism is comprehensively evaluated.

[0038] The stability model expression for the adaptive automatic detection mechanism is: In the formula, for The stability evaluation value of the adaptive automatic detection mechanism function at that time. This represents the number of static detection mechanisms performed within a fixed time window. To achieve the first within a fixed time window The deviation rate between the theoretical relay protection function test and the control template during the first static detection mechanism test. To achieve the first within a fixed time window The theoretical relay protection function test and the deviation rate after template optimization during the static detection mechanism test are compared. To achieve the first within a fixed time window The difference rate between the actual relay protection function test and the control template under the sub-static testing mechanism. , The weighting coefficients can be obtained using the least squares method. This represents the average of the adaptively automatically detected alienation rate within a fixed time window. The variance of the adaptive automatic detection of the alienation rate within a fixed time window. The number of adaptive automatic detections within a fixed time window. , For preset fixed constant terms, This is the variance threshold.

[0039] Furthermore, this solution proposes a system for automatic testing of substation relay protection functions, used to implement the method for automatic testing of substation relay protection functions as described above, including:

[0040] The knowledge base module is used to construct a knowledge base for relay protection based on the testing requirements of substation relay protection functions and to formulate standard relay protection function testing comparison templates.

[0041] The self-testing mechanism module is used to set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and to build an adaptive automatic detection mechanism based on the dynamic adjustment parameters of the substation equipment load.

[0042] A static detection module is used to extract abnormal signals based on the feedback of an adaptive automatic detection mechanism, and to use professional detection technology to specifically detect relay protection functions, thereby constructing a static detection mechanism.

[0043] The optimization and evaluation module is used to correct and optimize the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals based on the LSTM neural network algorithm; and to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism and the optimized historical data.

[0044] Preferably, the optimization and evaluation module includes:

[0045] The self-test result optimization unit is used to correct and optimize the internal action logic output of the relay protection device under the influence of test signals in the adaptive automatic detection mechanism based on the LSTM neural network algorithm.

[0046] A stability evaluation unit is used to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data.

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0048] This invention provides a method for automatically testing the relay protection function of a substation. By introducing multiple types of pulses into the secondary circuit of the relay protection device, different stimulus signals are simulated to test the function of the relay protection device. This allows for online automatic testing of the substation relay protection function without offline testing, thereby timely feedback and detection of substation relay protection function anomalies, improving the performance and reliability of the substation relay protection. To ensure the accuracy of the online automatic testing of the substation relay protection function, an adaptive trigger time interval for the pulse test signal is set to avoid the influence of equipment load and electromagnetic interference on the pulse signal, which could lead to abnormal prompts and wasted resources. Furthermore, a static detection mechanism is set up to automatically detect abnormalities based on the adaptive trigger time interval. Based on the LSTM neural network algorithm, the feedback and environmental data of the detection mechanism are used to improve the robustness and adaptability of the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals. This further enhances the accuracy of online automatic testing of substation relay protection functions. Finally, through the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data, a stability model of the adaptive automatic detection mechanism is constructed to comprehensively evaluate its stability, thereby ensuring the accuracy and reliability of overall performance. This achieves the goal of setting up an adaptive automatic detection mechanism to realize online automatic testing of substation relay protection functions, improving automatic detection efficiency and accuracy. Attached Figure Description

[0049] Figure 1 This is a flowchart of an automatic testing method for substation relay protection functions according to the present invention;

[0050] Figure 2 The self-testing device for simulating multiple types of pulses and the relay protection action logic signal acquisition device of the present invention are used to construct an adaptive automatic detection mechanism flowchart based on the dynamic adjustment parameters of the substation equipment load.

[0051] Figure 3 The flowchart of the present invention, based on the LSTM neural network algorithm, is used to correct and optimize the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals.

[0052] Figure 4 This is a flowchart illustrating the stability evaluation of the adaptive automatic detection mechanism based on static detection, adaptive automatic detection, and optimized historical data. Detailed Implementation

[0053] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0054] Reference Figure 1 As shown, a method for automatic testing of substation relay protection functions includes:

[0055] Based on the testing requirements of substation relay protection functions, a knowledge base for relay protection is constructed, and a standard test template for relay protection functions is developed.

