Method and apparatus for optimizing relay protection defect diagnosis measures

Abstract

A method and apparatus for optimizing relay protection defect diagnosis measures are disclosed. The method includes: collecting several substation anomaly information items; establishing a correlation matrix between defect points and defect diagnosis measures in an expert knowledge base; establishing an objective function based on the correlation matrix and the probability of each defect point; and solving for the optimal solution of the objective function to minimize the expected number of required defect diagnosis measures. The method and apparatus provided by this invention solve the problem that existing relay protection defect diagnosis methods fail to provide targeted and efficient defect diagnosis measures based on the probability of defect points. This helps field operators start with the diagnostic measures corresponding to the most likely defect points, locate defect points using the fewest diagnostic steps, achieve intelligent defect diagnosis, reduce defect investigation time, improve diagnostic efficiency, reduce the risk to power grid safety operation, and support the safe and stable operation of the power grid.

Description

Technical Field

[0001] This invention relates to the field of relay protection operation and maintenance technology, and more specifically, to a method and apparatus for optimizing relay protection defect diagnosis measures. Background Technology

[0002] Relay protection defect diagnosis is crucial for rapidly restoring device operation and improving the reliability of relay protection. Currently, relay protection defect diagnosis still relies heavily on on-site inspections or background message information, with insufficient application of online monitoring information and alarm information. Relay protection defect diagnosis and elimination also heavily depend on expert experience. Building an expert database is currently the main method for defect diagnosis. Based on the defect diagnosis expert database, an inference engine is constructed. Using user input data and the knowledge in the expert database, results are given according to certain inference rules. For cases where the rules are ambiguous, decision fusion (such as DS evidence theory) is used for processing. However, these cases are summaries of diagnostic methods for typical cases and have not yet extracted general rules for relay protection defect diagnosis. They can only provide maintenance personnel with some ideas and perspectives for defect diagnosis, but are insufficient to complete the entire defect diagnosis process. Data-driven methods mainly extract characteristic quantities that can characterize defects, such as device temperature, differential current, and optical port intensity, and use methods such as PCA and artificial neural networks to establish defect diagnosis rules and perform diagnosis. Secondary circuit defect diagnosis also requires modeling the connection relationship between equipment and circuit. Based on the topology model, a fault point location method based on Petri nets, semantic nets and other theories was established.

[0003] Relay protection defect diagnosis expert knowledge bases typically provide a set of defect diagnosis measures based on a single abnormal alarm message. On the one hand, for the same defect, multiple abnormal alarm messages are usually collected, and existing expert knowledge bases fail to comprehensively apply these multiple messages to propose targeted and efficient defect diagnosis measures. On the other hand, these expert knowledge bases do not consider the probability of each defect point, and the order of various defect diagnosis measures needs optimization. In extreme cases, placing the diagnosis measure corresponding to the most likely defect point last will severely reduce the efficiency of defect diagnosis. This results in time-consuming and inefficient defect diagnosis, failing to meet the higher demands placed on relay protection operation and maintenance for the safe operation of large AC / DC hybrid power grids. Summary of the Invention

[0004] In view of this, the present invention proposes a method and apparatus for optimizing relay protection defect diagnosis measures, aiming to solve the problem that existing relay protection defect diagnosis methods cannot provide targeted and efficient defect diagnosis measures based on the probability of the defect point.

[0005] In a first aspect, embodiments of the present invention provide a method for optimizing relay protection defect diagnosis measures, characterized in that the method includes: establishing an association matrix between defect points and defect diagnosis measures in an expert knowledge base; establishing an objective function based on the association matrix and the probability of each defect point; and solving for the optimal solution of the objective function to minimize the expected number of required defect diagnosis measures.

[0006] Furthermore, establishing the association matrix between defect points and defect diagnosis measures in the expert knowledge base includes: establishing an association matrix B = [b...] between defect points and necessary defect diagnosis measures in the expert knowledge base. kj ] K×J Among them, b kj =1 indicates that for defect point j, defect diagnosis measure k must be implemented before defect point j can be confirmed; b kj =0 indicates that the confirmation of defect point j does not require the implementation of defect diagnosis measure k; column vector b j =[b kj ] K×1 J represents the correlation vector of each defect diagnosis measure corresponding to defect point j; J represents the number of defect points, K represents the number of defect diagnosis measures, 1≤j≤J, 1≤k≤K, and J and K are both positive integers.

[0007] Further, establishing the objective function based on the correlation matrix and the probability of each defect point includes: based on the correlation matrix B = [b kj ] K×J The probability l of defect point j j Establish the following objective function:

[0008]

[0009] In a set of defect diagnosis measures sequence S, any element S(k) represents the execution sequence number of the k-th defect diagnosis measure.

[0010] Furthermore, the step of finding the optimal solution to the objective function to minimize the expected number of required defect diagnosis measures includes: using a nuclear decay algorithm to find the optimal solution to the objective function to minimize the expected number of required defect diagnosis measures.

[0011] Furthermore, the step of using the nuclear decay algorithm to solve the objective function to minimize the expected number of defect diagnosis measures includes: Step 1: Randomly combining K natural numbers as the initial order S of defect diagnosis measures. 0 Given an initial intensity C0, a half-life h, and a cutoff intensity C' for the radioactive particles, the intensity C of the radioactive particles in the r-th iteration is... r for:

[0012]

[0013] Where r is a positive integer, and the initial iteration r = 1:

[0014] Step 2: In the r-th iteration, set a random number α, 0 ≤ α ≤ 1; when α > 0.5, start from the current solution S 1 Two numbers are randomly selected from the given numbers and their positions are swapped to form the new solution S. 2 When α≤0.5, from the current solution S 1 Three numbers are randomly selected from the given numbers and their positions are swapped to form a new solution S. 2 If E(S) 2 )≤E(S 1 If a new solution is found, it is accepted; otherwise, it is determined according to probability p. iter Determine whether to accept the new solution; among which, If a new solution is received, it is taken as the current solution; this step is repeated g times, where g is a positive integer, and the solution obtained after repeating g times is taken as the optimal solution S for this iteration. best and the initial solution for the next iteration;

[0015] Step 3: Update the iteration number r = r + 1, when C r If the value is greater than C', return to step 2 and loop; otherwise, output the optimal solution S for each iteration. best As the final solution.

[0016] Furthermore, the statement according to probability p iter Determining whether to accept a new solution includes: generating random numbers 0 ≤ β ≤ 1, when β ≤ p iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0017] Furthermore, before establishing the association matrix between the defect points and the defect diagnosis measures in the expert knowledge base, the method further includes: collecting several alarm information items; obtaining several defect points based on the several alarm information items; and obtaining the probability of several defect points based on the several alarm information items and the several defect points.

[0018] Furthermore, the step of obtaining the probability of several defect points based on the several alarm information items and the several defect points includes: constructing a conditional probability matrix of the several defect points based on the several alarm information items and the several defect points; and obtaining the probability of the several defect points based on the conditional probability matrix and the importance of the several alarm information items.

[0019] Secondly, embodiments of the present invention also provide an apparatus for optimizing relay protection defect diagnosis measures, characterized in that the apparatus comprises: an association matrix construction unit for establishing an association matrix between defect points and defect diagnosis measures in an expert knowledge base; an objective function construction unit for establishing an objective function based on the association matrix and the probability of each defect point; and an optimization unit for solving the optimal solution of the objective function to minimize the expected number of required defect diagnosis measures.

[0020] Furthermore, the association matrix construction unit is also used to: establish an association matrix B = [b...] between defect points and necessary defect diagnosis measures in the expert knowledge base. kj ] K×J Among them, b kj =1 indicates that for defect point j, defect diagnosis measure k must be implemented before defect point j can be confirmed; b kj =0 indicates that the confirmation of defect point j does not require the implementation of defect diagnosis measure k; column vector b j =[b kj ] K×1 J represents the correlation vector of each defect diagnosis measure corresponding to defect point j; J represents the number of defect points, K represents the number of defect diagnosis measures, 1≤j≤J, 1≤k≤K, and J and K are both positive integers.

[0021] Furthermore, the objective function construction unit is also used to: based on the correlation matrix B = [b kj ] K×J The probability l of defect point j j Establish the following objective function:

[0022]

[0023] In a set of defect diagnosis measures sequence S, any element S(k) represents the execution sequence number of the k-th defect diagnosis measure.

