Power transmission line conductor icing monitoring method and device based on AI image recognition

Abstract

The invention relates to a power transmission line conductor icing monitoring method based on AI image recognition. The method comprises the following steps: acquiring microclimate, tension and conductor image data; the VGG16 deep convolutional neural network is improved and trained, the ice coating type of the wire is judged, and a recognition result of the ice coating type of the wire is obtained; calculating the icing thickness of the wire; and uploading the microclimate, the tension, the wire image data, the wire icing type identification result and the wire icing thickness to a system monitoring platform through a remote communication unit. The invention further provides a power transmission line wire icing monitoring system based on AI image recognition. According to the invention, through multi-sensor data fusion, the accuracy of icing monitoring is improved; an AI artificial intelligence module is adopted to innovatively fuse various features such as textures, edges and gray statistics, and feature deep fusion is performed through an improved VGG16 deep convolutional neural network, so that the icing type recognition accuracy is remarkably improved, and particularly, relatively high recognition stability is still kept in complex scenes such as rain and fog, night and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission line monitoring, and in particular to a method and device for monitoring ice coating on transmission line conductors based on AI image recognition. Background Art

[0002] Icing on transmission line conductors can cause insulator flashover, line tripping, line breakage, tower collapse, conductor galloping, and communication interruptions. Furthermore, ice damage to transmission lines is often accompanied by severe weather conditions, with mountains covered in ice and snow, resulting in obstructed transportation, disrupted communications, and difficult repairs. This often leads to prolonged and widespread power outages, resulting in significant economic losses and severe social impacts. Therefore, it is necessary to strengthen the monitoring and assessment of overhead transmission lines.

[0003] The transmission line conductor icing monitoring device can monitor icing in real time to ensure the safe operation of the power grid in rainy, snowy and freezing weather. Conductors will form different types of icing under different climate and terrain conditions, such as rime, rime, mixed rime and wet snow. Different icing types have different impacts on conductors, but currently the density of rime is ρ1 = 0.9g / cm 3 To calculate the ice thickness of the conductor, the calculated ice thickness of the conductor is inaccurate.

[0004] While the simulated conductor method overcomes the shortcomings of traditional weighing-based ice monitoring systems, such as the need for power outages for installation, the simulated conductor method only monitors simulated conductors. This can lead to false ice alarms after de-icing measures have been completed on the actual conductors. Furthermore, the simulated conductor method only monitors simulated conductors, making it difficult to visually determine the actual conductor's ice condition. Summary of the Invention

[0005] In order to solve the defect of being unable to realize real-time and effective monitoring of the icing status of transmission line conductors, the primary purpose of the present invention is to provide a transmission line conductor icing monitoring method based on AI image recognition, which uses AI image recognition to collect images of the monitored transmission line and perform image processing on them, and automatically identify the icing status and icing type of the monitored transmission line.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for monitoring ice coating on transmission line conductors based on AI image recognition, the method comprising the following steps in sequence:

[0007] (1) Obtain micro-meteorological data, tension data and conductor image data;

[0008] (2) improving the VGG16 deep convolutional neural network, training the improved VGG16 deep convolutional neural network, inputting the data obtained in step (1) into the trained VGG16 deep convolutional neural network, judging the wire icing type, and obtaining a wire icing type recognition result;

[0009] (3) Calculate the conductor ice thickness based on the conductor ice type identification results, micrometeorological data, and tension data;

[0010] (4) Upload the micro-meteorological data, tension data, conductor image data, conductor icing type identification results and conductor icing thickness to the system monitoring platform through the remote communication unit; if the conductor icing thickness is greater than the set warning value, an icing alarm message is sent to the system monitoring platform.

