An autonomous and controllable substation multi-modal perception operation and maintenance system and method

Abstract

The present invention belongs to the field of substations. To solve the problem that the security of information interaction in existing substations is poor and the autonomous control of substations cannot be achieved, a multi-modal perception operation and maintenance system and method for autonomous and controllable substations are provided. Among them, the multi-modal perception operation and maintenance system for autonomous and controllable substations includes a collection module for synchronously collecting the real-time operation data of each electrical equipment in the substation main machine and the substation standby machine, and / or transmitting the recording files of the substation main machine and the substation standby machine to the communication module together; the communication module is used for processing the received substation operation synchronization data and / or recording files through encryption and decryption algorithms and then transmitting them to the data analysis module; the data analysis module is used for using a large model in the power grid field to identify faults in the substation operation synchronization data, and retrieving matching solutions from the solution database according to the fault types, and it can realize the autonomous and controllable full-domain operation and maintenance of the substation.

Description

Technical Field

[0001] The present invention belongs to the field of substations, and particularly relates to an autonomous and controllable substation multi-modal perception operation and maintenance system and method. Background Art

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] In order to achieve more comprehensive monitoring and operation and maintenance of substations, current substation operation and maintenance systems access many intelligent devices, but there are the following problems: (1) Data collection between multiple substations is asynchronous, which leads to a complex process layer network, tight coupling of SCD (Substation Configuration Description) files, and duplicate configuration of similar devices, insufficient data sharing, resulting in resource waste and poor security of substation information interaction, making it impossible to achieve autonomous and controllable full-domain operation and maintenance of substations. (2) During the data synchronization process of multiple substations, the technology of isolating the local area network and the external network within the substation is adopted, and data is transmitted in plain text in the local area network, but there are still security risks in the data transmission of the local area network; (3) During the process of fault identification and processing, the training of large models is time-consuming and costly in terms of computing power, and the training and inference fusion of current large models do not retain more effective quantization information, and the effect of balancing training efficiency and quantization is not achieved. Summary of the Invention

[0004] In order to solve the technical problems existing in the above background art, the present invention provides an autonomous and controllable substation multi-modal perception operation and maintenance system and method, which can improve the poor security of substation information interaction and achieve autonomous and controllable full-domain operation and maintenance of substations.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] The first aspect of the present invention provides an autonomous and controllable substation multi-modal perception operation and maintenance system.

[0007] An autonomous and controllable substation multi-modal perception operation and maintenance system includes: a collection module, a communication module, a data analysis module, and a display module;

[0008] The collection module is used to synchronously collect the real-time operation data of each electrical device in the substation main machine and the substation standby machine, or / and transmit the recording files of the substation main machine and the substation standby machine to the communication module together;

[0009] The communication module is used to transmit the received substation operation synchronization data or / and recording files to the data analysis module after processing through encryption and decryption algorithms;

[0010] The data analysis module is used for: identifying faults in the operation synchronization data of each substation by using a large model in the power grid field, and retrieving a matching solution from the solution database according to the fault type; and querying whether there is a hash table in the recorded wave file that associates multiple recorded wave files in different formats with the fault object in advance, so as to determine the timeliness of the recorded wave file.

[0011] The display module is used to display the fault identification result, its corresponding solution, and the timeliness analysis result of the recorded wave file.

[0012] As an implementation, in the data analysis module, when there is no solution corresponding to the identified fault type in the solution database, search for the solution corresponding to the fault type with the highest similarity to the fault type description in the solution database and transmit it to the display module for display.

[0013] As an implementation, in the data analysis module, when there is no solution corresponding to the identified fault type in the solution database, store the finally determined solution and its matching fault type in the solution database and update the solution database.

[0014] As an implementation, the acquisition module includes:

[0015] A real-time database construction module, which is used to establish real-time databases for the substation main machine and the substation standby machine respectively;

[0016] A multi-machine interaction module, which is used to establish a communication connection between the substation main machine and the substation standby machine by using a secure bus with encryption authentication;

[0017] A communication link determination module, which determines the communication link between the substation main machine and the substation standby machine based on the communication connection and the real-time database, so that the substation main machine and the substation standby machine respectively send their own communication device lists to each other through the secure bus to achieve real-time data synchronization.

[0018] As an implementation, in the multi-machine interaction module, the substation main machine regularly sends its own communication device list to the corresponding substation standby machine, and the standby machine combines the acquisition device list of the substation main machine to obtain the standby machine's global device list; the substation standby machine marks and monitors the relevant devices in the obtained main machine device list, and the substation standby machine does not communicate with the marked relevant devices; at the same time, the substation standby machine regularly sends its own acquisition device list to the substation main machine, and the main machine combines the communication device list of the substation standby machine to obtain the main machine's global device list, and the substation main machine marks and monitors the relevant devices in the obtained standby machine device list, and the substation main machine does not communicate with the marked relevant devices.

[0019] As an implementation manner, in the multi-machine interaction module, when the substation host has a downtime fault, the substation host is offline from the safety bus, then the substation standby machine cannot receive the communication device list sent by the substation host, and the substation standby machine will take over the data communication in the communication device list to achieve uninterrupted data acquisition; meanwhile, a fault alarm prompt for the substation host is given, and at this time, the substation standby machine enters the single-machine operation state.

[0020] As an implementation manner, in the multi-machine interaction module, when the substation host recovers from the fault, it immediately sends a data synchronization request to the substation standby machine, and the substation standby machine sends the cached data to the substation host; after receiving the cached data, the substation host resumes sending the communication device list; after receiving the communication device list, the substation standby machine performs a difference set process with the standby machine global device list to restore the global communication list to the acquisition device list of the standby machine itself, and the substation host and the substation standby machine resume the parallel state of collecting data downward.

[0021] As an implementation manner, in the communication module, before data transmission, the debugging tool and the network gateway perform key negotiation to generate their respective public keys and private keys; the debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key by using the public key of the other party and their own private key;

[0022] During data communication, the debugging tool and the network gateway perform encrypted transmission and decrypted transmission on the power data to be transmitted based on the generated session key and the constructed encryption and decryption algorithm; among them, the construction process of the encryption and decryption algorithm includes:

[0023] On the basis of the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array;

[0024] Encrypt the data block based on the round key, and generate a hash message authentication code based on the given HMAC key, and merge the encrypted data and the hash message authentication code, which are stored in the second array;

[0025] Verify the hash message authentication code in the array, decrypt the data block, and store the generated status value in the third array.

[0026] As an implementation manner, in the communication module, the debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key by using the public key of the other party and their own private key, including:

[0027] The key pair generated by the debugging tool includes a first public key and a first private key, and the key pair generated by the network gateway includes a second public key and a second private key;

[0028] The debugging tool sends the first public key to the network gateway, and the network gateway sends the second public key to the debugging tool;

[0029] After receiving the first public key of the debugging tool, the network gateway calculates the session key by using the first public key of the debugging tool and the second private key of the network gateway;

[0030] After receiving the first public key of the network gateway, the debugging tool calculates the session key by using the first public key of the network gateway and the second private key of the debugging tool.

[0031] As an implementation, in the communication module, on the basis of the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array, including:

[0032] Define a first array with a preset size for storing the expanded key;

[0033] Customize the key data information, read the key data information byte by byte, and store them into the first array respectively;

[0034] Through a preset number of iterations, a new key value is generated each time using the previous key value and stored in the first array. In each iteration, linear transformation and non-linear transformation are applied to increase the complexity of the key, and the generated round keys are stored in the first array.

[0035] As an implementation, in the communication module, encrypt the data block based on the round key, and generate a hash message authentication code based on the given HMAC (Hash-based Message Authentication Code) key, including:

[0036] Define a second array with a preset size for storing intermediate data during the encryption process;

[0037] Read the input plaintext data byte by byte and store them into the array respectively;

[0038] Through a preset number of iterations, encrypt and transform each time using the current state value and the round key to generate a new state value;

[0039] Convert the final state value back to a byte array and store it in a temporary variable;

[0040] Use the given HMAC key to perform a hash operation on the encrypted data to generate a hash message authentication code.

[0041] As an implementation, in the communication module, verify the hash message authentication code in the array, decrypt the data block, and store the generated state value in the third array, including:

[0042] Define a third array with a preset size to store intermediate data during the decryption process;

[0043] Extract the data and the hash message authentication code;

[0044] Use the given hash message authentication code key to perform a hash operation on the extracted encrypted data to generate a new hash message authentication code; compare the generated hash message authentication code with the extracted hash message authentication code, and if they do not match, return verification failure;

[0045] Through iterative operations, each time use the current state value and the round key to perform a decryption transformation to generate a new state value, convert the final state value back to a byte array, and store it in the third array.

[0046] As an implementation, in the data analysis module, the deployment process of the large model in the power grid field is as follows:

[0047] Based on the relationship between the memory overheads of single parameter, single gradient, single optimizer state, the number of model parameters, and the number of worker nodes in data parallelism and the memory overhead of the parallel decomposition strategy for storage collaboration, with the memory overhead of the parallel decomposition strategy for storage collaboration being the smallest, determine the parallel decomposition strategy of the large model in the power grid field;

[0048] Based on the parallel decomposition strategy of the large model in the power grid field, use the hyperparameters to input the hardware device information and the deep learning framework, automatically allocate hardware resources, and then use the guidance statements to set the deep learning model structure and configure the parallel decomposition scheme, complete the automatic code translation, and achieve the automatic conversion and training of the deep learning model.

[0049] As an implementation, in the data analysis module, the memory overhead of a single parameter , the memory overhead of a single gradient , the memory overhead of a single optimizer state , the number of model parameters , and the number of worker nodes in data parallelism The relationship between these parameters and the memory overhead of the parallel decomposition strategy for storage collaboration is as follows: .

[0050] As an implementation, in the data analysis module, the number of model parameters is: ; where is the floating-point precision for calculation.

[0051] As an implementation manner, in the data analysis module, the parallel decomposition strategy of the large model in the power grid field includes inter-group data parallelism, intra-group model parallelism, pipeline parallelism, and optimizer parallelism strategies.

[0052] As an implementation manner, in the data analysis module, the intra-group model parallelism strategy is as follows: ; where is the memory overhead of the parallel decomposition strategy for storage collaboration, is the memory limit of a single device, is the intra-group device allocation;

[0053] The inter-group data parallelism strategy is as follows: ; where is the intra-group device allocation, is the inter-group device allocation, is the total number of available devices.

[0054] As an implementation manner, in the data analysis module, the training process of the large model in the power grid field is as follows:

[0055] Obtain the configuration parameter information of the large model to be trained;

[0056] Based on the obtained configuration parameter information, construct the loss function of the large model, calculate the gradient norm, and obtain the first-order gradient information perception score of the large model;

[0057] Add perturbations to the large model, construct the perturbed loss function of the large model, calculate the Hessian matrix, and obtain the Hessian gradient information perception score of the large model;

[0058] Perform adaptive weighted combination on the obtained first-order gradient information perception score and Hessian gradient information perception score of the large model to obtain the quantization perception score of the large model;

[0059] According to the obtained quantization perception score of the large model, determine the quantization perception training samples of the large model, and complete the distributed quantization perception training of the large model.

[0060] As an implementation manner, in the data analysis module, during the process of determining the quantization perception training samples of the large model, sort the weights in the obtained quantization perception score of the large model, determine the training samples according to the sorting, and perform data parallel processing on the determined samples to complete the distributed quantization perception training of the large model.

[0061] As an implementation, in the data analysis module, during the process of adaptive weighted combination, after normalizing the obtained first-order gradient information perception score and Hessian gradient information perception score, calculate the first-order gradient information perception score weight and Hessian gradient information perception score weight, perform adaptive weighted combination on the sample scores to obtain the quantization perception score of the large model, and determine the sample data suitable for the quantization perception training of the large model according to the quantization perception score of the large model.

