Operation evaluation method and system for relay protection equipment

Abstract

The invention provides an operation evaluation method and system for relay protection equipment, and relates to the field of power systems, and the method comprises the steps: taking the perception data of the relay protection equipment at the current moment as the input, and generating a state prediction value of the relay protection equipment based on a preset digital twinborn model; performing consistency verification on the state prediction value and the sensing data of the relay protection equipment at the next moment, and outputting a verification report; determining a health state probability and a network security risk value of the relay protection equipment according to the verification report; and determining the reliability of the relay protection equipment based on the health state probability and the network security risk value. Dynamic changes of the relay protection equipment are sensed in real time, state prediction is carried out, various conditions of the relay protection equipment in actual operation are reproduced truly, authenticity and accuracy of state prediction of the relay protection equipment are improved, state prediction values are corrected in combination with sensing data of the relay protection equipment at the next moment, and the relay protection equipment state prediction accuracy is improved. And reliable operation of a power system is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power systems, in particular to a method and system for operation evaluation of a relay protection device. BACKGROUND

[0002] The relay protection device is an important part of the power system, which is used to detect faults and quickly cut off the circuit to protect the equipment and personnel. The relay protection device is usually composed of relays, measurement elements, logic judgment elements and execution elements, which monitor the current, voltage, frequency and other parameters of the power system in real time. When abnormal conditions such as short circuit and overload occur, the relay protection device can quickly trigger the circuit breaker to trip and isolate the fault area.

[0003] The evaluation of the relay protection device is an important part of the reliability management of the power system, which aims to comprehensively evaluate the performance, operation state and management level of the device to ensure its reliable action in the event of power system faults. The related technology uses a remote control system for remote control, but since some relay protection functions depend on the actual working conditions on site, remote simulation testing may not completely reproduce the real fault scene, resulting in evaluation results deviating from the actual performance. Especially the simulation distortion of electrical quantities in fault scenes, since the dynamic changes between the operation state of the relay device itself and the operation environment are not considered, the evaluation results of the electrical quantities in the evaluation are obviously higher and cover up the device faults, and the relay device cannot be maintained in time, resulting in safety hazards. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a method and system for operation evaluation of a relay protection device, which solves the problem that remote simulation testing in the prior art may not completely reproduce the real fault scene, resulting in evaluation results deviating from the actual performance.

[0005] In a first aspect, the present application provides a method for operation evaluation of a relay protection device, comprising: using the perception data of the relay protection device at the current time as input, generating a state prediction value of the relay protection device based on a preset digital twin model; performing consistency check on the state prediction value and the perception data of the relay protection device at the next time and outputting a check report; determining the health state probability and the network security risk value of the relay protection device according to the check report; determining the reliability of the relay protection device based on the health state probability and the network security risk value.

[0006] In an embodiment, before the state prediction value of the relay protection device is generated based on the preset digital twin model, the method further comprises: determine a knowledge graph of the relay protection device according to static data of the relay protection device, a device model library, and a corresponding image library; perform semantic coding on the knowledge graph and obtain an ontology model of the relay protection device; train the ontology model by taking historical perception data and historical state data as input and obtain the digital twin model.

[0007] In an embodiment, the device model library includes a physical model and a mechanism model of the relay protection device, and the image library includes device images of the relay protection device. The semantic coding on the knowledge graph and the obtaining of the ontology model of the relay protection device specifically include: perform semantic matching on the static data of the relay protection device and physical parameters of the physical model and obtain a first semantic correspondence relationship; determine embedding vectors of the static data and the physical parameters according to the first semantic correspondence relationship and mark them as first embedding vectors; perform semantic matching on the static data of the relay protection device and mechanism parameters of the mechanism model and obtain a second semantic correspondence relationship; determine embedding vectors of the static data and the physical parameters according to the second semantic correspondence relationship and mark them as second embedding vectors; map the static data to a low-dimensional semantic space in a word embedding manner, obtain a distribution vector of the static data and mark it as a first distribution vector; perform feature extraction and semantic coding on the device images and obtain coded device images; map the coded device images to the low-dimensional semantic space corresponding to the static data, obtain a distribution vector of the device images and mark it as a second distribution vector; determine a semantic association matrix between the first distribution vector and the second distribution vector, and determine embedding vectors of the static data and the device images based on the semantic association matrix and mark them as third embedding vectors; determine a target weighted group by taking the first embedding vectors, the second embedding vectors, and the third embedding vectors as real-valued weights; add the target weighted group to the knowledge graph to obtain the ontology model.

[0008] In an embodiment, the training of the ontology model by taking historical perception data and historical state data as input and the obtaining of the digital twin model specifically include: determine a fault operation scenario corresponding to the ontology model from a pre-set fault database based on the historical perception data; determine running state data of the ontology model in the fault operation scenario by taking the historical perception data as input, and update the ontology model based on the similarity between the running state data and the historical state data and generate a new ontology model; When the similarity meets a preset update condition, the corresponding ontology model is taken as the digital twin model.

