Relay protection information modeling method and system based on secondary equipment big data platform

Abstract

The present invention provides a relay protection information modeling method and system based on a secondary big data platform, including the following steps: S01. Assign physical IDs to multiple substations in the region, and obtain all substation, equipment, and signal information in the region from the big data platform. All substations are denoted as sta1, sta2... staN, all equipment is denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN; S02. Correlate all the obtained substation, equipment, and signal information locally to form a complete hierarchical structure on the local platform; S03. Classify all signal points according to their names into alarm levels and signal categories; S04. Real-time obtain all signal data and waveform recording files in the big data platform, and perform alarms and classifications according to the preset alarm levels and signal classifications; S05. Organize the fault information during the fault time period.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid relay protection, and more specifically, to a relay protection information modeling method and system based on a secondary equipment big data platform. Background Art

[0002] With the rapid development of China's economy, the scale of China's power grid is also constantly expanding. In the monitoring and operation and maintenance of secondary equipment for relay protection, the types and quantities of various data platforms are increasing continuously. Among them, the modeling of most substations is based on SCD. Each platform can rely on the signal points defined in the SCD model file to carry out data transmission and model correspondence.

[0003] However, in the actual application process, it is necessary to ensure the consistency of the models of the receiving and sending parties during data transmission. For this purpose, it is necessary to introduce an SCD version control platform and synchronize the models of both parties at the same time. The process is relatively cumbersome and the system is more complex. In addition, when the data source itself is a data platform based on multiple data sources, it may not perform substation-end modeling based on SCD itself. For the party obtaining the data, it is necessary to obtain the correspondence between the substation-equipment-signal in order to actually use the obtained data. In the case of no SCD support, how to perform data modeling and complete the model correspondence between platforms has become a common concern.

[0004] "Research on the Modeling and Storage Method of Multi-source Data of Secondary Equipment in Smart Substations" (State Grid Beijing Maintenance Company, Beijing Zhongtai Huadian Technology Co., Ltd., School of Control and Computer Engineering, North China Electric Power University, Xiao Yongli, Liu Song, Jian Wei, Song Yaqi, September 2019) records the multi-source data and storage method of secondary equipment in smart substations. Specifically, see Figure 5 on page 4 of this document. This part of the content introduces the cooperation method of each layer of the data warehouse. It can be seen from Figure 5 that each database is for data communication between various devices inside the substation. However, for a region, such as a province, including multiple prefecture-level city substations, if the technology in the above-mentioned comparative document is adopted, it is necessary to perform independent modeling for each substation, and the information between substations cannot be quickly interconnected. And the actual situation is that there is data duplication between substations. If the software systems of each substation are independent, it will cause a large amount of resource waste. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to solve the problem of how to perform data modeling and complete the model correspondence between platforms in the prior art without SCD support.

[0006] The present invention realizes the solution of the above technical problems through the following technical means:

[0007] A relay protection information modeling method based on a secondary big data platform includes the following steps:

[0008] S01. Assign physical IDs to multiple substations in the region, and obtain all substation, device, and signal information in the region from the big data platform. All substations are denoted as sta1, sta2... staN, all devices are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN. The method for corresponding all substations in the big data platform to all substations in the region is as follows:

[0009] Obtain all substation information and protection device information according to the substation information interface and the protection device information interface, obtain the information of all intelligent terminals and merging units for auxiliary device information, and form a substation-device data structure of the secondary data platform according to the affiliated substation.

[0010] If there is an existing configuration file, read the associated information in the configuration file, parse the substation and device data information in the associated information, and associate it with the substation and device information in the big data platform system.

[0011] S02. Associate all the obtained substation, device, and signal information locally to form a complete hierarchical structure on the local platform.

[0012] S03. Classify all signal points according to their names into alarm levels and signal categories.

[0013] S04. Real-time obtain all signal data and waveform recording files in the big data platform, and perform alarms and classifications according to the preset alarm levels and signal classifications.

[0014] S05. Organize the fault information during the fault period.

[0015] The present invention combines the data of each system into a unified platform, which needs to ensure data integrity, accuracy, etc., and needs to perform two-level classification on alarm signals, including faults and anomalies, classifications, and alarm subtypes. Physical IDs are used to associate substations and devices one by one, and device signals are matched according to keywords to realize the model association mapping between systems. Compared with traditional technologies, it can save a large amount of resources, does not need to rely on SCD, and is suitable for regional substation monitoring.

[0016] Further, the specific method for obtaining signal data and waveform recording files from the big data platform in step S04 is as follows:

[0017] Input into the big data platform in units of time periods, obtain the file IDs of all waveform recording files within the input time period, use the file IDs as data to obtain the waveform recording files corresponding to the IDs; after obtaining the waveform recording files, according to the time marks in the waveform recording file names, regularize the waveform recordings belonging to the same fault, and place the waveform recording files of different types in the same folder during the fault; classify the waveform recordings generated by the oscillograph into the same category, classify the waveform recordings generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID, and then store them in the database.

[0018] Further, in step S01, if the substations and devices in the local platform change or are newly added, resulting in the mismatch of substation and device information between the local platform and the big data platform, then perform model association according to the following steps:

[0019] 1) Substation information association

[0020] Retrieve the set of substations in all big data platforms that are not associated with the substation information in the local platform, and match the substation information with the highest similarity to the current substation name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association;

[0021] 2) Protection device information association

[0022] Retrieve all protection devices of this substation in the big data platform, and match the protection device with the highest similarity to the current protection device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association;

[0023] 3) Auxiliary device information association

[0024] Retrieve all auxiliary devices of this substation in the big data platform, and match the auxiliary device with the highest similarity to the current auxiliary device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association;

[0025] The substation name matching algorithm based on feature statistics is specifically as follows:

[0026] First, define various thesauruses, including voltage level vocabulary, professional vocabulary for secondary equipment types, synonym dictionaries for equipment sets, equipment category dictionaries, and substation synonyms. Then, extract the voltage level based on the input name description to be extracted and the name type. If the voltage level is extracted, record it and remove the voltage level, and continue to extract the substation name. If successful, continue to extract the substation synonyms. If successful, record and remove the substation synonyms, and finally obtain the power station name. If the extraction of the substation name fails, extract the set information. If successful, record and remove the set information, continue to extract the secondary equipment type. If successful, record and remove the secondary equipment type, continue to extract the primary equipment type. If successful, record it, and finally obtain the equipment name;

[0027] For the obtained substation names, traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two substations with the highest similarity. Recursively perform this operation until all substations are associated;

[0028] For the obtained equipment names, traverse the secondary equipment in the two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two secondary equipment with the highest similarity. Recursively perform this operation until all equipment is associated;

[0029] The specific string matching algorithm is as follows:

[0030] According to the data modeling rules and historical data of the power dispatching control system, collect professional common words in the power system, and uniquely identify synonyms. Finally, obtain the keyword set. The string matching rules are as follows:

[0031] a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through equivalence rules, if there are inconsistencies, the string matching degree d w = 0, and it is directly determined that the strings do not match. If the keywords are exactly the same, continue to match according to Rule 2;