[0024] Furthermore, the optimization unit is also used to solve the objective function using a nuclear decay algorithm to minimize the expected number of required defect diagnostic measures.

[0025] Furthermore, the step of using the nuclear decay algorithm to solve the objective function to minimize the expected number of defect diagnosis measures includes: Step 1: Randomly combining K natural numbers as the initial order S of defect diagnosis measures. 0 Given an initial intensity C0, a half-life h, and a cutoff intensity C' for the radioactive particles, the intensity C of the radioactive particles in the r-th iteration is... r for:

[0026]

[0027] Where r is a positive integer, and the initial iteration r = 1:

[0028] Step 2: In the r-th iteration, set a random number α, 0 ≤ α ≤ 1; when α > 0.5, start from the current solution S 1 Two numbers are randomly selected from the given numbers and their positions are swapped to form the new solution S. 2 When α≤0.5, from the current solution S 1 Three numbers are randomly selected from the given numbers and their positions are swapped to form a new solution S. 2 If E(S) 2 )≤E(S 1 If a new solution is found, it is accepted; otherwise, it is determined according to probability p. iter Determine whether to accept the new solution; among which, If a new solution is received, it is taken as the current solution; this step is repeated g times, where g is a positive integer, and the solution obtained after repeating g times is taken as the optimal solution S for this iteration. best and the initial solution for the next iteration;

[0029] Step 3: Update the iteration number r = r + 1, when C r If the value is greater than C', return to step 2 and loop; otherwise, output the optimal solution S for each iteration. best As the final solution.

[0030] Furthermore, the statement according to probability p iter Determining whether to accept a new solution includes: generating random numbers 0 ≤ β ≤ 1, when β ≤ p iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0031] Furthermore, the device also includes: an information collection unit for collecting several alarm information items; a defect point acquisition unit for obtaining several defect points based on the several alarm information items; and a probability calculation unit for obtaining the probability of several defect points based on the several alarm information items and the several defect points.

[0032] Furthermore, the probability calculation unit is also used to: construct a conditional probability matrix for the plurality of defect points based on the plurality of alarm information and the plurality of defect points; and obtain the probability of the plurality of defect points based on the conditional probability matrix and the importance of the plurality of alarm information.

[0033] Thirdly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the methods provided in the embodiments of the present invention.

[0034] The method and apparatus for optimizing relay protection defect diagnosis measures provided in this invention construct a target model based on the correlation matrix between defect points and defect diagnosis measures in an expert knowledge base, and the probability of each defect point occurring. The optimal solution is then obtained from the objective function to minimize the expected number of required defect diagnosis measures. This method optimizes relay protection defect diagnosis measures and provides guidance to field operators. It helps them start with the diagnosis measures corresponding to the most likely defect points, locating the defects with the fewest diagnostic steps, achieving intelligent defect diagnosis, reducing defect investigation time, improving diagnostic efficiency, lowering the risk to power grid safety, and supporting the safe and stable operation of the power grid. Attached Figure Description

[0035] Figure 1 A flowchart of a method for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram illustrating the association between defect points and defect diagnosis measures in an expert knowledge base, provided as an exemplary embodiment of the present invention. (a) to (h) are schematic diagrams illustrating the association between defect points and defect diagnosis measures in an expert knowledge base when there are voltage abnormalities, current abnormalities, contact potential abnormalities of the control box (or intelligent terminal), abnormalities in the activation of other protection devices, hardware abnormalities, abnormalities in the channel of the line protection device, abnormalities in remote transmission (or other protection actions), and abnormalities in time synchronization, respectively.

[0037] Figure 3 A flowchart of a method for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention;

[0038] Figure 4 This is a schematic diagram illustrating the relay protection configuration of a line bay in a conventional substation with a double busbar connection, provided as an exemplary embodiment of the present invention.

[0039] Figure 5 This is a schematic diagram illustrating the relay protection configuration of a line bay in a smart substation with a double busbar connection, provided as an exemplary embodiment of the present invention.

[0040] Figure 6 A schematic diagram illustrating the variation of the expected number of defect diagnosis measures implemented as a function of the number of iteration steps, as provided in an exemplary embodiment of the present invention;

[0041] Figure 7 A schematic diagram of the structure of an apparatus for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention;

[0042] Figure 8 This is a schematic diagram of a device for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention. Detailed Implementation

[0043] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0044] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0045] Figure 1 The flowchart illustrates a method for optimizing relay protection defect diagnosis measures, as provided in an exemplary embodiment of the present invention.

[0046] like Figure 1 As shown, the method includes:

[0047] Step S101: Establish a correlation matrix between defect points and defect diagnosis measures in the expert knowledge base.

[0048] In this embodiment of the invention, when establishing the association between eight typical anomalies—relay protection voltage anomaly, current anomaly, contact potential anomaly of the operating box (or intelligent terminal), anomaly of other protection device output, hardware anomaly, line protection device channel anomaly, remote transmission (or other protection action) anomaly, and time synchronization anomaly—the defect points are correlated with the defect diagnosis measures in the expert knowledge base.

[0049] Figure 2 This is a schematic diagram illustrating the association between defect points and defect diagnosis measures in an expert knowledge base, provided as an exemplary embodiment of the present invention. (a) to (h) are schematic diagrams illustrating the association between defect points and defect diagnosis measures in the expert knowledge base when there are voltage anomalies, current anomalies, contact potential anomalies in the control box (or intelligent terminal), abnormal activation of other protection devices, hardware anomalies, abnormal channel activation of line protection devices, abnormal remote transmission (or other protection actions), and time synchronization anomalies, respectively. Figure 2 As shown, different columns represent different defect points, and different rows represent defect diagnosis measures. When several defect diagnosis measures are adopted and the defect point can be located, it indicates that these defect diagnosis measures are necessary to determine the defect point, and they are highlighted in the table by color.

[0050] It is important to understand that the association between the aforementioned defects and the defect diagnosis measures in the expert knowledge base can be established in advance. This step involves constructing an association matrix based on this association through data modeling.

[0051] Further, step S101 includes:

[0052] Establish a correlation matrix B = [b] between defect points and necessary defect diagnosis measures in the expert knowledge base. kj ] K×J ;

[0053] Among them, b kj =1 indicates that for defect point j, defect diagnosis measure k must be implemented before defect point j can be confirmed;

[0054] b kj =0 indicates that the confirmation of defect point j does not require the implementation of defect diagnosis measure k;

[0055] column vector b j =[b kj ] K×1 This represents the correlation vector between each defect diagnosis measure corresponding to defect point j;

[0056] J represents the number of defect points, K represents the number of defect diagnosis measures, 1≤j≤J, 1≤k≤K, and J and K are both positive integers.

[0057] Step S102: Establish the objective function based on the correlation matrix and the probability of each defect point.

[0058] Based on the relationship between the correlation matrix and the probability of each defect occurring, different types of objective functions can be established.

[0059] Further, step S102 includes:

[0060] According to the correlation matrix B = [b kj ] K×J The probability l of defect point j j Establish the following objective function:

[0061]

[0062] In a set of defect diagnosis measures sequence S, any element S(k) represents the execution sequence number of the k-th defect diagnosis measure.

[0063] Step S103: Solve for the optimal solution of the objective function to minimize the expected number of required defect diagnosis measures.

[0064] In this embodiment of the invention, the order of defect diagnosis measures is optimized based on different defect point probability distributions to minimize the expected number of required defect diagnosis measures. The condition is that for any given defect point, the defect point can only be determined after all associated defect diagnosis measures have been implemented. The optimal solution to the objective function can be obtained using algorithms such as nuclear decay algorithms, genetic algorithms, simulated annealing algorithms, hill-climbing algorithms, and particle swarm optimization algorithms.

[0065] Further, step S103 includes:

[0066] The optimal solution to the objective function is obtained by using the nuclear decay algorithm, so as to minimize the expected number of defect diagnosis measures required.

[0067] The nuclear decay algorithm is used to solve the optimization problem of the objective function. According to high-energy physics theory, the time required for the radioactivity intensity to reach half of its original value is called the half-life of the isotope, which reflects the law of development and evolution of things. The evolution of the state of radioactive high-energy particles is not instantaneous, but undergoes continuous changes, and its energy intensity changes with time according to an exponential distribution.