[0056] Set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and dynamically adjust parameters according to the substation equipment load to build an adaptive automatic detection mechanism;

[0057] Based on the feedback of the adaptive automatic detection mechanism, abnormal signals are extracted, and professional detection technology is used to specifically detect the relay protection function to build a static detection mechanism.

[0058] Based on the LSTM neural network algorithm, the output results of the internal action logic of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals are corrected and optimized.

[0059] The stability of the adaptive automatic detection mechanism is comprehensively evaluated based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data.

[0060] This can be explained by the fact that, compared to manual periodic inspections and offline testing, online automatic testing of substation relay protection functions has unique advantages in real-time performance and timeliness. By introducing multiple types of pulses into the secondary circuit of the relay protection device, different stimulus signals are simulated to test the device's functionality. This allows for online automatic testing of substation relay protection functions without offline testing and while ensuring the normal operation of the relay protection device. This effectively improves the efficiency and timeliness of automatic detection of substation relay protection functions. This solution utilizes the mechanism of online automatic testing of substation relay protection functions, the adaptive triggering time interval of the pulse test signal, and LSTM-based... Dynamic optimization of neural network algorithms avoids the impact of equipment load and electromagnetic interference on the stability of pulse signals, as well as the influence of other environmental factors and time differences on the internal action logic output of relay protection devices in the adaptive automatic detection mechanism under the influence of test signals. This improves the accuracy and reliability of online automatic testing of substation relay protection functions, providing a complete and efficient design approach for online automatic testing of substation relay protection functions and ensuring the accuracy and stability of automatic testing of substation relay protection functions. Here, time difference refers to the time difference between the feedback from the adaptive automatic detection mechanism and the detection of the static detection mechanism, which will cause changes in environmental data.

[0061] The construction of a knowledge base for relay protection and the formulation of standardized relay protection function testing templates based on the testing requirements of substation relay protection functions specifically include:

[0062] Based on the testing requirements of substation relay protection functions, a relay protection knowledge base is constructed. The relay protection knowledge base includes a standard comparison module and a data storage module. The standard comparison module is used to formulate standard relay protection function testing comparison templates, and the data storage module is used to store the testing information and monitoring data of substation relay protection functions.

[0063] It can be explained that establishing a relay protection knowledge base is an important standard and data analysis for realizing the automatic detection and evaluation of substation relay protection functions. It acquires crucial knowledge about relay protection devices, such as communication protocols, functional requirements, configuration methods, equipment parameters, working principles, operation methods, operation records, and fault reports, through substation industry standards, technical manuals, operation records, and expert experience. Furthermore, it transforms expert knowledge into structured data using expert scoring methods, thereby ensuring the comprehensiveness and standardization of the relay protection function detection comparison templates in the knowledge base. These templates use signal entities as nodes and relationships between entities as edges, describing the characteristics of entities and relationships through attributes. Based on protection principles, logical judgments, setting values, and pressure plate control, these templates constitute a complete relay protection function detection comparison template.

[0064] Reference Figure 2 As shown, the self-testing device simulating multiple types of pulses and the relay protection action logic signal acquisition device, which dynamically adjust parameters based on the substation equipment load, construct an adaptive automatic detection mechanism, specifically including:

[0065] In the secondary circuit of the relay protection device, a self-testing device simulating multiple types of pulses is set up, and different types of pulse test signals are superimposed on the signal under test through isolation devices;

[0066] Set up a relay protection action logic signal acquisition device to collect the internal action logic output of the relay protection device under the influence of test signals, and present it through indicator lights and data groups;

[0067] The system acquires dynamic load adjustment parameter data of substation equipment and electromagnetic interference data around relay protection devices, and then filters and normalizes the data using Kalman filtering algorithm and normalization formula.