[0032] b) Rule 2: Continuous digital data matching rule. In addition to keywords, string data may contain non-keyword data such as Chinese characters, numbers, and Greek letters. The attributives of substations are generally multiple Chinese characters. For example, s1 = Zhaoqiao Substation 500kV Voltage I Busbar…, s2 = Zhaoqiao Substation 500kV Voltage II Busbar…, the attributives of voltage are generally numbers, and the attributives of busbars are generally Greek letters or Chinese characters. If there are several consecutive Arabic numerals or Greek numerals, these several consecutive numbers are regarded as a whole;

[0033] After the keyword matching is successful, several consecutive digital data need to be matched; several consecutive digital data are regarded as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w = 0, and it is directly determined that the strings do not match; otherwise, continue the matching according to Rule 3;

[0034] c) Rule 3: Calculate the string matching degree. After both the keyword and the consecutive digital data are matched, calculate the string matching degree according to the string matching algorithm. However, Rule 3 needs to be followed, that is, regardless of the number of characters in the keyword, it is uniformly calculated as one character. Both "voltage" and "busbar" are calculated as one character; digital data is also calculated as one character;

[0035] d) Rule 4: String matching result determination rule. If the string matching degree d of strings s1 and s2 w is greater than the string matching degree threshold, it is determined that strings s1 and s2 match; otherwise, the two strings do not match.

[0036] Furthermore, the determination steps of the alarm level in step S04 are as follows:

[0037] 1) Obtain all signal data within a period of time and sort them according to the name length;

[0038] 2) Starting from the signal name with the longest name, set the keyword of the signal and the corresponding signal classification and alarm level for the keyword one by one. At this time, the manual classification of a single signal is completed;

[0039] 3) After a new keyword is set, it will automatically match all the remaining unmatched signals with the keyword, and classify the other signals that meet the keyword into the alarm level and signal classification of the keyword match;

[0040] 4) Repeat the above steps 2) and 3) until the remaining unmatched signals are zero, or the remaining signals are too short to have the meaning of keyword matching; at this time, a series of corresponding rules from the keyword to the alarm level and signal classification are formed;

[0041] 5) After the real-time uploaded signals are accessed, before each signal is stored, perform a keyword match according to the longest keyword match selection to determine its alarm level and classification;

[0042] 6) After determining the signal classification of the newly obtained signal, mark the alarm level and signal classification for the signal and store them in the database uniformly; it is convenient for subsequent query and use by category;

[0043] 7) For signal names without any matching keywords, a reminder will be given, and the user can expand the existing keyword correspondence rules at any time.

[0044] S05. Organize the data during the fault time period.

[0045] Furthermore, the method for organizing the fault information in step S05 is specifically as follows:

[0046] 1) Retrieve the protection start signal among all the sent alarms;

[0047] 2) After retrieving the protection start signal, record its time point as the start time. After retrieving for a period of time, retrieve the corresponding protection start reset signal and record its time point as the end time; thus, an action time period can be obtained.

[0048] 3) Query all the signals of the faulty device from the start time to the end time and record them as directly related signals;

[0049] 4) Query the signals that occurred within the time period of all the devices in the substation where the faulty device is located and record them as indirectly related signals;

[0050] 5) Query the oscillograms within the action time period of the substation where the faulty device is located. Record the oscillogram of the faulty device as directly related oscillogram, the oscillogram generated by the oscillograph as the whole station oscillogram, and the oscillograms of other devices as indirectly related oscillograms for this startup;

[0051] 6) If it is line protection, query the oscillogram file of the line protection on the opposite side of this line protection and record it as the oscillogram of the opposite side protection;

[0052] 7) According to the above organization method, a comprehensive and well-structured fault-related information report can be obtained to provide for other functions such as fault discrimination, fault analysis, and hidden fault identification.

[0053] The present invention also provides a relay protection information system based on a secondary big data platform, which is applied to the above method and includes a big data platform and a local platform; the local platform is communicatively connected to the big data platform, and the big data platform accesses the protection information platform, oscillogram platform, and secondary equipment online monitoring system within the region.

[0054] Furthermore, the local platform includes a model association tool, a configuration data module, a real-time data module, and a data processing module; the configuration data module is connected to the power grid big data platform to obtain all the substation information and corresponding device information in the big data platform for the model association tool to configure;

[0055] The association tool module is used to associate the local device model with the disordered data obtained from the big data platform, and save and modify the completed association relationship to complete the local substation-device hierarchical modeling for the real-time communication module to refer to;

[0056] The data communication module is used to complete real-time communication between the local platform and the big data platform, and obtain real-time signal change and waveform recording file information of the big data platform;

[0057] The data processing module is used to classify the acquired signal data by signal category and alarm level, and integrate multiple waveform recording files.

[0058] Further, the specific association method of the association tool module is as follows: Assign physical IDs to multiple substations in the region, and obtain all substation, device, and signal information in the region from the big data platform. All substations are denoted as sta1, sta2... staN, all devices are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN; The method for corresponding all substations in the big data platform to all substations in the region is as follows:

[0059] According to the substation information interface and the protection device information interface, obtain all substation information and protection device information. The auxiliary device information obtains the information of all intelligent terminals and merging units, and forms a substation-device data structure of the secondary data platform according to the affiliated substation.

[0060] If there is an existing configuration file, read the association information in the configuration file, parse the substation and device data information of the big data platform in the association information, and associate it with the substation and device information in the big data platform system.

[0061] Further, the local platform obtains waveform recording data from the big data platform. The specific method is as follows: Input the big data platform in units of time periods, obtain the file IDs of all waveform recording files within the input time period, and obtain the waveform recording file corresponding to the ID with the file ID as the data; After obtaining the waveform recording file, according to the time mark in the waveform recording file name, regularize the waveform recordings belonging to the same fault, and place the waveform recording files of different types in the same folder during the fault; Classify the waveform recordings generated by the waveform recorder into the same category, classify the waveform recordings generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID, and then store them in the database.

[0062] Further, if the substations and devices in the local platform change or are newly added, resulting in the mismatch of substation and device information between the local platform and the big data platform, the model association is carried out according to the following steps:

[0063] 1) Substation information association

[0064] Retrieve the set of substations in all big data platforms that are not associated with the substation information in the local platform, and match the substation information with the highest similarity to the current substation name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association;

[0065] 2) Association of protection device information

[0066] Retrieve all protection devices of this plant station in the big data platform, and according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, match the protection device with the highest similarity to the current protection device name and associate them;

[0067] 3) Association of auxiliary device information

[0068] Retrieve all auxiliary devices of this plant station in the big data platform, and according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, match the auxiliary device with the highest similarity to the current auxiliary device name and associate them;

[0069] The substation name matching algorithm based on feature statistics is specifically as follows:

[0070] First, define various thesauruses, including voltage level vocabulary, professional vocabulary of secondary equipment types, equipment set synonym dictionaries, equipment category dictionaries, and substation synonyms; then, according to the input name description to be extracted and the name type, extract the voltage level. If the voltage level is extracted, record it and remove the voltage level, and continue to extract the substation name. If the extraction is successful, continue to extract the substation synonyms. If the extraction is successful, record it and remove the substation synonyms. Finally, obtain the power station name; if the extraction of the substation name fails, extract the set information. If the extraction is successful, record it and remove the set information, and continue to extract the secondary equipment type. If the extraction is successful, record it and remove the secondary equipment type, and continue to extract the primary equipment type. If the extraction is successful, record it, and finally obtain the equipment name;

[0071] For the obtained substation name, traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two substations with the highest similarity. Recursively perform until all substations are associated;

[0072] For the obtained equipment name, traverse the secondary equipment in the two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two secondary equipment with the highest similarity. Recursively perform until all equipment is associated;

[0073] The string matching algorithm is specifically as follows:

[0074] According to the data modeling rules and historical data of the power dispatching control system, collect professional common words in the power system, and uniquely identify synonyms. Finally, obtain the keyword set; the string matching rules are as follows:

[0075] a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through equivalence rules and there are inconsistencies, the string matching degree d w = 0, and it is directly determined that the strings do not match; if the keywords are exactly the same, continue the matching according to Rule 2;

[0076] b) Rule 2: Continuous digital data matching rule. In addition to keywords, string data may contain non-keyword data such as Chinese characters, numbers, and Greek letters. The attributives of a substation are generally multiple Chinese characters. For example, s1 = Zhaoqiao Substation 500kV Voltage I Busbar…, s2 = Zhaoqiao Substation 500kV Voltage II Busbar…, the attributives of voltage are generally numbers, and the attributives of busbar are generally Greek letters or Chinese characters. If there are several consecutive Arabic numerals or Greek numerals, these several consecutive numerals are regarded as a whole;

[0077] After successful keyword matching, it is necessary to match several consecutive digital data; several consecutive digital data are regarded as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w = 0, and it is directly determined that the strings do not match; otherwise, continue the matching according to Rule 3;

[0078] c) Rule 3: Calculate the string matching degree. After the keywords and continuous digital data are both matched, calculate the string matching degree according to the string matching algorithm. However, it is necessary to follow Rule 3, that is, regardless of the number of characters in the keyword, it is uniformly calculated as one character. Both "voltage" and "busbar" are calculated as one character; digital data is also calculated as one character;

[0079] d) Rule 4: String matching result determination rule. If the string matching degree d of strings s1 and s2 w is greater than the string matching degree threshold, it is determined that strings s1 and s2 match; otherwise, the two strings do not match.

[0080] Furthermore, the method for the data processing module to sort out the fault information is specifically as follows:

[0081] 1) Retrieve the protection start signal from all the sent alarms;

[0082] 2) After retrieving the protection start signal, record its time point as the start time, and retrieve the corresponding protection start reset signal after a period of time, and record its time point as the end time; thus, an action time period can be obtained;

[0083] 3) Query all the signals from the start time to the end time of the faulty device and record them as directly related signals;

[0084] 4) Query the signals that occurred in all devices within the substation where the faulty device is located during the time period, and record them as indirectly related signals;

[0085] 5) Query the oscillograms within the substation where the faulty device is located that belong to the action time period. Record the oscillogram of the faulty device as directly related oscillogram, the oscillograms generated by the oscillograph as substation-wide oscillograms, and the oscillograms of other devices as indirectly related oscillograms for this startup;

[0086] 6) If it is line protection, query the oscillogram file of the line protection on the opposite side of this line protection and record it as the oscillogram of the opposite side protection;

[0087] 7) According to the above sorting method, a comprehensive and hierarchical fault-related information report can be obtained to be used by other functions such as fault discrimination, fault analysis, and hidden fault identification.

[0088] The advantages of the present invention are as follows:

[0089] By associating the big data platform with each substation, the data of each system are integrated into a unified platform. It is necessary to ensure the integrity, accuracy, etc. of the data, and it is also necessary to classify the alarm signals into two levels, including faults and anomalies, classification, and alarm sub-types. Use physical IDs to associate each substation and device one by one, and match the device signals according to keywords to realize the model association mapping between each system.

[0090] Define the data synchronization method, and use the periodic and changing method to automatically integrate the data of each system. Save the original data to realize traceability analysis. Use a relational database to improve the data processing efficiency. Description of the Drawings

[0091] Figure 1 It is the system structure diagram corresponding to adopting the modeling method of Embodiment 1 of the present invention;

[0092] Figure 2 It is the flow chart of the key information extraction algorithm based on statistical features in Embodiment 1 of the present invention;

[0093] Figure 3 It is the screenshot of the association result display interface in the modeling method of Embodiment 1 of the present invention;

[0094] Figure 4 It is the system structure schematic diagram in Embodiment 2 of the present invention. Detailed Embodiment

[0095] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0096] Embodiment 1

[0097] A relay protection information modeling method based on a secondary big data platform:

[0098] Step 1: Obtain all substation-equipment-signal information from the big data platform. All substations are denoted as sta1, sta2... staN, all equipment are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN.

[0099] Step 2: Correlate all the obtained substation, equipment, and signal information locally to form a complete hierarchical structure locally, as Figure 1 shown.

[0100] Step 3: Classify all signal points according to their names into alarm levels and signal categories. For individual signals that cannot be typically classified, it is allowed for the operator to manually specify or input information.

[0101] Step 4: Real-time obtain all signal data and waveform recording files, and display them according to the pre-built model. And it supports providing data sources for various advanced functions and prompting relevant personnel through information release.

[0102] Currently, only the telemetry and telecontrol data provided by all other systems are collected on the big data platform, and no interface for others to extract data is provided. This solution includes expanding its external information interface.

[0103] The interface design for obtaining is as follows:

[0104] 1. Taking the time period as the input, obtain the file IDs of all waveform recording files within the input time period;

[0105] 2. Taking the file ID as the data, obtain the waveform recording file corresponding to the ID.

[0106] Obtain the docId of all waveform recording files (including various formats in the waveform recording) within the time period in the big data platform through the above interface, and then download all files from the big data platform to the local through the docId.

[0107] When obtaining files, first send the time unit of the file to be obtained to the platform. After that, the platform will return all file ID numbers within the specified time range.

[0108] 1. According to the file ID numbers, obtain all the oscillogram files from the platform one by one.

[0109] 2. After obtaining the oscillogram files, according to the time markers in the oscillogram file names, regularize the oscillograms belonging to the same fault, and place the oscillogram files of different types in the same folder during the fault.

[0110] 3. Classify the oscillograms generated by the oscillograph into the same category, classify the oscillograms generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID. Then store them in the database.

[0111] 4. According to the time, the oscillogram of the oscillograph and the corresponding oscillogram of the protection device generated after a fault in a substation can be obtained on the local platform.

[0112] During the model correspondence process, since the big data platform has its own substation ID, device ID, and signal ID, all three of these IDs are unique within the system in the big data platform. The specific steps for obtaining data and model correspondence are as follows:

[0113] 1. Use the substation information interface to obtain all substation information.

[0114] 2. Use the protection device information interface to obtain all the information of the protection devices, use the auxiliary device information to obtain all the information of the intelligent terminals and merging units, and form the substation-device data structure of the secondary data platform according to the affiliated substation. Display it in the left tree structure on the interface, as Figure 3 shown.

