Semantic recognition method and system for distribution network scheduling and storage medium

Abstract

The embodiment of the invention provides a semantic recognition method and system for distribution network scheduling and a storage medium, and belongs to the technical field of transformers. The semantic recognition method comprises the steps of obtaining an infrared image of an area where a target distribution network transformer is located; the infrared image is preprocessed; constructing an image recognition model based on the full convolutional network; and inputting the infrared image into the image recognition model to obtain a pixel-level classification result of the infrared image. Accurate identification of the distribution network transformer is realized through acquisition and preprocessing of an infrared image, construction of an image identification model based on a full convolutional network and output of a pixel-level classification result. The method can be widely applied to the fields of electric power inspection, equipment maintenance and the like, and the working efficiency and safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and in particular to a semantic recognition method, system and storage medium for distribution network scheduling. Background Art

[0002] In the power system, the distribution transformer is an important distribution equipment, and its operating status is directly related to the stability and reliability of power supply. However, the traditional manual inspection method has problems such as low efficiency, high cost, and susceptibility to environmental influences, which makes it difficult to meet the efficient operation and maintenance needs of modern power systems. Therefore, there is an urgent need for a technical means to automatically identify the distribution transformer and accurately determine its operating status. Summary of the invention

[0003] The purpose of the embodiments of the present invention is to provide a semantic recognition method, system and storage medium for distribution network scheduling, which solves the problem of difficult identification and low identification efficiency of distribution network transformers.

[0004] In order to achieve the above object, an embodiment of the present invention provides a semantic recognition method for distribution network scheduling, the semantic recognition method comprising:

[0005] Obtain an infrared image of the area where the target distribution network transformer is located;

[0006] Preprocessing the infrared image;

[0007] Build an image recognition model based on a fully convolutional network;

[0008] The infrared image is input into the image recognition model to obtain a pixel-level classification result of the infrared image.

[0009] Optionally, acquiring an infrared image of a target distribution network transformer includes:

[0010] Acquire an image of a target distribution network transformer;

[0011] Acquire an image of the background area of ​​the target distribution transformer.

[0012] Optionally, preprocessing the infrared image includes:

[0013] Adjusting the size of the infrared image;

[0014] Performing normalization processing on the infrared image;

[0015] Using a median filter to remove noise from the infrared image;

[0016] Enhance the contrast of infrared images by histogram equalization.

[0017] Optionally, the image recognition model includes:

[0018] A first convolution unit, configured to perform convolution processing on the infrared image of the transformer to obtain a first feature map;

[0019] A first deconvolution unit, configured to perform a deconvolution process on the first feature map output by the first convolution unit to obtain a second feature map;

[0020] A second convolution unit, used for performing convolution processing on the second feature map output by the first deconvolution unit to obtain a third feature map;

[0021] A second deconvolution unit, used to perform deconvolution processing on the third feature map output by the second convolution unit to obtain a semantic segmentation result map;

[0022] An output unit classifies and outputs the semantic segmentation result map to obtain a pixel-level classification result of the infrared image.

[0023] Optionally, the image size of the semantic segmentation result image is the same as the image size of the infrared image of the distribution network transformer to be identified.

[0024] Optionally, an image recognition model is constructed based on a fully convolutional network, including:

[0025] Dividing the infrared images of the transformer into a training set and a validation set;

[0026] Injecting the training set into the image recognition model for training to update the parameters of the image recognition model;

[0027] Inputting the verification set into the image recognition model for training to calculate the performance index of the image recognition model;

[0028] Determining whether the performance indicator is greater than a first set threshold;

[0029] Under the condition that it is determined that the performance indicator is greater than the first set threshold, the image recognition model is saved.

[0030] Optionally, inputting the verification set into the image recognition model for training to calculate the performance index of the image recognition model includes:

[0031] Calculate the balanced F score of the image recognition model according to formula (1);

[0032]

[0033] Among them, F1 is the balanced F score, Precision is the precision, and Recall is the recall rate.

[0034] On the other hand, the present invention provides a semantic recognition method system for distribution network scheduling, the semantic recognition system comprising:

[0035] An image acquisition module, used to acquire infrared images of the area where the distribution network transformer to be identified is located;

[0036] An image processing module, used for preprocessing the infrared image;

[0037] An image input module, used for inputting the collected infrared image into an image recognition model for processing;

[0038] The result display module is used to display the pixel-level classification result of the output infrared image, and identify the image of the distribution network transformer to be identified from the infrared image.

