Intelligent substation optical fiber link online monitoring method based on domain knowledge graph

Abstract

The application discloses an intelligent substation optical fiber link online monitoring method based on a domain knowledge graph, belongs to the technical field of intelligent substation optical fiber link online monitoring, and comprises the following steps: exporting port information through SCD files and SPCD files, defining triad information according to needs, combining with virtual loop information, using a domain knowledge graph method to establish a virtual and real loop model, and completing intelligent substation full-link information display; meanwhile, a method for extracting interval layer equipment, process layer equipment and switch port optical power information is adopted to form an intelligent substation optical fiber link monitoring method, and an auxiliary means is provided for intelligent substation link breakage elimination.

Description

Technical Field

[0001] This invention relates to the field of online monitoring technology for fiber optic links in intelligent substations, and specifically to an online monitoring method for fiber optic links in intelligent substations based on domain knowledge graphs. Background Technology

[0002] Currently, there are few methods for monitoring the optical power of the device ports in intelligent stations. Maintenance and debugging personnel need to check the optical power of each port through the menus of the interval layer equipment and process layer equipment. It is not possible to monitor and predict changes in optical power or whether a link break will occur in real time.

[0003] Currently, most methods for handling link failures in intelligent stations are based on the process layer virtual loop, which is displayed and diagnosed through the SCD (system configuration file) file. There are few studies and results on the analysis of secondary loops based on the combination of virtual and physical loops, and very few methods for displaying full link information. At the same time, there is even less research on link failure prediction based on optical power and link failure processing based on optical power.

[0004] The establishment of SPCD (Fiber Optic Physical Loop Model File) files provides complete information on the physical ports of substations. Based on the requirements of the nine unified standards, the new switches have optical power monitoring functions, which provides conditions for the monitoring of fiber optic links in smart substations. Summary of the Invention

[0005] To overcome the shortcomings of the above technologies, this invention provides a method for exporting port information through SCD and SPCD files, defining triplet information as needed, combining virtual loop information, and using domain knowledge graph methods to establish a virtual-real loop model to complete the full-link information display of intelligent substations; at the same time, it extracts optical power information of bay layer devices, process layer devices, and switch ports.

[0006] The technical solution adopted by this invention to overcome its technical problems is:

[0007] A method for online monitoring of fiber optic links in intelligent substations based on domain knowledge graphs includes the following steps:

[0008] a) Obtain SPCD and SCD files;

[0009] b) Define ports and relationships in the SPCD and SCD files; c) Extract knowledge from the SPCD and SCD files. The virtual terminal connection table exported from the SCD contains the first and last devices and the receiving port number of the last device. Convert the virtual terminal connection table into a .CSV file. The port connection table exported from the SPCD contains the port number of the first and last device and the branch number. Convert the port connection table into a .CSV file. Input the code into the opened neo4j graph database, retrieve the .CSV file generated by the SPCD to form a knowledge graph 1 that needs to be checked by domain experts. Input the code into the opened neo4j graph database, retrieve the .CSV file generated by the SCD to form a knowledge graph 2 that needs to be checked by domain experts.

[0010] d) Compare the similarity between the information of a certain end device in Knowledge Graph 2 database and the entity port information in Knowledge Graph 1, and use the cosine similarity of string vectors to perform end port fusion and preliminary relationship fusion;

[0011] e) Conduct knowledge graph framework evaluation and knowledge graph relationship evaluation;

[0012] f) Complete the establishment of a knowledge graph-based intelligent substation full-link information display system;

[0013] g) The test and protection device, the intelligent control device, and the switch all use the debugging network port to read the optical power value of each optical port using the manufacturer's special tools. The optical power value of each optical port is identified and associated with the knowledge graph through the device port information and monitored. The optical power value of all optical ports is transmitted to the optical fiber link monitoring center interactive machine through the network cable, and then uploaded to the knowledge graph-based intelligent substation full-link information display system through the optical fiber link monitoring center interactive machine.