[0062] As an implementation, in the data analysis module, during the distributed quantization perception training of the large model, divide the determined large model quantization perception training samples into several subsets, process the divided subsets and calculate the subset gradients, aggregate all subset gradients in combination with the communication protocol to complete the update of the global model parameters; complete the distributed quantization perception training of the large model according to the updated model parameters.

[0063] As an implementation, in the data analysis module, perform backpropagation processing on the obtained large model loss function, calculate the partial derivative of the loss function after backpropagation processing with respect to the model parameters to obtain the model gradient; take the sum of the squares of all the obtained model gradients as the gradient norm, and take the square root of the gradient norm to obtain the first-order gradient information perception score of the large model.

[0064] As an implementation, in the data analysis module, calculate the second-order partial derivative of the constructed large model perturbed loss function with respect to the model parameters, take the obtained second-order partial derivative as the diagonal element of the Hessian matrix, calculate the sum of the squares of all the diagonal elements to obtain the Hessian gradient information perception score of the large model.

[0065] As an implementation, in the data analysis module, the process of determining the timeliness of the recorded wave file is as follows:

[0066] Judge whether the currently received recorded wave file exists in the preset hash table. If it exists, determine the fault object associated with the currently received recorded wave file in the hash table, otherwise end the query; wherein, the hash table is a hash table in which multiple recorded wave files in different formats are associated with fault objects, and multiple recorded wave files in different formats are associated with the same fault object;

[0067] Delete the key-value pair of the recorded wave file and the fault object in the hash table;

[0068] If the associated fault object is empty or the fault analysis is in progress, end the query, otherwise judge whether all the preset multiple fault recorded wave files in different formats exist in the currently received recorded wave file; if all the preset types of fault recorded wave files exist, start the fault processing thread, otherwise continue to obtain the recorded wave file.

[0069] As an implementation, in the data analysis module, it further includes: comparing the SCD files of each substation with the reference SCD file, and the process is as follows:

[0070] Read the content of each sub-node in the two SCD files to be compared;

[0071] Compare the content of each sub-node in the two SCD files to be compared, and store the comparison result in a data structure; among them, the comparison process of the content of each sub-node in the two SCD files to be compared includes:

[0072] Take the first SCD file as the reference file, traverse the specified nodes of the first SCD file, and search in the second SCD file for the corresponding nodes and the node data with differences according to the unique attribute identifier of the node and the check code calculated based on the node content, and take the node data with differences as change items and store them in the data structure;

[0073] Take the second SCD file as the reference file, traverse the specified nodes of the second SCD file, search for the corresponding nodes in the first SCD file, and take the node data not found as new items and store them in the data structure.

[0074] As an implementation, the data analysis module is further configured with an intelligent alarm module, and the process of automatic modeling of the intelligent alarm module is as follows:

[0075] Build an initial basic information library; among them, the initial basic information library includes a characteristic signal library, a characteristic equipment library, and a fault alarm information expert library;

[0076] Expand the initial basic information library according to the set rules; the process is as follows: classify the signals in the substation based on the preset rules, abstract the required signal types and store them in the characteristic signal library; classify the primary and secondary equipment based on the preset rules, abstract the required primary and secondary equipment types and store them in the characteristic equipment library; form fault alarm information based on the preset conditions for typical faults to occur and the sequence of signal occurrence, and store the corresponding solutions for various fault alarm information in the fault alarm information expert library together;

[0077] Classify the signals / equipment included in the selected actual interval / device according to the expanded characteristic signal library and characteristic equipment library, and then generate instances of fault distribution and alarm inference and form an intelligent alarm configuration information file according to the expanded fault alarm information expert library;

[0078] Publish the intelligent alarm configuration information file to the intelligent alarm module to obtain the final basic information library.

[0079] As an implementation, the data analysis module is further configured with a multi-thread management module, and the multi-thread management module is configured to:

[0080] Use multiple threads to separately execute hardware monitoring, alarm information monitoring, and control operations of the substation;

[0081] Adopt an independent acquisition method to obtain the hardware monitoring information and alarm information monitoring information of the substation monitoring system platform in real time according to the set sampling frequency;

[0082] Obtain the issued control operation items, perform corresponding control operations and detections on the substation monitoring system to obtain monitoring results; the control operation items are adaptively selected and determined according to the monitoring information obtained by independent acquisition.

[0083] The second aspect of the present invention provides a working method for an autonomous and controllable substation multi-modal perception operation and maintenance system.

[0084] A working method for an autonomous and controllable substation multi-modal perception operation and maintenance system includes:

[0085] Synchronously collect the real-time operation data of each electrical equipment in the substation main machine and the substation standby machine, or / and transmit the recording wave files of the substation main machine and the substation standby machine together through an encryption and decryption algorithm;

[0086] Obtain the decrypted operation synchronization data of each substation and / or the recording wave file;

[0087] Use a large model in the power grid field to identify faults in the operation synchronization data of each substation, and retrieve the matching solutions from the solution database; and query whether there is a hash table associating multiple recording wave files in different formats with the fault object in the recording wave file, so as to determine the timeliness of the recording wave file;

[0088] Display the fault identification results, their corresponding solutions, and the timeliness analysis results of the recording wave file.

[0089] The beneficial effects of the present invention are:

[0090] (1) The present invention provides an autonomous and controllable substation multi-modal perception operation and maintenance system, which synchronously collects the real-time operation data of each electrical equipment in the substation main machine and the substation standby machine or / and transmits the recording wave files of the substation main machine and the substation standby machine together through an encryption and decryption algorithm, uses a large model in the power grid field to identify faults in the operation synchronization data of each substation and retrieve the matching solutions, and determines the timeliness of the recording wave file by querying the hash table associating multiple recording wave files in different formats with the fault object, realizing the autonomous and controllable full-domain operation and maintenance of the substation while ensuring the secure information interaction of the substation.

[0091] (2) The present invention provides a real-time data synchronization method for multi-machine substations based on a secure bus, which sends its own communication device list to the other party through the secure bus, reducing the investment in substation digital infrastructure; at the same time, the present invention fully considers disaster tolerance processing when the monitoring host fails and gives an alarm prompt, minimizing the data loss caused by failures to the greatest extent.

[0092] (3) The present invention uses a fusion encryption method of sm2 and sm4 for encryption, and re-implements the standard sm4 algorithm after improvement, and adds an hmac data authentication function for communication, ensuring the security and efficiency in the data transmission process.

[0093] (4) The present invention proposes a parallel decomposition strategy for the coordination of computing and storage resources. According to the computing power and storage capacity limitations of available computing resources, a series of appropriate parallel decomposition strategies are selected, and finally an optimal large model distributed training strategy is constructed to improve the training efficiency and reduce the training cost; an automatic generation technology for large model training in the power grid field is proposed, which automatically converts the currently applied traditional artificial intelligence models into large models that can be distributed for training and completes the automatic configuration of hardware resources, thus realizing the elastic expansion and agile deployment of large model functions.

[0094] (5) The present invention combines the gradient criterion and the Hessian matrix, considers the first-order gradient information perception and the Hessian gradient information perception, reasonably sets the weights of the scoring function through an adaptive weighting mechanism, selects the sample data that is most sensitive to quantization-aware training, and realizes distributed training in the core sample selection and quantization perception process, achieving parallel acceleration of the large model while improving the quantization training efficiency and quantization effect of the large model.

[0095] The advantages of the additional aspects of the present invention will be partially given in the following description, partially will become obvious from the following description, or will be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] The schematic diagrams in the specification forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0097] Figure 1 It is a schematic diagram of the structure of the autonomous and controllable substation multi-modal perception operation and maintenance system according to an embodiment of the present invention;

[0098] Figure 2 It is a schematic diagram of the structure of the acquisition module according to an embodiment of the present invention;

[0099] Figure 3 It is a diagram of the secure communication process according to an embodiment of the present invention;

[0100] Figure 4 It is the deployment process of the large model in the power grid field of the embodiment of the present invention;

[0101] Figure 5 It is the training process of the large model in the power grid field of the embodiment of the present invention;

[0102] Figure 6 It is the process of automatic modeling of the intelligent alarm module of the embodiment of the present invention;

[0103] Figure 7 It is a schematic diagram of the Communication data of the communication service model of the embodiment of the present invention;

[0104] Figure 8 It is a schematic diagram of the IED data of the communication service model of the embodiment of the present invention;

[0105] Figure 9 It is a schematic diagram of the Data Type Templates data of the communication service model of the embodiment of the present invention;

[0106] Figure 10 It is a flowchart of the working method of the autonomous and controllable substation multi-modal perception operation and maintenance system of the embodiment of the present invention. Detailed implementation manners

[0107] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0108] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further descriptions of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0109] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "include" and / or "comprise" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0110] According to Figure 1 , the embodiment of the present invention provides an autonomous and controllable substation multi-modal perception operation and maintenance system, which includes: a collection module 101, a communication module 102, a data analysis module 103, and a display module 104;

[0111] The acquisition module 101 is used to synchronously acquire the real-time operation data of each electrical device in the substation main machine and the substation standby machine, and / or transmit the recording files of the substation main machine and the substation standby machine to the communication module together;

[0112] The communication module 102 is used to process the received substation operation synchronization data and / or recording files through encryption and decryption algorithms and then transmit them to the data analysis module;

[0113] The data analysis module 103 is used to: identify faults in the substation operation synchronization data by using a large model in the power grid field, and retrieve the corresponding solutions from the solution database according to the fault type; and query whether there is a hash table associating multiple recording files in different formats with the fault object in the pre-created recording files, so as to determine the timeliness of the recording files;

[0114] The display module 104 is used to display the fault identification results and their corresponding solutions, as well as the timeliness analysis results of the recording files.

[0115] In one or more embodiments, in the data analysis module 103, when there is no solution corresponding to the identified fault type in the solution database, find the solution corresponding to the fault type with the highest similarity to the fault type description from the solution database and transmit it to the display module for display.

[0116] The similarity here can be implemented by the distance between two vectors, which will not be elaborated here.

[0117] In the data analysis module 103, when there is no solution corresponding to the identified fault type in the solution database, store the finally determined solution and its matching fault type in the solution database and update the solution database.

[0118] Figure 2 It is a schematic structural diagram of the acquisition module in an embodiment of the present invention. As Figure 2 shown, the acquisition module 101 includes:

[0119] A real-time database construction module 1011, which is used to establish the real-time databases of the substation main machine and the substation standby machine respectively;

[0120] A multi-machine interaction module 1012, which is used to establish a communication connection between the substation main machine and the substation standby machine by using a secure bus with encryption authentication;

[0121] A communication link determination module 1013, which determines the communication links of the substation main machine and the substation standby machine based on the communication connection and the real-time database, so that the substation main machine and the substation standby machine respectively send their own communication device lists to each other through the secure bus to achieve real-time data synchronization.

[0122] The real-time database here is used to store real-time, dynamic, and changing data. The real-time database exists in the computer memory and is volatile storage. Commonly, shared memory, Redis, etc. can be used as the real-time database. The historical database stores non-volatile and persistent information such as various statistics, historical values, events, etc. Commonly, relational databases such as mysql and sqllite databases can be used as the historical database.

[0123] The IPC (Inter-Process Communication, data bus) service, that is, IPC is used to send data, commands, etc. of this node to other nodes through the network TCP or UDP, realizing the functions of data transmission and command transmission.

[0124] In this embodiment, the "real-time data" includes analog quantities (such as current and voltage), status quantities (signals), and original data such as metering transmitted through communication messages, as well as indirect data and related alarm event information generated after processing the original data (such as logical operations, filtering, etc.). By serializing these data, they can be transmitted among nodes through the bus, thereby achieving the purpose of data synchronization.

