A transformer fault intelligent auxiliary analysis method, recording medium and system
Through the combination of message middleware and deep learning model, a fault library of multi-level classification labels is established, which solves the problems of redundant data and low analysis efficiency in transformer fault detection, and achieves efficient and accurate fault identification and system reliability.
Patent Information
- Application Number
- CN202210329133.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-30
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2042-03-30
AI Technical Summary
The prior art has problems such as redundant data, low analysis efficiency, easy to misjudgment or misjudgment in transformer fault detection, and the system is easily stagnated when the massive data is transmitted simultaneously.
Message middleware is used to build a network architecture, establish a fault library of multi-level classification tags, and fault identification and optimization are carried out through deep learning models, fault identification module is installed for classification marking and retrieval, and massive data is processed using asynchronous transmission.
It improves the efficiency and accuracy of fault analysis, reduces the misjudgment rate, overcomes the bottleneck of the synchronous transmission of massive data, and improves the reliability of the system.
Smart Images

Figure CN114692755B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of transformer fault detection, and discloses an intelligent auxiliary analysis method for transformer faults, as well as a recording medium and a system capable of storing and executing the method program. Background Art
[0002] In order to ensure the safe and reliable operation of the power system, the power company has strengthened the monitoring methods and means for the operation of power transformers. Based on the promotion of various technical means such as oil chromatography, ultra-high frequency, core grounding current, and ultrasound, a large amount of monitoring data has been generated. By analyzing these different types of data with different detection principles, the on-site operation status of the transformer is judged. At present, various types of data are redundant and numerous, resulting in a large workload for joint analysis, low efficiency, and easy misjudgment or omission. In addition, the synchronous transmission of massive analysis data will cause the entire system to stagnate when a subsystem failure in the transmission link is suspended. Summary of the invention
[0003] In view of the above problems, the present invention provides a transformer fault intelligent auxiliary analysis method, and the specific scheme includes the following steps:
[0004] S1. Build a network architecture that uses message middleware to complete information transmission and storage between subsystems;
[0005] S2. Establish a fault library that identifies transformer faults through multi-level classification labels, batch upload transformer fault information and its corresponding fault status collected in the same way, establish a fault deep learning model by category, and continuously optimize the fault library of the corresponding category;
[0006] S3. Install a multi-level fault identification module in the network architecture of S1;
[0007] S4. The transformer faults to be monitored obtained on site are classified according to the multi-level classification labels and uploaded to the fault identification module. The fault identification module retrieves similar faults in the fault library and outputs the corresponding fault status.
[0008] Searching in a fault library with multi-level classification identification saves search time and reduces the use of test resources; using a network architecture to process graphic information reduces manual work, improves efficiency and reduces the misjudgment rate; using message middleware in the transmission process can overcome the fault bottleneck encountered in the synchronous transmission of massive data through asynchronous transmission, thereby improving the reliability of the entire system.
[0009] Preferably, the method further includes uploading the transformer faults to be monitored and their actual fault states obtained on-site to the deep learning model to optimize the fault library.
[0010] This method allows the original information data on site to be reused and optimizes the deep learning model in real time, which will be more conducive to improving the accuracy of fault diagnosis.
[0011] Another solution of the present invention is to provide a non-transitory readable recording medium for storing one or more programs including a plurality of instructions. When the instructions are executed, the processing circuit will execute the above-mentioned transformer fault intelligent auxiliary analysis method.
[0012] Another embodiment of the present invention is to provide a transformer fault intelligent auxiliary analysis system, including a processing circuit and a memory electrically coupled thereto, wherein the memory is configured to store at least one program, wherein the program includes a plurality of instructions, and the processing circuit runs the program to execute the above-mentioned transformer fault intelligent analysis method. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 A schematic diagram of a network architecture connection in a transformer fault intelligent auxiliary analysis method according to an embodiment of the present invention; DETAILED DESCRIPTION
[0014] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making innovative efforts are within the scope of protection of the present invention.
[0015] See also Figure 1 The network architecture used in the present invention to implement fault information processing is named as: deep learning fault identification system, including:
[0016] (1) Fault data storage service
[0017] The transformer standard fault database has been established and continuously deepened, and the data format is consistent with the existing information system (PMS) of the State Grid. Multi-level classification labels are set. Each level of label represents a label classification, such as the region where the equipment is located, and the corresponding labels at this level are such as "Jiangsu", "Henan", etc. The secondary label represents a specific classification of the upper label, such as "regional subordinate substation", the "Jiangsu" secondary label corresponds to "Xuyi Station", "Taizhou Station", "Dongwu Station", and the "Henan" secondary label corresponds to "Nanyang Station", "Southern Henan Station". The secondary secondary labels of the above secondary labels further represent further classifications, for example, the "Dongwu Station" secondary label corresponds to "Phase A of Main Transformer No. 1". There is a subordinate relationship between the upper and secondary labels. When the user uses the label library to annotate the sample library, the upper label is displayed as a label classification, and the secondary label is displayed as an option under the label classification for the user to choose. This function not only ensures that the faults used in the platform can match the current standard fault atlas library, but also can be flexibly adjusted to adapt to the changes in the standard atlas library in the future for classification.
[0018] 1) Storage Files
[0019] Based on the above classification method, the faults and corresponding data files (such as XML files) are stored, and the deep learning model files are stored separately.
[0020] 2) File reading
[0021] Read the above faults and their data files or call the deep learning model files.
[0022] (2) Web Application
[0023] 1) Fault entry
[0024] According to user needs, various types of collected fault information can be uploaded in batches at regular intervals or in real time.