[0009] In an embodiment, the perception data includes real-time electrical quantities, real-time mechanical quantities, and real-time network security data; the state prediction value includes electrical prediction quantities, mechanical prediction quantities, and network security prediction data; the consistency check of the state prediction value and the next-time perception data of the relay protection device specifically includes: performing correlation check and / or parameter deviation rate check on the next-time real-time electrical quantities and the electrical prediction quantities; and / or performing synchronization check and / or contact stroke error check on the next-time real-time mechanical quantities and the mechanical prediction quantities; and / or performing communication link reliability check on the next-time real-time network security data and the network security prediction data.

[0010] In an embodiment, the check report includes electrical quantity deviation, mechanical quantity deviation, functional logic deviation, communication error rate, transmission delay, number of illegal packets, number of encryption failures, and packet compliance rate of the relay protection device; the determination of the health state probability and the network security risk value of the relay protection device according to the check report specifically includes: determining the health state probability of the relay protection device through a preset Bayesian classification model according to the electrical quantity deviation, the mechanical quantity deviation, and the functional logic deviation; determining the risk weight of each of the communication error rate, the transmission delay, the number of illegal packets, the number of encryption failures, and the packet compliance rate based on a preset network security rule, and determining the network security risk value through weighted summation based on the risk weight.

[0011] In an embodiment, the determination of the reliability of the relay protection device based on the health state probability and the network security risk value specifically includes: converting the health state probability into a health base score of the relay protection device through weighted summation, and converting the health base score into a health correction score based on a predicted state transition function; obtaining a network security base score by subtracting the network security risk value from a preset maximum security score, and converting the network security base score into a network security correction score based on a preset penalty function; obtaining a reliability score through weighted summation based on the health correction score and the network security correction score.

[0012] In an embodiment, the expression of the state transition function is:

[0013] wherein, is a health correction score, is a health base score, is a probability of transition from state i to state j, is a trend sensitivity coefficient.

[0014] In an embodiment, the expression of the penalty function is:

[0015]

[0016] wherein, is a cyber security correction score, is a cyber security base score, is a penalty coefficient; is a probability of the relay protection device being in a fault warning state, is a cyber security risk value.

[0017] In a second aspect, the present application provides a running evaluation system for a relay protection device, characterized in that it comprises a processor and a memory; wherein the memory stores a computer program, and the computer program is used to load and execute the running evaluation method for the relay protection device according to any one of the first aspect by the processor.

[0018] In the running evaluation method and system for the relay protection device in the embodiment, the digital twin model can perceive the dynamic changes of the relay protection device in real time and make state prediction, truly reproduce various conditions of the relay protection device in actual operation, improve the authenticity and accuracy of the state prediction of the relay protection device, and at the same time, correct the state prediction value in combination with the perception data of the next moment of the relay protection device, which can effectively avoid that the digital twin simulation distorts and covers up the real fault of the relay protection device, and at the same time, evaluate the reliability of the relay protection device in combination with the health state and the network security of the relay protection device, effectively eliminate the security risks, and guarantee the reliable operation of the power system. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0020] Figure 1A flowchart of a method for operation evaluation of a relay protection device is provided for an embodiment.

[0021] Figure 2 A structural diagram of an electronic device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0022] The specific embodiments of the present application will be described in detail hereinafter with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the description of the present application, all the other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the scope of the present application.

[0023] In the description of the present application, unless explicitly defined and limited, the terms such as "arrange", "mount", "connect" and the like shall be understood in a broad sense, for example, can be fixedly connected, or can be detachably connected, or integrally connected; can be mechanically connected, or can be electrically connected; can be directly connected, or can be indirectly connected through an intermediate medium. The specific meanings of the above terms can be understood according to the specific circumstances by those of ordinary skill in the art.

[0024] The terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship when the product of the present application is usually placed, and are only for the convenience of description and simplification of description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0025] The terms "first", "second", "third" and the like are only for distinguishing similar attributes of elements, and do not indicate or imply relative importance or a particular order.

[0026] The terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, in addition to including the listed elements, other elements not explicitly listed can also be included.

[0027] As Figure 1 shown, the present embodiment provides a method for operation evaluation of a relay protection device, comprising: Step S10: taking the perception data of the relay protection device at the current time as input, generating a state prediction value of the relay protection device based on a preset digital twin model; Step S20: performing consistency check on the state prediction value and the perception data of the relay protection device at the next time and outputting a check report; Step S30: determining a health state probability and a network security risk value of the relay protection device according to the check report; Step S40: determining the reliability of the relay protection device based on the health state probability and the network security risk value.