[0011] In step (2), the improvement of the VGG16 deep convolutional neural network refers to: adding a self-attention mechanism module after the convolution layer of the third, fourth and fifth convolutional blocks of the VGG16 deep convolutional neural network, and the self-attention mechanism module strengthens the representation ability of the key area by adaptively calculating the spatial position weights of the feature map; adding a channel attention mechanism, i.e., SE module, at the end of each convolutional block of the VGG16 deep convolutional neural network, and the SE module adaptively recalibrates the feature channel by learning the mutual dependence between channels.

[0012] In step (2), the training of the improved VGG16 deep convolutional neural network specifically refers to: obtaining 2000 wire image data of different ice coverage degrees in different time periods and different climatic conditions to form a training data set, preprocessing the wire image data, extracting a multidimensional feature vector, the multidimensional feature vector includes the texture features, edge features, grayscale statistical features and morphological change features of the image, standardizing the extracted multidimensional feature vector, and performing dimensionality reduction through principal component analysis, and then using the improved VGG16 deep convolutional neural network as a classifier for training.

[0013] In step (3), the calculation formula for the conductor ice thickness is:

[0014]

[0015] Where: δ is the ice thickness of the conductor, d is the conductor diameter, l is the length of the simulated conductor, ΔT is the temperature difference between the simulated conductor and the actual conductor, K and β are both empirical coefficients, and v is the wind speed; ρ nis the conductor ice density. When n=1, it represents the conductor ice type as rime. When n=2, it represents the conductor ice type as mixed rime. When n=3, it represents the conductor ice type as rime. N1 represents the tension collected by the tension sensor unit before simulating the conductor ice coverage. N2 represents the tension collected by the tension sensor unit after simulating the conductor ice coverage.

[0016] Another object of the present invention is to provide a device for monitoring ice coating on transmission line conductors based on AI image recognition, comprising:

[0017] Data acquisition unit, used to collect micro-meteorological data, tension data and conductor image data;

[0018] A data monitoring terminal is used to receive and process the data transmitted by the data acquisition unit and upload the processed data to the system monitoring platform;

[0019] Simulated conductor, used to simulate the state of actual transmission conductor;

[0020] The power supply unit is used to supply power to the data acquisition unit and the data monitoring terminal.

[0021] The data acquisition unit is installed on the cross arm of the transmission angle steel tower, and the data acquisition unit includes:

[0022] The micro-meteorological sensing unit uses micro-meteorological sensors to collect meteorological data around the angle steel tower;

[0023] The tension sensing unit uses two high-precision tension sensors installed at both ends of the simulated wire to collect the weight of the simulated wire;

[0024] The first image monitoring unit uses a high-pixel fixed-focus lens and a starlight-level night vision lens for 24-hour uninterrupted monitoring of analog wires;

[0025] The second image monitoring unit uses a zoom network camera to monitor the actual transmission lines 24 hours a day.

[0026] The data monitoring terminal includes a main control and AI processing unit and a communication unit; the main control and AI processing unit includes:

[0027] The AI artificial intelligence module is used to determine the type of conductor ice coverage based on actual conductor images and transmit the information to the data analysis and processing module;

[0028] Data analysis and processing module, used to calculate the ice thickness of the conductor;

[0029] The communication unit includes:

[0030] Local communication module, used for local connection debugging;

[0031] Remote communication module, used to communicate with the system monitoring platform.

[0032] It can be seen from the above technical solution that the beneficial effects of the present invention are as follows: First, the present invention breaks through the limitations of traditional single monitoring methods through multi-sensor data fusion and improves the accuracy of ice monitoring: on the one hand, a dual image monitoring unit is combined with a micro-meteorological sensor unit and a tension sensor unit to realize all-weather multi-dimensional data acquisition, a high-pixel fixed-focus lens is used to ensure the capture of simulated wire details, and a zoom camera dynamically tracks the actual wire status. Assisted by meteorological parameters and weight change data, a comprehensive analysis system covering physical properties, morphological changes and environmental factors is constructed; on the other hand, ice type image recognition is performed through the AI artificial intelligence module, and the ice thickness of the wire is corrected according to different ice types, making ice monitoring more accurate; second, the AI artificial intelligence module is used to innovatively fuse multiple features such as texture, edge, and grayscale statistics, and the improved VGG16 deep convolutional neural network is used for deep feature fusion, which significantly improves the accuracy of ice type recognition, especially in complex scenes such as rain, fog, and night. It still maintains a high recognition stability, solving the recognition distortion problem caused by lighting changes and weather interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the method of the present invention;