[0014] h) Monitor the optical power of all ports with a specified sampling period, set a sampling warning value, and do not store the normal optical power of the ports. If the optical power of a port drops to the specified warning value e, the system will detect the power loss. pd If the monitoring mechanism is triggered, the monitoring mechanism stores the optical power values ​​of all ports in the circuit where the port is located and plots the monitoring curve. The monitoring curve data is uploaded to the intelligent substation full-link information display system and transmitted to the designated client. i) If there is a link break alarm in the m-th circuit, and if the last device cannot receive the virtual circuit information of the circuit, it is determined that the receiving port of the last device is faulty, the n-th optical fiber circuit of the circuit is faulty, and the transmitting power port of the next-level device of the last device is faulty.

[0015] Furthermore, step b) includes the following steps:

[0016] b-1) Treat the full-site ports in the SPCD file as entities in the knowledge graph triples, and treat the virtual loop information in the SCD file and the link number in the SPCD file as relations;

[0017] b-2) Convert the SCD file into a triplet data format as follows: <First end transpose, virtual loop information, last end device port number>, device name / board information: intelligent electronic device name + voltage level + line name + device type + port number;

[0018] b-3) Convert the SPCD file into a triplet data format: <first port, fiber optic branch number, last port>. The port information consists of: intelligent electronic device name + voltage level + line name + device type + port number. In step b-2), the device type is either a process layer device, a bay layer device, or an ODF. In step b-2), only the end subscription device has a port number. In step b-3), the device type is either a process layer device, a bay layer device, or an ODF. Further, in step c), knowledge extraction is performed by device area, which includes main transformer equipment, 220kV equipment, 110kV equipment, and common equipment.

[0019] Furthermore, step d) includes the following steps:

[0020] d-1) Extract the port information string of a certain end device from Knowledge Graph 2 database and vectorize it with the port information strings of various ports in Knowledge Graph 1 database to form string vectors respectively. and string vectors Through formula The cosine similarity sim of the two string vectors is calculated. The maximum value of the cosine similarity sim is selected to obtain the two most similar ports. The original port information of knowledge graph 2 is replaced with the original port information of knowledge graph 1. The port information of knowledge graph 1 is calculated with all the end port information of knowledge graph 2, and the port information with the same port information is replaced to complete the end port fusion.

[0021] d-2) Connect the tail ports of Knowledge Graph 1 and Knowledge Graph 2 to obtain a new knowledge graph. Change the pointers of triples in Knowledge Graph 2, swap the tail ports of triples in Knowledge Graph 2, and force the head and tail ports in the new knowledge graph to connect, forming a virtual one-way connected graph. The newly formed virtual one-way connected graph has no virtual loop information from port 5 to port 1. Establish the adjacency matrix of the directed graph, and the matrix elements... v i Let v be the i-th element in the adjacency matrix of the directed graph. jLet i be the j-th element in the adjacency matrix of the directed graph, i = {1, 2, ..., N}, j = {1, 2, ..., N}, and N be the total number of nodes in the adjacency matrix of the directed graph. The out-degree and in-degree of the adjacency matrix of the directed graph are both 1. The sum of all elements in the adjacency matrix of the directed graph is the sum of the number of edges of the graph. Take the union of the relations of each triple in the cycle, replace each branch relation in the cycle, and delete the triples in each branch relation that do not have branch number information to complete the initial relation fusion.

[0022] Furthermore, step e) includes the following steps:

[0023] e-1) The knowledge graph framework evaluation adopts the Manhattan distance vector group measurement method. Chinese characters are vectorized and branch numbers and port numbers are presented as hexadecimal numbers. The first and last port information and branch numbers of each branch in a certain loop in Knowledge Graph 1 and the new knowledge graph are vectorized to form two vector groups. The Manhattan distance of each vector in the two vector groups is calculated. If the calculated vector is zero, the loop framework structure is correct. If the calculated vector is not zero, the inconsistent port information is manually found and modified.

[0024] e-2) The knowledge graph relationship evaluation uses the string vector cosine similarity method. The virtual loop information of knowledge graph 2 is vectorized and used as the reference vector. The relationship of each branch in the loop in the new knowledge graph is formed into a vector, and the comparison result with the reference vector is calculated. If the comparison result is all 1, the loop framework structure is correct. If the comparison result is not all 1, the virtual loop information related to the branch in the new knowledge graph is modified manually.

[0025] In step g), optical splitter 2 is used to split the optical power of the test device, the intelligent device, and the switch. One of the split optical power channels is amplified by SOA and then continues to be transmitted through the link. The other optical power channel is amplified by SOA and then continues to be monitored for optical power at the link port. The optical fiber output from SOA for monitoring needs to be connected to the switch, and the optical power value is read through the switch.