[0068] Furthermore, the nuclear decay algorithm is used to solve for the optimal solution of the objective function, so as to minimize the expected number of required defect diagnosis measures, including:

[0069] Step 1: Randomly combine K natural numbers to form the initial sorting S of defect diagnosis measures. 0 Given an initial intensity C0, a half-life h, and a cutoff intensity C' for the radioactive particles, the intensity C of the radioactive particles in the r-th iteration is... r for:

[0070]

[0071] Where r is a positive integer, and the initial iteration r = 1:

[0072] Step 2: In the r-th iteration, set a random number α, 0≤α≤1;

[0073] When α > 0.5, from the current solution S 1 Two numbers are randomly selected from the given numbers and their positions are swapped to form the new solution S. 2 ;

[0074] When α≤0.5, from the current solution S 1 Three numbers are randomly selected from the given numbers and their positions are swapped to form a new solution S. 2 ;

[0075] If E(S) 2 )≤E(S 1 If a new solution is found, it is accepted; otherwise, it is determined according to probability p. iterDetermine whether to accept the new solution; among which, If a new solution is received, then the new solution will be used as the current solution;

[0076] This step is repeated g times, where g is a positive integer. The solution obtained after repeating this step g times is taken as the optimal solution S for this iteration. best and the initial solution for the next iteration;

[0077] Step 3: Update the iteration number r = r + 1, when C r If the value is greater than C', return to step 2 and loop; otherwise, output the optimal solution S for each iteration. best As the final solution.

[0078] In the above embodiments, when using the nuclear decay algorithm to find the optimal ranking of relay protection defect diagnosis measures, the nuclear decay algorithm searches randomly. In order to search for a better solution near the current solution, in the early stages of iteration, the nuclear decay algorithm is allowed to accept a suboptimal solution so that a better solution can be searched in the vicinity of the suboptimal solution. As the iteration progresses, the probability of accepting a suboptimal solution decreases so that it eventually converges to the searched optimal solution.

[0079] Furthermore, according to probability p iter Determining whether to accept the new solution includes:

[0080] Generate random numbers 0 ≤ β ≤ 1, when β ≤ p iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0081] In E(S) 2 )>E(S 1 Under the condition that 0 ≤ β ≤ 1, generate random numbers when β ≤ p. iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0082] Relying solely on expert knowledge bases for defect diagnosis requires sequentially implementing the necessary diagnostic measures for each defect point to pinpoint the specific defects within each typical anomaly. Figure 2 As shown, to locate each defect point among the eight typical anomalies, it is necessary to complete all the defect diagnosis measures highlighted in color in the column where the defect point is located in this table. Since there is not a one-to-one correspondence between defect points and defect diagnosis measures, the optimal sorting method for defect diagnosis measures cannot be directly determined based on the probability of the defect point and the association between the defect point and the defect diagnosis measures in the expert knowledge base.

[0083] The above embodiments construct a target model based on the correlation matrix between defect points and defect diagnosis measures in the expert knowledge base, and the probability of each defect point occurring. The optimal solution is then obtained from the objective function to minimize the expected number of required defect diagnosis measures. The method provided in this invention optimizes relay protection defect diagnosis measures and provides it to field operators. This guides them to start with the diagnostic measures corresponding to the most likely defect points, locate the defect points using the fewest diagnostic steps, achieve intelligent defect diagnosis, reduce defect investigation time, improve diagnostic efficiency, reduce the risk to power grid safety operation, and support the safe and stable operation of the power grid.

[0084] Figure 3 The flowchart illustrates a method for optimizing relay protection defect diagnosis measures, as provided in an exemplary embodiment of the present invention.

[0085] like Figure 3 As shown, the method includes steps S301-306, wherein steps S304-306 are related to... Figure 1 Steps S101-103 are the same as those in step S304, and will not be repeated here. Before step S304, the following steps are also included:

[0086] Step S301: Collect several alarm messages.

[0087] In this embodiment of the invention, the alarm information can be online abnormal alarm information of the relay protection of the substation within a preset time interval. It can be one item, two items or more items, and can be used as the basis for analyzing the scope and cause of the defect.

[0088] Step S302: Based on several alarm messages, obtain several defect points.

[0089] In this embodiment of the invention, the range of defect points that cause each individual alarm message can be considered as the defect range associated with that alarm message. The defect range associated with an individual alarm message generally starts from the device that issued the abnormal message (the abnormal object is an analog quantity, a digital quantity, etc.) and ends at that device, encompassing a closed-loop path including the hardware / ports of both ends, intermediate devices / ports, secondary circuits, and channels. Based on each alarm message, a closed-loop path of the defect range can be obtained. The node faults included in this closed-loop path are the defect points. A closed-loop path of a defect range can include several defect points.

[0090] Specifically, Figure 4 This is a schematic diagram illustrating the relay protection configuration of a line bay in a conventional substation with a double busbar connection, provided as an exemplary embodiment of the present invention. Figure 4 As shown, based on the object classification of alarm information, taking the defect range of conventional substation relay protection as an example, the closed-loop path of the defect range associated with alarm information is generally as follows:

[0091] 1) When the object of the alarm information is voltage (e.g., TV disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: voltage transformer secondary winding → voltage secondary circuit (via terminal box / control cabinet) → voltage paralleling device in voltage paralleling panel (with air switch at the inlet) → voltage secondary circuit (entering the terminal block of the protection panel) → voltage switching device (with air switch at the outlet) → protection device sampling board → protection device CPU board; for bus protection devices, the above closed-loop path does not include the voltage switching device; when the collected voltage is line voltage and the substation wiring form is 3 / 2 connection, the above closed-loop path does not include voltage paralleling panel, voltage paralleling device, and voltage switching device, as shown by line segment ① in the figure;

[0092] 2) When the object of the alarm information is current (e.g., TA disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: current transformer secondary winding → current secondary circuit (via terminal box / control cabinet) → current secondary circuit (entering the terminal block of the protection panel) → protection device sampling board → protection device CPU board, as shown by line segment mark ② in the figure;

[0093] 3) When the object of the alarm information is the contact potential of the control box (or intelligent terminal) (e.g., trip position abnormal alarm), the closed-loop path of the defect range associated with the alarm information generally consists of two parts. One part is the path of the switching quantity, and the other part is the path of the related current quantity. The path of the current quantity is as described above. The path of the switching quantity is: control box contact → switching quantity secondary circuit (via terminal box / control cabinet) → switching quantity secondary circuit (entering the terminal block of the protection panel) → protection device input board → protection device CPU board, as shown by line segment ③ in the figure.

[0094] 4) When the alarm information is related to the start failure or tripping signal of other protection devices (e.g., tripping input abnormal alarm), the closed-loop path of the defect range associated with the alarm information is generally: source protection device CPU board → source protection device output board → secondary circuit of switch quantity (via the terminal block of the source protection panel) → secondary circuit of switch quantity (via the terminal box / control cabinet) → secondary circuit of switch quantity (terminal block of the receiving protection panel) → receiving protection device input board → receiving protection device CPU board, as shown by line segment ④ in the figure;

[0095] 5) When the object of the alarm information is device hardware (e.g., power supply module abnormality alarm, input module abnormality alarm, RAM memory abnormality alarm), taking the power supply module abnormality alarm as an example, the closed-loop path of the defect range associated with this alarm information is generally: DC power supply → DC circuit → device power supply module → device CPU board, as shown by line segment label ⑤ in the figure.

[0096] 6) When the object of the alarm information is a device channel (e.g., channel abnormality alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking a multiplexed fiber optic channel as an example): CPU board of the opposite line protection device → management board of the opposite line protection device → secondary circuit → screen of the opposite multiplexed interface device → coaxial cable → digital distribution frame of the opposite side → communication channel (including communication screens on both sides) → digital distribution frame of this side → coaxial cable → screen of the current multiplexed interface device of this side → secondary circuit → management board of the current line protection device → CPU board of the current line protection device, as shown by line segment mark ⑥ in the figure;

[0097] 7) When the object of the alarm information is a remote transmission or remote trip signal (e.g., a long-term remote input abnormal alarm), the closed-loop path of the defect range associated with the alarm information includes two parts. One part is the output of other protection devices on this side and related switching circuits, and the other part is the output of other protection devices on the opposite side and related switching circuits. Specifically, it includes: bus protection, circuit breaker protection, high-resistance protection and other protection CPU plug-in → other protection output plug-in → switching secondary circuit → line protection device input plug-in → line protection device CPU plug-in and other protection CPU plug-in of bus protection, circuit breaker protection, high-resistance protection and other protection on the opposite side → other protection output plug-in → switching secondary circuit → line protection device input plug-in → line protection device CPU plug-in, as shown by line segment label ⑦ in the figure;

[0098] 8) When the object of the alarm information is the time synchronization signal (e.g., time synchronization anomaly alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking B code time synchronization as an example): satellite → antenna → time synchronization source → secondary circuit → protection device management board → protection device CPU board, as shown by line segment mark ⑧ in the figure.