[0115] 3. If there is already a configuration file, read the associated information in the configuration file, parse the substation and device data information of the big data platform in the associated information, and perform an associated display with the substation and device information in the big data platform system.

[0116] 4. If the substation and device models in the local platform change or are newly added, resulting in a situation where the substation and device model information between the local platform and the big data platform does not correspond, then perform model association according to the following steps:

[0117] 1) Substation information association

[0118] Retrieve the set of substations in all big data platforms that are not associated with the substation information in the local platform. According to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, match the substation information with the highest similarity to the current substation name and perform the association.

[0119] 2) Association of protection device information

[0120] Retrieve all the protection devices of this substation in the big data platform, and match the protection device with the highest similarity to the current protection device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association.

[0121] 3) Association of auxiliary device information

[0122] Retrieve all the auxiliary devices of this substation in the big data platform, and match the auxiliary device with the highest similarity to the current auxiliary device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association.

[0123] 5. Method 1: Use the key information extraction algorithm based on statistical features. The specific details of the algorithm are as follows:

[0124] 1) Define various thesauruses

[0125] Define the voltage level thesaurus: Define the vocabulary that identifies the voltage level of the substation, such as "220kV", "500kV", etc.

[0126] Define the secondary equipment type thesaurus: Define the professional vocabulary that identifies the equipment type, such as words like "protection", "merging unit", "intelligent terminal", etc.

[0127] Define the thesaurus of synonyms for equipment sets: Define the thesaurus of synonyms for sets A and B of equipment. For example, the synonyms of "Set A" are "First Set", "Main A", etc.

[0128] Define the thesaurus of equipment category words: Define the equipment types. The synonyms of "main transformer" are "transformer", the synonyms of "busbar" are "bus", and the synonyms of "line" are "wire", for example.

[0129] Define the synonyms of substations: Define the synonyms of substations. The synonyms of "substation" are "station", "transformer substation", etc.

[0130] 2) Name extraction

[0131] As Figure 2 shown, when inputting a name description to be extracted and the name type (identifying the substation name or equipment name), the system can automatically extract the voltage level information in the description first. If successful, record the voltage level and remove the voltage level information. At present, voltage level information exists in both substation names and equipment names, and it can be extracted normally, and generally there is no failure.

[0132] ① For the substation name description, extract and record the substation synonyms and remove the substation synonyms.

[0133] ②For the description of the device name, extract the information of the device set, secondary device type, and primary device type, record it, and remove such information.

[0134] The remaining description is the name. The obtained substation name attributes and device name attributes are as follows:

[0135] Substation name attributes:

[0136] Attribute Name Type Meaning Voltage Level String Identify the substation voltage level Substation Synonym String Identify the extracted substation synonyms Substation Name String Identify the extracted substation name

[0137] Device name attributes:

[0138] Attribute Name Type Meaning Voltage Level String Identify the substation voltage level Secondary Equipment Type String Identify the secondary equipment type Secondary Equipment Set String Identify the extracted secondary equipment set Primary Equipment Type String Identify the primary equipment type Secondary Equipment Name String Identify the extracted secondary equipment name

[0139] 3) Substation information association

[0140] Traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair respectively, and associate the two substations with the highest similarity. Recursively perform this operation until all substations are associated.

[0141] 4) Device information association

[0142] Traverse the secondary devices in two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair respectively, and associate the two secondary devices with the highest similarity. Recursively perform this operation until all devices are associated.

[0143] Method 2: Use the string matching rule algorithm. The specific details of the algorithm are as follows:

[0144] 1) Keywords of device signal names

[0145] According to the data modeling rules of the power dispatching control system and the characteristics of historical data, there are some professional common words in the string data, such as substation, voltage, kV, bus, etc. The keywords of the power regulation system data mainly include substation, transformer, station, voltage, voltage level, kV, kV voltage, AC, bus, line, wire, disconnecting switch, switch, knife switch, circuit breaker, current transformer, voltage transformer, reactor, capacitor, etc. Integrate the keywords into a set to form a keyword set.

[0146] 2) Keyword equivalence rules

[0147] In the keywords of the regulation data, there are cases where multiple words have the same meaning. For example, voltage, kV, and kV voltage all represent voltage. To address this issue, a keyword equivalence rule is proposed in this paper. This rule processes keywords with the same meaning in an equivalent relationship.

[0148] In the keywords of power dispatching control system data, the same semantics may be expressed by different words. For example, a substation can be represented by "substation", "transformer substation" or "substation", and voltage can be represented by "kV", "kV voltage" or "voltage level". To solve the problem of multiple expressions for the same semantics, equivalence rules are used to match the keywords. The keyword set A of the power dispatching control system data = {"substation", "transformer substation", "substation", "voltage", "voltage level", "kV", "kV voltage", "AC", "busbar", "line", "switch", "disconnector", "isolating switch", "circuit breaker", "reactor", "current transformer", "voltage transformer", "capacitor"}.

[0149] To achieve reasonable matching of strings, equivalent keywords need to be uniquely identified. Based on the power dispatching control system data, the unique identifier set T of keywords is summarized as T = {"transformer substation", "voltage", "AC", "busbar", "line", "switch", "disconnector", "isolating switch", "circuit breaker", "reactor", "current transformer", "voltage transformer", "capacitor",...}.

[0150]

[0151] 3) Overall string matching rule

[0152] To speed up the string matching speed and accuracy, during the matching process of two strings in the dispatching control system data, the string matching rules are formulated and the string matching degree calculation method is proposed according to the characteristics of the control string data. Among them, the string matching steps and rules are as follows:

[0153] a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through the equivalence rule, if there is an inconsistency, the string matching degree d w = 0, and it is directly determined that the strings do not match; if the keywords are exactly the same, continue to match according to Rule 2.

[0154] b) Rule 2: Continuous digital data matching rule. In addition to keywords, string data may contain non-keyword data such as Chinese characters, numbers, and Greek letters. The attributives of a substation are generally multiple Chinese characters. For example, s1 = Zhaoqiao Substation 500kV Voltage I Busbar..., s2 = Zhaoqiao Substation 500kV Voltage II Busbar..., the attributives of voltage are generally numbers, and the attributives of busbar are generally Greek letters or Chinese characters. If there are several consecutive Arabic numerals or Greek numerals, these several consecutive numerals are regarded as a whole.

[0155] After the keyword matching is successful, several consecutive digital data need to be matched; several consecutive digital data are regarded as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w=0, directly determine that the string does not match; otherwise, continue matching according to rule three.

[0156] c) Rule 3: Calculate string matching. After the keyword and the continuous digital data are matched, the string matching algorithm is used to calculate the string matching. However, Rule 3 must be followed, that is, no matter how many characters the keyword contains, it is calculated as one character. "Voltage" and "busbar" are both calculated as one character; digital data is also calculated as one character.

[0157] d) Rule 4: String matching result judgment rule. If the string matching degree of strings s1 and s2 is d w If it is greater than the string matching threshold, then the strings s1 and s2 are determined to be matched; otherwise, the two strings do not match.