[0028] In the operation evaluation method for the relay protection device, the digital twin model can perceive the dynamic changes of the relay protection device in real time and make state prediction, truly reproduce various conditions of the relay protection device in actual operation, improve the authenticity and accuracy of the state prediction of the relay protection device, correct the state prediction value in combination with the perception data of the next moment of the relay protection device, effectively avoid the distortion of the digital twin simulation to cover up the real failure of the relay protection device, and evaluate the reliability of the relay protection device in combination with the health state and network security of the relay protection device, effectively eliminate safety hazards, and ensure the reliable operation of the power system.

[0029] Step S10: taking the perception data of the relay protection device at the current moment as input, and generating a state prediction value of the relay protection device based on a preset digital twin model.

[0030] The perception data is the device operation parameters and network security data collected in real time by various sensors and network detection units deployed on the relay protection device. The device operation parameters include electrical quantities (current, voltage, power, frequency, harmonic content, etc. reflecting the electrical characteristics of the device) and mechanical quantities (operation parameters of mechanical components of the device, such as travel, pressure, and action time of relay contacts, opening and closing speed of circuit breakers, mechanical vibration amplitude, etc.).

[0031] The digital twin model can accurately simulate various extreme fault scenarios and simulate the influence of aging (such as relay contact wear and sensor precision decline) of the relay protection device on the protection performance based on the perception data of the relay protection device at the current moment. Through real-time data driving, the action characteristics of the relay protection device are accurately reflected, errors caused by environmental factors in testing are avoided, potential defects are found in advance, and safety maintenance of the relay protection device is performed in a timely manner.

[0032] Before the state prediction value of the relay protection device is generated based on the preset digital twin model, it further includes: Step S101: determining a knowledge graph of the relay protection device according to static data of the relay protection device, a device model library, and a corresponding image library; Step S102: performing semantic coding on the knowledge graph and obtaining an ontology model of the relay protection device; Step S103: taking historical perception data and historical state data as input, training the ontology model and obtaining the digital twin model.

[0033] In step S101, the static data refers to the basic data of the relay protection device in the design, manufacturing and deployment stages, including device basic parameters, device manufacturing data and power grid associated data. The device basic parameters include physical attributes such as rated voltage, rated current, rated frequency, size, weight, installation method, electrical parameters such as action threshold (overcurrent / overvoltage / zero sequence current constant), return coefficient, action time constant, power consumption characteristics, and industry standards such as IEC 61850 and GB / T 14285. The device manufacturing data includes schematic diagrams such as secondary circuit wiring diagram, logic control diagram, chip-level hardware architecture diagram, core component materials (such as relay contact silver alloy composition, PCB board substrate FR-4 parameters), material properties such as temperature resistance performance index, and manufacturer information such as device model, factory number, design life, warranty period, firmware version number. The power grid associated data includes access system parameters such as voltage level (10kV / 110kV) and protection range (line / transformer / bus protection type), and network topology data such as associated switch device number, CT / PT transformation ratio, and adjacent protection device coordination relationship.

[0034] The device model library contains standardized model libraries of geometric, physical and logical characteristics of relay protection devices and their associated systems, usually including physical models and mechanism models. The physical model contains the three-dimensional geometric mechanism of the relay shell, terminal row, display screen and other entity components, and integrates the finite element structure of material mechanics performance (elastic modulus, Poisson's ratio), thermal conductivity coefficient, electromagnetic shielding efficiency and other parameters. The mechanism model integrates electrical characteristic units, dynamic characteristic units, logic protection units and communication protocol units, etc. The electrical characteristic unit is based on Kirchhoff's law to establish an equivalent circuit model, including coil inductance, contact resistance, and capacitor filter circuit parameters. The dynamic characteristic unit is provided with differential equations for describing the action process of the relay protection device (such as relay pull-in time-current curve, A / D conversion delay rule of sampling circuit); the logic protection unit is provided with protection criterion execution logic rules described in the form of state machine; the communication protocol unit integrates IEC 61850 communication service mapping protocol, GOOSE message transmission delay protocol, Modbus data interaction timing protocol, etc.

[0035] The image library contains device images of the relay protection device, including two-dimensional engineering drawings (such as the circuit diagram of the secondary circuit of the relay) and three-dimensional rendering diagrams (such as the internal armature movement mechanism of the relay), fault waveform diagrams (such as voltage / current waveform curve diagram, phase vector diagram, frequency spectrum analysis diagram) and other visual graphics.

[0036] The knowledge graph of the relay protection device is constructed by "entity-relation-attribute" triplets. The intelligent knowledge base has three parts of entity type, relation definition and construction method. For example, the device entity of the entity type is a relay, the relation definition is "relay-connection-contact group", and the construction method is "extracting the connection relation of the relay and the contact group from the technical manual and the fault report by NLP technology". The knowledge graph of the relay protection device can be constructed by "entity-relation-attribute" triplets.