[0099] Specifically, Figure 5 This is a schematic diagram illustrating the relay protection configuration of a line bay in a smart substation with a double busbar connection, provided as an exemplary embodiment of the present invention. Figure 5 As shown, based on the object classification of alarm information, taking the defect range of conventional substation relay protection as an example, the closed-loop path of the defect range associated with alarm information is generally as follows:

[0100] 1) When the object of the alarm information is voltage (e.g., TV disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: voltage transformer secondary winding → voltage analog circuit (including air switch) → bus merging unit → cascaded fiber optic circuit → branch merging unit → direct sampling fiber optic circuit → branch protection device; for the voltage of the input bus protection device, the voltage of the transformer low-voltage winding, and the line voltage, the closed-loop path is: voltage transformer secondary winding → voltage analog circuit (including air switch) → merging unit → direct sampling fiber optic circuit → protection device, as shown by line segment ① in the figure;

[0101] 2) When the object of the alarm information is current (e.g., TA disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: current transformer secondary winding → current analog quantity loop → merging unit → direct acquisition fiber optic loop → protection device, as shown by line segment mark ② in the figure.

[0102] 3) When the object of the alarm information is the contact potential of the control box (or smart terminal) (e.g., trip position abnormal alarm), the closed-loop path of the defect range associated with the alarm information generally consists of two parts. One part is the path of the switching quantity, and the other part is the path of the related current quantity. The path of the current quantity is as described above, and the path of the switching quantity is specifically: smart terminal input contact → smart terminal → GOOSE point-to-point loop → protection device, as shown by line segment ③ in the figure.

[0103] 4) When the object of the alarm information is the start failure or failure trip signal of other protection devices (e.g., failure trip input abnormal alarm), the closed-loop path of the defect range associated with the alarm information is generally: source protection device CPU board → source protection device optical port → GOOSE networking loop → receiving protection device optical port → receiving protection device CPU board, as shown by line segment ④ in the figure.

[0104] 5) When the object of the alarm information is device hardware (e.g., power supply module abnormality alarm, input module abnormality alarm, RAM memory abnormality alarm), taking the power supply module abnormality alarm as an example, the closed-loop path of the defect range associated with this alarm information is generally: DC power supply → DC circuit → device power supply module → device CPU board, as shown by line segment label ⑤ in the figure.

[0105] 6) When the object of the alarm information is a device channel (e.g., channel abnormality alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking a multiplexed fiber optic channel as an example): CPU board of the opposite line protection device → management board of the opposite line protection device → secondary circuit → screen of the opposite multiplexed interface device → coaxial cable → digital distribution frame of the opposite side → communication channel (including communication screens on both sides) → digital distribution frame of this side → coaxial cable → screen of the current multiplexed interface device of this side → secondary circuit → management board of the current line protection device → CPU board of the current line protection device, as shown by line segment mark ⑥ in the figure;

[0106] 7) When the object of the alarm information is a remote transmission or remote trip signal (e.g., a long-term remote input abnormal alarm), the closed-loop path of the defect range associated with the alarm information includes two parts. One part is the output of other protection devices on this side and related switching circuits, and the other part is the output of other protection devices on the opposite side and related switching circuits. The specific path includes: bus protection, circuit breaker protection, intelligent terminal and other protection CPU plug-in → other protection switching terminal → GOOSE networking circuit → line protection device optical port → line protection device CPU plug-in and bus protection, circuit breaker protection, intelligent terminal and other protection CPU plug-in on the opposite side → other protection switching terminal → GOOSE networking circuit → line protection device optical port → line protection device CPU plug-in, as shown by line segment label ⑦ in the figure;

[0107] 8) When the object of the alarm information is the time synchronization signal (e.g., time synchronization anomaly alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking B code time synchronization as an example): satellite → antenna → time synchronization source → secondary circuit → protection device management board → protection device CPU board, as shown by line segment mark ⑧ in the figure.

[0108] Step S303: Based on several alarm messages and several defect points, obtain the probability of several defect points.

[0109] In this embodiment of the invention, when new alarm information appears within a certain time interval or under the condition that historical alarm information has not been eliminated, the alarm information is correlated, indicating that these alarm information are caused by at least one identical defect, and the elimination of one or more defects means the recurrence of related multiple alarm signals. Each alarm event can be plotted as a closed-loop path of the defect range associated with the alarm, and a fault at any node on the closed-loop path can cause the alarm signal to occur. The intersection of the cause and location of one or more defects within the defect range of each alarm is most likely to occur. Based on several alarm information and several defect points, the intersection of defect points is calculated to define the scope of defect diagnosis, comprehensively determine the probability of each defect point, and provide a basis for selecting a reasonable combination of relay protection defect diagnosis measures. Compared with analyzing and diagnosing according to the defect range of any single alarm, the defect diagnosis process is more targeted, reducing the workload of operators analyzing multiple alarms that appear in a short period, grasping the main issues, improving the efficiency of defect diagnosis, and enhancing the intelligence level of defect diagnosis.

[0110] Further, step S303 includes:

[0111] Based on several alarm messages and several defect points, construct a conditional probability matrix for several defect points;

[0112] Based on the conditional probability matrix and the importance of several alarm messages, the probability of several defect points is obtained.

[0113] Specifically, there is I alarm messages f i (1≤i≤i), there are a total of J possible defect points u. j (1≤j≤J). Based on the i-th alarm message, the conditional probability of J defect points is known to be P(u...). j |f i Using I alarm messages as rows of a matrix and J possible defect points as columns, we construct matrix A = [a ij ] I×J =[P(u j |f i )] I×J Considering the importance of I alarm messages, set a weight vector w = [w i ] 1×I Then, the probability size l' of the defect point that integrates I alarm messages can be calculated using A and w:

[0114]

[0115] l' is a row vector containing J elements, determined by A and w. The operator is used for calculation. First, the vectors involved in the operation are element-wise minimized, then the elements are maximized to obtain the result, which is a single number. The result is l'. j (1≤j≤J), which represents the probability of the j-th defect point, and its importance is related to the importance w of the I alarm messages. i (1≤i≤I) and the conditional probability P(u) of each defect point after the alarm information is generated. j |f i The relationship between (1≤i≤I, 1≤j≤J) is:

[0116]

[0117] Due to ∑ 1≤j≤J l' j It is usually not equal to 1, so the row vector l' needs to be normalized, that is...

[0118]

[0119] The obtained l is the probability of J defect points that combine I alarm information.

[0120] Example 1

[0121] The relay protection device simultaneously reported "CT disconnection" and "TWJ abnormality". Defects related to the CT disconnection alarm information include... Figure 2 As shown in (b), the defects related to TWJ abnormal alarm information are as follows: Figure 2 As shown in (d), by integrating them, the defect points are shown in the table below:

[0122] j Defect Name 1 Virtual terminal error (merging unit, smart terminal to protection device) 2 Protection device malfunction 3 Incorrect transformer ratio setting 4 Fiber optic loop anomaly (from merging unit, smart terminal to protection device) 5 Merging unit failure 6 Terminal box / control cabinet to merging unit fault 7 CT secondary winding to terminal box / control cabinet fault 8 CT body malfunction 9 Smart terminal malfunction 10 TWJ circuit fault 11 Switch auxiliary contact fault

[0123] First, calculate the conditional probability of each defect point under the condition that two online self-test alarm messages occur simultaneously, as the basis for ranking the defect diagnosis measures.