[0026] Preferably, the warning value e in step h) pd It is -30dB.

[0027] Preferably, in step h), 5G transmission technology is used to remotely and quickly transmit data to the designated client.

[0028] The beneficial effects of this invention are: by exporting port information through SCD and SPCD files, defining triplet information as needed, and combining it with virtual loop information, a virtual and real loop model is established using domain knowledge graph methods to complete the full-link information display of the intelligent substation; at the same time, a method for extracting optical power information of bay layer equipment, process layer equipment, and switch ports is formed to create an optical fiber link monitoring method for intelligent substations, and at the same time, it provides an auxiliary means for eliminating link failures in intelligent substations. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention;

[0030] Figure 2 This is a schematic diagram of the end port information fusion of the present invention;

[0031] Figure 3 This is a preliminary schematic diagram of the relationship fusion of the present invention;

[0032] Figure 4 This is a schematic diagram of the knowledge graph for the full information fusion of optical fiber loops in this invention;

[0033] Figure 5 This is a schematic diagram illustrating an example of the knowledge graph framework evaluation of the present invention;

[0034] Figure 6 This is a schematic diagram illustrating an example of knowledge graph relationship evaluation in this invention;

[0035] Figure 7 This is an expanded diagram of the optical fiber physical loop port information of the present invention. Detailed Implementation

[0036] The following is in conjunction with the appendix Figure 1 To be continued Figure 7 The present invention will be further described below.

[0037] As attached Figure 1 As shown, an online monitoring method for fiber optic links in intelligent substations based on domain knowledge graphs includes the following steps:

[0038] a) Obtain the SPCD and SCD files. The SCD file provides information on the virtual loops subscribed to by the device, the receiving device port information, the sending device name, and the information flow direction. The SPCD file provides a topology diagram of the fiber optic loop links and ports within the substation. To facilitate modeling, it is agreed that the switches use multicast technology, which provides the conditions for the correspondence between virtual and physical loops.

[0039] b) Entity Definition: The virtual subgraph in the SCD file is a directed graph, but it only provides information on the first and last devices and port information for transmitting virtual loop information. It lacks port information for the entire link transmitting virtual loop information. The SPCD file, however, provides a port connection topology diagram for the entire intelligent station, including port information for bay layer devices, process layer devices, switches, and optical distribution units (ODF). Port and relationship definitions are performed on both the SPCD and SCD files.

[0040] c) Knowledge Extraction: The theory of the online monitoring system for fiber optic links in intelligent substations based on knowledge graphs originates from SPCD and SCD file databases, both of which are relational databases. Knowledge extraction is performed on the SPCD and SCD files. The virtual terminal connection table exported from the SCD contains the start and end devices and the receiving port number of the end device. The virtual terminal connection table is converted into a .CSV file. The port connection table exported from the SPCD contains the port number of the start and end devices and the branch number. The port connection table is converted into a .CSV file. Code is entered into the opened neo4j graph database, and the .CSV file generated by the SPCD is retrieved to form knowledge graph 1, which needs to be reviewed by domain experts. Code is also entered into the opened neo4j graph database, and the .CSV file generated by the SCD is retrieved to form knowledge graph 2, which also needs to be reviewed by domain experts.

[0041] d) Knowledge fusion: Based on the similarity comparison between the information of a certain end device in Knowledge Graph 2 database and the entity port information in Knowledge Graph 1, the cosine similarity of string vectors is used to perform end port fusion and preliminary relationship fusion.

[0042] e) Quality Assessment: The knowledge graph framework and relationships are evaluated. Before the quality assessment, given the characteristics of the database, many port and virtual loop information in each interval are similar. Dictionaries are defined for each device area, represented using one-hot encoding, and vectors are formed. Vector comparisons are then performed using these dictionary vectors.

[0043] f) Complete the establishment of a knowledge graph-based intelligent substation full-link information display system.