[0125] In this embodiment, both the substation main machine and the substation standby machine are deployed with SCADA (Supervisory Control And Data Acquisition) and communication front-end programs. The communication device program is used to communicate with various devices with network communication capabilities in the substation, such as protection and measurement control devices, protocol conversion devices, and on-line monitoring devices. The unprocessed data collected through data communication is the original data. The SCADA system processes the original data through arithmetic operations, inversion, dead zone judgment, mutation filtering, etc. according to preset thresholds, coefficients, flag bits, etc., and converts it into engineering data that can be displayed, alarmed, and recognized by humans. At the same time, after processing, the original data and engineering data are stored in the real-time database for use by other subsystems (or modules), and the non-real-time engineering data with statistical significance is stored in the historical database.

[0126] Before performing security bus encryption authentication, start the substation main machine, and realize reliable connections between various substation main machines based on the security bus. At the same time, the substation standby machine initiates a connection to the substation main machine; the substation standby machine sends a registration request message to the substation main machine. Based on time authentication, the substation main machine passes the authentication if and only if the certificates match, completing the successful authentication of the security bus, and the substation main machine and the substation standby machine can receive data messages from each other.

[0127] It should be noted that the security bus encryption can also adopt the ssh encryption tunnel method, so that the bus itself does not need to perform data authentication and encryption, and the ssh tunnel provides authentication and encryption services.

[0128] In the multi-machine interaction module 1012, the substation host regularly sends its own communication device list to the corresponding substation standby machine. The standby machine combines the acquisition device list of the substation host to obtain the standby machine global device list. The substation standby machine marks and monitors the relevant devices in the obtained host device list, and the substation standby machine does not communicate with the marked relevant devices. At the same time, the substation standby machine regularly sends its own acquisition device list to the substation host. The host combines the communication device list of the substation standby machine to obtain the host global device list. The substation host marks and monitors the relevant devices in the obtained standby machine device list, and the substation host does not communicate with the marked relevant devices.

[0129] In the multi-machine interaction module 1012, in this embodiment, assuming the total data scale is S, when the substation host and the substation standby machine start, real-time databases with an S-scale are established in full volume respectively. The substation host and the substation standby machine are respectively responsible for the communication links of the next S / N devices / modules according to the pre-configured data communication ratio (in this embodiment, N = 2).

[0130] After the communication link is established, the substation host and the substation standby machine communicate with the devices / modules they are responsible for. In the substation, communication protocols such as DLT / 860 mms, CMS, modbus, cdt, and 103 are generally used. The substation host and the substation standby machine respectively process data on the scale of S / N. Data processing includes but is not limited to business processes such as storing in the real-time database, storing in the historical database, statistical processing, inversion processing, coefficient processing, change dead zone processing, and associated telecontrol error.

[0131] After the real-time database is established, the substation host and the substation standby machine respectively send their communication device lists to the bus; that is, the device lists L1, L2,..., LN responsible for communication by all nodes are recorded on each machine. Generally, when N = 2, two device lists, L1 and L2, are stored on each machine.

[0132] In this embodiment, after the substation host processes the data of S / N scale, it sends the processed result to the substation standby via the security bus. To ensure performance, the method of real-time library address offset is adopted to transmit only the necessary information. Before synchronizing the data, the real-time library determines the data attributes to be synchronized according to the pre-configured settings of the real-time library, which usually include the real-time library address, value, time, quality and other related attributes, simply referred to as the data synchronization packet (SyncPacket, SP). The substation host only needs to send the SP to the standby. Considering the data message length limit of the security bus or the operating system, it can be sent in frames. Generally, 500 data packets can be used as one frame for sending. At the same time, the substation standby also uses the same mechanism to send to the substation host.

[0133] After the substation host and the substation standby are each responsible for the data of S / N after processing, they only need to parse the processed message sent by the other party into individual SP data and immediately store it in the real-time library and the historical library according to the address, without the need for further data processing. Although each machine is only responsible for processing the data of S / N scale, the data stored in the real-time library and the historical database of each machine is the full-scale data S. Data storage redundancy is achieved, but data acquisition and processing are parallel.

[0134] The substation host will regularly send its communication device list L1 to the standby. The data communication list structure includes the address, IP, and communication status of the downstream communication. After receiving the list L1, the substation standby performs a union operation on its own acquisition device list L2 and combines them into a global device list LG. In LG, all devices from L1 are marked, and the substation standby only monitors the status of these devices and does not perform communication processing. Similarly, the substation standby also sends the device list L2 to the host at regular intervals, and the host also performs the corresponding union and monitoring work.

[0135] When the substation host fails due to a crash, since the substation host is detached from the bus, the substation standby will not be able to receive the communication device list L1 sent by the other party. Therefore, the substation standby will take over the communication of the downstream data in the device list L1, ensure uninterrupted data acquisition, and display an alarm message on the interface to remind the duty personnel to handle it immediately to recover the fault. At this time, the substation standby enters the single-machine operation state.

[0136] When the substation standby takes over the device list L1, it will immediately classify the data from the device list L1 and the data from the device list L2, and cache the event information SOE and the historical library information HIS; the cache size is determined by the preset. If the host does not recover for a long time, a cache overflow alarm will be reported; the cache size can be set by the average fault handling time * average data load.

[0137] When the substation host recovers from a fault, it immediately sends a data synchronization request to the substation standby; the substation standby sends the cached data to the host; after the substation host finishes receiving the cache, it resumes sending the device list L1; after the substation standby receives the device list L1, it performs a difference set operation on the device list L1 and its own global list LG, and the communication list is restored to the device list L2, and the substation host and the substation standby resume the parallel state of the downstream acquisition.

[0138] After data distribution is carried out using the security bus in this embodiment, assuming the scale of the accessed data is S and the number of monitoring computer nodes is N, the amount of data accessed by each monitoring host can be reduced to S / N. According to the typical configuration of general substations of 220 kV and above, N = 2, so the amount of data accessed by each monitoring host is reduced to S / 2, and each monitoring host can retain half of the CPU computing power, reducing the investment in the substation digital infrastructure; at the same time, the present invention fully considers the disaster recovery processing when a certain monitoring host fails and gives an alarm prompt, minimizing the data loss caused by the failure to the greatest extent.

[0139] In this embodiment, the security bus technology is used to connect multiple monitoring hosts in parallel into a large data processing center, and end-to-end encryption authentication technology is provided to ensure that the computer nodes accessing the security bus have legal authorization.

[0140] In the communication module, as Figure 3 shown, the secure communication process is as follows:

[0141] Before data transmission, the debugging tool and the network gateway perform key negotiation to generate their respective public keys and private keys; the debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key using the public key of the other party and their own private key.

[0142] Specifically, in the communication module, the debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key using the public key of the other party and their own private key, including:

[0143] The key pair generated by the debugging tool includes the first public key and the first private key, and the key pair generated by the network gateway includes the second public key and the second private key;

[0144] The debugging tool sends the first public key to the network gateway, and the network gateway sends the second public key to the debugging tool;

[0145] After the network gateway receives the first public key of the debugging tool, it calculates the session key using the first public key of the debugging tool and the second private key of the network gateway;

[0146] After the debugging tool receives the first public key of the network gateway, it calculates the session key using the first public key of the network gateway and the second private key of the debugging tool.

[0147] In this embodiment, taking the communication parties as the debugging tool and the network gateway as an example, it shows that a secure key exchange needs to be carried out before the data transmission starts.

[0148] The key exchange process is as follows:

[0149] Step 001: Generate key pairs: The debugging tool and the network gateway respectively generate their own key pairs, including public keys and private keys. Assume the key pair generated by the debugging tool is (public key T, private key t), and the key pair generated by the network gateway is (public key G, private key g);

[0150] Step 002: Public key exchange: The debugging tool sends its public key T to the network gateway, and the network gateway sends its public key G to the debugging tool.

[0151] Step 003: Calculate the session key: After the network gateway receives the public key T of the debugging tool, it calculates the session key K using the public key T of the debugging tool and its own private key g. Similarly, after the debugging tool receives the public key G of the network gateway, it calculates the same session key K using the public key G of the network gateway and its own private key t.

[0152] Step 004: Confirm the session key: The debugging tool and the network gateway confirm that the session key K has been successfully generated and only the two of them know this session key.

[0153] Through the above key exchange process, the debugging tool and the network gateway successfully generate a shared session key K for subsequent data encryption and decryption operations. In this way, even in an open network environment, the secure transmission of data can be ensured.

[0154] During data communication, the debugging tool and the network gateway encrypt and decrypt the power data to be transmitted based on the generated session key and the constructed encryption and decryption algorithms; among them, the construction process of the encryption and decryption algorithms includes:

[0155] Based on the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array;

[0156] Encrypt the data block based on the round keys, and generate a hash message authentication code based on the given HMAC key. Merge the encrypted data and the hash message authentication code and store them in the second array;

[0157] Verify the hash message authentication code in the array, decrypt the data block, and store the generated status value in the third array.

[0158] Among them, the encryption process includes:

[0159] Once the session key is generated, both communicating parties can use this session key in combination with the ours-sm4-hmac algorithm to encrypt the data to be transmitted.

[0160] The debugging tool encrypts the plaintext data M to be transmitted using the session key K and the constructed encryption algorithm to generate the ciphertext C, expressed as: ;

[0161] The debugging tool sends the ciphertext C to the gateway.

[0162] Among them, the decryption process includes:

[0163] After receiving the ciphertext C, the gateway decrypts it using the shared session key K and the ours-sm4-hmac algorithm to restore the original plaintext data M, expressed as: ;

[0164] Suppose the debugging tool needs to send an encrypted message to the gateway. First, the debugging tool encrypts the plaintext message "Hello, Gateway!" using the shared session key K and the ours-sm4-hmac algorithm to generate the ciphertext C. Then, the debugging tool sends the ciphertext C to the gateway. After receiving the ciphertext C, the gateway decrypts it using the same session key K and the ours-sm4-hmac algorithm to restore the original plaintext message "Hello, Gateway!".

[0165] In one or more embodiments, in the communication module, on the basis of the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array, including:

[0166] Define a first array with a preset size for storing the expanded key;

[0167] Customize the key data information, read the key data information byte by byte, and store them into the first array respectively;

[0168] Through a preset number of iterations, each time a new key value is generated using the previous key value and stored in the first array. In each iteration, linear transformation and non-linear transformation are applied to increase the complexity of the key, and the generated round keys are stored in the first array.

[0169] This embodiment takes generating 64 round keys through the round key generation function on the basis of the original sm4_key_schedule as an example, and specifically includes the following steps:

[0170] Step 011, Initialize the array, define an array K with a size of 68 for storing the expanded key.

[0171] Step 012, Key Loading and Initial Transformation: Customize the key data information. In this embodiment, a 128-bit key (16 bytes) is selected. From the input 128-bit key (16 bytes), 4 32-bit integers are read byte by byte and stored in K[0] to K[3] respectively. Perform an exclusive OR operation on these 4 integers and mix them with the fixed parameter FK.

[0172] Step 013, Generate Round Keys: Through 32 iterative operations, each time use the previous 4 key values to generate a new key value and store it in K. In each iteration, apply linear transformation and non-linear transformation to increase the complexity of the key.

[0173] Step 014, Output Round Keys: Store the generated 64 round keys in the output array rk.

[0174] Generate round keys through the above function for subsequent encryption and decryption processes.