[0034] Figure 2 This is a structural block diagram of the device of the present invention. DETAILED DESCRIPTION

[0035] like Figure 1 As shown, a method for monitoring ice coating on transmission line conductors based on AI image recognition includes the following steps in sequence:

[0036] (1) Obtain micro-meteorological data, tension data and conductor image data;

[0037] (2) improving the VGG16 deep convolutional neural network, training the improved VGG16 deep convolutional neural network, inputting the data obtained in step (1) into the trained VGG16 deep convolutional neural network, judging the wire icing type, and obtaining a wire icing type recognition result;

[0038] (3) Calculate the conductor ice thickness based on the conductor ice type identification results, micrometeorological data, and tension data;

[0039] (4) Upload the micro-meteorological data, tension data, conductor image data, conductor icing type identification results and conductor icing thickness to the system monitoring platform through the remote communication unit; if the conductor icing thickness is greater than the set warning value, an icing alarm message is sent to the system monitoring platform.

[0040] In step (2), the improvement of the VGG16 deep convolutional neural network refers to: adding a self-attention mechanism module after the convolution layer of the third, fourth and fifth convolutional blocks of the VGG16 deep convolutional neural network, and the self-attention mechanism module strengthens the representation ability of the key area by adaptively calculating the spatial position weights of the feature map; adding a channel attention mechanism, i.e., SE module, at the end of each convolutional block of the VGG16 deep convolutional neural network, and the SE module adaptively recalibrates the feature channel by learning the mutual dependence between channels.

[0041] In step (2), the training of the improved VGG16 deep convolutional neural network specifically refers to: obtaining 2,000 wire image data of different degrees of ice coverage in different time periods and different climatic conditions to form a training data set, preprocessing the wire image data, extracting a multidimensional feature vector, the multidimensional feature vector includes the texture features, edge features, grayscale statistical features and morphological change features of the image, standardizing the extracted multidimensional feature vector, and reducing the dimension through principal component analysis, and then using the improved VGG16 deep convolutional neural network as a classifier for training. The improved VGG16 deep convolutional neural network finally outputs the probability score of each category through the Softmax layer, selects the category with the highest probability as the recognition result, and thus determines whether the wire is covered with ice and the specific type of ice coverage.

[0042] In step (3), the AI module is used to obtain information about whether the wire is covered with ice. If not, the tension and N1 of the simulated wire monitored by two high-precision tension sensors under the condition of no ice are recorded. If ice is present, the type of icing on the wire is transmitted, including rime, rime, and mixed rime, and the tension and N2 of the simulated wire monitored by two high-precision tension sensors after icing are recorded.

[0043] The calculation formula for the ice thickness of the conductor is:

[0044]

[0045] Where: δ is the ice thickness of the conductor, d is the conductor diameter, l is the length of the simulated conductor, ΔT is the temperature difference between the simulated conductor and the actual conductor, K and β are both empirical coefficients, and v is the wind speed; ρ n is the conductor ice density. When n=1, it represents the conductor ice type as rime. When n=2, it represents the conductor ice type as mixed rime. When n=3, it represents the conductor ice type as rime. N1 represents the tension collected by the tension sensor unit before simulating the conductor ice coverage. N2 represents the tension collected by the tension sensor unit after simulating the conductor ice coverage.