[0037] In step S102, the knowledge graph is semantically encoded to obtain the ontology model of the relay protection device, specifically including: Step S1021: semantically matching the static data of the relay protection device with the physical parameters of the physical model to obtain a first semantic correspondence relationship; determining the embedding vectors of the static data and the physical parameters according to the first semantic correspondence relationship and recording them as first embedding vectors; Step S1022: semantically matching the static data of the relay protection device with the mechanism parameters of the mechanism model to obtain a second semantic correspondence relationship; determining the embedding vectors of the static data and the physical parameters according to the second semantic correspondence relationship and recording them as second embedding vectors; Step S1023: mapping the static data to a low-dimensional semantic space in the form of word embedding to obtain a distribution vector of the static data and recording it as a first distribution vector; performing feature extraction and semantic encoding on the device image to obtain an encoded device image; mapping the encoded device image to the low-dimensional semantic space corresponding to the static data to obtain a distribution vector of the device image and recording it as a second distribution vector; determining the semantic association matrix between the first distribution vector and the second distribution vector, and determining the embedding vectors of the static data and the device image based on the semantic association matrix and recording them as third embedding vectors; Step S1024: taking the first embedding vector, the second embedding vector and the third embedding vector as real-valued weights to determine a target weighted group; adding the target weighted group to the knowledge graph to obtain the ontology model.

[0038] In step S1021, the physical model parameters in the physical model are associated with the static data in the knowledge graph through semantic mapping rules. By querying the synonyms and hyponyms defined in the semantic dictionary and merging them into the knowledge graph, a first semantic correspondence between the static data and the physical model parameters can be established. For example, although "rated current" in the static data and "current rating" in the physical model are slightly different in expression, they have the same essential meaning. Through natural language processing technology combined with a semantic dictionary, the texts of the static data and the physical parameters are segmented, tagged, and named entity recognized, and concepts with the same semantics are associated to obtain the first semantic correspondence.

[0039] The embedding vector is a representation method for converting text information into a numerical vector for computer processing and analysis. According to the first semantic correspondence, a word embedding model (such as Word2Vec, GloVe, etc.) is used to convert each word in the static data and the physical parameters into a low-dimensional vector. For example, for the word "rated current", a fixed-length vector is obtained after processing by the word embedding model, and then these vectors are combined to obtain the embedding vector of the static data and the physical parameters, i.e., the first embedding vector.

[0040] In step S1022, the mechanism model describes the working principle and internal mechanism of the relay protection device, and its mechanism parameters include action time, action threshold, protection logic, etc. Similarly, the static data and the mechanism parameters of the mechanism model are semantically matched to find their correspondence. For example, "action current setting value" in the static data and "overcurrent protection action current threshold" in the mechanism model are related concepts, and a second semantic correspondence is established through semantic matching. Similar to the method of determining the first embedding vector, according to the second semantic correspondence, the word embedding model is used to convert the static data and the mechanism parameters into a second embedding vector.

[0041] In step S103, the text information in the static data is mapped to a low-dimensional semantic space in the form of word embedding. The word embedding model represents each word as a vector, and then obtains the distribution vector of the entire static data, i.e., the first distribution vector, by averaging, summing, etc. The distribution vector can reflect the semantic features of the static data and has certain distribution rules in the low-dimensional space.

[0042] The computer vision technology (such as convolutional neural network, CNN) can be used to extract the color, texture, shape and other features of the image, and then the features are semantically coded to convert the image features into semantic information of the text description (for example, converting the device image into the text description of "the relay protection device shell is square, and the indicator light flashes"). Then the text description is processed by word embedding to map it to the same low-dimensional semantic space as the static data, and the distribution vector of the device image, i.e. the second distribution vector, is obtained.

[0043] The semantic association matrix between the first distribution vector and the second distribution vector can be constructed by calculating the cosine similarity between the two vectors, and each element in the semantic association matrix represents the semantic association degree of the static data and the device image in a certain dimension. Based on the semantic association matrix, the original vector information of the static data and the device image is combined to determine the embedding vector of the static data and the device image, i.e. the third embedding vector.

[0044] In step S1024, the first embedding vector, the second embedding vector and the third embedding vector are used as real-valued weights, and the three embedding vectors are fused by weighted summation to obtain a target weighted group. Since the knowledge graph already contains the entity, relationship and attribute information of the relay protection device, the addition of the target weighted group enables the knowledge graph to more accurately represent the semantic information and internal relationship of the device, and the constructed ontology model of the relay protection device can be used for knowledge reasoning, fault diagnosis and state prediction, providing strong support for remote simulation testing and operation and maintenance management of the relay protection device. It can be understood that the weight of each embedding vector can be adjusted according to its importance in the model.