[0175] Among them, in the communication module, encrypt the data block based on the round key and generate a hash message authentication code based on the given HMAC key, including:

[0176] Define a second array with a preset size to store intermediate data during the encryption process;

[0177] Read byte by byte from the input plaintext data and store them in the array respectively;

[0178] Through a preset number of iterations, each time use the current state value and the round key for encryption transformation to generate a new state value;

[0179] Convert the final state value back to a byte array and store it in a temporary variable;

[0180] Use the given HMAC key to perform a hash operation on the encrypted data to generate a hash message authentication code.

[0181] Modify the encryption function to support 64-round encryption. Encrypt the data block through the encryption function and generate an HMAC to ensure the integrity and authenticity of the data, specifically including the following steps:

[0182] Step 021: Initialize the array: Define an array X with a size of 68 to store intermediate data during the encryption process;

[0183] Step 022: Data Loading: From the input 128-bit plaintext data, read 4 32-bit integers byte by byte and store them in X[0] to X[3] respectively;

[0184] Step 023: Round Key Encryption: Through 64 iterative operations, each time use the current state value and the round key for encryption transformation to generate a new state value;

[0185] Step 024: Data Output: Convert the final state value back to a byte array and store it in the temporary variable encrypted_data.

[0186] Step 025: Generate HMAC: Use the given HMAC key to perform a hash operation on the encrypted data to generate a 32-byte HMAC value.

[0187] Step 026: Combine Output: Combine the encrypted data and the HMAC value and store them in the output array output.

[0188] In the communication module, verifying the hash message authentication code in the array and decrypting the data block, and storing the generated state value in a third array includes:

[0189] Define a third array with a preset size to store intermediate data during the decryption process;

[0190] Extract the data and the hash message authentication code;

[0191] Use the given hash message authentication code key to perform a hash operation on the extracted encrypted data to generate a new hash message authentication code; compare the generated hash message authentication code with the extracted hash message authentication code, if they do not match, return verification failure;

[0192] Through iterative operations, each time use the current state value and the round key for decryption transformation to generate a new state value, convert the final state value back to a byte array, and store it in the third array.

[0193] For example, modify the decryption function to support 64 rounds of decryption, specifically including:

[0194] Step 031: Initialize the Array: Define an array X with a size of 68 to store intermediate data during the decryption process;

[0195] Step 032: Extract Data and HMAC, Extract the first 16 bytes from the input data as the encrypted data, and the last 32 bytes as the HMAC value.

[0196] Step 033: Verify HMAC, including:

[0197] Use the given HMAC key to perform a hash operation on the extracted encrypted data to generate a new HMAC value; compare the generated HMAC value with the extracted HMAC value, if they do not match, return verification failure.

[0198] Step 034: Round key decryption: Through 64 iterative operations, each time use the current state value and the round key to perform a decryption transformation to generate a new state value.

[0199] Step 035: Data output: Convert the final state value back to a byte array and store it in the output array output.

[0200] This embodiment uses a combined encryption method of sm2 and sm4 for encryption, and re - implements the standard sm4 algorithm with improvements, and adds an hmac data authentication function for communication, ensuring the security and efficiency during the data transmission process.

[0201] In one or more embodiments, in the data analysis module, as Figure 4 shown, the deployment process of the large - model in the power grid field is as follows:

[0202] S401: Based on the relationship between the memory overhead of a single parameter, the memory overhead of a single gradient, the memory overhead of a single optimizer state, the number of model parameters, and the number of worker nodes in data parallelism and the memory overhead of the parallel decomposition strategy for storage cooperation, with the memory overhead of the parallel decomposition strategy for storage cooperation being the smallest, determine the parallel decomposition strategy of the large - model in the power grid field;

[0203] S402: Based on the parallel decomposition strategy of the large - model in the power grid field, use the hyperparameters to input the hardware device information and the deep - learning framework, automatically allocate hardware resources, and then use the guiding statements to set the deep - learning model structure and configure the parallel decomposition plan to complete the automatic code translation, realizing the automatic conversion and training of the deep - learning model.

[0204] Specifically, in the data analysis module, during the current large - model training process, due to the significant increase in data such as model parameters, gradients, and optimizer states, the memory overhead limit is much greater than the computing overhead limit. Therefore, a parallel decomposition strategy for computing - storage cooperation is proposed. From the perspective of memory constraints, on the premise of improving the model training efficiency, based on the sharding technology of model parameters, gradients, and optimizer states, the parallel decomposition strategy of the large - model in the power grid field of this embodiment adopts the parallel decomposition strategy for storage cooperation.

[0205] Memory overhead of a single parameter 、Memory overhead of a single gradient 、Memory overhead of a single optimizer state 、Number of model parameters and the number of worker nodes in data parallelism The relationship between these parameters and the memory overhead of the parallel decomposition strategy for storage cooperation is as follows: .

[0206] This embodiment designs an optimal pipeline for improving the model training efficiency based on the model sharding strategy, combined with the forward and backward propagation calculation time, communication time, and model structure of the model.

[0207] Among them, the number of model parameters is: ; among them, is the floating-point precision for calculation.

[0208] The number of model parameters is automatically obtained by the deep learning framework.

[0209] In this embodiment, the parallel decomposition strategy of the large model in the power grid field includes inter-group data parallelism, intra-group model parallelism, pipeline parallelism, and optimizer parallelism strategies.

[0210] Among them, the intra-group model parallelism strategy is: ; among them, is the memory overhead of the parallel decomposition strategy for storage cooperation, is the memory limit of a single device, is the intra-group device allocation.

[0211] The inter-group data parallelism strategy is: ; among them, is the intra-group device allocation, is the inter-group device allocation, is the total number of available devices.

[0212] To solve the problems of difficult functional elastic expansion and agile iterative update of the large model in the power grid field, the automatic generation technology for training the large model in the power grid field has been studied, which can automatically convert the artificial intelligence model trained and run in the traditional single-machine and single-card environment in the power grid field into a large model in the multi-machine and multi-card environment, and can complete automated hardware resource configuration and training. Currently, it supports the deep learning frameworks PyTorch and TensorFlow, and is implemented through hyperparameter input and guidance statements.

[0213] Hyperparameter input is mainly used for framework selection and hardware resource configuration, and supports inputting the framework name, computing power resource information, and network transmission information through hyperparameters.

[0214] The guidance statement is to insert some simple guidance statements into the code of the traditional artificial intelligence model, which is used to implement the parallel decomposition strategy of computing and storage cooperation. The guidance statement identifies the parallel decomposition strategy adopted by the model, the specific number of devices for each strategy, and the model structure, so as to automatically translate the code of the small model in the single-machine and single-card environment into the code of the large model in the multi-machine and multi-card environment.

[0215] Finally, the automatic configuration and execution of the hardware environment are realized by using the hyperparameter input and the code translation result.

[0216] This embodiment adopts a parallel decomposition strategy for computing-storage collaboration, which can make full use of computing and storage resources. When using multiple (such as 4) GPUs to complete the distributed training of large models, the memory overhead can be reduced by 50% compared with that before applying this strategy, and the computing time can achieve an acceleration ratio of more than 5.4 times compared with a single GPU. Therefore, the comprehensive cost calculation can be reduced by about 52%.

[0217] This embodiment adopts an automatic generation technology for the training of large models in the power grid field, which can achieve agile iterative updates of large models and elastic expansion of functions, and shorten the R & D cycle of large models. Suppose that experienced large model R & D engineers need to go through five stages, namely, constructing a dataset, designing a domain model, adapting computing power resources, fine-tuning a general large model, and optimizing parameters, which takes a total of 3 months. Using the automatic generation technology proposed in this project, the processes of domain model design, computing power resource adaptation, and general large model fine-tuning can be greatly reduced, and the R & D cycle can be shortened by more than 50%.

[0218] In some other embodiments, in the data analysis module, as Figure 5 shown, the training process of the large model in the power grid field is as follows:

[0219] S501: Obtain the configuration parameter information of the large model to be trained;

[0220] S502: Based on the obtained configuration parameter information, construct the loss function of the large model, calculate the gradient norm, and obtain the first-order gradient information perception score of the large model;

[0221] S503: Add perturbations to the large model, construct the loss function with perturbations of the large model, calculate the Hessian matrix, and obtain the Hessian gradient information perception score of the large model;

[0222] S504: Perform adaptive weighted combination on the obtained first-order gradient information perception score and Hessian gradient information perception score of the large model to obtain the quantization perception score of the large model;

[0223] S505: Determine the quantization perception training samples of the large model according to the obtained quantization perception score of the large model, and complete the distributed quantization perception training of the large model.

[0224] In the specific implementation process, in the data analysis module, during the process of determining the quantization perception training samples of the large model, sort the weights in the obtained quantization perception score of the large model, determine the training samples according to the sorting, and perform data parallel processing on the determined samples to complete the distributed quantization perception training of the large model.

[0225] In the data analysis module, during the process of adaptive weighted combination, after performing normalization processing on the obtained first-order gradient information perception score and Hessian gradient information perception score, calculate the first-order gradient information perception score weight and Hessian gradient information perception score weight, perform adaptive weighted combination on the sample scores, obtain the quantization perception score of the large model, and determine the sample data suitable for the quantization perception training of the large model according to the quantization perception score of the large model.

[0226] In the data analysis module, during the process of distributed quantization perception training of the large model, divide the determined large model quantization perception training samples into several subsets, process the divided subsets and calculate the subset gradients, aggregate all subset gradients in combination with the communication protocol, and complete the update of the global model parameters; complete the distributed quantization perception training of the large model according to the updated model parameters.

[0227] In the data analysis module, perform backpropagation processing on the obtained large model loss function, calculate the partial derivative of the loss function after backpropagation processing with respect to the model parameters to obtain the model gradient; take the square sum of all the obtained model gradients as the gradient norm, and take the square root of the gradient norm to obtain the first-order gradient information perception score of the large model.

[0228] In this embodiment, the first-order gradient is used to reflect the loss function, that is, the change speed of the model parameters. When calculating the gradient of a certain input sample, it is actually evaluating the sensitivity of the model parameters to the loss change caused by the sample; if the gradient norm of the sample is large, it means that the sample has a more significant impact on the model parameters, indicating that the model is more sensitive to the sample.

[0229] In this embodiment, for the loss function perform backpropagation and calculate the gradient ; where represents the parameters of the model, i represents the index of the model parameters.

[0230] Calculate the square sum of the gradients as the gradient norm:

[0231] ;

[0232] Take the square root to get:

[0233] .

[0234] In the data analysis module, calculate the second-order partial derivative of the constructed large model perturbed loss function with respect to the model parameters, take the obtained second-order partial derivative as the diagonal element of the Hessian matrix, calculate the square sum of all the diagonal elements, and obtain the Hessian gradient information perception score of the large model.

[0235] During the process of quantization-aware training (QAT), the model needs to adapt to quantization errors; by calculating and analyzing the gradient norm, the top K samples are selected, and the samples that have the greatest impact on the results during quantization-aware training are identified. The obtained samples are used as a subset and re-participated in the model training process so that the model can still maintain a high accuracy after quantization.

[0236] In this embodiment, the parameter weights are perturbed and the increase in the training loss is observed.

[0237] Assume a given parameter perturbation and adopt the second-order Taylor approximation of the training loss. Assume a given parameter and apply a small perturbation such that The change in the training loss can be expressed as:

[0238]

[0239] where represents the expectation of the gradient of the loss function with respect to the parameter and represents the i-th value of the Hessian matrix of the loss, and is the second derivative of the loss function with respect to the parameter

[0240] Since the target model is a model that tends to converge, assume ; we get:

[0241]

[0242] Therefore, quantifies the impact of the perturbation caused by the training samples on the model training loss, the larger the

[0243] value, the higher the sensitivity of the sample to the full-precision model, and the higher the information richness contained in the sample. Therefore, the greater the possibility of being selected as a core sample. In actual calculations, directly obtaining the Hessian matrix score is very complex and computationally intensive, which will undoubtedly increase the additional computational cost during quantization. Therefore, this embodiment uses the Fisher information matrix (F) as an effective approximation; that is, the sum of the squares of the gradients is used to approximate the sum of the squares of the diagonal elements of the Hessian matrix.