[0068] Based on the dynamic adjustment of substation equipment load and electromagnetic interference around relay protection devices, the load fluctuation characteristics and electromagnetic interference characteristics affecting the detection of pulse test signals are extracted respectively.

[0069] The load fluctuation characteristics include: voltage, current, active / reactive power, load rate, and harmonic components; the electromagnetic interference characteristics include: interference amplitude, frequency distribution, pulse width, and frequency of occurrence.

[0070] The load fluctuation characteristics and electromagnetic interference characteristics are used as inputs, and the pulse test signal trigger time step is used as the output.

[0071] With the goal of minimizing pulse test signal fluctuations, a loss function for stabilizing pulse test signals is constructed based on the mean square error formula.

[0072] Based on historical data of relay protection function detection and / or by conducting test experiments, training sets, test sets, and validation sets are constructed for training neural network algorithms, respectively.

[0073] Based on the collected dynamic adjustment parameters of substation equipment load and electromagnetic interference data around the relay protection device, combined with the trained neural network model, the pulse test signal trigger time step is output.

[0074] Based on the frequency requirements of automatic testing of substation relay protection functions, a pulse test signal trigger time reference value is set, and a time step is superimposed to determine the adaptive trigger time interval of the pulse test signal.

[0075] This can be explained by the fact that when introducing multiple types of pulses into the secondary circuit of a relay protection device to simulate different stimulus signals for testing the device's functionality, the pulse signal strength is generally not set very high to ensure that the pulse signal does not interfere with the device's function. Therefore, it is crucial to consider the impact of substation equipment load and electromagnetic interference on the pulse signal to avoid instability caused by these factors, which could lead to erroneous logical outputs, system evaluation errors, and wasted resources. Thus, online automatic testing of substation relay protection functions should be avoided during periods when equipment load and electromagnetic interference are active, filtering out erroneous information and improving the accuracy of online automatic testing. To ensure that the relay protection operation is unaffected by the pulse test signal, hardware such as isolation transformers and signal injection modules are used in the secondary circuit to superimpose the test signal without affecting the normal transmission of the original signal. This is achieved through the relay protection device's self-test interface. The action logic signal acquisition device collects signals from these interfaces to reflect the internal logic state without triggering actual actions. Secondly, it simulates different stimulus signals, including short-circuit pulses (including metallic / impedance type), high-frequency oscillation waves, switching operation surges, CT saturation distortion, PT resonant waves, and composite fault pulses. Through cyclic testing, it obtains the internal action logic output of the relay protection device under the influence of these pulses. This is then compared with the relay protection function detection reference template in the relay protection knowledge base to determine the relay protection function. Specifically, to avoid errors from single measurements, an automatic detection cycle number needs to be set. Multiple sets of tests using the same type of pulse are conducted, and the confidence level is set using a normal distribution to obtain a more accurate relay protection function judgment result. Since automatic detection is affected by the cycle number, a detection time occurs when completing an automatic detection. Therefore, this solution sets a pulse test signal trigger time reference value by pre-setting the cycle number value to avoid the influence of automatic detection time on the adaptive trigger time interval.

[0076] The aforementioned feedback based on the adaptive automatic detection mechanism, extracting abnormal signals, and using specialized detection techniques to specifically detect relay protection functions, constitutes a static detection mechanism that includes:

[0077] Based on the feedback results of the adaptive automatic detection mechanism, the abnormal signals output by the internal action logic of the relay protection device under the influence of the test signal are obtained.

[0078] Using professional testing techniques and combining abnormal signals, we conduct targeted tests to detect whether there are any abnormalities in the relay protection function, and record the professional test results.

[0079] The static detection mechanism, which utilizes specialized detection techniques and combines them with abnormal signals, specifically detects whether there are any abnormalities in the relay protection function and records the results. It complements and verifies the adaptive automatic detection mechanism. When the adaptive automatic detection mechanism reports an abnormality in its internal action logic output, the static detection mechanism uses specialized techniques to conduct targeted in-depth detection of the abnormal point. For example, it verifies whether the action threshold of the protection device under a specific fault waveform meets the design requirements, ensuring the authenticity of the abnormal signal and preventing the data fed back by the adaptive automatic detection mechanism from deviating from the actual detection data. This provides a more reliable basis for the evaluation of the relay protection function and improves the accuracy and reliability of the entire detection system.