[0158] It should be noted that the plant and equipment information of all secondary big data platforms is directly listed. For newly associated equipment, you can manually check and adjust the association. You must first associate the plant and station before you can associate the equipment under the corresponding plant and station. A device on a local platform can only be associated with one secondary big data platform device. If the association is repeated, the previous association should be cancelled.

[0159] This solution uses the existing interface of the secondary big data platform to obtain all plant and equipment information of the big data platform. A corresponding hierarchical relationship between plant and equipment is formed locally. The information of each device and plant can be viewed through the system interface.

[0160] Data synchronization and storage, after the model is matched, the method for processing the matched data is as follows:

[0161] 1) The acquired data are divided into six types of signals and recording files: alarm, soft pressure plate, remote signal, switch quantity, analog quantity, and protection event.

[0162] 2) Signal data does not need to transmit status information, only change information needs to be transmitted. Acquisition is performed periodically.

[0163] 3) After acquiring the data, the signal data is directly stored in the database according to the following structure, the recorded data is stored as a local file, and the relevant information of the recorded file is recorded in the database.

[0164] Signal data storage structure:

[0165] Field Name Description id Signal ID station_id Substation ID ied_id Equipment ID name Signal Name value Value time Occurrence Time level Alarm Level

[0166] Recording data storage structure:

[0167]

[0168] Regarding Step 3 above, for signal data, keyword matching is adopted for classifying the alarm levels. The matching method is exact containment matching, and the steps are as follows:

[0169] 1. Obtain all signal data within a period of time and sort them according to the name length

[0170] 2. Starting from the signal name with the longest name, set the keyword of this signal and the corresponding signal classification and alarm level one by one. At this time, the manual classification of a single signal is completed

[0171] 3. After completing the setting of a new keyword, it will automatically match all the remaining unmatched signals with this keyword, and classify the other signals that meet this keyword into the alarm level and signal classification of keyword matching

[0172] 4. Repeat the above two steps until the remaining unmatched signals are zero, or the remaining signals are too short to have the meaning of keyword matching. At this time, a series of corresponding rules from keywords to alarm levels and signal classifications are formed

[0173] 5. After accessing the signals sent in real time, before storing each signal, we will perform a keyword matching according to the longest keyword matching selection to determine its alarm level and classification

[0174] 6. After determining the classification of the newly obtained signal, mark the alarm level and signal classification of this signal and store them in the database uniformly. It is convenient for subsequent query and use by category

[0175] 7. For signal names without any matching keywords, a reminder will be given, and the user can expand the existing corresponding rules of keywords at any time

[0176] Analyze the stock data and all model data in advance to form the corresponding rule relationship between keywords and alarm levels. After obtaining data once in a cycle, perform rule matching on the signals to form the corresponding alarm levels and store them in the database. In this way, it is possible to complete the modeling of signals locally when only the signal name is known without relying on any other information

[0177] The device signals are classified according to fault signals and alarm signals. Among them, the fault signals are severe alarms, including categories such as power self-check abnormality, differential current self-check abnormality, DSP channel self-check error, RAM error, FPGA error, EPROM abnormality, etc. The alarm signals are abnormal alarms, including categories such as input / output abnormality, management board abnormality, position jump abnormality, GOOSE configuration error, GOOSE reception and transmission mismatch, GOOSE disconnection, link abnormality, etc

[0178] The specific method for sorting out fault information is as follows: After obtaining the signal data and the oscillogram file data, the data within a specific fault time period is sorted out. Considering that the occurrence time of some faults may just straddle the cycle time node of data acquisition, resulting in incomplete information acquisition, the time for processing data lags behind the time for data acquisition by one cycle.

[0179] For the sorting out of fault data, it is mainly queried through two signal points: protection start and start reset. After querying a valid set of start - reset information, all historical information within the time period from start to reset, as well as the oscillogram files of the protection and the oscillograph, are obtained. The steps are as follows:

[0180] 1. Retrieve the protection start signal among all the sent - up alarms

[0181] 2. After retrieving the protection start signal, record its time point as the start time. After retrieving for a period of time, retrieve the corresponding protection start reset signal and record its time point as the end time. Thus, an action time period can be obtained.

[0182] 3. Query all the signals within the time period from the start time to the end time of this device and record them as directly related signals.

[0183] 4. Query the signals that occurred within the time period for all the devices in this substation and record them as indirectly related signals.

[0184] 5. Query the oscillograms within the substation that belong to the action time period. Record the oscillogram of this device as directly related oscillogram, the oscillogram generated by the oscillograph as the substation - wide oscillogram, and the oscillograms of other devices as indirectly related oscillograms for this start.

[0185] 6. If it is line protection, query the oscillogram file of the line protection on the opposite side of this line protection and record it as the opposite - side protection oscillogram

[0186] 7. According to the above sorting - out method, a comprehensive and well - structured fault - related information report can be obtained to be used by other functions such as fault discrimination, fault analysis, and hidden - fault identification.

[0187] Based on the above principles, after expanding the original change - position signal and oscillogram data attributes, the following attributes are added on the basis of the original attributes of the change - position signal and oscillogram data:

[0188] Data attributes added to the change - position signal:

[0189] Attribute Name Type Meaning Data Type String Identify that the data belongs to a changed signal or a waveform recording file Occurrence Moment Time Identify the moment when this signal changes Relative Time Stamp Time Time relative to the protection start moment when the signal changes Data Label String Identify that the data belongs to directly or indirectly related signals Equipment Physical ID String Uniquely identify the equipment

[0190] Data attributes added to the oscillogram file:

[0191] Attribute Name Type Meaning Data Type String Identify that the data belongs to a changed signal or a waveform recording file Occurrence Moment Time Identify the moment when the waveform recording file is generated Relative Time Stamp Time Time relative to the protection start moment when the waveform recording is generated Data Label String Identify that the data belongs to directly or indirectly related waveform recordings Equipment Physical ID String Uniquely identify the equipment Equipment Type Type Identify the equipment type This Opposite Substation Type Identify that the waveform recording file belongs to this substation or the opposite substation of the line

[0192] Embodiment 2

[0193] As Figure 4 shown, the modeling method includes a configuration data module, a model association tool, a real-time data communication module, and a data processing module on the local platform.

[0194] The layout at the actual site is as shown in the figure. The local platform is connected to the big data platform, and the big data platform is connected to systems such as the protection information management system, waveform recording system, and secondary monitoring system, collecting a large amount of information data of secondary equipment. However, its content is disordered and chaotic for the outside world and is difficult to be directly applied to fault analysis and processing. Therefore, the present invention is also applicable to processing data without model correspondence in communication with other platforms.

[0195] Through communication with the external big data platform and its own data processing, the present invention can first obtain a large amount of power grid operation data and store it locally. By modeling the substation and sorting out the data, it can be stored locally for users to consult from the connected workstations, and the data can be made hierarchical and more suitable for other functions carried out locally.