[0044] g) The testing device, intelligent control device, and switch can all read the optical power values ​​of each optical port through the debugging network port using the manufacturer's dedicated tools. Since the ODF cannot read the port optical power, to minimize distortion, a two-way optical splitter is used. After splitting, one path is amplified by an SOA (Optical State Amplifier) ​​before continuing link transmission, while the other path is amplified by the SOA before continuing link port optical power monitoring. The fiber optic cable output from the SOA for monitoring needs to be connected to a switch, and the optical power value is read through the switch. Port optical power and the knowledge graph are identified, associated, and monitored through the device ports using port information. All port optical power is transmitted via network cable to the fiber optic link monitoring center switch and finally uploaded to the fiber optic link monitoring system.

[0045] h) Monitor the optical power of all ports with a specified sampling period, set a sampling warning value, and do not store the normal optical power of the ports. If the optical power of a port drops to the specified warning value e, the system will detect the power loss. pd If the signal is received, the monitoring mechanism is triggered. The monitoring mechanism stores the optical power values ​​of all ports in the circuit containing that port and plots the monitoring curves. The monitoring curve data is then uploaded to the intelligent substation full-link information display system and transmitted to the designated client.

[0046] i) If a link break alarm occurs in the m-th loop, and if the last-end device cannot receive the virtual loop information for that loop, then the last-end device's receiving port is considered faulty, the n-th fiber loop in that loop is faulty, and the transmitting power port of the upstream device of the last-end device is faulty. Based on the above principle, assume that the loop has n branches, and there are n transmitting ports and n receiving ports; the normal optical power of each port is recorded as 1, and the optical power exceeding the lower limit is recorded as 0; establish a transmitting matrix T = (T1, T2, ..., T...). i ,...,T n The receiving matrix R = (R1, R2, ..., R) i ,...,R n ), with R i The adjacent upstream branch is denoted as L. i , where n represents the number of branches. T i R represents the optical power of the i-th transmitting port in this loop. i Let Y represent the optical power of the i-th receiving port in the loop, where i ∈ (1, 2, ..., n). When a link break occurs, if all elements in the transmitting and receiving matrices are zero, it indicates that the originating transmitting port cannot transmit, and the link break occurs at the originating device; if not all elements in the transmitting and receiving matrices are zero, denoted as Y = T × R. T The calculated value Y plus 1 represents the port location where power cannot be received. The break point is the receiving port, the adjacent upstream branch, or the previous level transmitting port.

[0047] As attached Figure 7As shown, if R3 = 0 and all other matrices are 1, then If port R3 receives no optical power or the fiber in tributary L3 is damaged; if R3=0, T3=0, and all other matrices are 1, then... Then port T3 cannot transmit optical power.

[0048] In one specific embodiment of the present invention, step b) includes the following steps:

[0049] b-1) Treat the full-site ports in the SPCD file as entities in the knowledge graph triples, and treat the virtual loop information (the "internal signal description" of the subscription device) and the link number in the SPCD file as relations;

[0050] b-2) Convert the SCD file into a triplet data format as follows: <First end transpose, virtual loop information, last end device port number>, device name / board information: Intelligent electronic device name (IEDname) + voltage level + line name + device type + port number (only the last subscription device has this information).

[0051] b-3) Convert the SPCD file into a triplet data format: <first port, fiber optic branch number, last port>. The port information consists of: Intelligent Electronic Device Name (IEDname) + Voltage Level + Line Name + Device Type + Port Number.

[0052] In step b-2), the device type is either a process layer device, a bay layer device, or an ODF. In step b-2), only the end subscription device has a port number. In step b-3), the device type is either a process layer device, a bay layer device, or an ODF. In a specific embodiment of the present invention, knowledge extraction in step c) is performed across different equipment areas, including main transformer equipment, 220kV equipment, 110kV equipment, and common equipment.

[0053] In a specific embodiment of the present invention, step d) includes the following steps:

[0054] d-1) Extract the port information string of a certain end device from Knowledge Graph 2 database and vectorize it with the port information strings of various ports in Knowledge Graph 1 database to form string vectors respectively. and string vectors Through formula The cosine similarity *sim* between two string vectors is calculated. The maximum value of *sim* is selected to identify the two most similar ports. The original port information of Knowledge Graph 2 is replaced with the original port information of Knowledge Graph 1. The similarity of the port information in Knowledge Graph 1 is calculated using all the terminal port information of Knowledge Graph 2, and ports with identical information are replaced, thus completing the terminal port fusion. (See attached) Figure 2As shown, after processing, "A: vectorized string, B: vectorized character, their union (vector, character, string), after vectorization A = (1,1,1,1,1,1) B = (1,1,1,1,1,0), then similarity is calculated," the two most similar ports are determined, and the original port information in Knowledge Graph 2 is replaced with the original port information in Knowledge Graph 1. Similarly, all the end port information in Knowledge Graph 2 is used to calculate the similarity of the port information in Knowledge Graph 1, and the same port information is replaced, completing the fusion of end port information.