[0124] Let i = 1 represent a CT disconnection alarm and i = 2 represent a TWJ anomaly alarm, and obtain the conditional probability matrix A = [a] for each defect point when the self-test alarm information occurs. ij ] I×J =[P(u j |f i )] 2×11 :

[0125]

[0126] In the above formula, A 1,10 =A 2,10 =0. In fact, this case selects the intelligent substation line protection device for defect diagnosis, and the 10th defect point, i.e., the "TWJ circuit fault" at j=10, only occurs in conventional substations. The intelligent substation uses GOOSE networking to transmit switching signals, so A 1,10 =A 2,10 =0.

[0127] Let w = [0.45, 0.55], then... It can be calculated that:

[0128] l'=[0.18,0.17,0.03,0.15,0.22,0.11,0.12,0.03,0.21,0,0.26];

[0129] Normalizing l' yields:

[0130] l=[0.122,0.115,0.020,0.101,0.149,0.074,0.081,0.020,0.142,0.000,0.176];

[0131] l represents the conditional probability of each defect point under the condition that the two online self-test alarm messages, "CT disconnection" and "TWJ abnormality," occur simultaneously. For example, the probability of the defect point "virtual terminal error (merging unit, intelligent terminal to protection device)" is 0.122.

[0132] The defect diagnosis measures are shown in the table below:

[0133] k Defect diagnostic measures 1 Check the virtual terminal connection 2 Restored after replacing the line protection module 3 Check the protection device ratio 4 Fiber optic loop inspection 5 The merging unit outputs SV messages normally. 6 The analog input of the merging unit is normal. 7 The current at the terminal box / control cabinet is normal. 8 Both sets of protection reported CT line breakage. 9 The smart terminal output matches the actual position of the switch. 10 The smart terminal input matches the actual position of the switch. 11 The auxiliary contact of the switch is consistent with the actual position.

[0134] Establish a correlation matrix B between defect diagnosis measures and defect points. 11×11 :

[0135]

[0136] Take S 0 =[1,2,3,4,5,6,7,8,9,10,11],C0=1,C'=0.03,h=40,g=10,The expected number of defect diagnosis measures implemented varies with the number of iteration steps as follows: Figure 6 As shown.

[0137] When the initial defect diagnosis measures are ordered, i.e., implemented one by one according to k from smallest to largest, the expected number of defect diagnosis measures to be implemented is 5.888, meaning that an average of 5.888 defect diagnosis measures are needed to obtain a defect diagnosis conclusion. The algorithm of this embodiment optimizes the implementation order of defect diagnosis measures, resulting in the following optimized implementation order:

[0138] S best =[11,1,2,4,5,6,7,8,10,9,3];

[0139] Based on the optimization results, the 11th defect diagnosis measure, "the auxiliary contact of the switch is consistent with the actual position," was implemented first, followed by the 1st defect diagnosis measure, "check the virtual terminal connection," and so on. The expected number of defect diagnosis measures implemented was ultimately reduced to 5.029. This demonstrates that by optimizing the defect diagnosis measures, the most likely defect point was located using the fewest diagnostic steps, achieving intelligent defect diagnosis, reducing defect investigation time, and lowering the risk to safe power grid operation.

[0140] The optimal defect diagnosis measures, ranked as follows, are as follows when both "CT disconnection" and "TWJ abnormality" online self-check alarms occur simultaneously:

[0141] k Defect diagnostic measures order 1 Check the virtual terminal connection 2 2 Recovery after replacing the line protection module 3 3 Check the protection device ratio 11 4 Fiber optic loop inspection 4 5 The merging unit outputs SV messages normally. 5 6 The analog input of the merging unit is normal. 6 7 The current at the terminal box / control cabinet is normal. 7 8 Both sets of protection reported CT line breakage. 8 9 The smart terminal output matches the actual position of the switch. 10 10 The smart terminal input matches the actual position of the switch. 9 11 The auxiliary contact of the switch is consistent with the actual position. 1

[0142] Figure 7 This is a schematic diagram of a device for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention.

[0143] like Figure 7 As shown, the device includes:

[0144] The association matrix construction unit 701 is used to establish an association matrix between defect points and defect diagnosis measures in the expert knowledge base.

[0145] In this embodiment of the invention, when establishing the association between eight typical anomalies—relay protection voltage anomaly, current anomaly, contact potential anomaly of the operating box (or intelligent terminal), anomaly of other protection device output, hardware anomaly, line protection device channel anomaly, remote transmission (or other protection action) anomaly, and time synchronization anomaly—the defect points are correlated with the defect diagnosis measures in the expert knowledge base.

[0146] Figure 2 This is a schematic diagram illustrating the association between defect points and defect diagnosis measures in an expert knowledge base, provided as an exemplary embodiment of the present invention. (a) to (h) are schematic diagrams illustrating the association between defect points and defect diagnosis measures in the expert knowledge base when there are voltage anomalies, current anomalies, contact potential anomalies in the control box (or intelligent terminal), abnormal activation of other protection devices, hardware anomalies, abnormal channel activation of line protection devices, abnormal remote transmission (or other protection actions), and time synchronization anomalies, respectively. Figure 2 As shown, different columns represent different defect points, and different rows represent defect diagnosis measures. When several defect diagnosis measures are adopted and the defect point can be located, it indicates that these defect diagnosis measures are necessary to determine the defect point, and they are highlighted in the table by color.

[0147] It is important to understand that the association between the aforementioned defects and the defect diagnosis measures in the expert knowledge base can be established in advance. This unit aims to construct an association matrix based on this association through data modeling.

[0148] Furthermore, the association matrix construction unit 701 is also used for:

[0149] Establish a correlation matrix B = [b] between defect points and necessary defect diagnosis measures in the expert knowledge base. kj ] K×J ;

[0150] Among them, b kj =1 indicates that for defect point j, defect diagnosis measure k must be implemented before defect point j can be confirmed;

[0151] b kj =0 indicates that the confirmation of defect point j does not require the implementation of defect diagnosis measure k;

[0152] column vector b j =[b kj ] K×1 This represents the correlation vector between each defect diagnosis measure corresponding to defect point j;

[0153] J represents the number of defect points, K represents the number of defect diagnosis measures, 1≤j≤J, 1≤k≤K, and J and K are both positive integers.

[0154] The objective function construction unit 702 is used to build an objective function based on the correlation matrix and the probability of each defect point.

[0155] Based on the relationship between the correlation matrix and the probability of each defect occurring, different types of objective functions can be established.

[0156] Furthermore, the objective function construction unit 702 is also used for:

[0157] According to the correlation matrix B = [b kj ] K×J The probability l of defect point j j Establish the following objective function:

[0158]

[0159] In a set of defect diagnosis measures sequence S, any element S(k) represents the execution sequence number of the k-th defect diagnosis measure.

[0160] The optimization unit 703 is used to solve for the optimal solution of the objective function so as to minimize the expected number of required defect diagnosis measures.

[0161] In this embodiment of the invention, the order of defect diagnosis measures is optimized based on different defect point probability distributions to minimize the expected number of required defect diagnosis measures. The condition is that for any given defect point, the defect point can only be determined after all associated defect diagnosis measures have been implemented. The optimal solution to the objective function can be obtained using algorithms such as nuclear decay algorithms, genetic algorithms, simulated annealing algorithms, hill-climbing algorithms, and particle swarm optimization algorithms.

[0162] Furthermore, the optimization unit 703 is also used for:

[0163] The optimal solution to the objective function is obtained by using the nuclear decay algorithm, so as to minimize the expected number of defect diagnosis measures required.

[0164] The nuclear decay algorithm is used to solve the optimization problem of the objective function. According to high-energy physics theory, the time required for the radioactivity intensity to reach half of its original value is called the half-life of the isotope, which reflects the law of development and evolution of things. The evolution of the state of radioactive high-energy particles is not instantaneous, but undergoes continuous changes, and its energy intensity changes with time according to an exponential distribution.