[0196] The configuration data module is connected to the power grid big data platform to obtain all substation information and corresponding equipment information in the big data platform for configuration by the model association tool. The substation information includes name, commissioning time, voltage level, substation type, and substation ID of the big data platform, and the equipment information includes equipment name, dispatching name, equipment type, commissioning time, voltage level, manufacturer, physical ID, equipment ID of the big data platform, etc.

[0197] The association tool module only runs locally and is used to associate the local equipment model with the disordered data obtained from the big data platform, and save and modify the completed association relationship to complete the substation-equipment hierarchical modeling locally for reference by the real-time communication module.

[0198] The data communication module is used to complete real-time communication between platforms and obtain information such as real-time signal changes and waveform recording files (including remote signals, remote measurements, waveform recordings, protection events, etc.) of the big data platform.

[0199] The data processing module is used to classify the obtained signal data by signal type and alarm level, and integrate multiple waveform recording files to form comprehensive and hierarchical data. And the processed data is stored and forwarded to other required modules.

[0200] The specific association method of the associated tool module is as follows: Assign physical IDs to multiple substations in the area, and obtain all substation, device, and signal information in the area from the big data platform. All substations are denoted as sta1, sta2... staN, all devices are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN. The method for corresponding all substations in the big data platform to all substations in the area is as follows:

[0201] According to the substation information interface and protection device information interface, obtain all substation information and protection device information. For auxiliary device information, obtain the information of all intelligent terminals and merging units, and form the substation-device data structure of the secondary data platform according to the affiliated substation.

[0202] If there is an existing configuration file, read the association information in the configuration file, parse the substation and device data information in the big data platform in the association information, and associate it with the substation and device information in the big data platform system.

[0203] The local platform obtains the oscillogram data from the big data platform. The specific method is as follows: Input the big data platform in units of time periods, obtain the file IDs of all oscillogram files within the input time period, and obtain the oscillogram files corresponding to the IDs with the file IDs as data. After obtaining the oscillogram files, according to the time marks in the oscillogram file names, regularize the oscillograms belonging to the same fault, and place the oscillogram files of different types in the same fault in the same folder. Classify the oscillograms generated by the oscillograph into the same category, classify the oscillograms generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID, and then store them in the database.

[0204] If the substations and devices in the local platform change or are newly added, resulting in the mismatch of substation and device information between the local platform and the big data platform, the model association shall be carried out according to the following steps:

[0205] 1) Substation information association

[0206] Retrieve the set of substations in the big data platform that are not associated with the substation information in the local platform, and match the substation information with the highest similarity to the current substation name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association.

[0207] 2) Protection device information association

[0208] Retrieve all protection devices of this substation in the big data platform, and match the protection device with the highest similarity to the current protection device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association.

[0209] 3) Auxiliary device information association

[0210] Retrieve all auxiliary devices of this plant station in the big data platform, and match the auxiliary device with the highest similarity to the current auxiliary device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform association;

[0211] The substation name matching algorithm based on feature statistics is specifically as follows:

[0212] First, define various thesauruses, including voltage level vocabulary, professional vocabulary of secondary equipment types, equipment set synonym dictionaries, equipment category dictionaries, and substation synonyms; then, extract the voltage level according to the input name description to be extracted and the name type. If the voltage level is extracted, record it and remove the voltage level, and continue to extract the substation name. If the extraction is successful, continue to extract the substation synonyms. If the extraction is successful, record and remove the substation synonyms, and finally obtain the power station name; if the extraction of the substation name fails, extract the set information. If the extraction is successful, record and remove the set information, continue to extract the secondary equipment type. If the extraction is successful, record and remove the secondary equipment type, and continue to extract the primary equipment type. If the extraction is successful, record it, and finally obtain the equipment name;

[0213] For the obtained substation name, traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two substations with the highest similarity. Recursively perform this operation until all substations are associated;

[0214] For the obtained equipment name, traverse the secondary equipment in the two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two secondary equipment with the highest similarity. Recursively perform this operation until all equipment is associated;

[0215] The string matching algorithm is specifically as follows:

[0216] According to the data modeling rules and historical data of the power dispatching control system, collect professional common words in the power system, and uniquely identify synonyms, and finally obtain the keyword set; the string matching rules are as follows:

[0217] a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through the equivalence rule, if there are inconsistencies, the string matching degree d w = 0, and directly determine that the strings do not match; if the keywords are exactly the same, continue to match according to Rule 2;

[0218] b) Rule 2: Continuous digital data matching rule. In addition to keywords, string data may contain non-keyword data such as Chinese characters, numbers, Greek letters, etc. The attributive of a substation is generally multiple Chinese characters, for example, s1 = Zhaoqiao substation 500kV voltage I bus..., s2 = Zhaoqiao substation 500kV voltage II bus..., the attributive of voltage is generally a number, and the attributive of the bus is generally a Greek letter or Chinese character. If several consecutive Arabic numerals or Greek numerals appear, these consecutive numbers are treated as a whole;

[0219] After the keyword is successfully matched, it is necessary to match several consecutive digital data; several consecutive digital data are treated as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w =0, directly determine that the string does not match; otherwise, continue matching according to rule 3;

[0220] c) Rule 3: Calculate string matching. After the keyword and the continuous digital data are matched, the string matching algorithm is used to calculate the string matching. However, Rule 3 must be followed, that is, no matter how many characters the keyword contains, it is calculated as one character. "Voltage" and "busbar" are calculated as one character; digital data is also calculated as one character;

[0221] d) Rule 4: String matching result judgment rule. If the string matching degree of strings s1 and s2 is d w If it is greater than the string matching threshold, then the strings s1 and s2 are determined to be matched; otherwise, the two strings do not match.

[0222] The method for collating fault information by the data processing module is as follows:

[0223] 1) Retrieve the protection start signal from all the alarms sent up;

[0224] 2) After the protection start signal is retrieved, its time point is recorded as the start time, and after a period of time, the corresponding protection start reset signal is retrieved, and its time point is recorded as the end time; thus, an action time period can be obtained;

[0225] 3) Query all signals from the start time to the end time of the faulty device and record them as directly related signals;

[0226] 4) Query the signals generated by all devices in the substation where the faulty device is located within the time period and record them as indirect related signals;

[0227] 5) Query the wave recordings in the substation where the faulty device is located within the action time period, record the wave recordings of the faulty device as directly related wave recordings, the wave recordings generated by the wave recorder as the whole station wave recordings, and the wave recordings of other equipment as indirectly related wave recordings of this startup;

[0228] 6) If it is line protection, query the oscillogram file of the line protection on the opposite side of this line protection, and record it as the oscillogram of the opposite side protection;

[0229] 7) According to the above sorting method, a comprehensive and well-structured fault-related information report can be obtained to be used by other functions such as fault discrimination, fault analysis, and hidden fault identification.