[0080] Reference Figure 3 As shown, the specific steps for correcting and optimizing the internal action logic output of the relay protection device under the influence of test signals in the adaptive automatic detection mechanism based on the LSTM neural network algorithm include:

[0081] Historical data from adaptive automatic detection mechanisms, static detection mechanisms, and environmental monitoring are acquired, and the data are filtered and normalized using filtering algorithms and normalization formulas.

[0082] Based on historical data from the adaptive automatic detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the theoretical relay protection function detection and the comparison template is calculated and denoted as . ;

[0083] Based on historical data from the static detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the actual relay protection function detection and the comparison template is calculated and denoted as . ;

[0084] by With the goal of minimizing the error, a loss function for the detection distortion rate of relay protection function is constructed based on the mean square error formula.

[0085] Based on environmental monitoring data, an environmental loss function for relay protection function testing is constructed with the goal of minimizing the error of environmental parameters between theoretical and actual relay protection function testing.

[0086] Based on the loss function of the distortion rate of relay protection function detection and the environmental loss function during relay protection function detection, a comprehensive loss function for relay protection function detection is constructed.

[0087] In the hidden state, constraints are set to store historical environmental disturbance features and standard data as the cell state.

[0088] Based on the LSTM neural network algorithm and combined with the detection data of the adaptive automatic detection mechanism, the output results of the internal action logic of the relay protection device under the influence of the test signal in the adaptive automatic detection mechanism are corrected and optimized.

[0089] It can be explained that by setting an adaptive trigger time interval for the pulse test signal, the impact of equipment load and electromagnetic interference on the pulse test signal can be reduced. However, it is still impossible to avoid the error between the theoretical data and the actual detection data of the internal action logic output of the relay protection device under the influence of the test signal. That is, there is an error between the detection data of the adaptive automatic detection mechanism and the detection data of the static detection mechanism. Therefore, this solution uses an LSTM neural network algorithm, combined with theoretical data, actual detection data, environmental data, and a relay protection function detection comparison template, to analyze the output results of the internal action logic of the relay protection device under the influence of the test signal in the adaptive automatic detection mechanism. The adaptive automatic detection mechanism is modified and optimized to improve the robustness and adaptability of the internal action logic output of the relay protection device under the influence of test signals. The historical data of the adaptive automatic detection mechanism includes: internal action logic output data of the relay protection device under the influence of various types of pulse signals and corresponding pulses, and internal action logic output data of the relay protection device triggered by various types of pulse signals. The historical data of the static detection mechanism includes: standardized test data of the action characteristics of the relay protection device under standard pulse excitation. The historical environmental monitoring data includes: temperature, humidity, wind speed, lightning strike density, ice thickness, vibration intensity, equipment load parameters, and electromagnetic interference characteristics.

[0090] The expression for the alienation rate is: In the formula, The variation rate value of the relay protection function test and comparison template. The number of sets of relay protection function test parameters and reference template standard parameters. For the first The weights of each parameter, Relay protection function test One parameter data, For comparison template number Standard parameter data;

[0091] The relay protection function test comparison template is composed of knowledge of protection principles, logical judgment, setting values, and control panels. Therefore, when making differential comparisons, the results need to be quantified. For example, if logical errors exist... The value is set to a fixed value; if The value in Within the standard range, then The difference was set to 0 to ensure the comparability and accuracy of the data. Specifically, the difference rate between the theoretical relay protection function test and the control template was [not specified]. The difference between the actual relay protection function test and the control template It is the overall conceptual value obtained based on the alienation rate expression, while , It is the first The specific alienation rate value corresponding to each static detection mechanism indicates that the alienation rate value in each static detection mechanism follows the alienation rate expression;

[0092] The environmental loss function for electrical protection function detection is also based on the mean square error formula. By analyzing the changes and fluctuations of different environmental parameters, the environmental loss function for electrical protection function detection is constructed, thereby reducing the data deviation problem caused by the environmental deviation between theoretical detection and actual detection.