[0165] Furthermore, the nuclear decay algorithm is used to solve for the optimal solution of the objective function, so as to minimize the expected number of required defect diagnosis measures, including:

[0166] Step 1: Randomly combine K natural numbers to form the initial sorting S of defect diagnosis measures. 0 Given an initial intensity C0, a half-life h, and a cutoff intensity C' for the radioactive particles, the intensity C of the radioactive particles in the r-th iteration is...r for:

[0167]

[0168] Where r is a positive integer, and the initial iteration r = 1:

[0169] Step 2: In the r-th iteration, set a random number α, 0≤α≤1;

[0170] When α > 0.5, from the current solution S 1 Two numbers are randomly selected from the given numbers and their positions are swapped to form the new solution S. 2 ;

[0171] When α≤0.5, from the current solution S 1 Three numbers are randomly selected from the given numbers and their positions are swapped to form a new solution S. 2 ;

[0172] If E(S) 2 )≤E(S 1 If a new solution is found, it is accepted; otherwise, it is determined according to probability p. iter Determine whether to accept the new solution; among which, If a new solution is received, then the new solution will be used as the current solution;

[0173] This step is repeated g times, where g is a positive integer. The solution obtained after repeating this step g times is taken as the optimal solution S for this iteration. best and the initial solution for the next iteration;

[0174] Step 3: Update the iteration number r = r + 1, when C r If the value is greater than C', return to step 2 and loop; otherwise, output the optimal solution S for each iteration. best As the final solution.

[0175] In the above embodiments, when using the nuclear decay algorithm to find the optimal ranking of relay protection defect diagnosis measures, the nuclear decay algorithm searches randomly. In order to search for a better solution near the current solution, in the early stages of iteration, the nuclear decay algorithm is allowed to accept a suboptimal solution so that a better solution can be searched in the vicinity of the suboptimal solution. As the iteration progresses, the probability of accepting a suboptimal solution decreases so that it eventually converges to the searched optimal solution.

[0176] Furthermore, according to probability p iter Determining whether to accept the new solution includes:

[0177] Generate random numbers 0 ≤ β ≤ 1, when β ≤ p iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0178] In E(S) 2 )>E(S 1Under the condition that 0 ≤ β ≤ 1, generate random numbers when β ≤ p. iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

[0179] Relying solely on expert knowledge bases for defect diagnosis requires sequentially implementing the necessary diagnostic measures for each defect point to pinpoint the specific defects within each typical anomaly. Figure 2 As shown, to locate each defect point among the eight typical anomalies, it is necessary to complete all the defect diagnosis measures highlighted in color in the column where the defect point is located in this table. Since there is not a one-to-one correspondence between defect points and defect diagnosis measures, the optimal sorting method for defect diagnosis measures cannot be directly determined based on the probability of the defect point and the association between the defect point and the defect diagnosis measures in the expert knowledge base.

[0180] The above embodiments construct a target model based on the correlation matrix between defect points and defect diagnosis measures in the expert knowledge base, and the probability of each defect point occurring. The optimal solution is then obtained from the objective function to minimize the expected number of required defect diagnosis measures. The method provided in this invention optimizes relay protection defect diagnosis measures and provides it to field operators. This guides them to start with the diagnostic measures corresponding to the most likely defect points, locate the defect points using the fewest diagnostic steps, achieve intelligent defect diagnosis, reduce defect investigation time, improve diagnostic efficiency, reduce the risk to power grid safety operation, and support the safe and stable operation of the power grid.

[0181] Figure 8 This is a schematic diagram of a device for optimizing relay protection defect diagnosis measures, provided as an exemplary embodiment of the present invention.

[0182] like Figure 8 As shown, the device includes units 801-806, wherein units 804-806 and Figure 7 Units 701-703 are the same as those in the previous section, and will not be described again here. This device also includes:

[0183] Information collection unit 801 is used to collect several alarm information items.

[0184] In this embodiment of the invention, the alarm information can be online abnormal alarm information of the relay protection of the substation within a preset time interval. It can be one item, two items or more items, and can be used as the basis for analyzing the scope and cause of the defect.

[0185] The defect acquisition unit 802 is used to obtain a number of defect points based on a number of alarm messages.

[0186] In this embodiment of the invention, the range of defect points that cause each individual alarm message can be considered as the defect range associated with that alarm message. The defect range associated with an individual alarm message generally starts from the device that issued the abnormal message (the abnormal object is an analog quantity, a digital quantity, etc.) and ends at that device, encompassing a closed-loop path including the hardware / ports of both ends, intermediate devices / ports, secondary circuits, and channels. Based on each alarm message, a closed-loop path of the defect range can be obtained. The node faults included in this closed-loop path are the defect points. A closed-loop path of a defect range can include several defect points.

[0187] Specifically, Figure 4 This is a schematic diagram illustrating the relay protection configuration of a line bay in a conventional substation with a double busbar connection, provided as an exemplary embodiment of the present invention. Figure 4 As shown, based on the object classification of alarm information, taking the defect range of conventional substation relay protection as an example, the closed-loop path of the defect range associated with alarm information is generally as follows:

[0188] 1) When the object of the alarm information is voltage (e.g., TV disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: voltage transformer secondary winding → voltage secondary circuit (via terminal box / control cabinet) → voltage paralleling device in voltage paralleling panel (with air switch at the inlet) → voltage secondary circuit (entering the terminal block of the protection panel) → voltage switching device (with air switch at the outlet) → protection device sampling board → protection device CPU board; for bus protection devices, the above closed-loop path does not include the voltage switching device; when the collected voltage is line voltage and the substation wiring form is 3 / 2 connection, the above closed-loop path does not include voltage paralleling panel, voltage paralleling device, and voltage switching device, as shown by line segment ① in the figure;

[0189] 2) When the object of the alarm information is current (e.g., TA disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: current transformer secondary winding → current secondary circuit (via terminal box / control cabinet) → current secondary circuit (entering the terminal block of the protection panel) → protection device sampling board → protection device CPU board, as shown by line segment mark ② in the figure;

[0190] 3) When the object of the alarm information is the contact potential of the control box (or intelligent terminal) (e.g., trip position abnormal alarm), the closed-loop path of the defect range associated with the alarm information generally consists of two parts. One part is the path of the switching quantity, and the other part is the path of the related current quantity. The path of the current quantity is as described above. The path of the switching quantity is specifically: control box contact → switching quantity secondary circuit (via terminal box / control cabinet) → switching quantity secondary circuit (entering the terminal block of the protection panel) → protection device input board → protection device CPU board, as shown by line segment ③ in the figure.

[0191] 4) When the alarm information is related to the start failure or tripping signal of other protection devices (e.g., tripping input abnormal alarm), the closed-loop path of the defect range associated with the alarm information is generally: source protection device CPU board → source protection device output board → secondary circuit of switch quantity (via the terminal block of the source protection panel) → secondary circuit of switch quantity (via the terminal box / control cabinet) → secondary circuit of switch quantity (terminal block of the receiving protection panel) → receiving protection device input board → receiving protection device CPU board, as shown by line segment ④ in the figure;

[0192] 5) When the object of the alarm information is device hardware (e.g., power supply module abnormality alarm, input module abnormality alarm, RAM memory abnormality alarm), taking the power supply module abnormality alarm as an example, the closed-loop path of the defect range associated with this alarm information is generally: DC power supply → DC circuit → device power supply module → device CPU board, as shown by line segment label ⑤ in the figure.

[0193] 6) When the object of the alarm information is a device channel (e.g., channel abnormality alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking a multiplexed fiber optic channel as an example): CPU board of the opposite line protection device → management board of the opposite line protection device → secondary circuit → screen of the opposite multiplexed interface device → coaxial cable → digital distribution frame of the opposite side → communication channel (including communication screens on both sides) → digital distribution frame of this side → coaxial cable → screen of the current multiplexed interface device of this side → secondary circuit → management board of the current line protection device → CPU board of the current line protection device, as shown by line segment mark ⑥ in the figure;

[0194] 7) When the object of the alarm information is a remote transmission or remote trip signal (e.g., a long-term remote input abnormal alarm), the closed-loop path of the defect range associated with the alarm information includes two parts. One part is the output of other protection devices on this side and related switching circuits, and the other part is the output of other protection devices on the opposite side and related switching circuits. The specific path includes: bus protection, circuit breaker protection, high-resistance protection and other protection CPU plug-in → other protection output plug-in → switching secondary circuit → line protection device input plug-in → line protection device CPU plug-in and other protection CPU plug-in of bus protection, circuit breaker protection, high-resistance protection and other protection on the opposite side → other protection output plug-in → switching secondary circuit → line protection device input plug-in → line protection device CPU plug-in, as shown by line segment label ⑦ in the figure;

[0195] 8) When the object of the alarm information is the time synchronization signal (e.g., time synchronization anomaly alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking B code time synchronization as an example): satellite → antenna → time synchronization source → secondary circuit → protection device management board → protection device CPU board, as shown by line segment mark ⑧ in the figure.