[0045] After obtaining the ontology model of the relay protection device, since the ontology model is established according to the static data of the relay protection device, in order to make the digital twin model reflect the recent running state of the relay protection device, the ontology model needs to be trained in combination with the recent historical perception data and historical state data of the relay protection device, so that the fault diagnosis and state prediction of the digital twin model can be fitted to the recent state of the relay protection device.

[0046] Step S103: training the ontology model and obtaining the digital twin model by taking the historical perception data and the historical state data as input.

[0047] The historical perception data usually collects the perception data of the relay protection device in the past week (or month). The perception data is the device operation parameters and network security data collected in real time by various sensors and network detection units deployed on the relay protection device. The device operation parameters include electrical quantities (current, voltage, power, frequency, harmonic content, etc. reflecting the electrical characteristics of the device) and mechanical quantities (operation parameters of mechanical components of the device, such as travel of relay contacts, pressure, action time, opening and closing speed of circuit breakers, mechanical vibration amplitude, etc.).

[0048] The historical state data is a labeled record of the historical operation state of the device, including normal operation state and abnormal state such as fault, defect and early warning. The state data usually includes three types of running state labels (such as normal operation, overload operation, abnormal temperature rise, protection misoperation / refusal, etc.), fault attributes (such as coil aging, poor contact of contacts, logic chip failure, fault occurrence time and repair record, etc.), and artificial verification data (such as fault diagnosis conclusion, insulation resistance test value, action characteristic verification data, etc.).

[0049] The historical perception data and the historical state data are used as inputs to train the ontology model and obtain the digital twin model, specifically including: Step S1031: Based on the historical perception data, the corresponding fault operation scene of the ontology model is determined from a pre-set fault database; Step S1032: The historical perception data is used as input to determine the running state data of the ontology model in the fault operation scene, and the ontology model is updated based on the similarity of the running state data and the historical state data to generate a new ontology model; Step S1033: When the similarity meets the pre-set update condition, the corresponding ontology model is used as the digital twin model.

[0050] In step S1031, the fault database contains internal faults (such as protection device hardware failure, logic chip anomaly), external faults (such as transmission line short circuit, ground fault) and communication faults (such as SV message interruption, GOOSE link anomaly) and other fault scenes. The preprocessed historical perception data and the feature vectors in the fault database are matched by a pattern recognition algorithm (such as support vector machine SVM, random forest), so that the corresponding fault operation scene of the current data can be identified.

[0051] In step S1032, the ontology model predicts the running state data at the next time based on the historical perception data at the previous time, and then compares the historical state data at the next time with the corresponding running state data in terms of similarity (such as root mean square error, cosine similarity). If the similarity is lower than a threshold value, it indicates that the model simulation has a deviation, and the trainable parameters in the ontology model (such as the error compensation coefficient in the mechanism model) need to be adjusted through the back propagation algorithm; or the knowledge graph of the ontology model is manually intervened to correct (such as supplementing the missing fault logic rules), and then the semantic encoding is re-performed.

[0052] In step S1033, when the similarity meets the maximum threshold range or its value converges, it is determined that the update condition is reached and the ontology model is terminated. Taking the root mean square error (RMSE) as an example, if the RMSE of all key state parameters is less than 3% of the threshold value in the last N iterations, and the logical criterion accuracy is greater than 99.5%, the ontology model is terminated, and the ontology model at this time is taken as the digital twin model.

[0053] The embodiment drives the model to traverse the preset fault library through the historical data, and calibrates the model parameters based on the historical state data, so as to ensure that the digital twin model is highly consistent with the real device in terms of time sequence response and logical output, and continuously iterates the model through the actual running data, solves the characteristic deviation problem of the relay protection device caused by long-term operation, and improves the reliability of remote testing.

[0054] Step S20: consistency checking of the state prediction value and the perception data of the relay protection device at the next time and output of a checking report.

[0055] The consistency checking of the state prediction value and the perception data of the relay protection device at the next time specifically includes: performing correlation checking and / or parameter deviation rate checking on the real-time electrical quantity at the next time and the electrical prediction quantity waveform; and / or performing synchronization checking and / or contact stroke error checking on the real-time mechanical quantity at the next time and the mechanical prediction quantity action time; and / or performing communication link reliability checking on the real-time network security data at the next time and the network security prediction data.

[0056] The correlation checking is to judge the consistency of the real-time electrical quantity and the electrical quantity waveform predicted by the digital twin model in shape and trend through quantitative analysis of the similarity between them. The waveform overall similarity can be calculated by the Pearson correlation coefficient, and the similarity value ranges from -1 to 1. The closer the absolute value is to 1, the stronger the correlation is, and the higher the consistency is. It is commonly used for consistency checking of voltage, current, power and other waveform data.