[0055] d-2) Connect the tail ports of Knowledge Graph 1 and Knowledge Graph 2 to obtain a new knowledge graph. Change the pointers of triples in Knowledge Graph 2, swap the tail ports of triples in Knowledge Graph 2, and force the head and tail ports in the new knowledge graph to connect, forming a virtual one-way connected graph. The newly formed virtual one-way connected graph has no virtual loop information from port 5 to port 1. Establish the adjacency matrix of the directed graph, and the matrix elements... v i Let v be the i-th element in the adjacency matrix of the directed graph. j Let i be the j-th element in the adjacency matrix of a directed graph, where i = {1, 2, ..., N} and j = {1, 2, ..., N}, and N is the total number of nodes in the adjacency matrix. The out-degree and in-degree of the adjacency matrix are both 1, and the sum of all elements in the adjacency matrix equals the sum of the edges in the graph. Satisfying these conditions indicates that the connected graph is correctly connected and the triplet relationships are correct. Take the union of the relationships of each triplet in the cycle, and replace each branch relationship in the cycle. After replacement, delete the triples in each branch relationship that lack branch number information, completing the initial relationship fusion.

[0056] As attached Figure 3 and attached Figure 4 As shown, an example of an adjacency matrix is ​​provided:

[0057] The new graph obtained by transforming Knowledge Graph 1 and Knowledge Graph 2 is shown in the attached figure. Figure 3 As shown, the adjacency matrix is ​​represented as:

[0058]

[0059] The matrix has an out-degree and in-degree of 1, 5 vertices, and a sum of 5 elements, indicating that the loop is unidirectionally connected.

[0060] In one specific embodiment of the present invention, step e) includes the following steps:

[0061] e-1) As defined by the entity, port information consists of: Intelligent Electronic Device Name (IEDname) + Voltage Level + Line Name + Device Type (e.g., Process Layer Device, Bay Layer Device, ODF, etc.) + Port Number. The information such as "Intelligent Electronic Device Name (IEDname) + Voltage Level + Line Name," "Device Type (e.g., Process Layer Device, Bay Layer Device, ODF, etc.)," ​​and "Port Number" are unique. To reduce the dimensionality of the vector, knowledge framework evaluation is performed separately for each equipment area (e.g., main transformer area, 220kV equipment area, etc.). Branch numbers and port numbers are presented in hexadecimal for easier computer calculation. Similarity evaluation is performed on the framework information of each loop. The first and last port information and branch numbers of each branch in a loop in Knowledge Graph 1 and the new Knowledge Graph are vectorized to form two vector groups. The Manhattan distance is calculated for each vector in the two vector groups. If the calculated vector is zero, the loop framework structure is correct; if the calculated vector is not zero, inconsistent port information is manually searched and corrected.

[0062] As attached Figure 5 As shown, a direct-jump circuit is used as an example:

[0063] According to the definition dictionary, "-" is defined as zero, "PL2203A:220kV Lingxian II Line Protection Set A", "7-D", "IL2203A:220kV Lingxian II Line Intelligent Terminal Set A", "3-A", "11", "7-E", "PL2203A:220kV Lingxian II Line Protection Set A" = (1, 0, 0, 0, 0, 0, ...) and "IL2203A:220kV Lingxian II Line Intelligent Terminal Set A" = (0, 1, 0, 0, 0, ...)

[0064] "7-D" = (0, 1, 1, 1, 0, 1, 1, 0, 1)

[0065] "7-E"=(0,1,1,1,0,1,1,1,0)

[0066] "3-A" = (0, 0, 1, 1, 0, 1, 0, 1, 0)

[0067] "11" = (0, 0, 0, 0, 1, 0, 1, 1)

[0068] Perform Manhattan distance group metric calculation:

[0069]

[0070] Calculation results:

[0071]

[0072] The above results indicate that there is an error in the circuit, which needs to be checked.