[0244] Since H is the Hessian matrix of the negative log-likelihood loss, H is equivalent to the Fisher information matrix, that is, we get:

[0245]

[0246] In this embodiment, the score based on the Hessian matrix is denoted as:

[0247]

[0248] Weighted combination of the gradient score and the Hessian matrix score can make the core sample selection method more reasonable. Different weights can reflect the different importance of gradient information and Hessian information in sample selection.

[0249] In this embodiment, a weighting coefficient is introduced to adjust the relative importance of these two scores, and the gradient score and the Hessian matrix score are respectively normalized:

[0250]

[0251]

[0252] Among them, is the mean of the first-order gradient scores, is the standard deviation of the first-order gradient information scores, is the mean of the Hessian scores, is the standard deviation of the first-order gradient scores; the variances of the first-order gradient scores and the Hessian matrix scores are respectively:

[0253]

[0254]

[0255] The weights of the gradient and Hessian scores are obtained:

[0256]

[0257]

[0258] Combining the first-order gradient weight and the Hessian matrix weight, the final total score is obtained:

[0259] .

[0260] In this embodiment, the influence of various types of samples is balanced, and the sample with the highest score is comprehensively selected as the selected core sample.

[0261] In some other embodiments, in the data analysis module, the process of determining the timeliness of the recorded wave file is as follows:

[0262] Determine whether the currently received waveform recording file exists in the preset hash table. If it exists, determine the fault object associated with the currently received waveform recording file in the hash table; otherwise, end the query. Among them, the hash table is a hash table in which multiple waveform recording files in different formats are associated with fault objects, and multiple waveform recording files in different formats are associated with the same fault object.

[0263] Delete the key-value pair of the waveform recording file and the fault object in the hash table.

[0264] If the associated fault object is empty or a fault analysis is in progress, end the query; otherwise, determine whether all multiple preset different-format fault waveform recording files exist in the currently received waveform recording file. If all preset types of fault waveform recording files exist, start the fault processing thread; otherwise, continue to obtain waveform recording files.

[0265] In some embodiments, after receiving a list of fault waveform recording files including dat files, cfg files, and hdr files, create a fault object, then create a hash table in which three file names are associated with the fault object. The three file names are all associated with the same fault object. Finally, request these three files from the platform in sequence and wait for the platform to push the requested files to the App.

[0266] When the file pushed by the platform is received, obtain the fault object corresponding to the file in the hash table. If the obtained fault object is not empty and has not started running, then verify that if all three files are received, start the analysis process of the fault object. The timeliness is relatively strong and can meet the high real-time requirements for fault analysis of the intelligent alarm App.

[0267] For example, the process of determining the timeliness of the waveform recording file may include:

[0268] S1. The intelligent alarm App subscribes to the fault information of relevant devices from the main-auxiliary integrated platform.

[0269] Optionally, the intelligent alarm App is an APP client with functions such as data processing and transmission; the main-auxiliary integrated platform is equipment such as the integrated monitoring platform for main and auxiliary equipment of an intelligent substation; subscribing can be understood as the intelligent alarm App receiving information in the main-auxiliary integrated platform. The relevant devices and fault information are all relevant equipment and fault information in the substation, and the specific equipment and information content can be determined according to monitoring requirements or other requirements, which will not be elaborated here.

[0270] S2. After a fault occurs, the protection or waveform recording device will generate a waveform recording signal and file. After the background collects the waveform recording file of the device, it will push a file list to the intelligent alarm App.

[0271] S3. The intelligent alarm App processes the received file list, divides it into one or more fault groups, and sequentially determines whether the faults have been processed. If they have been processed, the data processing ends. If not, step S4 is performed.

[0272] It can be understood that after the intelligent alarm App processes the received file list and divides it into one or more fault groups, during the grouping process, a member for recording the fault time is created for the fault group. There is a map set in the App, and the map records the processed fault groups. The map is a key-value pair associating the fault time with the grouping time. Whether a fault has been processed is determined based on whether the map contains the fault time of the fault. If the map contains the time, it means it has been processed. If the map does not contain the time, it means it has not been processed. At the same time, the time of the fault needs to be added to the map.

[0273] S4. Create a fault object respectively according to the file list of each fault.

[0274] S5. Check whether the fault object is valid. If it is valid, step S6 is performed. Otherwise, the processing ends.

[0275] Optionally, the criterion for whether a fault object is valid is that it must simultaneously meet the naming conforming to the naming specification of the fault file, conforming to the subscribed faults, and the list includes at least three files. Optionally, the naming specification ends with —_F or _f, and the three files are dat file, cfg file, and hdr file.

[0276] S6. Associate the fault timeout signal and the processing function. The purpose of step S6 is that if the intelligent alarm App does not receive the files sent by the main-auxiliary integrated platform for a long time, the created fault object should be deleted. For example, if it does not receive the files sent by the main-auxiliary integrated platform for 10 minutes or longer.

[0277] S7. The intelligent alarm App sequentially makes asynchronous requests to the main-auxiliary integrated platform for fault files according to the file list, that is, sequentially requests the dat file, cfg file, and hdr file.

[0278] S8. Add all the requested files to the hash table. The hash table is composed of file name - fault object key-value pairs. Optionally, create the dat file - fault object key-value pair, cfg file - fault object key-value pair, and hdr file - fault object key-value pair sequentially.

[0279] S9. The intelligent alarm App receives the request file sent back by the platform, parses it, first determines whether the file name is included in the hash table. If not, it ends. If included, it obtains the fault object associated with the file in the hash table, and then deletes the file-fault object key-value pair in the hash table. If the fault object is empty or is undergoing fault parsing, it ends. Otherwise, it proceeds to step S10.

[0280] S10. Check whether the three files of the fault object have all been summoned, that is, whether all have been received. If all have been received, start the fault handling thread. Otherwise, continue to wait for the request file sent by the platform.

[0281] This embodiment can, after the main-auxiliary integrated platform transmits the fault recording file to the intelligent alarm App, perform fault analysis in a timely manner, improve the time efficiency of fault analysis, and meet the requirements of high real-time performance of high-reliability substations.

[0282] In some other embodiments, in the data analysis module, it further includes: comparing the SCD files of each substation with the reference SCD file. The process is as follows:

[0283] 1) Read the content of each sub-node in the two SCD files to be compared; among them, the content of each sub-node in the two SCD files to be compared includes: nodes such as Substation, Communication, IED, and DataTypeTemplates of the data type model and all sub-nodes thereunder.

[0284] 2) Compare the content of each sub-node in the two SCD files to be compared, and store the comparison result in a data structure; among them, the comparison process of the content of each sub-node in the two SCD files to be compared includes:

[0285] Take the first SCD file as the reference file, traverse the specified nodes of the first SCD file, and according to the unique attribute identifier of the node and the check code calculated based on the node content, search in the second SCD file for whether there is a corresponding node and the node data with differences, and take the node data with differences as the change items and store them in the data structure.

[0286] Take the second SCD file as the reference file, traverse the specified nodes of the second SCD file, search for the corresponding nodes in the first SCD file, and take the node data not searched as the new items and store them in the data structure.

[0287] The following takes the two SCD files A and B to be compared as an example to illustrate the specific comparison method. Take file A as the reference file and traverse the specified nodes of file A. Taking one of the nodes to be compared, node a, as an example, the steps are as follows:

[0288] Step a: Determine whether the node a to be compared has a unique identification attribute;

[0289] Step b: If it does, search for the corresponding node b in File B according to the value of the unique identification attribute of the node a to be compared. If it exists, compare the attribute values of the node b corresponding to the node a to be compared. If there are differences in the attribute values, store them in the change item of the data structure. If not, it means that the corresponding node b in File B has been deleted, and store it in the deletion item of the data structure;

[0290] Step c: If not, first delete all the attributes of the node a to be compared, and calculate the check code C1 of the node a to be compared. Then traverse the corresponding nodes in File B, calculate the check code C2 in the same way, compare C1 and C2, and determine whether C1 and C2 are equal. If they are equal, the two are the same node, and further compare the attribute values of the nodes. If there are differences in the attribute values of the nodes, store them in the change item of the data structure. Otherwise, there is no node in File B with the same check code C1, which means that the node with the same check code C1 in File B has been deleted, and store it in the deletion item of the data structure;

[0291] In this embodiment, the calculation methods of the check code C1 and the check code C2 of the node a to be compared are as follows: Extract the content sequence for calculating the CRC check code, and the calculation rules include:

[0292] (1) Delete the spaces, line breaks, carriage returns, and list symbols between elements and attributes;

[0293] (2) Arrange all the extracted element attributes in alphabetical order a-z;

[0294] (3) Elements without child elements and assignments end with " / >";

[0295] (4) Convert the sequence to a UTF-8 sequence, calculate the four-byte CRC-32 check code, and for those with less than four bytes, fill the high bytes with 0x0. CRC calculation parameters: number of bits Width = 32; generation item Poly = 04C11DB7; initialization value Init = FFFFFFFF; whether the data to be measured is reversed RefIn = True; whether the calculated value is reversed RefOut = True; output data exclusive OR item XorOut = FFFFFFFF.

[0296] Step d: Repeat Step a - Step c until all the specified nodes in File A have been compared.

[0297] Finally, take File B as the reference file, traverse the specified nodes, and search for the corresponding nodes in File A. If they do not exist, it means that the nodes do not exist in File A and are new nodes, and store them as new items in the new item of the data structure.

[0298] It should be noted that in this embodiment, the check code can be a CRC (Cyclic Redundancy Check) check code, parity check, MD5, SHA1, etc. There is no specific limitation in this embodiment, and it can be selected according to the actual situation.

[0299] In one or more embodiments, the data analysis module is further configured with an intelligent alarm module. As Figure 6 shown, the process of automatic modeling of the intelligent alarm module is as follows:

[0300] S601: Construct an initial basic information library; wherein, the initial basic information library includes a characteristic signal library, a characteristic equipment library, and a fault alarm information expert library;

[0301] S602: Expand the initial basic information library according to the set rules; the process is as follows: Classify the signals in the substation based on the preset rules, abstract the required signal types and store them in the characteristic signal library; Classify the primary and secondary equipment based on the preset rules, abstract the required primary and secondary equipment types and store them in the characteristic equipment library; Form fault alarm information based on the preset conditions for the occurrence of typical faults and the sequence of signal occurrence, and store the corresponding solutions for various fault alarm information in the fault alarm information expert library together.

[0302] In step S602: Classify the signals in the substation according to the fault alarm discrimination conditions.

[0303] In the characteristic signal library, the classification rules of the type signals are filtered by name keywords or reference signals, where the filtering rules support logical operations and priorities.

[0304] S603: Classify the signals / devices included in the selected actual interval / device according to the expanded characteristic signal library and characteristic equipment library, and then generate instances of fault distribution and alarm inference and form an intelligent alarm configuration information file according to the expanded fault alarm information expert library.

[0305] In step S603, automatically classify the measuring points included in the interval according to the characteristic signal library, and then instantiate the criterion signals depending on the fault analysis type associated with the interval.

[0306] The generated instances are the models required by the intelligent alarm analysis module for real-time monitoring of the discrimination of actual faults or anomalies.

[0307] In the specific implementation process, after classifying the equipment or signals, the types of signals required for fault or alarm judgment can be bound to the actual equipment and signals included in the interval or device according to the fault alarm information expert library, and instances of fault distribution and alarm inference are generated.