[0057] Parameter deviation rate check is to evaluate the deviation degree of state prediction value by calculating the deviation percentage of perception data and state prediction value predicted by digital twin model. It is commonly used for consistency check of data such as protection action threshold, sampling frequency, signal delay time, etc.

[0058] Synchronization check is to verify whether the real-time mechanical quantity and the action timing predicted by the model are consistent for the mechanical action time of the relay protection device. It is commonly used for consistency check of data such as breaker opening and closing time, relay contact response time, etc.

[0059] Contact stroke error check monitors the displacement of the contact stroke or mechanism in real time through displacement sensors, checks the displacement stroke of the mechanical components, compares the real-time stroke data with the predicted theoretical stroke, and detects the wear, jamming or deformation of the mechanical components. It is commonly used for consistency check of mechanical structures such as relay contacts and circuit breaker operating mechanisms.

[0060] Communication link reliability check includes basic index check, compliance check and anti-interference ability check of error rate, packet loss rate and transmission delay, to evaluate the stability, integrity and anti-interference ability of the communication link. It is commonly used for consistency check of communication data of protection devices and transformers, switches and background systems.

[0061] In the check report, the basic information of the relay protection device is included, as well as the check purpose, check method, check result and handling suggestion of abnormal check items of various check items.

[0062] Step S30: determining the health status probability and network security risk value of the relay protection device according to the check report.

[0063] The check report includes the electrical quantity deviation (usually the data checked by correlation check and parameter deviation rate check), mechanical quantity deviation (usually the data checked by contact stroke error check), functional logic deviation (usually the data checked by synchronization check), communication error rate, transmission delay, number of illegal packets, number of encryption failures and packet compliance rate of the relay protection device. By extracting and analyzing the check results of various items in the check report, the health status probability and network security risk value of the relay protection device can be evaluated.

[0064] The determination of the health status probability and network security risk value of the relay protection device according to the check report specifically includes: Step S301: determining the health status probability of the relay protection device through a preset Bayesian classification model according to the electrical quantity deviation, the mechanical quantity deviation and the functional logic deviation; Step S302: Based on the preset network security rule, the risk weight of each of the communication error rate, the transmission delay, the number of illegal packets, the number of encryption failures and the packet compliance rate is determined, and the network security risk value is determined by weighted summation based on the risk weight.

[0065] In step S301, the calculation formula of the Bayesian classification model is:

[0066] wherein is the kth health state, is the electrical quantity deviation; is the mechanical quantity deviation; is the functional logic deviation; is the prior probability of the health state (obtained according to the historical operation data of the relay protection device); is the probability of the deviation feature under the condition that the relay protection device is in the health state , is the probability of the deviation feature under the condition that the relay protection device is in the health state , is the probability of the deviation feature under the condition that the relay protection device is in the health state , is the health state, is the pre-warning state, and is the fault state.

[0067] In step S302, the network security rule has a corresponding scoring level for each of the communication error rate, the transmission delay, the number of illegal packets, the number of encryption failures and the packet compliance rate, each scoring level has a different risk weight, and the corresponding network security risk value can be determined by weighted summation of each risk weight.

[0068] Step S40: The reliability of the relay protection device is determined based on the health state probability and the network security risk value.

[0069] Since the "health base score" only reflects the static evaluation of the current or historical state of the relay protection device, but the health state (such as aging and wear) of the relay protection device is a gradual process and has a certain time sequence correlation. At the same time, the "network security risk value" is a result obtained in a linear manner by weighted summation, but the influence of network security events has a nonlinear characteristic (such as a single encryption failure may have no effect, but high-frequency encryption failures may cause the data leakage risk to increase sharply). Therefore, the health base score and the network security risk value need to be corrected to adapt to the dynamic evolution law of the health state of the relay protection device and the network security, and to avoid evaluation deviation caused by relying only on current data. ​

[0070] determining the reliability of the relay protection device based on the health state probability and the network security risk value, specifically comprising: Step S401: converting the health state probability into a health base score of the relay protection device through weighted summation, and converting the health base score into a health correction score based on a predicted state transition function; Step S402: obtaining a network security base score by subtracting the network security risk value from a preset maximum security score, and converting the network security base score into a network security correction score based on a preset penalty function; Step S403: obtaining a reliability score through weighted summation based on the health correction score and the network security correction score.

[0071] In step S401, the categories of health state include (health state), (pre-warning state), and (fault state). The corresponding calculation formula of the weighted summation method is: , wherein is the base score of the health state, the pre-warning state, and the fault state, which can be adjusted adaptively according to historical fault detection results and factory settings.

[0072] Then, the health base score is corrected by introducing a state transition trend through a hidden Markov chain. The expression of the state transition function is:

[0073] wherein, is the health correction score, is the health base score, is the probability of transitioning from state i to state j, is a trend sensitivity coefficient (the value range is 0.1-0.2).