[0073] e-2) The knowledge graph relationship evaluation uses the string vector cosine similarity method. The virtual loop information of knowledge graph 2 is vectorized and used as the reference vector. The relationship of each branch in the loop in the new knowledge graph is formed into a vector, and the comparison result with the reference vector is calculated. If the comparison result is all 1, the loop framework structure is correct. If the comparison result is not all 1, the virtual loop information related to the branch in the new knowledge graph is modified manually.

[0074] As attached Figure 6 As shown in the example:

[0075] Figure 6 Example diagram of knowledge graph relationship evaluation

[0076] If the virtual circuit information string in diagram 2 is "Circuit breaker A phase trip position Circuit breaker B phase trip position Circuit breaker C phase trip position Lockout reclosing Low pressure lockout reclosing", and the new virtual circuit information string is "Circuit breaker A phase trip position Circuit breaker B phase trip position Circuit breaker C phase trip position Lockout reclosing Low pressure lockout reclosing", after vectorization, A = (1, ... ,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1), B=(1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1), the similarity calculation result is: The requirements are not met and an inspection is needed.

[0077] In step g), optical splitter 2 is used to split the optical power of the test device, the intelligent device, and the switch. One of the split optical power channels is amplified by SOA and then continues to be transmitted through the link. The other optical power channel is amplified by SOA and then continues to be monitored for optical power at the link port. The optical fiber output from SOA for monitoring needs to be connected to the switch, and the optical power value is read through the switch.

[0078] In a specific embodiment of the present invention, the warning value e in step h) pd The value is -30dB. In step h), 5G transmission technology is used to remotely and quickly transmit data to the designated client.