[0230] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for relay protection information modeling based on a secondary big data platform, characterized in that Including the following steps: S01. Assign physical IDs to multiple substations in the area, and obtain all substation, device, and signal information in the area from the big data platform. All substations are denoted as sta1, sta2... staN, all devices are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN. The method for corresponding all substations in the big data platform to all substations in the area is as follows: Obtain all substation information and protection device information according to the substation information interface and protection device information interface. Obtain the information of all intelligent terminals and merging units for auxiliary device information. Form a substation-device data structure of the secondary data platform according to the affiliated substation. If there is an existing configuration file, read the associated information in the configuration file, parse the substation and device data information in the associated information, and associate it with the substation and device information in the big data platform system. S02. Associate all the obtained substation, device, and signal information locally, and form a complete hierarchical structure on the local platform. S03. Classify all signal points according to the alarm level and signal category by name. S04. Real-time obtain all signal data and recording files in the big data platform, and perform alarm and classification according to the preset alarm level and signal classification. S05. Organize the fault information during the fault time period.

2. The relay protection information modeling method based on the secondary big data platform according to claim 1, wherein The specific method for obtaining signal data and recording files from the big data platform in step S04 is as follows: Input the big data platform in units of time periods, obtain the file IDs of all recording files within the input time period, use the file IDs as data to obtain the recording files corresponding to the IDs. After obtaining the recording files, according to the time marks in the recording file names, regularize the recordings belonging to the same fault, and place the recording files of different types in the same fault in the same folder. Classify the recordings generated by the recorder into the same category, classify the recordings generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID, and then store them in the database.

3. The relay protection information modeling method based on the secondary big data platform according to claim 1, characterized in that, In step S01, if the substations and devices in the local platform change or are newly added, resulting in the mismatch of substation and device information between the local platform and the big data platform, the model association is performed according to the following steps: 1) Substation information association Retrieve the set of substations in the big data platform that are not associated with the substation information in the local platform, and match the substation information with the highest similarity to the current substation name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. 2) Protection device information association Retrieve all protection devices of this substation in the big data platform, and match the protection device with the highest similarity to the current protection device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. 3) Auxiliary device information association Retrieve all auxiliary devices of this substation in the big data platform, and match the auxiliary device with the highest similarity to the current auxiliary device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. The substation name matching algorithm based on feature statistics is specifically as follows: First, define various thesauruses, including voltage level vocabulary, professional vocabulary of secondary equipment types, equipment set synonym dictionaries, equipment category dictionaries, and substation synonyms. Then, extract the voltage level according to the input name description to be extracted and the name type. If the voltage level is extracted, record it and remove the voltage level, and continue to extract the substation name. If the extraction is successful, continue to extract the substation synonyms. If the extraction is successful, record and remove the substation synonyms. Finally, obtain the power station name. If the extraction of the substation name fails, extract the set information. If the extraction is successful, record and remove the set information, and continue to extract the secondary equipment type. If the extraction is successful, record and remove the secondary equipment type, and continue to extract the primary equipment type. If the extraction is successful, record it, and finally obtain the equipment name. For the obtained substation names, traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair respectively, and associate the two substations with the highest similarity. Recursively perform this operation until all substations are associated. For the obtained equipment names, traverse the secondary equipment in the two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair respectively, and associate the two secondary equipment with the highest similarity. Recursively perform this operation until all equipment is associated. The string matching algorithm is specifically as follows: According to the data modeling rules and historical data of the power dispatching control system, collect professional common words in the power system, and uniquely identify synonyms. Finally, obtain the keyword set. The string matching rules are as follows: a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through the equivalence rule, if there are inconsistencies, the string matching degree d w = 0, and it is directly determined that the strings do not match; if the keywords are exactly the same, continue the matching according to Rule 2; b) Rule two: Continuous digital data matching rule. Except for keywords, string data may contain non-keyword data such as Chinese characters, numbers, and Greek letters. The attributives of substations are generally multiple Chinese characters. For example, s1 = Zhaoqiao Substation 500kV Voltage I Busbar…, s2 = Zhaoqiao Substation 500kV Voltage II Busbar… The attributives of voltages are generally numbers, and the attributives of busbars are generally Greek letters or Chinese characters. If several consecutive Arabic numerals or Greek numerals appear, these consecutive numerals are regarded as a whole. After the keyword matching is successful, several consecutive digital data need to be matched; several consecutive digital data are regarded as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w = 0, and it is directly determined that the string does not match; otherwise, continue the matching according to Rule 3; c) Rule three: Calculate the string matching degree. After both the keywords and the continuous digital data are matched, calculate the string matching degree according to the string matching algorithm. However, it is necessary to follow Rule three, that is, regardless of the number of characters in the keyword, it is uniformly calculated as one character, and both "voltage" and "busbar" are calculated as one character. Numerical data is also calculated as one character. d) Rule Four: String matching result determination rule. If the string matching degree d of strings s1 and s2 w is greater than the string matching degree threshold, then it is determined that strings s1 and s2 are matched; otherwise, the two strings do not match.

4. The relay protection information modeling method based on the secondary big data platform according to claim 1, characterized in that The determination steps of the alarm level in step S04 are as follows: 1) Obtain all signal data within a period of time and sort them according to the name length. 2) Starting from the signal name with the longest name, set the keyword of the signal and the corresponding signal classification and alarm level one by one. At this time, the manual classification of a single signal is completed. 3) After setting a new keyword, it will automatically match all the remaining unmatched signals with this keyword, and classify the other signals that match this keyword into the alarm level and signal classification of the keyword match. 4) Repeat the above steps 2) and 3) until the remaining unmatched signals are zero, or the remaining signals are too short to have the meaning of keyword matching; at this time, a series of corresponding rules from keywords to alarm levels and signal classifications are formed; 5) After accessing the signals sent in real time, before each signal is stored, perform a keyword matching on it according to the longest keyword matching selection to determine its alarm level and classification; 6) After determining the classification of the newly obtained signal, mark the signal with the alarm level and signal classification, and uniformly store it in the database; it is convenient for subsequent query and use by category; 7) For signal names without any matching keywords, a reminder will be given, and the user can expand the existing keyword corresponding rules at any time.

5. The relay protection information modeling method based on the secondary big data platform according to claim 1, wherein The method for sorting out the fault information in step S05 is specifically as follows: 1) Retrieve the protection start signal among all the sent alarms; 2) After retrieving the protection start signal, record its time point as the start time, and after retrieving for a period of time, retrieve the corresponding protection start reset signal and record its time point as the end time; thus, an action time period can be obtained; 3) Query all the signals of the faulty device from the start time to the end time and record them as directly related signals; 4) Query all the signals that occurred within the time period of all the devices in the substation where the faulty device is located and record them as indirectly related signals; 5) Query the oscillograms within the action time period of the substation where the faulty device is located. Record the oscillogram of the faulty device as the directly related oscillogram, the oscillogram generated by the oscillograph as the whole station oscillogram, and the oscillograms of other devices as the indirectly related oscillograms of this startup; 6) If it is a line protection, query the oscillogram file of the line protection on the opposite side of the line protection and record it as the oscillogram of the opposite side protection; 7) According to the above sorting method, a comprehensive and well-structured fault-related information report can be obtained to provide for other fault discrimination, fault analysis, and hidden fault identification functions.

6. The relay protection information system based on the secondary big data platform is characterized in that, Applied to the method described in any one of claims 1 to 5, it includes a big data platform and a local platform; the local platform is communicatively connected to the big data platform, and the big data platform accesses the protection information platform, oscillogram platform, and secondary equipment online monitoring system within the region.