[0074] In step S402, the network security risk value SR is first converted into a security base score SCS through the formula , and then the security base score is corrected by coupling effect. The expression of the penalty function is:

[0075]

[0076] wherein, is the network security correction score, is the network security base score, is a penalty coefficient; is the probability of the relay protection device being in a fault pre-warning state, is a network security risk value.

[0077] In step S403, the reliability score weighted sum formula is: , and are weight coefficients of the health correction score and the network security correction score respectively, and , which can be adaptively set according to requirements.

[0078] When the reliability score deviates greatly from the actual historical experience, the reliability score can be further combined with the hidden Markov chain to update the state transition matrix :

[0079] wherein is a reliability sensitivity coefficient, and the value range is generally 0.3. The lower the score is, the higher the probability of transition to the fault state is.

[0080] In summary, in the operation evaluation method for the relay protection device in the embodiment, on the one hand, by establishing a digital twin model, the dynamic changes of the relay protection device are perceived in real time and state prediction is performed, various conditions of the relay protection device in actual operation are truly reproduced, the authenticity and accuracy of the state prediction of the relay protection device are improved, and the state prediction value is corrected in combination with the perception data of the next moment of the relay protection device, so that the distortion of the digital twin simulation is effectively avoided to cover up the real failure of the relay protection device, and the reliable operation of the power system is ensured. On the other hand, the health state probability is combined with the network security risk value to construct a multi-dimensional reliability evaluation system. Even if the physical state of the device is normal, the network security warning will be triggered, the protection function failure caused by network attacks is avoided, and the demand of the smart grid for the "information-physical" fusion security is met.

[0081] Based on the same inventive concept as the above embodiment, the embodiment further provides an operation evaluation system for a relay protection device, which further comprises a processor and a memory; wherein the memory stores a computer program, and the computer program is used to load and execute the operation evaluation method for the relay protection device as described above by the processor.

[0082] As shown in Figure 2 , based on the same inventive concept as the above embodiment, the embodiment further provides a computer readable storage medium, which stores instructions, and the instructions are used to load and execute the operation evaluation method for the relay protection device as described above by the processor.

[0083] In the embodiments of the mobile terminal and the computer readable storage medium provided in the present application, all the technical features of the embodiments of the control method are included, and the description and explanation content is basically the same as that of the embodiments of the method, which will not be repeated here.

[0084] The embodiments of the present application also provide a computer program product, which comprises computer program code, and when the computer program code is run on a computer, the computer is caused to execute the method in various possible embodiments.

[0085] The embodiments of the present application also provide a chip, which comprises a memory and a processor, the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the device installed with the chip executes the method in various possible embodiments.

[0086] The above-mentioned serial numbers of the embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments.

[0087] In the present application, for the same or similar term concept, technical scheme and / or application scene description, generally only the first time is described in detail, and for the sake of brevity, the repeated description is not repeated, and for the understanding of the technical scheme of the present application, the same or similar term concept, technical scheme and / or application scene description which is not described in detail can be referred to the related description before.

[0088] In the present application, the description of each embodiment has its own emphasis, and the part not described or recorded in a certain embodiment can be referred to the related description of other embodiments.

[0089] The technical features of the technical scheme of the present application can be combined arbitrarily, in order to make the description simple, all possible combinations of the technical features in the above-mentioned embodiments are not described, however, as long as the combination of these technical features does not exist contradictory, it should be considered as the range recorded in the present application.

[0090] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, can also be through hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application essentially or say the part of the prior art to make contributions can be embodied in the form of software products, the computer software product is stored in the above-mentioned storage medium, including a number of instructions to make a terminal device execute the method of each embodiment of the present application. The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, any equivalent structure or equivalent flow transformation using the content of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

[0091] It should be noted that each embodiment in the specification adopts a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts of each embodiment can be referred to.

[0092] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for evaluating the operation of a relay protection device, characterized in that: include: Taking the current perception data of the relay protection device as input, generating a state prediction value of the relay protection device based on a preset digital twin model; Performing consistency check on the state prediction value and the sensing data of the relay protection device at the next moment and outputting a check report; Determining the health status probability and network security risk value of the relay protection equipment according to the inspection report; The reliability of the relay protection device is determined based on the health state probability and the network security risk value.

2. The operation evaluation method for relay protection equipment according to claim 1 is characterized in that: Before generating the state prediction value of the relay protection device based on the preset digital twin model, the method further includes: Determining a knowledge graph of the relay protection device according to the static data of the relay protection device, the device model library, and the corresponding image library; Performing semantic encoding on the knowledge graph and obtaining an ontology model of the relay protection device; Taking historical perception data and historical state data as input, the ontology model is trained to obtain the digital twin model.