[0196] Specifically, Figure 5This is a schematic diagram illustrating the relay protection configuration of a line bay in a smart substation with a double busbar connection, provided as an exemplary embodiment of the present invention. Figure 5 As shown, based on the object classification of alarm information, taking the defect range of conventional substation relay protection as an example, the closed-loop path of the defect range associated with alarm information is generally as follows:

[0197] 1) When the object of the alarm information is voltage (e.g., TV disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: voltage transformer secondary winding → voltage analog circuit (including air switch) → bus merging unit → cascaded fiber optic circuit → branch merging unit → direct sampling fiber optic circuit → branch protection device; for the voltage of the input bus protection device, the voltage of the transformer low-voltage winding, and the line voltage, the closed-loop path is: voltage transformer secondary winding → voltage analog circuit (including air switch) → merging unit → direct sampling fiber optic circuit → protection device, as shown by line segment ① in the figure;

[0198] 2) When the object of the alarm information is current (e.g., TA disconnection alarm), the closed-loop path of the defect range associated with the alarm information is generally: current transformer secondary winding → current analog quantity loop → merging unit → direct acquisition fiber optic loop → protection device, as shown by line segment mark ② in the figure.

[0199] 3) When the object of the alarm information is the contact potential of the control box (or smart terminal) (e.g., trip position abnormal alarm), the closed-loop path of the defect range associated with the alarm information generally consists of two parts. One part is the path of the switching quantity, and the other part is the path of the related current quantity. The path of the current quantity is as described above. The path of the switching quantity includes: smart terminal input contact → smart terminal → GOOSE point-to-point loop → protection device, as shown by line segment ③ in the figure.

[0200] 4) When the object of the alarm information is the start failure or failure trip signal of other protection devices (e.g., failure trip input abnormal alarm), the closed-loop path of the defect range associated with the alarm information is generally: source protection device CPU board → source protection device optical port → GOOSE networking loop → receiving protection device optical port → receiving protection device CPU board, as shown by line segment ④ in the figure.

[0201] 5) When the object of the alarm information is device hardware (e.g., power supply module abnormality alarm, input module abnormality alarm, RAM memory abnormality alarm), taking the power supply module abnormality alarm as an example, the closed-loop path of the defect range associated with this alarm information is generally: DC power supply → DC circuit → device power supply module → device CPU board, as shown by line segment label ⑤ in the figure.

[0202] 6) When the object of the alarm information is a device channel (e.g., channel abnormality alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking a multiplexed fiber optic channel as an example): CPU board of the opposite line protection device → management board of the opposite line protection device → secondary circuit → screen of the opposite multiplexed interface device → coaxial cable → digital distribution frame of the opposite side → communication channel (including communication screens on both sides) → digital distribution frame of this side → coaxial cable → screen of the current multiplexed interface device of this side → secondary circuit → management board of the current line protection device → CPU board of the current line protection device, as shown by line segment mark ⑥ in the figure;

[0203] 7) When the object of the alarm information is a remote transmission or remote trip signal (e.g., a long-term remote input abnormal alarm), the closed-loop path of the defect range associated with the alarm information includes two parts. One part is the output of other protection devices on this side and related switching circuits, and the other part is the output of other protection devices on the opposite side and related switching circuits. The specific path includes: bus protection, circuit breaker protection, intelligent terminal and other protection CPU plug-in → other protection switching terminal → GOOSE networking circuit → line protection device optical port → line protection device CPU plug-in and bus protection, circuit breaker protection, intelligent terminal and other protection CPU plug-in on the opposite side → other protection switching terminal → GOOSE networking circuit → line protection device optical port → line protection device CPU plug-in, as shown by line segment label ⑦ in the figure;

[0204] 8) When the object of the alarm information is the time synchronization signal (e.g., time synchronization anomaly alarm), the closed-loop path of the defect range associated with the alarm information is generally (taking B code time synchronization as an example): satellite → antenna → time synchronization source → secondary circuit → protection device management board → protection device CPU board, as shown by line segment mark ⑧ in the figure.

[0205] The probability calculation unit 803 is used to obtain the probability of several defect points based on several alarm messages and several defect points.

[0206] In this embodiment of the invention, when new alarm information appears within a certain time interval or under the condition that historical alarm information has not been eliminated, the alarm information is correlated, indicating that these alarm information are caused by at least one identical defect, and the elimination of one or more defects means the recurrence of related multiple alarm signals. Each alarm event can be plotted as a closed-loop path of the defect range associated with the alarm, and a fault at any node on the closed-loop path can cause the alarm signal to occur. The intersection of the cause and location of one or more defects within the defect range of each alarm is most likely to occur. Based on several alarm information and several defect points, the intersection of defect points is calculated to define the scope of defect diagnosis, comprehensively determine the probability of each defect point, and provide a basis for selecting a reasonable combination of relay protection defect diagnosis measures. Compared with analyzing and diagnosing according to the defect range of any single alarm, the defect diagnosis process is more targeted, reducing the workload of operators analyzing multiple alarms that appear in a short period, grasping the main issues, improving the efficiency of defect diagnosis, and enhancing the intelligence level of defect diagnosis.

[0207] Furthermore, the probability calculation unit 803 is also used for:

[0208] Based on several alarm messages and several defect points, construct a conditional probability matrix for several defect points;

[0209] Based on the conditional probability matrix and the importance of several alarm messages, the probability of several defect points is obtained.

[0210] Specifically, there is I alarm messages f i (1≤i≤i), there are a total of J possible defect points u. j (1≤j≤J). Based on the i-th alarm message, the conditional probability of J defect points is known to be P(u...). j |f i Using I alarm messages as rows of a matrix and J possible defect points as columns, we construct matrix A = [a ij ] I×J =[P(u j |f i )] I×J Considering the importance of I alarm messages, set a weight vector w = [w i ] 1×I Then, the probability size l' of the defect point that integrates I alarm messages can be calculated using A and w:

[0211]

[0212] l' is a row vector containing J elements, determined by A and w. The operator is used for calculation. First, the vectors involved in the operation are element-wise minimized, then the elements are maximized to obtain the result, which is a single number. The result is l'. j (1≤j≤J), which represents the probability of the j-th defect point, and its importance is related to the importance w of the I alarm messages. i (1≤i≤I) and the conditional probability P(u) of each defect point after the alarm information is generated. j |f i The relationship between (1≤i≤I, 1≤j≤J) is:

[0213]

[0214] Due to ∑ 1≤j≤J l' j It is usually not equal to 1, so the row vector l' needs to be normalized, that is...

[0215]

[0216] The obtained l is the probability of J defect points that combine I alarm information.

[0217] Example 2

[0218] The relay protection device simultaneously reported "CT disconnection" and "TWJ abnormality". Defects related to the CT disconnection alarm information include... Figure 2 As shown in (b), the defects related to TWJ abnormal alarm information are as follows: Figure 2 As shown in (d), by integrating them, the defect points are shown in the table below:

[0219]

[0220]

[0221] First, calculate the conditional probability of each defect point under the condition that two online self-test alarm messages occur simultaneously, as the basis for ranking the defect diagnosis measures.

[0222] Let i = 1 represent a CT disconnection alarm and i = 2 represent a TWJ anomaly alarm, and obtain the conditional probability matrix A = [a] for each defect point when the self-test alarm information occurs. ij ] I×J =[P(u j |f i )] 2×11 :

[0223]

[0224] In the above formula, A 1,10 =A 2,10=0. In fact, this case selects the intelligent substation line protection device for defect diagnosis, and the 10th defect point, i.e., the "TWJ circuit fault" at j=10, only occurs in conventional substations. The intelligent substation uses GOOSE networking to transmit switching signals, so A 1,10 =A 2,10 =0.

[0225] Let w = [0.45, 0.55], then... It can be calculated that:

[0226] l'=[0.18,0.17,0.03,0.15,0.22,0.11,0.12,0.03,0.21,0,0.26];

[0227] Normalizing l' yields:

[0228] l=[0.122,0.115,0.020,0.101,0.149,0.074,0.081,0.020,0.142,0.000,0.176];

[0229] l represents the conditional probability of each defect point under the condition that the two online self-test alarm messages, "CT disconnection" and "TWJ abnormality," occur simultaneously. For example, the probability of the defect point "virtual terminal error (merging unit, intelligent terminal to protection device)" is 0.122.