7. The relay protection information system based on the secondary big data platform according to claim 6, characterized in that, The local platform includes a model association tool, a configuration data module, a real-time data module, and a data processing module; the configuration data module is connected to the power grid big data platform to obtain all the substation information and corresponding device information in the big data platform for the model association tool to configure; The association tool module is used to associate the device model of the local platform with the disordered data obtained from the big data platform, and save and modify the completed association relationship to complete the local substation-equipment hierarchical modeling for the real-time communication module to refer to; The data communication module is used to complete the real-time communication between the local platform and the big data platform, and obtain the real-time signal change and oscillogram file information of the big data platform; The data processing module is used to divide the obtained signal data into signal categories and alarm levels, and integrate multiple oscillogram files.

8. The relay protection information system based on the secondary big data platform according to claim 6, characterized in that The specific association method of the associated tool module is as follows: Assign physical IDs to multiple substations in the area, and obtain all substation, device, and signal information in the area from the big data platform. All substations are denoted as sta1, sta2... staN, all devices are denoted as dev1, dev2... devN, and all signals are denoted as sig1, sig2... sigN. The method for corresponding all substations in the big data platform to all substations in the area is as follows: Obtain all substation information and protection device information according to the substation information interface and protection device information interface. Obtain the information of all intelligent terminals and merging units according to the auxiliary device information, and form the substation-device data structure of the secondary data platform according to the affiliated substation. If there is an existing configuration file, read the association information in the configuration file, parse the substation and device data information in the big data platform in the association information, and associate it with the substation and device information in the big data platform system.

9. The relay protection information system based on the secondary big data platform according to claim 8, characterized in that The local platform obtains the oscillogram data from the big data platform. The specific method is as follows: Input the big data platform in units of time periods, obtain the file IDs of all oscillogram files within the input time period, and obtain the oscillogram file corresponding to the ID with the file ID as the data. After obtaining the oscillogram file, regularize the oscillograms belonging to the same fault according to the time mark in the oscillogram file name, and place the oscillogram files of different types in the same folder during the fault. Classify the oscillograms generated by the oscillograph into the same category, classify the oscillograms generated by the protection device itself into the same category, and classify them according to the corresponding substation ID and device ID, and then store them in the database.

10. The relay protection information system based on the secondary big data platform according to claim 9, characterized in that, If the substations and devices in the local platform change or are newly added, resulting in the mismatch between the local platform and the substation and device information in the big data platform, the model association shall be carried out according to the following steps: 1) Substation information association Retrieve the set of substations in the big data platform that are not associated with the substation information in the local platform, and match the substation information with the highest similarity to the current substation name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. 2) Protection device information association Retrieve all protection devices of this substation in the big data platform, and match the protection device with the highest similarity to the current protection device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. 3) Auxiliary device information association Retrieve all auxiliary devices of this substation in the big data platform, and match the auxiliary device with the highest similarity to the current auxiliary device name according to the substation name matching algorithm based on feature statistics or the string matching rule algorithm, and perform the association. The substation name matching algorithm based on feature statistics is specifically as follows: First, define various thesaurus, including voltage level vocabulary, professional vocabulary of secondary equipment types, synonym dictionary of equipment sets, equipment category dictionary, and substation synonyms. Then, according to the input name description to be extracted and the name type, extract the voltage level. If the voltage level is extracted, record it and remove the voltage level, and continue to extract the substation name. If successful, continue to extract the substation synonyms. If successful, record and remove the substation synonyms, and finally obtain the power station name. If the extraction of the substation name fails, extract the set information. If successful, record and remove the set information, continue to extract the secondary equipment type. If successful, record and remove the secondary equipment type, continue to extract the primary equipment type. If successful, record it, and finally obtain the equipment name; For the obtained substation names, traverse all unassociated substation information in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two substations with the highest similarity. Recursively perform this operation until all substations are associated; For the obtained equipment names, traverse the secondary equipment in the two associated substations in the big data platform and the local platform, calculate the similarity of attributes between each pair, and associate the two secondary equipment with the highest similarity. Recursively perform this operation until all equipment is associated; The specific string matching algorithm is as follows: According to the data modeling rules and historical data of the power dispatching control system, collect professional common words in the power system, and uniquely identify synonyms. Finally, obtain the keyword set. The string matching rules are as follows: a) Rule 1: Keyword matching rule. When the keywords in two strings are converted into unique identifiers through the equivalence rule and there are inconsistencies, the string matching degree d w = 0, and it is directly determined that the strings do not match; if the keywords are exactly the same, continue the matching according to Rule 2; b) Rule two: Continuous digital data matching rule. Except for keywords, the string data may contain non-keyword data such as Chinese characters, numbers, and Greek letters. The attributives of substations are generally multiple Chinese characters. For example, s1 = Zhaoqiao Substation 500kV Voltage I Busbar…, s2 = Zhaoqiao Substation 500kV Voltage II Busbar… The attributives of voltage are generally numbers, and the attributives of busbars are generally Greek letters or Chinese characters. If there are several consecutive Arabic numerals or Greek numerals, these several consecutive numbers are regarded as a whole; After the keyword matching is successful, several consecutive digital data need to be matched; several consecutive digital data are regarded as a whole and matched according to the digital data comparison method. If they are not equal, the string matching degree d w = 0, and it is directly determined that the string does not match; otherwise, continue the matching according to Rule 3; c) Rule three: Calculate the string matching degree. After the keywords and continuous digital data are both matched, calculate the string matching degree according to the string matching algorithm. However, it is necessary to follow Rule three, that is, regardless of how many characters the keyword contains, it is uniformly calculated as one character, and both "voltage" and "busbar" are calculated as one character. Numerical data is also calculated as one character; d) Rule Four: String matching result determination rule. If the string matching degree d of strings s1 and s2 w is greater than the string matching degree threshold, then it is determined that strings s1 and s2 are matched; otherwise, the two strings do not match.

11. The relay protection information system based on the secondary big data platform according to claim 6, characterized in that, The specific method for the data processing module to sort out the fault information is as follows: 1) Retrieve the protection start signal in all the sent alarms; 2) After retrieving the protection start signal, record its time point as the start time, and retrieve the corresponding protection start reset signal after a period of time, and record its time point as the end time. Thus, an action time period can be obtained; 3) Query all the signals within the start time to the end time of the fault device, and record them as directly related signals; 4) Query all the signals that occurred within the time period of all the devices in the substation where the fault device is located, and record them as indirectly related signals; 5) Query the oscillograms within the substation where the faulty device is located during the action time period. Record the oscillogram of the faulty device as the directly related oscillogram, the oscillograms generated by the oscillograph as the substation-wide oscillograms, and the oscillograms of other devices as the indirectly related oscillograms for this startup; 6) If it is line protection, query the oscillogram file of the line protection on the opposite side of this line protection and record it as the oscillogram of the opposite side protection; 7) According to the above sorting method, a comprehensive and well-structured fault-related information report can be obtained for use by other fault discrimination, fault analysis, and hidden fault identification functions.

Images