Relay protection device debugging method and system based on model prediction

The method integrates relay protection device debugging using a pre-trained prediction model to automate parameter calculation, improving efficiency by directly obtaining test parameters through a human-computer interface.

CN119068386APending Publication Date: 2024-12-03EAST CHINA POWER TRANSMISSION & TRANSFORMATION ENG

Patent Information

Application Number
CN202411044471.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Existing technologies for debugging relay protection devices are inefficient, as they require manual calculation of debugging parameters and extensive on-site verification, which is time-consuming and laborious.

Method used

A method and system using a pre-trained prediction model to integrate debugging categories of relay protection devices, utilizing a data set processed with video images and instruction manuals, and employing a Yolov5 structure with GIOU_Loss for model training, enabling direct input of fixed values to obtain test parameters.

Benefits of technology

The method significantly enhances the efficiency of relay protection device debugging by automating parameter calculation and reducing on-site verification time through human-computer interaction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068386A_ABST
    Figure CN119068386A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of relay protection equipment debugging, and relates to a relay protection device debugging method and system based on model prediction, and the method comprises the following specific steps: S1, making a data set, and carrying out the preprocessing of collected data; s2, labeling the data in the data set; s3, splitting the labeled data set into a training set and a test set; s4, using the data of the training set and the test set to train a model based on a yolov5 structure to obtain a prediction model; s5, calculating the difference between the prediction frame and the real frame by using GIOULoss as a loss function; s6, verifying the trained model on a test set; s7, inputting the picture of the relay protection device into the prediction model, and outputting the model; and S8, calling data in the data set according to an output result to obtain debugging parameters. According to the method, debugging categories of relay protection devices of different models are integrated through a pre-trained prediction model, teaching materials and specifications are referred, and a fixed value to be verified is directly input to obtain test parameters.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of relay protection equipment debugging, and in particular to a relay protection device debugging method and system based on model prediction. Background Art

[0002] Relay protection devices are devices that protect power systems when faults (overcurrent, short circuit, etc.) occur during the operation of the power system. However, these devices must be debugged before they are put into use to ensure that they can perform protective actions normally when faults occur. Currently, operators test relay protection devices according to the textbooks or instructions of different types of devices. There are many types of tests, and different fixed value checks require the calculation of debugging parameters by themselves through corresponding formulas. In the actual installation and use process, a lot of time and energy is required for on-site verification, which is time-consuming and laborious. Summary of the invention

[0003] The purpose of the present invention is to address the problems existing in the background technology and to propose a relay protection device debugging method and system based on model prediction. The debugging categories of relay protection devices of different types are integrated through a pre-trained prediction model, and textbooks and instructions are referred to to increase interactivity. The constants to be verified are directly input to directly obtain the test parameters.

[0004] The technical solution of the present invention is a relay protection device debugging method based on model prediction, comprising the following specific steps: S1, create a data set and pre-process the collected data; S11, classifying the relay protection devices according to their types, and saving the video image texts and teaching materials or instruction manual texts of the same category in the same directory; S12, setting a fixed frame rate to cut the video image text; S2. Label the data in the data set; the labeled video text image takes the model of the relay protection device as the true value; The cut video text images are annotated by manual annotation; or, Use SAM for automatic annotation; manually verify the automatic annotation results; S3, splitting the labeled data set into a training set and a test set; preferably, splitting the data set into a training set and a test set in a ratio of 4:1; S4, using the data of the training set and the test set to train the model based on the yolov5 structure to obtain a prediction model; S5. Use GIOU_Loss as the loss function to calculate the difference between the predicted box and the true box; S6. Verify the trained model on the test set; S7, inputting the picture of the relay protection device into the prediction model, and the prediction model outputs the product model; S8. Call the data in the data set to obtain debugging parameters according to the output result.

[0005] In an optional embodiment, the data collected in S1 includes video images of different types of relay protection devices and teaching materials or instructions for the relay protection devices of the same type; and video image texts of different types and multiple viewing angles are obtained.

[0006] In an optional embodiment, after completing 500 trainings in S5, the model with the smallest loss in the test set is selected as the optimal model.

[0007] In an optional embodiment, the optimal model is tested with a test set, and the results show that the recognition accuracy reaches 0.92240.