[0308] In one or more embodiments, the user can also add or delete the faults and alarm types automatically bound to the intervals or devices, supporting personalized customization. At the same time, it also supports the intelligent alarm modeling replication operation of intervals or devices. The alarm module of the interval or device can be copied, and then the device can be replaced to form a new interval or device alarm configuration.

[0309] The basic information library supports user extension and one-key import / export operations. It can be used as the initial information library for subsequent projects, reducing the workload of on-site implementation.

[0310] Among them, in the intelligent alarm configuration tool, an intelligent alarm virtual interval is created and bound to the actual physical interval. The actual physical interval refers to the interval in the substation, which is a combination of actual electrical equipment in the substation; the intelligent alarm virtual interval is a processing unit established in the intelligent alarm analysis module for advanced applications such as fault judgment and abnormal analysis. It can be bound to the actual physical interval, transformer, or bus. In addition, it also needs to be bound to other information required for advanced application analysis, such as interval or associated acquisition device information.

[0311] Based on the primary equipment included in the interval and the measuring points bound to the primary equipment, identify the IEDs where these measuring points are located and automatically bind them to the intelligent alarm virtual interval.

[0312] Based on the IED bound to the interval and information such as the voltage level of the interval, find the fault alarm types associated with this type of interval in the fault alarm information library, and construct the alarm types that this interval needs to process.

[0313] Automatically classify the measuring points included in the interval according to the characteristic signal library, and then instantiate the criterion signals depending on the fault analysis types associated with this interval.

[0314] Complete the modeling of intelligent alarm analysis through the binding of device type, fault type, and signals.

[0315] S604: Publish the intelligent alarm configuration information file to the intelligent alarm module to obtain the final basic information library.

[0316] In this embodiment, according to the extended characteristic signal library and characteristic equipment library, classify the signals / devices included in the selected actual interval / device, and then generate instances of fault distribution and alarm inference and form an intelligent alarm configuration information file according to the extended expert library of fault alarm information. Furthermore, publish it to the intelligent alarm module to obtain the final basic information library, which supports rapid modeling, reduces the workload of on-site construction personnel; improves the actual utilization rate of intelligent alarms and reduces the workload of monitoring personnel; supports the import / export of the basic information library, improves the iteration rate of the basic information library, and is conducive to expanding the types of fault analysis and alarm inference processing.

[0317] In one or more embodiments, the data analysis module is further configured with a multi-thread management module, and the multi-thread management module is configured to:

[0318] Use multiple threads to separately execute hardware monitoring, alarm information monitoring, and control operations of the substation;

[0319] Adopt an independent acquisition method to obtain hardware monitoring information and alarm information monitoring information of the substation monitoring system platform in real time according to the set sampling frequency;

[0320] Obtain the issued control operation items, perform corresponding control operations and detections on the substation monitoring system, and obtain monitoring results; the control operation items are adaptively selected and determined according to the monitoring information obtained by independent acquisition.

[0321] In this embodiment, the use of multiple threads can execute hardware monitoring, alarm information monitoring, and control operations in parallel, improving the operating efficiency of the system. Each thread works independently and will not block each other, enabling the system to respond and process various monitoring tasks in real time; the multi-threads have clear division of labor and each performs its own duties. Even if a certain thread fails, other threads can still continue to run, ensuring the overall stability and reliability of the system.

[0322] The independent acquisition method is: adopt the method of actively monitoring and uploading the hardware status and alarm information;

[0323] Optionally, the hardware status monitoring includes data acquisition of CPU load, main memory, physical memory, virtual memory, hard disk usage, network card usage, motherboard temperature, optical drive usage, and USB device status to ensure the health of the hardware status of the substation monitoring background;

[0324] Optionally, the alarm information monitoring includes local and remote login information, operation input echo, logout operation, screen unlocking, abnormal network access, illegal port external connection, key directory / file change, and user permission change, etc.;

[0325] In the actual device, there will be multiple IPs and ports enabled, but not all of them need to be monitored for alarms. Adding the normally used IPs and ports to the white list can eliminate the alarms corresponding to these IPs and ports.

[0326] For a further technical solution, when the monitoring data is automatically collected and uploaded, configure the network and port white list to achieve flexible monitoring; in order to achieve flexible monitoring, the configuration of the network and port white list can be dynamically managed through a configuration file or a management interface, including the following steps:

[0327] Step a1: Define a configuration file containing the allowed network and port information;

[0328] Specifically, the monitoring software is equipped with a configuration tool that can graphically configure device information, including local or remote IP, port, and network whitelist, as well as the switch options for each monitoring item, which are finally stored in the configuration file.

[0329] Step a2: After the first monitoring software starts monitoring, it reads the configuration file and dynamically sets the firewall rules according to the configuration file.

[0330] Specifically, different firewall rules are set according to the different services, the confidentiality and security levels of the information transmitted through the corresponding ports, etc. provided in the configuration file, providing fine-grained access control.

[0331] In the above solution, configuring the network and setting the port whitelist to restrict data transmission to only specific network addresses and ports can improve the security and reliability of data transmission and prevent unauthorized access or data transmission.

[0332] The monitoring data is automatically collected and uploaded. The TCP / IP protocol can be used to ensure the reliable transmission of data. Through the confirmation and retransmission mechanism, packet loss can be effectively avoided.

[0333] Furthermore, the automatic collection and upload of monitoring data includes the following steps:

[0334] Step b1: Set the triggering mechanism for collecting monitoring data to collect the hardware status and alarm information.

[0335] Step b2: Sub-packet, package, and compress the data, and divide the priority of data transmission.

[0336] Specifically, the data is sub-packeted according to the set size for easy network transmission and retransmission. Each packet contains a sequence number and a check code to ensure data integrity. Further, the sub-packeted data is compressed to reduce the amount of data transmitted, thereby improving the transmission speed.

[0337] Optionally, different priorities are set according to the importance and real-time requirements of the data. High-priority data such as alarm information is preferentially transmitted to ensure the real-time nature of key data.

[0338] In this embodiment, data with high real-time requirements is preferentially scheduled to ensure its transmission in the first time.

[0339] Step b3: Set buffers at the sending end and the receiving end respectively. Place the data packets to be sent in the buffer according to the priority, and send the data in the sending-end buffer to the receiving end.

[0340] In this step, using buffers at the sending end and the receiving end to manage the sending and receiving of data can prevent data loss caused by instantaneous network jitter.

[0341] Step b4: After the receiving end receives the data packet, it sends an acknowledgment message, and the sending end determines whether retransmission is required based on the acknowledgment message;

[0342] Step b5: Set a timeout period. If the acknowledgment message is not received within the specified time, the sending end resends the data packet;

[0343] Steps b4 to b5 above avoid data packet loss through the retransmission mechanism and the timeout retransmission mechanism, improving the accuracy of data transmission.

[0344] A further technical solution is to dynamically adjust the data sending rate according to the network bandwidth during data transmission to ensure good real-time performance even when the network bandwidth fluctuates.

[0345] To achieve smooth data transmission, further, a flow control mechanism is set at the sending end and the receiving end to prevent network congestion caused by too fast data sending rate;

[0346] The process of adaptive selection is implemented in the network security master station. The network security master station automatically selects the optimal control operation item according to the current monitoring data and preset rules, and issues the corresponding control command for operation.

[0347] Set the terminal of the monitoring software in this embodiment to interact with the network security master station for control operations. The terminal of the monitoring software itself acts as a client and can not actively perform control operations. The master station performs adaptive selection of control operation items and issues them. After the monitoring software receives the corresponding message, the monitoring software executes the corresponding operation and returns the corresponding operation result.

[0348] For example, if it is detected that a certain hardware index exceeds the preset threshold, the network security master station selects to execute the corresponding control operation, such as parameter viewing or vulnerability scanning, and issues it to the monitoring software. The monitoring software executes the parameter viewing or vulnerability scanning process to obtain the detection result;

[0349] Among them, the control operation items include parameter viewing, parameter setting, baseline verification, vulnerability scanning, and active network disconnection operation;

[0350] The adaptive selection is based on a preset rule library, which contains the optimal control operation items corresponding to different monitoring information. Specifically, the process of adaptive selection is as follows:

[0351] Step c1: Data collection and preprocessing: Collect the hardware monitoring information and alarm information of the substation monitoring system and perform data preprocessing;

[0352] Step c2: Feature extraction: Extract key features from the preprocessed monitoring data and perform standardization processing;

[0353] Step c3, Construction of the rule base: Construct a rule base that includes different monitoring information and optimal control operation items;

[0354] Optionally, the rule base includes a series of predefined rules, and each rule corresponds to specific monitoring information and optimal control operation items;

[0355] Optionally, the rule format can be: IF <monitoring condition> THEN <control operation>

[0356] For example: IF CPU usage > 90% AND memory occupancy > 80% THEN perform parameter viewing operation.

[0357] The rule base is continuously updated and optimized based on historical data and expert experience to ensure the effectiveness and accuracy of the rules

[0358] Step c4, Rule matching: Check each rule in the rule base in sequence according to the current feature set, perform rule matching, and determine whether the monitoring condition is met;

[0359] Step c5, Select the optimal operation: Select the optimal control operation item in the rules that meet the conditions based on the priority;

[0360] Step c6, Command issuance: Generate and issue a control command to execute the corresponding control operation.

[0361] An implementable implementation manner, in step c4, the rule matching step in the adaptive selection process includes using a decision tree algorithm to check the monitoring conditions of each rule in sequence, find the matching leaf node, and execute the corresponding control operation item;

[0362] An alternative technical solution, in step c4, the rule matching step may also include using a Bayesian network to calculate the posterior probability according to the current feature set and select the control operation item with the highest probability for execution.

[0363] Another alternative technical solution, in step c4, the rule matching step may also include using fuzzy logic, through a fuzzy inference system, to calculate the fuzzy membership degree according to the current feature set and select the control operation item with the highest membership degree.

[0364] The monitoring method of this embodiment strengthens the self - safety monitoring of the substation monitoring system, monitors the software and hardware of the substation monitoring system in real - time, helps to early - warn potential risks, and reduces the probability of safety accidents.

[0365] For a further technical solution, the monitoring software in this embodiment is written in C++. The cross-platform development of the monitoring software has significant advantages and can run on multiple platforms. Secondly, it supports multiple compilers and has good support for different operating systems. For different hardware, 80% of the functions can be used normally, and only a small number of devices with different drivers and hardware interfaces need to be further adapted. The functions are modular, the development is flexible and efficient, and the adaptation efficiency is high, which solves the compatibility problems of different hardware products. It can solve the software and hardware compatibility and adapt to the devices and systems of different manufacturers and different periods.

[0366] In one or more embodiments, the acquisition module 101 can be implemented by using a CMS client, the data analysis module 103 is implemented by using a CMS server, and the CMS encrypted remote control communication process of the communication module 102 includes:

[0367] Step A: Set the encrypted transmission mode on both the CMS server and the CMS client, and load the corresponding database for configuration. Among them, the encrypted transmission mode can be specifically set according to the actual situation.

[0368] Step B: Determine whether a connection is established with the CMS service port on the CMS client. If so, send a remote control command to the CMS server.

[0369] If the CMS client fails to establish a connection with the CMS service port, test the network connection volume with the CMS server (for example, ping the CMS server IP), and periodically detect the CMS service port. If the CMS server is detected, connect to the CMS server.

[0370] It should be noted here that in other embodiments, other existing command operations can also be used to test the network connection volume with the CMS server.

[0371] If the TCP / IP port is not monitored, exit the main process of the CMS server module.

[0372] In this embodiment, the remote control message structure includes reference, ctlVal, operTm, origin_orCat, origin_orIdent, ctlNum, t, test, check, and AddCause.