3. The operation evaluation method for relay protection equipment according to claim 2 is characterized in that: The device model library includes a physical model and a mechanism model of the relay protection device; the image library includes device images of the relay protection device; and the semantic encoding of the knowledge graph to obtain the ontology model of the relay protection device specifically includes: Performing semantic matching on the static data of the relay protection device and the physical parameters of the physical model to obtain a first semantic correspondence; determining an embedding vector of the static data and the physical parameters based on the first semantic correspondence and recording the embedding vector as a first embedding vector; Performing semantic matching on the static data of the relay protection device and the mechanism parameters of the mechanism model to obtain a second semantic correspondence; determining an embedding vector of the static data and the physical parameter according to the second semantic correspondence and recording the embedding vector as a second embedding vector; Mapping the static data to a low-dimensional semantic space in a word embedding manner to obtain a distribution vector of the static data and recording it as a first distribution vector; performing feature extraction and semantic encoding on the device image to obtain an encoded device image; mapping the encoded device image to a low-dimensional semantic space corresponding to the static data to obtain a distribution vector of the device image and recording it as a second distribution vector; determining a semantic association matrix between the first distribution vector and the second distribution vector, and determining an embedding vector for the static data and the device image based on the semantic association matrix and recording it as a third embedding vector; The first embedding vector, the second embedding vector, and the third embedding vector are used as real-valued weights to determine a target weighted group; and the target weighted group is added to the knowledge graph to obtain the ontology model.

4. The operation evaluation method for relay protection equipment according to claim 2 is characterized in that: The method of using historical perception data and historical state data as input to train the ontology model and obtain the digital twin model specifically includes: Based on the historical perception data, determining the fault operation scenario corresponding to the ontology model from a preset fault database; Taking the historical perception data as input, determining the operating state data of the ontology model in the fault operation scenario, and updating the ontology model based on the similarity between the operating state data and the historical state data to generate a new ontology model; When the similarity meets the preset update condition, the corresponding ontology model is used as the digital twin model.

5. The operation evaluation method for relay protection equipment according to claim 1 is characterized in that: The sensing data includes real-time electrical quantities, real-time mechanical quantities and real-time network security data; the state prediction value includes electrical prediction quantities, mechanical prediction quantities and network security prediction data; The consistency check between the state prediction value and the sensing data of the relay protection device at the next moment specifically includes: Performing correlation check and / or parameter deviation rate check on the real-time electrical quantity at the next moment and the electrical predicted quantity waveform; and / or Perform synchronization check and / or contact stroke error check on the real-time mechanical quantity at the next moment and the mechanical predicted quantity action time; and / or The real-time network security data at the next moment is used to perform a communication link reliability check with the network security prediction data.

6. The operation evaluation method for relay protection equipment according to claim 1 is characterized in that: The inspection report includes the electrical quantity deviation, mechanical quantity deviation, functional logic deviation, communication bit error rate, transmission delay, number of illegal messages, number of encryption failures, and message compliance rate of the relay protection device; and determining the health status probability and network security risk value of the relay protection device based on the inspection report specifically includes: Determining the health status probability of the relay protection device through a preset Bayesian classification model according to the electrical quantity deviation, the mechanical quantity deviation, and the functional logic deviation; Based on preset network security rules, the risk weight of each of the communication bit error rate, the transmission delay, the number of illegal messages, the number of encryption failures and the message compliance rate is determined, and the network security risk value is determined by weighted summation based on the risk weights.

7. An operation evaluation method for relay protection equipment according to any one of claims 1 to 6, characterized in that: Determining the reliability of the relay protection device based on the health status probability and the network security risk value specifically includes: Converting the health state probability into a basic health score of the relay protection device by weighted summation, and converting the basic health score into a modified health score based on a predicted state transition function; Obtaining a basic network security score by subtracting the network security risk value from a preset maximum security score, and converting the basic network security score into a modified network security score based on a preset penalty function; A reliability score is obtained by weighted summation based on the health modified score and the network security modified score.

8. The operation evaluation method for relay protection equipment according to claim 7, characterized in that: The expression of the state transfer function is: in, Corrected score for health, Score for health base, The probability of transitioning from state i to state j is, is the trend sensitivity coefficient.

9. The operation evaluation method for relay protection equipment according to claim 7, characterized in that: The expression of the penalty function is: in, Corrected score for cybersecurity, Score for Cybersecurity Basics, is the penalty coefficient; is the probability that the relay protection device is in a fault warning state, is the network security risk value.

10. An operation evaluation system for relay protection equipment, characterized in that: It comprises a processor and a memory; wherein the memory stores a computer program, and the computer program is used for the processor to load and execute an operation evaluation method for relay protection equipment as described in any one of claims 1 to 9.

Images