[0079] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for online monitoring of optical fiber links in an intelligent substation based on a domain knowledge graph, characterized in that, Includes the following steps: a) Obtain SPCD and SCD files; b) Define ports and relationships for SPCD files and SCD files; c) Extract knowledge from SPCD and SCD files. The virtual terminal connection table exported from SCD contains the first and last devices and the receiving port number of the last device. Convert the virtual terminal connection table into a .CSV file. The port connection table exported from SPCD contains the port number of the first and last device and the branch number. Convert the port connection table into a .CSV file. Enter the code into the opened neo4j graph database and retrieve the .CSV file generated by SPCD to form knowledge graph 1 that needs to be checked by domain experts. Enter the code into the opened neo4j graph database and retrieve the .CSV file generated by SCD to form knowledge graph 2 that needs to be checked by domain experts. d) Compare the similarity between the information of a certain end device in Knowledge Graph 2 database and the entity port information in Knowledge Graph 1, and use the cosine similarity of string vectors to perform end port fusion and preliminary relationship fusion; e) Conduct knowledge graph framework evaluation and knowledge graph relationship evaluation; f) Complete the establishment of a knowledge graph-based intelligent substation full-link information display system; g) The test and protection device, the intelligent control device, and the switch all use the debugging network port to read the optical power value of each optical port using the manufacturer's special tools. The optical power value of each optical port is identified and associated with the knowledge graph through the device port information and monitored. The optical power value of all optical ports is transmitted to the optical fiber link monitoring center interactive machine through the network cable, and then uploaded to the knowledge graph-based intelligent substation full-link information display system through the optical fiber link monitoring center interactive machine. h) Monitoring of all port optical power is defined as a sampling period, setting a sampling warning value, the normal optical power of the port is not stored, if the optical power of a port decreases to the specified warning value The monitoring mechanism is triggered, the monitoring mechanism stores the optical power values of all ports of the loop where the port is located and draws a monitoring curve, uploads the monitoring curve data to the intelligent substation full-link information display system, and transmits to the designated client; i) if the first If the chain alarm of the loop is on, and if the end device does not receive the dummy loop information of the loop and the device before the upper level device of the end device is normal, the receiving port of the upper level device is normal, it is determined that the receiving port of the end device is faulty, or the end device is adjacent to the upstream branch fault, or the sending port of the upper level device of the end device is faulty. 2.The method of claim 1, wherein, Step b) includes the following steps: b-1) Treat the full-site ports in the SPCD file as entities in the knowledge graph triples, and treat the virtual loop information in the SCD file and the link number in the SPCD file as relations; b-2) Convert the SCD file into a triplet data format as follows: <First-end device, virtual loop information, last-end device port number>, device name / board information: intelligent electronic device name + voltage level + line name + device type + port number; b-3) Convert the SPCD file into a triplet data format: <first port, fiber optic branch number, last port>. The port information consists of: intelligent electronic device name + voltage level + line name + device type + port number. 3.The method of claim 1, wherein the method further comprises: In step b-2), the device type is: process layer device, interval layer device, or ODF. In step b-2), only the end subscription device has a port number. In step b-3), the device type is: process layer device, interval layer device, or ODF. 4.The method of claim 1, wherein the method further comprises: In step c), knowledge extraction is performed in different equipment areas, including main transformer equipment, 220kV equipment, 110kV equipment, and public equipment. 5.The method of claim 2, wherein, Step d) includes the following steps: d-1) Extract the port information string of a certain end device from Knowledge Graph 2 database and vectorize it with the port information strings of various ports in Knowledge Graph 1 database to form string vectors respectively. and string vectors Through formula Calculate the cosine similarity between the two string vectors. Select cosine similarity The maximum value is used to obtain the two most similar ports. The original port information of knowledge graph 2 is replaced with the original port information of knowledge graph 1. The similarity of the port information of knowledge graph 1 is calculated using all the end port information of knowledge graph 2, and the same port information is replaced to complete the end port fusion. d-2) Connect the tail ports of Knowledge Graph 1 and Knowledge Graph 2 to obtain a new knowledge graph. Change the pointers of triples in Knowledge Graph 2, swap the tail ports of triples in Knowledge Graph 2, and force the head and tail ports in the new knowledge graph to connect, forming a virtual one-way connected graph. The newly formed virtual one-way connected graph has no virtual loop information from port 5 to port 1. Establish the adjacency matrix of the directed graph, and the matrix elements... , Let the adjacency matrix of a directed graph be the th One element, Let the adjacency matrix of a directed graph be the th One element, , , Let be the total number of nodes in the adjacency matrix of the directed graph. The out-degree and in-degree of the adjacency matrix of the directed graph are both 1. The sum of all elements in the adjacency matrix of the directed graph is the sum of the number of edges in the graph. Take the union of the relations of each triple in the cycle, replace each branch relation in the cycle, and delete the triples in each branch relation that do not have branch number information to complete the initial relation fusion. 6.The method of claim 5, wherein, Step e) includes the following steps: e-1) The knowledge graph framework evaluation adopts the Manhattan distance vector group measurement method. Chinese characters are vectorized and branch numbers and port numbers are presented as hexadecimal numbers. The first and last port information and branch numbers of each branch in a certain loop in Knowledge Graph 1 and the new knowledge graph are vectorized to form two vector groups. The Manhattan distance of each vector in the two vector groups is calculated. If the calculated vector is zero, the loop framework structure is correct. If the calculated vector is not zero, the inconsistent port information is manually found and modified. e-2) The knowledge graph relationship evaluation uses the string vector cosine similarity method. The virtual loop information of knowledge graph 2 is vectorized and used as the reference vector. The relationship of each branch in the loop in the new knowledge graph is formed into a vector, and the comparison result with the reference vector is calculated. If the comparison result is all 1, the loop framework structure is correct. If the comparison result is not all 1, the virtual loop information related to the branch in the new knowledge graph is modified manually. 7.The smart substation optical fiber link online monitoring method based on domain knowledge graph according to claim 1, characterized in that: In step g), the optical splitter 2 is used to split the optical power of the test device, the intelligent device, and the switch. One of the split optical power channels is amplified by the SOA and then continues to transmit the optical power through the link. The other optical power channel is amplified by the SOA and then continues to monitor the optical power of the link port. The optical fiber used for monitoring after passing through the SOA needs to be connected to the switch, and the optical power value is read through the switch.

8. The method for online monitoring of fiber optic links in intelligent substations based on domain knowledge graphs according to claim 1, characterized in that: The pre-alarm value in step h) is -30 dB.

9. The method for online monitoring of fiber optic links in intelligent substations based on domain knowledge graphs according to claim 1, characterized in that: In step h), 5G transmission technology is used to remotely and quickly transmit data to the designated client.

Images