[0046] like Figure 2 As shown, the device includes:

[0047] Data acquisition unit, used to collect micro-meteorological data, tension data and conductor image data;

[0048] A data monitoring terminal is used to receive and process the data transmitted by the data acquisition unit and upload the processed data to the system monitoring platform;

[0049] Simulated conductor, used to simulate the state of actual transmission conductor;

[0050] The power supply unit is used to supply power to the data acquisition unit and the data monitoring terminal.

[0051] The data acquisition unit is installed on the cross arm of the transmission angle steel tower, and the data acquisition unit includes:

[0052] The micro-meteorological sensing unit uses micro-meteorological sensors to collect meteorological data around the angle steel tower;

[0053] The tension sensing unit uses two high-precision tension sensors installed at both ends of the simulated wire to collect the weight of the simulated wire;

[0054] The first image monitoring unit uses a high-pixel fixed-focus lens and a starlight-level night vision lens for 24-hour uninterrupted monitoring of analog wires;

[0055] The second image monitoring unit uses a zoom network camera to monitor the actual transmission lines 24 hours a day.

[0056] The data monitoring terminal includes a main control and AI processing unit and a communication unit; the main control and AI processing unit includes:

[0057] The AI module is used to determine the type of conductor ice coverage based on actual conductor images and transmit the information to the data analysis and processing module. The AI module uses the RK3568 artificial intelligence algorithm core board, model RK3568, which has a built-in improved VGG16 deep convolutional neural network.

[0058] The data analysis and processing module is used to calculate the ice thickness of conductors. Based on the ice type identification results transmitted by the AI module and combined with the micrometeorological and tension data transmitted by the data acquisition module, the data analysis and processing module calculates the conductor ice thickness. The module then uploads the micrometeorological data, tension data, image data, conductor ice type, and thickness to the system monitoring platform via the remote communication module. The data analysis and processing module also determines the conductor ice thickness and, if the conductor ice thickness exceeds the alarm value, sends an ice alarm to the system monitoring platform. The data processing and analysis module uses the Qualcomm MSM8953 system-on-chip (SoC), model MSM-8953-B-792NSP-TR-01-AB, which integrates an Adreno 506 GPU and a Hexagon 546 DSP.

[0059] The communication unit includes:

[0060] Local communication module, used for local connection debugging; including WiFi communication and wired connection communication;

[0061] Remote communication module, used to communicate with the system monitoring platform, including optical fiber communication and 4G communication.

[0062] The present invention breaks through the limitations of traditional single monitoring methods through multi-sensor data fusion and improves the accuracy of ice monitoring: on the one hand, a dual-image monitoring unit is combined with a micro-meteorological sensor unit and a tension sensor unit to realize all-weather multi-dimensional data collection, a high-pixel fixed-focus lens is used to ensure the capture of simulated wire details, and a zoom camera dynamically tracks the actual wire status. Assisted by meteorological parameters and weight change data, a comprehensive analysis system covering physical properties, morphological changes and environmental factors is constructed; on the other hand, ice type image recognition is performed through the AI artificial intelligence module, and the ice thickness of the wire is corrected according to different ice types, making ice monitoring more accurate; the AI artificial intelligence module is used to innovatively fuse multiple types of features such as texture, edge, and grayscale statistics, and the improved VGG16 deep convolutional neural network is used for deep feature fusion, which significantly improves the accuracy of ice type recognition, especially in complex scenes such as rain, fog, and night. It still maintains a high recognition stability, and solves the recognition distortion problem caused by lighting changes and weather interference.

[0063] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for monitoring ice coating on transmission line conductors based on AI image recognition, characterized by: The method comprises the following steps in sequence: (1) Obtain micro-meteorological data, tension data and conductor image data; (2) improving the VGG16 deep convolutional neural network, training the improved VGG16 deep convolutional neural network, inputting the data obtained in step (1) into the trained VGG16 deep convolutional neural network, judging the wire icing type, and obtaining a wire icing type recognition result; (3) Calculate the conductor ice thickness based on the conductor ice type identification results, micrometeorological data, and tension data; (4) Uploading micro-meteorological data, tension data, conductor image data, conductor ice type identification results and conductor ice thickness to the system monitoring platform through the remote communication unit; If the ice thickness of the conductor is greater than the set warning value, an ice alarm message will be sent to the system monitoring platform.