[0025] 2) Resource retrieval annotation
[0026] Search all fault sample files stored in the platform by conditions. The search conditions include: ① Tag type ② Sample serial number ③ Sample marking status (entered, marked, reviewed, rejected) ④ Entry user ⑤ Marking user ⑥ Review user ⑦ Entry time period ⑧ Sample name (keyword)
[0027] Feature annotation is selected for the sample to be trained, and the relevant annotation information is generated into an XML file as information data to bind the sample.
[0028] 3) Set up training set
[0029] The sample group selected for deep learning model training is generally a collection of labels of the same category and level. The training set corresponds to an exclusive task name and directly calls the set path without taking up additional storage space.
[0030] 4) Tag Management
[0031] Dedicated pages are used to set up and manage calibrated tags, such as setting tag names or determining tag levels.
[0032] 5) Model management
[0033] Used to call and manage deep learning model files, which have a separate storage path.
[0034] By setting a good training set, the deep learning model is called to continuously learn the features of the training set and continuously optimize the model.
[0035] 6) Scene Management
[0036] Centralized management of added scenes (a batch, a type of transformer or a substation), including functions such as viewing basic scene information, editing and modifying scene names or labels and descriptions, deleting scenes, searching scenes, and setting management permissions.
[0037] 7) Upload function
[0038] Upload a task (sample calibration, sample training or fault identification).
[0039] 8) Analysis and identification module
[0040] Based on the trained deep learning model, the test set is added to the task, and the fault identification results are output through model recognition.
[0041] 9) System log
[0042] Record key operations of system users.
[0043] 10) System data
[0044] Statistics platform data.
[0045] 11) User Management
[0046] Manage platform user accounts and permissions.
[0047] (3) Message middleware (including sending and receiving asynchronous messages)
[0048] 1) Send messages to the deep learning system (sample labeling learning messages, fault analysis messages).
[0049] 2) Receive messages sent by the deep learning system (models generated by the deep learning system, fault analysis results).
[0050] (4) Deep Learning
[0051] 1) Data analysis
[0052] Parse the message received from the message middleware to obtain label information and the download address of the fault sample and model.
[0053] 2) Download the file
[0054] Download fault samples, models, and corresponding configuration files from the file storage server.
[0055] 3) Prepare data
[0056] Convert fault samples and label information into a trainable and recognizable format.
[0057] 4) Network training
[0058] Using the prepared training set, call the deep learning model to train the deep neural network.
[0059] 5) Upload the model
[0060] Upload the trained model and its related configuration files to the file service storage.
[0061] 6) Target Identification
[0062] Invoke the deep learning model to identify faults in the samples.
[0063] 7) Generate Report
[0064] Generate a target recognition result report and upload it to the file storage server.
[0065] 8) Result feedback
[0066] Send the results of deep learning processing to the middleware.
[0067] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computers, available storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.
[0068] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0069] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0071] Compiling the above method steps into a program and then storing it on a hard disk or other non-transitory storage medium constitutes the technical solution of "a non-transitory readable recording medium" of the present invention; and electrically connecting the storage medium to a computer processor, which can complete real-time detection of various types of transformer faults in substations through data processing, constitutes the technical solution of "a transformer fault intelligent auxiliary analysis system" of the present invention.
[0072] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. 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.
Claims
1. A transformer fault intelligent auxiliary analysis method, characterized in that The following steps are involved: S1. Build a network architecture that uses message middleware to complete the transmission and storage of various fault information between subsystems; The network architecture includes a fault data storage service module, a WEB application module, a message middleware, and a deep learning system; The fault data storage service module stores data in a fault library that identifies transformer faults through multi-level classification labels, stores deep learning model files, and can also read faults and their data files or call deep learning model files; The WEB application module includes a fault entry function, a resource retrieval and annotation function, a training set setting function, a label management function, a model management function, a scenario management function, an upload function, an analysis and identification module, a system log, a system data, and a user management function. The user searches all fault sample files stored in the platform according to conditions through the resource retrieval and annotation function, selects feature annotations for samples to be trained, selects a sample group for deep learning model training through the training set setting function, and the sample group is a collection of labels of the same class and level. The model management function is used to call the deep learning model through the set training set, so that it continuously learns the features of the training set and continuously optimizes the model. The upload function is used to upload sample calibration tasks, sample training tasks or fault identification tasks. In the analysis and identification module, based on the trained deep learning model, the test set is added to the task, and the fault identification result is output through model identification. The message middleware sends sample labeling learning messages and fault analysis messages to the deep learning system, and receives the learning model and fault analysis results sent by the deep learning system; The deep learning system parses the message received from the message middleware, obtains label information, and the download address of the fault sample and model, downloads the fault sample, model and corresponding configuration file from the file storage server, converts the fault sample and label information into a trainable and recognizable format, uses the prepared training set to call the deep learning model to train the deep neural network, uploads the trained model and its related configuration file to the file service storage, and the deep learning system can also call the deep learning model to identify the fault in the sample, generate a target recognition result report, upload it to the file storage server, and send the result of the deep learning processing to the message middleware; S2. Classify the transformer faults to be monitored obtained on site according to the multi-level classification labels and upload them to the network architecture, and output the corresponding fault status.
2. A non-transitory readable recording medium for storing one or more programs including a plurality of instructions, characterized in that: When the instruction is executed, the processing circuit will be caused to execute the transformer fault intelligent auxiliary analysis method described in claim 1.
3. A transformer fault intelligent auxiliary analysis system, comprising a processing circuit and a memory electrically coupled thereto, characterized in that: The memory configuration stores at least one program, the program includes a plurality of instructions, and the processing circuit runs the program to execute the transformer fault intelligent auxiliary analysis method according to claim 1.