[0230] The defect diagnosis measures are shown in the table below:

[0231]

[0232]

[0233] Establish a correlation matrix B between defect diagnosis measures and defect points. 11×11 :

[0234]

[0235] Take S 0 =[1,2,3,4,5,6,7,8,9,10,11],C0=1,C'=0.03,h=40,g=10,The expected number of defect diagnosis measures implemented varies with the number of iteration steps as follows: Figure 6 As shown.

[0236] When the initial defect diagnosis measures are ordered, i.e., implemented one by one according to k from smallest to largest, the expected number of defect diagnosis measures to be implemented is 5.888, meaning that an average of 5.888 defect diagnosis measures are needed to obtain a defect diagnosis conclusion. The algorithm of this embodiment optimizes the implementation order of defect diagnosis measures, resulting in the following optimized implementation order:

[0237] S best =[11,1,2,4,5,6,7,8,10,9,3];

[0238] Based on the optimization results, the 11th defect diagnosis measure, "the auxiliary contact of the switch is consistent with the actual position," was implemented first, followed by the 1st defect diagnosis measure, "check the virtual terminal connection," and so on. The expected number of defect diagnosis measures implemented was ultimately reduced to 5.029. This demonstrates that by optimizing the defect diagnosis measures, the most likely defect point was located using the fewest diagnostic steps, achieving intelligent defect diagnosis, reducing defect investigation time, and lowering the risk to safe power grid operation.

[0239] The optimal defect diagnosis measures, ranked as follows, are as follows when both "CT disconnection" and "TWJ abnormality" online self-check alarms occur simultaneously:

[0240] k Defect diagnostic measures order 1 Check the virtual terminal connection 2 2 Restored after replacing the line protection module 3 3 Check the protection device ratio 11 4 Fiber optic loop inspection 4 5 The merging unit outputs SV messages normally. 5 6 The analog input of the merging unit is normal. 6 7 The current at the terminal box / control cabinet is normal. 7 8 Both sets of protection reported CT line breakage. 8 9 The smart terminal output matches the actual position of the switch. 10 10 The smart terminal input matches the actual position of the switch. 9 11 The auxiliary contact of the switch is consistent with the actual position. 1

[0241] The present invention also provides a computer-readable storage medium storing one or more programs that, when executed by one or more processors, implement any of the above-described methods for optimizing relay protection defect diagnosis measures.

[0242] The invention has been described with reference to a few embodiments. However, as will be known to those skilled in the art, and as defined in the appended claims, other embodiments besides those disclosed above fall equivalently within the scope of the invention.

[0243] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the art, unless otherwise expressly defined herein. All references to “a / the / the [device, component, etc.]” ​​are openly interpreted as at least one instance of said device, component, etc., unless otherwise expressly stated. The steps of any method disclosed herein need not be performed in the exact order disclosed unless explicitly stated otherwise.

[0244] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0245] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0246] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0247] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0248] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing relay protection defect diagnosis measures, characterized in that, The method includes: Establish a correlation matrix between defect points and defect diagnosis measures in the expert knowledge base; Based on the correlation matrix and the probability of each defect point, an objective function is established; Solve for the optimal solution of the objective function to minimize the expected number of defect diagnostic measures required; The establishment of the association matrix between defect points and defect diagnosis measures in the expert knowledge base includes: Establish a correlation matrix between defect points and necessary defect diagnosis measures in the expert knowledge base. B =[ b kj ] K×J ; in, b kj =1 indicates that for defect points j Defect diagnosis measures must be implemented. k Defects j Only then can it be confirmed; b kj =0 indicates a defect. j Confirmation indicates that no defect diagnostic measures need to be implemented. k ; column vector b j =[ b kj ] K×1 Representation and Defect Points j Correlation vectors for each corresponding defect diagnosis measure; J Indicates the number of defects. K Indicates the number of defect diagnostic measures, 1≤ j ≤ J ,1≤ k ≤ K , J , K All are positive integers; The step of establishing the objective function based on the correlation matrix and the probability of each defect point includes: According to the correlation matrix B =[ b kj ] K×J and defects j probability of occurrence l j Establish the following objective function: ； Among them, the ranking of a set of defect diagnosis measures S any of its elements S ( k ) indicates the first k The execution sequence number of the defect diagnosis measure; The step of finding the optimal solution to the objective function to minimize the expected number of required defect diagnosis measures includes: The optimal solution to the objective function is obtained by using the nuclear decay algorithm, so as to minimize the expected number of defect diagnosis measures required.

2. The method according to claim 1, characterized in that, The method of using the nuclear decay algorithm to solve for the optimal solution of the objective function, so as to minimize the expected number of required defect diagnosis measures, includes: Step 1: Put K A random combination of natural numbers is used as the initial ordering of defect diagnosis measures. S 0 Set the initial intensity of radioactive particles C 0. Half-life h Cut-off strength C' Then the first r Radioactive particle intensity in the next iteration C r for: ； in, r For positive integers, the initial iteration r =1: Step 2: r In the next iteration, a random number is set. α , 0≤ α ≤1; when α When the value is greater than 0.5, start from the current solution. S 1 Randomly select two numbers and swap their positions to obtain a new solution. S 2 ; when α When ≤0.5, start from the current solution S 1 Randomly select three numbers and swap their positions to obtain a new solution. S 2 ; like E ( S 2 )≤ E ( S 1 If a new solution is found, it is accepted; otherwise, it is determined by probability. p iter Determine whether to accept the new solution; among which, If a new solution is received, then the new solution will be used as the current solution. This step is repeated. g Second-rate, g Positive integers, repeated g The solution obtained subsequently is taken as the optimal solution for this iteration. S best and the initial solution for the next iteration; Step 3: Update the number of iterations r = r +1, when C r > C' If the condition is met, return to step 2 and repeat the loop; otherwise, output the optimal solution for each iteration. S best As the final solution.

3. The method according to claim 2, characterized in that, According to probability p iter Determining whether to accept the new solution includes: Generate random numbers 0≤ β ≤1, when β ≤ p iter If a new solution is accepted, it should be accepted; otherwise, it should be rejected.

4. The method according to any one of claims 1-3, characterized in that, Before establishing the association matrix between defect points and defect diagnosis measures in the expert knowledge base, the following steps are also included: Collect several alarm information items; Based on the aforementioned alarm information, several defect points are identified. Based on the aforementioned alarm information and the aforementioned defect points, the probability of several defect points is obtained.

5. The method according to claim 4, characterized in that, Based on the aforementioned alarm information and the aforementioned defect points, the probability of several defect points is obtained, including: Based on the aforementioned alarm information and the aforementioned defect points, construct a conditional probability matrix for the aforementioned defect points; Based on the conditional probability matrix and the importance of the alarm information, the probability of the possible defects is obtained.

6. A device for optimizing relay protection defect diagnosis measures, characterized in that, The device includes: The association matrix construction unit is used to establish an association matrix between defect points and defect diagnosis measures in the expert knowledge base; The objective function construction unit is used to establish an objective function based on the correlation matrix and the probability of each defect point; An optimization unit is used to solve for the optimal solution of the objective function so as to minimize the expected number of required defect diagnosis measures; The association matrix construction unit is further configured to: Establish a correlation matrix between defect points and necessary defect diagnosis measures in the expert knowledge base. B =[ b kj ] K×J ; in, b kj =1 indicates that for defect points j Defect diagnosis measures must be implemented. k Defects j Only then can it be confirmed; b kj =0 indicates a defect. j Confirmation indicates that no defect diagnostic measures need to be implemented. k ; column vector b j =[ b kj ] K×1 Representation and Defect Points j Correlation vectors for each corresponding defect diagnosis measure; J Indicates the number of defects. K Indicates the number of defect diagnostic measures, 1≤ j ≤ J ,1≤ k ≤ K , J , K All are positive integers; The objective function construction unit is further configured to: According to the correlation matrix B =[ b kj ] K×J and defects j probability of occurrence l j Establish the following objective function: ； Among them, the ranking of a set of defect diagnosis measures S any of its elements S ( k ) indicates the first k The execution sequence number of the defect diagnosis measure; The optimization unit is also used for: The optimal solution to the objective function is obtained by using the nuclear decay algorithm, so as to minimize the expected number of defect diagnosis measures required.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-5.

Images