[0008] A relay protection device debugging system based on model prediction, which is debugged using the above method, includes a data acquisition module, a data processing module, a database, a prediction module, an input module and an output module; The data acquisition module uses the camera to collect image information of the relay protection device; The data processing module transcodes the collected image information and sends it to the prediction module; The prediction module recognizes the image information and outputs the prediction result; The user inputs the set value to be verified through the input module, and the output module retrieves the teaching material or instruction manual data of the relay protection device of this model from the database according to the prediction results to output the test parameters.

[0009] A computer program product comprises a computer program / instruction, which implements the steps of the above method when executed by a processor.

[0010] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention integrates the debugging categories of relay protection devices of different types through a pre-trained prediction model, refers to the contents of teaching materials and instructions, and performs human-computer interaction through a set input module. The fixed value to be verified can be directly input through the system to directly obtain the test parameters, thereby greatly improving the efficiency of on-site equipment installation and debugging. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention; Figure 2 Schematic diagram of the system structure of an embodiment of the present invention. DETAILED DESCRIPTION

[0012] Example 1 like Figure 1 As shown, a relay protection device debugging method based on model prediction proposed by the present invention includes the following specific steps: S1. Create a data set and preprocess the collected data; the collected data includes video images of different types of relay protection devices and textbooks or instructions for the relay protection devices of this type; obtain different types of video image texts from multiple perspectives; among them, the textbook uses "Relay Protection Debugging and Circuit Fault Troubleshooting Examples" edited by Hui Jie, China Electric Power Press, 2020.1; the textbook introduces the principles, protection logic verification, faults and troubleshooting methods of different types of relay protection devices through three parts: line protection, transformer protection and busbar protection; the instruction manual retrieves the instruction manual text published on the Internet and saves it together with the textbook text; S11. Classify the relay protection devices according to their types, and save the video image texts and textbook or manual texts of the same category in the same directory; the textbook text only intercepts the content in the directory associated with the model, and merges it with the manual into a text text for saving. For example, when the model of the relay protection device is PCS-931, the page number is 3-28 through the textbook query, so only the content of pages 3-28 is merged and saved with the manual of the model; S12. Set a fixed frame rate to cut the video image text. Python+Opencv can be commonly used to cut the video into frame images for saving; S2. Label the data in the data set; the labeled video text image takes the model of the relay protection device as the true value; The cut video text images are annotated by manual annotation; or, Use SAM for automatic labeling; manually verify the automatic labeling results; the efficiency of data labeling can be greatly improved through automatic labeling and manual verification.

[0013] S3, splitting the labeled data set into a training set and a test set; preferably, splitting the data set into a training set and a test set in a ratio of 4:1; S4, using the data of the training set and the test set to train the model based on the yolov5 structure to obtain a prediction model; Commonly, yolov5 has 4 models. For the s, m, l, and x models of YOLOv5, the configuration files are the same. The main difference is that they have different model sizes and computational complexities. Among them, yolov5-s has the smallest model complexity and fast detection speed. Its layout follows the overall layout of the YOLO (You only look once) series. The entire network structure is also divided into four parts: input, Backbone, Neck, and output. In this embodiment, the yolov5-s model is selected for training, which is conducive to improving the detection speed.

[0014] S5. Use GIOU_Loss as the loss function to calculate the difference between the predicted box and the true box. After completing 500 trainings in S5, the model with the smallest loss in the test set is selected as the optimal model.

[0015] S6. Verify the trained model on the test set. Use the test set to test the optimal model. The results show that the recognition accuracy reaches 0.92240.

[0016] S7, inputting the picture of the relay protection device into the prediction model, and the prediction model outputs the product model; S8. Call the data in the data set to obtain debugging parameters according to the output result.

[0017] In this embodiment, the model of the user's relay protection device is quickly identified by establishing a prediction model, and the data in the data set is retrieved according to the model. The constant value to be verified can be directly input to directly obtain the test parameters, which greatly improves efficiency.