[0373] Step C: Listen on the TCP / IP port at the CMS server to determine if a remote control command is received. If so, continue to determine the positive response. If it is a remote control message, sign it, put the signature content into the remote control structure, encode the remote control message and send the message, and then forward the remote control success response to the CMS client.

[0374] If the CMS server has a positive response, sign the remote control message using the SM2 algorithm to generate a 64-bit hash value, and put the 64-bit hash value into the remote control structure.

[0375] Specifically, the 64-bit hash value is put into the origin_orident field in the remote control structure.

[0376] If the CMS server has a negative response, give a remote control negative response.

[0377] Among them, the remote control commands include: remote control selection, remote control execution, and remote control cancellation.

[0378] Step D: After the CMS client receives the remote control response, verify the signature of the remote control message, and then determine whether the signature verification passes. If so, the remote control is successful; otherwise, the remote control fails.

[0379] After the remote control is successful, perform corresponding response operations according to the actual request operations corresponding to the remote control commands at the CMS server.

[0380] After the signature verification passes, the CMS server performs remote control positive (negative) response operations according to the remote control actual request operations. For positive response, the switch is in the closed or open position, and the remote control ends. Otherwise, directly return a negative response, and the remote control ends.

[0381] This embodiment realizes the remote control encryption communication process during the CMS communication of the substation monitoring system, encrypts the application layer, and performs ciphertext transmission during the remote control communication process. It realizes that after intercepting the message, the message cannot be replayed to attack the equipment in the substation, greatly improving the security and reliability in the substation.

[0382] In some alternative embodiments, when the data analysis module 103 is implemented using the CMS server, the SylixOS operating system can be used to build the CMS server.

[0383] The following gives the specific process of building the CMS server using the SylixOS operating system, including:

[0384] Step 01, pre-build the SCL model, and the SCL model includes Communication data, IED data, and DataTypeTemplates data;

[0385] Step 02, use a configuration file to configure the CMS server parameters in the SylixOS operating system, where the configuration file includes the file name of the SCL model;

[0386] Step 03, parse the SCL model based on the file name of the SCL model in the configuration file, and store the parsed content in a tree form;

[0387] Step 04, apply for the space required for the CMS server to run, and execute the initialization program and the registration program;

[0388] Step 05, start the CMS service by calling the service interface, and perform service configuration on the CMS service.

[0389] For the convenience of understanding the present invention, the principle of the method for constructing a CMS protocol server based on the SylixOS operating system of the present invention is described below, in combination with the process of establishing a CMS service for a protection and control device using the SylixOS operating system in the embodiment, to further describe the method for constructing a CMS protocol server based on the SylixOS operating system provided by the present invention.

[0390] The CMS protocol is a substation core communication protocol with complete independent intellectual property rights, and is widely used in substation secondary systems and equipment. Taking a configurable hierarchical distributed heterogeneous protection and control device as an example, it uses the substation configuration description language (SCL) as the model language to interact with the data of the protection and control device and the gateway. Specifically, the method for constructing a CMS protocol server based on the SylixOS operating system includes:

[0391] Specifically, the SCL model structure includes:

[0392] <header>Part: This part is used to identify the SCL file and its historical version information. Each entry in the historical version information can record its version, revision version, reason for modification, modifier, modification time, and modification content, etc.

[0393] <header>

[0394] <history>

[0395] <hitem revision="1.5" version="6.7" when="2021-07-12 10:26:49" who="" why="" what="" / >

[0396] < / history>

[0397] < / header>

[0398] <substation>Part: Used to describe the functional structure of the substation. This part is not considered for protection and measurement and control devices. Substation is not an essential content, and this part does not need to be constructed.

[0399] <communication>Part, such as Figure 7 As shown: mainly includes the relevant configurations for communications such as IP, MAC, IP-SUBNET, etc. In the protection and measurement control device, it is necessary to configure the station control layer and process layer subnets. Among them, the station control layer contains the basic information of Network A and Network B, and the process layer subnet needs to configure information such as the MAC address, VLAN-ID, VLAN-PRIORITY, APPID, and physical connection of GOOSE. Among them, the physical connection corresponds to the board card number and port number in the protection and measurement control device.

[0400] <ied>Part, such as Figure 8 shown: Describes the basic information of the protection and measurement control device. Nodes such as logical devices and logical nodes are all included in the node AccessPoint. Logical devices include several categories such as common LD, protection, waveform recording, measurement, and control. The logical device LD includes logical nodes such as LLN0, general LPHD, alarm signal, GOOSE alarm, device temperature, power supply voltage monitoring, protection function blocking, and time synchronization status self-checking.

[0401] <Data Type Templates> part, such as Figure 9 shown: This part mainly describes the logical node types, including data types and data compositions. It can be referenced by logical nodes in the IED, provides a basis for the organization of data in the data set, and thus affects the content of data transmission in the report. Logical node types include several categories such as general LLN0, protection LLN0, measurement LLN0, protection and measurement control LPHD, overcurrent protection, line overcurrent protection, protection trip, etc.

[0402] The CMS protocol server of the protection and measurement control device based on the SylixOS operating system is guided by the communication message (CMS) protocol specification of the substation secondary system in DL / T860. It directly maps the communication service model to TCP / IP, processes the operation requests from the client in a listening manner, processes the corresponding data according to the standard, and decides whether to send a message according to the specific operation. In the SylixOS operating system, the server call is made in the form of a library.

[0403] The specific process of step 02 includes:

[0404] Step 021: Pre-configure the server in the SylixOS operating system using a configuration file. The configuration file includes information such as ASDU, APDU, whitelist, SCL file name, whether to enable Log and the maximum number of Log entries, file root path, etc.

[0405] Step 022: Parse the data model according to the SCL file name in the configuration file, and store it in a tree form after parsing. The parsing content includes:

[0406] Communication parameters such as the local IP address. This communication parameter needs to be consistent with the local content to start normally, otherwise the startup will fail.

[0407] Basic content of the data set, including information on remote signal and remote measurement points.

[0408] Report control block information, including the data set referenced by the report control block, trigger mode, whether it is urcb or brcb, etc.

[0409] Control class information, including basic information of control points, control modes, etc.

[0410] Basic information of other control blocks except the report control block.

[0411] Replacement class information, including replacement points, replacement values, etc.

[0412] In the protection and measurement and control device based on the SylixOS operating system, only 1 CPU core is allocated to the server. Therefore, when starting parsing, the number of threads started will be reduced to avoid resource exhaustion caused by too many threads, and the CPU utilization rate will be as high as 100%.

[0413] Step 023: During the initialization of SylixOS, space required for the server to run needs to be applied for, and variables used need to be initialized; at the same time, user information callbacks, data information callbacks, and association callbacks need to be registered. The user information callback is used for registering user initialization callbacks, service startup and exit; the data information callback is used for callbacks during data reading and writing; the association callback is used for callbacks when the client connection is disconnected. At the same time, during initialization, it is possible to obtain whether the startup is successful through corresponding interfaces. The device uses the information returned by this interface for the next operation. If it fails, the fault light needs to be lit for prompt. The device can obtain the IEDname through the corresponding interface. When there is a recorded wave file, the recorded wave file needs to be named. The device can obtain / write the information corresponding to a ref (a reference mechanism) by using read / write interfaces, data definition reading interfaces, etc.

[0414] Step 024: Call the service interface to start the CMS service. A listening service is enabled in the service to listen for connected clients; when a client connects, the basic information of the client needs to be judged in detail to determine whether it is in the whitelist and whether the maximum number of connections has been exceeded. The server starts all basic services, including association services, read / write services, definition acquisition services, report services, log services, control services, setting services, replacement services, file services, and remote procedure call services. For example, in the report service, the server stores the trigger method for the client and sends a report to the client when the established conditions are met. If the data trigger method is data change and the client enables the corresponding report control block, a report will be sent to the client when the data changes. In the control service, the server saves and judges the control mode and control sequence of the client. When the operation sequence and control mode correspond, corresponding commands will be sent to the protection and measurement and control device. During the service process, log information is printed and output to a file. In the protection and measurement and control device based on the SylixOS operating system, it is not appropriate to output too much information to the terminal.

[0415] Step 025: Exit the service and release the space applied for by the server in sequence.

[0416] Such as Figure 10 As shown in the figure, this embodiment provides a working method for an autonomous and controllable substation multi-modal perception operation and maintenance system, including:

[0417] S701: Synchronously collect the real-time operation data of each electrical equipment in the substation main machine and the substation standby machine, or / and perform encryption and decryption algorithm transmission on the recording files of the substation main machine and the substation standby machine together;

[0418] S702: Obtain the decrypted substation operation synchronization data and / or recording files;

[0419] S703: Use the large model in the power grid field to identify faults in the substation operation synchronization data, and retrieve the matching solutions from the solution database according to the fault type; and query whether there is a hash table associating multiple recording files in different formats with the fault object pre-created in the recording file, so as to determine the timeliness of the recording file;

[0420] S704: Display the fault identification results, their corresponding solutions, and the timeliness analysis results of the recording files.

[0421] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.< / ied> < / communication> < / substation> < / header>

Claims

1. An autonomous and controllable substation multi-modal perception operation and maintenance system, characterized in that, Including: A collection module, a communication module, a data analysis module, and a display module; The collection module is used to synchronously collect the real-time operation data of each electrical device in the substation main machine and the substation standby machine, and / or transmit the waveform recording files of the substation main machine and the substation standby machine to the communication module together; The communication module is used to transmit the received synchronous operation data of each substation and / or waveform recording files to the data analysis module after being processed by encryption and decryption algorithms; The data analysis module is used for: using a large model in the power grid field to identify faults in the synchronous operation data of each substation, and retrieving matching solutions from the solution database according to the fault type; and querying whether there is a hash table associated with multiple waveform recording files of different formats pre-created for the fault object in the waveform recording file, so as to determine the timeliness of the waveform recording file; and comparing the SCD files of each substation with the reference SCD file, taking the first SCD file as the reference file, traversing the specified nodes of the first SCD file, and according to the unique attribute identifier of the node and the checksum calculated based on the node content, searching in the second SCD file for the corresponding nodes and the different node data, and storing the different node data as change items in the data structure; When there is a unique attribute identifier for the comparison node, search for the node corresponding to the comparison node in the second SCD file according to the value of the unique identifier attribute of the comparison node to be compared. If it exists, compare the attribute values of the node corresponding to the comparison node to be compared. If the attribute values are different, store them in the change items of the data structure; When there is no unique attribute identifier for the comparison node, first delete all the attributes of the comparison node, calculate the checksum C1 of the comparison node, then traverse the corresponding nodes in the second SCD file, calculate the checksum C2, compare C1 and C2, and judge whether C1 and C2 are equal. If they are equal, the two are the same node, and further compare the attribute values of the nodes. If the attribute values of the nodes are different, store them in the change items of the data structure; The display module is used to display the fault identification results and their corresponding solutions, as well as the timeliness analysis results of the waveform recording files.

2. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, wherein, In the data analysis module, when there is no solution corresponding to the identified fault type in the solution database, search for the solution corresponding to the fault type with the highest similarity to the fault type description in the solution database and transmit it to the display module for display.

3. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that In the data analysis module, when there is no solution corresponding to the identified fault type in the solution database, store the finally determined solution and its matching fault type in the solution database and update the solution database.

4. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, The collection module includes: A real-time database construction module, which is used to establish the real-time databases of the substation main machine and the substation standby machine respectively; A multi-machine interaction module, which is used to establish a communication connection between the substation main machine and the substation standby machine using a secure bus with encryption authentication; A communication link determination module, which determines the communication links between the substation main unit and the substation standby unit based on the communication connection and the real-time database, enabling the substation main unit and the substation standby unit to send their respective communication device lists to each other via the security bus to achieve real-time data synchronization.

5. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 4, characterized in that In the multi-machine interaction module, the substation main unit regularly sends its own communication device list to the corresponding substation standby unit. The standby unit combines the acquisition device list of the substation main unit to obtain the standby unit's global device list. The substation standby unit marks and monitors the relevant devices in the obtained main unit device list, and the substation standby unit does not communicate with the marked relevant devices. At the same time, the substation standby unit regularly sends its own acquisition device list to the substation main unit. The main unit combines the communication device list of the substation standby unit to obtain the main unit's global device list. The substation main unit marks and monitors the relevant devices in the obtained standby unit device list, and the substation main unit does not communicate with the marked relevant devices.

6. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 5, characterized in that, In the multi-machine interaction module, when a downtime fault occurs in the substation main unit, the substation main unit goes offline from the security bus. Then the substation standby unit cannot receive the communication device list sent by the substation main unit, and the substation standby unit will take over the data communication in the communication device list to achieve uninterrupted data acquisition. At the same time, a fault alarm prompt for the substation main unit is given. At this time, the substation standby unit enters the single-machine operation state.

7. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 5, characterized in that, In the multi-machine interaction module, when the substation main unit recovers from the fault, it immediately sends a data synchronization request to the substation standby unit. The substation standby unit sends the cached data to the substation main unit. After receiving the cached data, the substation main unit resumes sending the communication device list. After receiving the communication device list, the substation standby unit performs a difference set operation with the standby unit's global device list to restore the global communication list to the substation standby unit's own acquisition device list, and the substation main unit and the substation standby unit resume the parallel state of data acquisition downward.

8. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, In the communication module, before data transmission, the debugging tool and the network gateway perform key negotiation to generate their respective public keys and private keys. The debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key using the other party's public key and their own private key. During data communication, the debugging tool and the network gateway encrypt and decrypt the power data to be transmitted based on the generated session key and the constructed encryption and decryption algorithm. Among them, the construction process of the encryption and decryption algorithm includes: Based on the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array. Encrypt the data block based on the round keys, and generate a hash message authentication code based on the given HMAC key. Combine the encrypted data and the hash message authentication code and store them in the second array. Verify the hash message authentication code in the array, decrypt the data block, and store the generated status value in the third array.

9. The self - controllable substation multi - modal perception operation and maintenance system according to claim 8, characterized in that In the communication module, the debugging tool and the network gateway pass their respective public keys to each other through the public key exchange method, and at the same time calculate the shared session key using the other party's public key and their own private key, including: The key pair generated by the debugging tool includes a first public key and a first private key, and the key pair generated by the network gateway includes a second public key and a second private key; The debugging tool sends the first public key to the network gateway, and the network gateway sends the second public key to the debugging tool; After receiving the first public key of the debugging tool, the network gateway calculates the session key using the first public key of the debugging tool and the second private key of the network gateway; After receiving the first public key of the network gateway, the debugging tool calculates the session key using the first public key of the network gateway and the second private key of the debugging tool.

10. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 8, characterized in that, In the communication module, on the basis of the standard SM4 algorithm, additional rounds are added to expand the key to generate round keys, which are stored in the first array, including: Define a first array with a preset size for storing the expanded key; Customize the key data information, read the key data information byte by byte, and store them into the first array respectively; Through a preset number of iterations, generate a new key value each time using the previous key value and store it in the first array. In each iteration, apply linear transformation and non-linear transformation to increase the complexity of the key, and store the generated round keys in the first array.

11. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 8, characterized in that In the communication module, encrypt the data block based on the round key, and generate a hash message authentication code based on the given HMAC key, including: Define a second array with a preset size for storing intermediate data during the encryption process; Read the input plaintext data byte by byte and store them into the array respectively; Through a preset number of iterations, perform an encryption transformation each time using the current state value and the round key to generate a new state value; Convert the final state value back to a byte array and store it in a temporary variable; Perform a hash operation on the encrypted data using the given HMAC key to generate a hash message authentication code.

12. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 8, characterized in that In the communication module, verify the hash message authentication code in the array, decrypt the data block, and store the generated state value in the third array, including: Define a third array with a preset size for storing intermediate data during the decryption process; Extract the data and the hash message authentication code; ​ ​ 13. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, ​ ​ Based on the parallel decomposition strategy of the large model in the power grid field, the hyperparameters are used to input the hardware device information and the deep learning framework, and the hardware resources are automatically allocated. Then, the guidance statements are used to set the deep learning model structure and configure the parallel decomposition scheme, complete the automatic translation of the code, and achieve the automatic conversion and training of the deep learning model.

14. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 13, characterized in that, In the data analysis module, the memory overhead of a single parameter , the memory overhead of a single gradient , the memory overhead of a single optimizer state , the number of model parameters , and the number of worker nodes in data parallelism The relationship between these parameters and the memory overhead of the parallel decomposition strategy for storage collaboration is as follows: .

15. The independently controllable substation multi-modal perception operation and maintenance system according to claim 14, characterized in that, In the data analysis module, the number of model parameters is ; where is the precision of floating-point numbers for calculation.

16. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 13, characterized in that, In the data analysis module, the parallel decomposition strategy of the large model in the power grid field includes inter-group data parallelism, intra-group model parallelism, pipeline parallelism, and optimizer parallelism strategies.

17. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 16, characterized in that, In the data analysis module, the in-group model parallel strategy is as follows: ; where is the memory overhead of the parallel decomposition strategy for storage cooperation, is the memory limit of a single device, is the in-group device allocation volume; The inter-group data parallel strategy is as follows: ; where is the allocation quantity of devices within a group, is the allocation quantity of devices between groups, is the total number of available devices.

18. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, In the data analysis module, the training process of the large model in the power grid field is as follows: Obtain the configuration parameter information of the large model to be trained; Based on the obtained configuration parameter information, construct the loss function of the large model, calculate the gradient norm, and obtain the first-order gradient information perception score of the large model; Add perturbations to the large model, construct the perturbed loss function of the large model, calculate the Hessian matrix, and obtain the Hessian gradient information perception score of the large model; Perform adaptive weighted combination on the obtained first-order gradient information perception score and Hessian gradient information perception score of the large model to obtain the quantization perception score of the large model; According to the obtained quantization perception score of the large model, determine the quantization perception training samples of the large model, and complete the distributed quantization perception training of the large model.

19. The independently controllable substation multi-modal perception operation and maintenance system according to claim 18, characterized in that, In the data analysis module, during the process of determining the quantization perception training samples of the large model, sort the weights in the obtained quantization perception score of the large model, determine the training samples according to the sorting, perform data parallel processing on the determined samples, and complete the distributed quantization perception training of the large model.

20. The independently controllable substation multi-modal perception operation and maintenance system according to claim 18, characterized in that, In the data analysis module, during the process of adaptive weighted combination, after performing normalization processing on the obtained first-order gradient information perception score and Hessian gradient information perception score, calculate the weights of the first-order gradient information perception score and the Hessian gradient information perception score, perform adaptive weighted combination on the sample scores to obtain the quantization perception score of the large model, and determine the sample data suitable for the quantization perception training of the large model according to the quantization perception score of the large model.

21. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 18, characterized in that, In the data analysis module, during the process of distributed quantization perception training of the large model, divide the determined quantization perception training samples of the large model into several subsets, process the divided subsets and calculate the subset gradients, aggregate all subset gradients in combination with the communication protocol, and complete the update of the global model parameters; complete the distributed quantization perception training of the large model according to the updated model parameters.

22. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 18, characterized in that, In the data analysis module, perform backpropagation processing on the obtained loss function of the large model, calculate the partial derivative of the loss function after backpropagation processing with respect to the model parameters to obtain the model gradient; take the sum of the squares of all the obtained model gradients as the gradient norm, and take the square root of the gradient norm to obtain the first-order gradient information perception score of the large model.

23. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 18, characterized in that, In the data analysis module, calculate the second-order partial derivative of the constructed perturbed loss function of the large model with respect to the model parameters, take the obtained second-order partial derivative as the diagonal element of the Hessian matrix, calculate the sum of the squares of all the diagonal elements, and obtain the Hessian gradient information perception score of the large model.

24. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, In the data analysis module, the process of determining the timeliness of the oscillogram file is as follows: Judge whether the currently received oscillogram file exists in the preset hash table. If it exists, determine the fault object associated with the currently received oscillogram file in the hash table; otherwise, end the query. Among them, the hash table is a hash table in which multiple oscillogram files in different formats are associated with fault objects, and multiple oscillogram files in different formats are associated with the same fault object; Delete the key-value pair of the oscillogram file and the fault object in the hash table; If the associated fault object is empty or the fault analysis is in progress, end the query; otherwise, judge whether all multiple preset fault oscillogram files in different formats exist in the currently received oscillogram file. If all preset types of fault oscillogram files exist, start the fault processing thread; otherwise, continue to obtain the oscillogram file.

25. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that In the data analysis module, during the process of comparing the SCD files of each substation with the reference SCD file, Read the content of each sub-node in the two SCD files to be compared; Compare the content of each sub-node in the two SCD files to be compared, and store the comparison result in the data structure; Take the second SCD file as the reference file, traverse the specified node of the second SCD file, search for the corresponding node in the first SCD file, and store the node data not found as a new item in the data structure.

26. The autonomous and controllable substation multi-modal perception operation and maintenance system according to claim 1, characterized in that, The data analysis module is also configured with an intelligent alarm module. The process of automatic modeling of the intelligent alarm module is as follows: Build an initial basic information library. Among them, the initial basic information library includes a characteristic signal library, a characteristic equipment library, and a fault alarm information expert library; Expand the initial basic information library according to the set rules. The process is as follows: Classify the signals in the substation based on the preset rules, abstract the required signal types and store them in the characteristic signal library; Classify the primary and secondary equipment based on the preset rules, abstract the required primary and secondary equipment types and store them in the characteristic equipment library; Form fault alarm information based on the preset conditions for typical faults to occur and the sequence of signal occurrence, and store the corresponding solutions for various fault alarm information in the fault alarm information expert library together; According to the expanded characteristic signal library and characteristic equipment library, classify the signals / equipment included in the selected actual interval / device, and then generate instances of fault distribution and alarm inference and form an intelligent alarm configuration information file according to the expanded fault alarm information expert library; Publish the intelligent alarm configuration information file to the intelligent alarm module to obtain the final basic information library.

27. The multi-modal perception operation and maintenance system for a self-controlled substation according to claim 1, characterized in that The data analysis module is also configured with a multi-thread management module, and the multi-thread management module is configured as: Use multiple threads to separately execute the hardware monitoring, alarm information monitoring, and control operations of the substation; Adopt an independent acquisition method to obtain the hardware monitoring information and alarm information monitoring information of the substation monitoring system platform in real time according to the set sampling frequency; Obtain the issued control operation items, perform corresponding control operations and detections on the substation monitoring system to obtain the monitoring results. The control operation items are adaptively selected and determined according to the monitoring information obtained by independent acquisition.

28. A working method of an autonomous and controllable substation multi-modal perception operation and maintenance system according to any one of claims 1-27, characterized in that, Include: Synchronously collect the real-time operation data of each electrical device in the substation main unit and the substation standby unit, and / or perform encryption and decryption algorithm transmission on the oscillogram files of the substation main unit and the substation standby unit together; Obtain the decrypted synchronous operation data of each substation and / or oscillogram files; Use a large model in the power grid field to identify faults in the synchronous operation data of each substation, and retrieve the matching solutions from the solution database according to the fault types; and query whether there is a hash table associating multiple oscillogram files in different formats pre-created with the fault objects in the oscillogram files, so as to determine the timeliness of the oscillogram files; Display the fault identification results, their corresponding solutions, and the timeliness analysis results of the oscillogram files.

Images