[0093] Reference Figure 4 As shown, the comprehensive evaluation of the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data specifically includes:

[0094] The static detection mechanism, the adaptive automatic detection mechanism, and the optimized historical data are obtained, and the data are normalized using a normalization formula.

[0095] The heterogeneity rates of static detection mechanism, adaptive automatic detection mechanism and optimized historical data and control templates were obtained respectively;

[0096] Based on the static detection mechanism, the adaptive automatic detection mechanism, and the optimized alienation rate, a stability model of the adaptive automatic detection mechanism is established.

[0097] Based on the stability model of the adaptive automatic detection mechanism, the stability of the adaptive automatic detection mechanism is comprehensively evaluated.

[0098] The stability model expression for the adaptive automatic detection mechanism is: In the formula, for The stability evaluation value of the adaptive automatic detection mechanism function at that time. This represents the number of static detection mechanisms performed within a fixed time window. To within a fixed time window The deviation rate between the theoretical relay protection function test and the control template during the first static detection mechanism test. To within a fixed time window The theoretical relay protection function test and the deviation rate after template optimization during the static detection mechanism test are compared. To within a fixed time window The difference rate between the actual relay protection function test and the control template under the sub-static testing mechanism. , The weighting coefficients can be obtained using the least squares method. This represents the average of the adaptively automatically detected alienation rate within a fixed time window. The variance of the adaptive automatic detection of the alienation rate within a fixed time window. The number of adaptive automatic detections within a fixed time window. , For preset fixed constant terms, This is the variance threshold.

[0099] This can be explained by the fact that when evaluating the stability of the adaptive automatic detection mechanism, it is necessary to consider whether static detection mechanism behavior occurs within a pre-set fixed time window. If no static detection mechanism behavior occurs or / or the number of static detection mechanism behaviors is too small, the data on the difference between the actual relay protection function detection and the reference template will be insufficient, thus failing to pass the test. Stability evaluation value of the adaptive automatic detection mechanism function The formula is used to evaluate the stability of the adaptive automatic detection mechanism. Therefore, it is necessary to set a stability evaluation mode for the adaptive automatic detection mechanism based on the number of times the static detection mechanism behavior is triggered within a preset fixed time window, thereby improving the accuracy and reliability of the stability evaluation of the adaptive automatic detection mechanism. , For preset fixed constant terms, The variance threshold can be optimized using machine learning algorithms with a large amount of sample data.

[0100] Furthermore, based on the same inventive concept as the aforementioned method for automatic testing of substation relay protection functions, this solution proposes a system for automatic testing of substation relay protection functions, comprising:

[0101] The knowledge base module is used to construct a knowledge base for relay protection based on the testing requirements of substation relay protection functions and to formulate standard relay protection function testing comparison templates.

[0102] The self-testing mechanism module is used to set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and to build an adaptive automatic detection mechanism based on the dynamic adjustment parameters of the substation equipment load.

[0103] A static detection module is used to extract abnormal signals based on the feedback of an adaptive automatic detection mechanism, and to use professional detection technology to specifically detect relay protection functions, thereby constructing a static detection mechanism.

[0104] The optimization and evaluation module is used to correct and optimize the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals based on the LSTM neural network algorithm; and to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism and the optimized historical data.

[0105] The optimization and evaluation module includes:

[0106] The self-test result optimization unit is used to correct and optimize the internal action logic output of the relay protection device under the influence of test signals in the adaptive automatic detection mechanism based on the LSTM neural network algorithm.

[0107] A stability evaluation unit is used to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data.

[0108] In summary, the advantages of this invention are: setting up an adaptive automatic detection mechanism to realize online automatic testing of substation relay protection functions, thereby improving automatic detection efficiency and accuracy.