2. The method for monitoring ice coating on transmission line conductors based on AI image recognition according to claim 1, characterized in that: In step (2), the improvement of the VGG16 deep convolutional neural network refers to: adding a self-attention mechanism module after the convolution layer of the third, fourth and fifth convolutional blocks of the VGG16 deep convolutional neural network, and the self-attention mechanism module strengthens the representation ability of the key area by adaptively calculating the spatial position weights of the feature map; adding a channel attention mechanism, i.e., SE module, at the end of each convolutional block of the VGG16 deep convolutional neural network, and the SE module adaptively recalibrates the feature channel by learning the mutual dependence between channels.

3. The method for monitoring ice coating on power transmission line conductors based on AI image recognition according to claim 1, characterized in that: In step (2), the training of the improved VGG16 deep convolutional neural network specifically refers to: obtaining 2000 wire image data of different ice coverage degrees in different time periods and different climatic conditions to form a training data set, preprocessing the wire image data, extracting a multidimensional feature vector, the multidimensional feature vector includes the texture features, edge features, grayscale statistical features and morphological change features of the image, standardizing the extracted multidimensional feature vector, and performing dimensionality reduction through principal component analysis, and then using the improved VGG16 deep convolutional neural network as a classifier for training.

4. The method for monitoring ice coating on transmission line conductors based on AI image recognition according to claim 1, characterized in that: In step (3), the calculation formula for the conductor ice thickness is: Where: δ is the ice thickness of the conductor, d is the conductor diameter, l is the length of the simulated conductor, ΔT is the temperature difference between the simulated conductor and the actual conductor, K and β are both empirical coefficients, and v is the wind speed; ρ n is the conductor ice density. When n=1, it represents the conductor ice type as rime. When n=2, it represents the conductor ice type as mixed rime. When n=3, it represents the conductor ice type as rime. N1 represents the tension collected by the tension sensor unit before simulating the conductor ice coverage. N2 represents the tension collected by the tension sensor unit after simulating the conductor ice coverage.

5. A device for implementing the method for monitoring ice coating on transmission line conductors based on AI image recognition according to any one of claims 1 to 4, characterized in that: include: Data acquisition unit, used to collect micro-meteorological data, tension data and conductor image data; A data monitoring terminal is used to receive and process the data transmitted by the data acquisition unit and upload the processed data to the system monitoring platform; Simulated conductor, used to simulate the state of actual transmission conductor; The power supply unit is used to supply power to the data acquisition unit and the data monitoring terminal.

6. The device according to claim 5, characterized in that: The data acquisition unit is installed on the cross arm of the transmission angle steel tower, and the data acquisition unit includes: The micro-meteorological sensing unit uses micro-meteorological sensors to collect meteorological data around the angle steel tower; The tension sensing unit uses two high-precision tension sensors installed at both ends of the simulated wire to collect the weight of the simulated wire; The first image monitoring unit uses a high-pixel fixed-focus lens and a starlight-level night vision lens for 24-hour uninterrupted monitoring of analog wires; The second image monitoring unit uses a zoom network camera to monitor the actual transmission lines 24 hours a day.

7. The device according to claim 5, characterized in that: The data monitoring terminal includes a main control and AI processing unit and a communication unit; the main control and AI processing unit includes: The AI artificial intelligence module is used to determine the type of conductor ice coverage based on actual conductor images and transmit the information to the data analysis and processing module; Data analysis and processing module, used to calculate the ice thickness of the conductor; The communication unit includes: Local communication module, used for local connection debugging; Remote communication module, used to communicate with the system monitoring platform.

Images