[0018] Example 2 A relay protection device debugging system based on model prediction, which is debugged using the method of embodiment 1, and includes a data acquisition module, a data processing module, a database, a prediction module, an input module and an output module; The data acquisition module uses the camera to collect image information of the relay protection device based on machine vision recognition; The data processing module transcodes the collected image information and sends it to the prediction module; The prediction module recognizes the image information, compares it with the frame image with the true value, and quickly outputs the prediction result; The user inputs the set value to be verified through the input module, and the output module retrieves the teaching material or instruction manual data of the relay protection device of this model from the database according to the prediction results to output the test parameters.

[0019] In this embodiment, the user can directly take a photo of the relay protection device, and the system will directly output the product model after processing; the user can input the constant value to be verified according to the required test category and directly obtain the test parameters, which greatly improves efficiency.

[0020] Example 3 A computer program product comprises a computer program / instruction, which implements the steps of the above method when executed by a processor.

[0021] In this embodiment, the computer program / instructions can be downloaded by a computer or mobile phone; when the user runs the program on the computer or mobile phone, the camera of the electronic device is directly called to collect image data, analyze and process it, and output the results.

[0022] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A relay protection device debugging method based on model prediction, It is characterized in that The specific steps include: S1, create a data set and pre-process the collected data; S2, label the data in the dataset; S3, split the labeled data set into a training set and a test set; S4, using the data of the training set and the test set to train the model based on the yolov5 structure to obtain a prediction model; S5. Use GIOU_Loss as the loss function to calculate the difference between the predicted box and the true box; S6. Verify the trained model on the test set; S7, inputting the picture of the relay protection device into the prediction model, and the prediction model outputs the product model; S8. Call the data in the data set to obtain debugging parameters according to the output result.

2. The relay protection device debugging method based on model prediction according to claim 1, It is characterized in that The data collected in S1 include video images of different types of relay protection devices and teaching materials or instructions for the relay protection devices of the same type; and video image texts of different types and multiple viewing angles are obtained.

3. The relay protection device debugging method based on model prediction according to claim 2, It is characterized in that The steps of data preprocessing in S1 include: S11, classifying the relay protection devices according to their types, and saving the video image texts and teaching materials or instruction manual texts of the same category in the same directory; S12, setting a fixed frame rate to cut the video image text.

4. The relay protection device debugging method based on model prediction according to claim 1, It is characterized in that The step of labeling data in S2 includes: labeling the cut video text image by manual labeling; or, Use SAM for automatic annotation; verify the automatic annotation results manually.

5. The relay protection device debugging method based on model prediction according to claim 4, It is characterized in that The annotated video text image takes the model of the relay protection device as the true value.

6. The relay protection device debugging method based on model prediction according to claim 1, It is characterized in that In S3, the data set is split into training set and test set in a ratio of 4:

1.

7. The relay protection device debugging method based on model prediction according to claim 1, It is characterized in that In S5, after completing 500 training cycles, the model with the smallest loss in the test set is selected as the optimal model.

8. According to the relay protection device debugging method based on model prediction according to claim 7, It is characterized in that The optimal model was tested using the test set, and the results showed that the recognition accuracy reached 0.92240.

9. A relay protection device debugging system based on model prediction, using the method according to any one of claims 1 to 8 for debugging, It is characterized in that It includes data acquisition module, data processing module, database, prediction module, input module and output module; The data acquisition module uses the camera to collect image information of the relay protection device; The data processing module transcodes the collected image information and sends it to the prediction module; The prediction module recognizes the image information and outputs the prediction result; The user inputs the set value to be verified through the input module, and the output module retrieves the teaching material or instruction manual data of the model of relay protection device in the database according to the prediction results to output the test parameters.

10. A computer program product comprising a computer program / instructions, It is characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Citation Information

Patent Citations

  • Microcomputer relay protection automatic test system

    CN106199258A

  • Detection system and method for relay protection device

    CN107976214A

  • Power equipment information management system and method based on two-dimensional code identification technology

    CN108090530A

  • Wheat ear detection method based on improved YOLOv5

    CN114973002A

  • Communication system work order performance method and system

    US20040254757A1

Cited By

  • Power transmission line relay protection setting value prediction method, system and platform based on large language model

    CN120453994A

  • A method, system, and platform for predicting transmission line relay protection settings based on large language models.

    CN120453994B