[0109] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for automatically testing the relay protection function of a substation, characterized in that, include: Based on the testing requirements of substation relay protection functions, a knowledge base for relay protection is constructed, and a standard test template for relay protection functions is developed. Set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and dynamically adjust parameters according to the substation equipment load to build an adaptive automatic detection mechanism; Based on the feedback of the adaptive automatic detection mechanism, abnormal signals are extracted, and professional detection technology is used to specifically detect the relay protection function to build a static detection mechanism. Based on the LSTM neural network algorithm, the output results of the internal action logic of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals are corrected and optimized. Based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data, the stability of the adaptive automatic detection mechanism is comprehensively evaluated. The self-testing device and relay protection action logic signal acquisition device, which simulate multiple types of pulses, are configured to dynamically adjust parameters based on the substation equipment load to construct an adaptive automatic detection mechanism, specifically including: In the secondary circuit of the relay protection device, a self-testing device simulating multiple types of pulses is set up, and different types of pulse test signals are superimposed on the signal under test through isolation devices; Set up a relay protection action logic signal acquisition device to collect the internal action logic output of the relay protection device under the influence of test signals, and present it through indicator lights and data groups; The system acquires dynamic load adjustment parameter data of substation equipment and electromagnetic interference data around relay protection devices, and then filters and normalizes the data using Kalman filtering algorithm and normalization formula. Based on the dynamic adjustment of substation equipment load and electromagnetic interference around relay protection devices, the load fluctuation characteristics and electromagnetic interference characteristics affecting the detection of pulse test signals are extracted respectively. The load fluctuation characteristics include: voltage, current, active / reactive power, load rate, and harmonic components; the electromagnetic interference characteristics include: interference amplitude, frequency distribution, pulse width, and frequency of occurrence. The load fluctuation characteristics and electromagnetic interference characteristics are used as inputs, and the pulse test signal trigger time step is used as the output. With the goal of minimizing pulse test signal fluctuations, a loss function for stabilizing pulse test signals is constructed based on the mean square error formula. Based on historical data of relay protection function detection and / or by conducting test experiments, training sets, test sets, and validation sets are constructed for training neural network algorithms, respectively. Based on the collected dynamic adjustment parameters of substation equipment load and electromagnetic interference data around the relay protection device, combined with the trained neural network model, the pulse test signal trigger time step is output. Based on the frequency requirements of automatic testing of substation relay protection functions, a pulse test signal trigger time reference value is set, and a time step is superimposed to determine the adaptive trigger time interval of the pulse test signal.

2. The method for automatic testing of substation relay protection functions according to claim 1, characterized in that, The construction of a knowledge base for relay protection and the formulation of standardized relay protection function testing templates based on the testing requirements of substation relay protection functions specifically include: Based on the testing requirements of substation relay protection functions, a relay protection knowledge base is constructed. The relay protection knowledge base includes a standard comparison module and a data storage module. The standard comparison module is used to formulate standard relay protection function testing comparison templates, and the data storage module is used to store the testing information and monitoring data of substation relay protection functions.

3. The method for automatic testing of substation relay protection functions according to claim 2, characterized in that, The aforementioned feedback based on the adaptive automatic detection mechanism, extracting abnormal signals, and using specialized detection techniques to specifically detect relay protection functions, constitutes a static detection mechanism that includes: Based on the feedback results of the adaptive automatic detection mechanism, the abnormal signals output by the internal action logic of the relay protection device under the influence of the test signal are obtained. Using professional testing techniques and combining abnormal signals, we conduct targeted tests to detect whether there are any abnormalities in the relay protection function, and record the professional test results.

4. The method for automatic testing of substation relay protection functions according to claim 3, characterized in that, The method for correcting and optimizing the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals, based on the LSTM neural network algorithm, specifically includes: Historical data from adaptive automatic detection mechanisms, static detection mechanisms, and environmental monitoring are acquired, and the data are filtered and normalized using filtering algorithms and normalization formulas. Based on historical data from the adaptive automatic detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the theoretical relay protection function detection and the comparison template is calculated and denoted as . ; Based on historical data from the static detection mechanism and the relay protection function detection comparison template, the discrepancy rate between the actual relay protection function detection and the comparison template is calculated and denoted as . ; by With the goal of minimizing the error, a loss function for the detection distortion rate of relay protection function is constructed based on the mean square error formula. Based on environmental monitoring data, an environmental loss function for relay protection function testing is constructed with the goal of minimizing the error of environmental parameters between theoretical and actual relay protection function testing. Based on the loss function of the distortion rate of relay protection function detection and the environmental loss function during relay protection function detection, a comprehensive loss function for relay protection function detection is constructed. In the hidden state, constraints are set to store historical environmental disturbance features and standard data as the cell state. Based on the LSTM neural network algorithm and combined with the detection data of the adaptive automatic detection mechanism, the output results of the internal action logic of the relay protection device under the influence of the test signal in the adaptive automatic detection mechanism are corrected and optimized.

5. The method for automatic testing of substation relay protection functions according to claim 4, characterized in that, The comprehensive evaluation of the stability of the adaptive automatic detection mechanism based on static detection mechanism, adaptive automatic detection mechanism, and optimized historical data specifically includes: The static detection mechanism, the adaptive automatic detection mechanism, and the optimized historical data are obtained, and the data are normalized using a normalization formula. The heterogeneity rates of static detection mechanism, adaptive automatic detection mechanism and optimized historical data and control templates were obtained respectively; Based on the static detection mechanism, the adaptive automatic detection mechanism, and the optimized alienation rate, a stability model of the adaptive automatic detection mechanism is established. Based on the stability model of the adaptive automatic detection mechanism, the stability of the adaptive automatic detection mechanism is comprehensively evaluated. The stability model expression for the adaptive automatic detection mechanism is: In the formula, for The stability evaluation value of the adaptive automatic detection mechanism function at that time. This represents the number of static detection mechanisms performed within a fixed time window. To within a fixed time window The deviation rate between the theoretical relay protection function test and the control template during the first static detection mechanism test. To within a fixed time window The theoretical relay protection function test and the deviation rate after template optimization during the static detection mechanism test are compared. To within a fixed time window The difference rate between the actual relay protection function test and the control template under the sub-static testing mechanism. , The weighting coefficients are obtained by solving using the least squares method. This represents the average of the adaptively automatically detected alienation rate within a fixed time window. The variance of the adaptive automatic detection of the alienation rate within a fixed time window. The number of adaptive automatic detections within a fixed time window. , For preset fixed constant terms, This is the variance threshold.

6. A system for automatic testing of substation relay protection functions, characterized in that, A method for automatically testing the relay protection function of a substation as described in any one of claims 1-5, comprising: The knowledge base module is used to construct a knowledge base for relay protection based on the testing requirements of substation relay protection functions and to formulate standard relay protection function testing comparison templates. The self-testing mechanism module is used to set up a self-testing device that simulates multiple types of pulses and a relay protection action logic signal acquisition device, and to build an adaptive automatic detection mechanism based on the dynamic adjustment parameters of the substation equipment load. A static detection module is used to extract abnormal signals based on the feedback of an adaptive automatic detection mechanism, and to use professional detection technology to specifically detect relay protection functions, thereby constructing a static detection mechanism. The optimization and evaluation module is used to correct and optimize the internal action logic output of the relay protection device in the adaptive automatic detection mechanism under the influence of test signals based on the LSTM neural network algorithm; and to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism and the optimized historical data.

7. The system for automatic testing of substation relay protection functions according to claim 6, characterized in that, The optimization and evaluation module includes: The self-test result optimization unit is used to correct and optimize the internal action logic output of the relay protection device under the influence of test signals in the adaptive automatic detection mechanism based on the LSTM neural network algorithm. A stability evaluation unit is used to comprehensively evaluate the stability of the adaptive automatic detection mechanism based on the static detection mechanism, the adaptive automatic detection mechanism, and optimized historical data.

Images