[0039] On the other hand, the computer-readable storage medium stores instructions, and the instructions are used to be read by a machine so that the machine executes any of the semantic recognition methods described above.

[0040] Through the above technical scheme, the present invention provides a semantic recognition method, system and storage medium for distribution network scheduling. First, an infrared image of the area where the target distribution network transformer is located is obtained, and then the infrared image is preprocessed. Finally, the infrared image is input into the image recognition model to obtain the pixel-level classification result of the infrared image. Compared with the prior art, the present invention uses infrared images to identify distribution network transformers, while traditional methods mainly rely on visible light images. Infrared images can provide richer information, especially in the case of insufficient light or blocked vision, and can still clearly display the location and status of the equipment, thereby improving the accuracy and robustness of recognition. Through the acquisition, preprocessing, construction of an image recognition model based on a fully convolutional network and the output of pixel-level classification results of infrared images, accurate identification of distribution network transformers is achieved. This method can be widely used in fields such as power inspection and equipment maintenance to improve work efficiency and safety.

[0041] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0043] Figure 1 is a flow chart of a semantic recognition method according to an embodiment of the present invention;

[0044] Figure 2is a flow chart of obtaining an infrared image of a target distribution network transformer according to an embodiment of the present invention;

[0045] Figure 3 is a flow chart of preprocessing infrared images according to one embodiment of the present invention;

[0046] Figure 4 is a flowchart of training an image recognition model according to one embodiment of the present invention. DETAILED DESCRIPTION

[0047] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0048] Figure 1 FIG. 1 is a flow chart of a semantic recognition method according to an embodiment of the present invention. In the flow chart, the semantic recognition method includes:

[0049] In step S1, an infrared image of the area where the target distribution network transformer is located is obtained.

[0050] In step S2, the infrared image is preprocessed.

[0051] In step S3, an image recognition model is constructed based on a fully convolutional network.

[0052] In step S4, the infrared image is input into the image recognition model to obtain the pixel-level classification result of the infrared image.

[0053] Compared with the prior art, the present invention uses infrared images to identify distribution network transformers, while traditional methods mainly rely on visible light images. Infrared images can provide richer information, especially in the case of insufficient light or obstructed vision, and can still clearly display the location and status of the equipment, thereby improving the accuracy and robustness of recognition. Through the acquisition and preprocessing of infrared images, the construction of an image recognition model based on a fully convolutional network, and the output of pixel-level classification results, accurate identification of distribution network transformers is achieved. This method can be widely used in fields such as power inspection and equipment maintenance to improve work efficiency and safety.

[0054] In this implementation, the specific content of the acquired infrared image may be various ones known to those skilled in the art. In one example of the present invention, for example Figure 2 As shown, it may include:

[0055] In step S11, an image of a target distribution network transformer is acquired.

[0056] In step S12, an image of the background area of ​​the target distribution network transformer is acquired.

[0057] In this implementation, in order to obtain a relatively high-quality infrared image of the transformer, the infrared image of the transformer needs to be preprocessed. The preprocessing method may be a variety of methods known to those skilled in the art. In one example of the present invention, for example Figure 3 As shown, it may include:

[0058] In step S21, the size of the infrared image is adjusted to a standard size for subsequent processing.

[0059] In step S22, the infrared image is normalized to normalize the image pixel values ​​to between [0, 1] to reduce the brightness difference between different images.

[0060] In step S23, a median filter is used to remove the noise of the infrared image. The median filter is used to remove random noise in the image and improve the image quality.

[0061] In step S24, the contrast of the infrared image is enhanced by histogram equalization. The contrast of the image is enhanced to make the transformer more prominent.

[0062] In this embodiment, the specific structure of the constructed image recognition model can be various known to those skilled in the art. On the premise of identifying the target distribution network transformer, it can be various known to those skilled in the art. In one example of the present invention, it can include a first convolution unit, a first deconvolution unit, a second convolution unit and an output unit. The first convolution unit is used to perform convolution processing on the infrared image of the transformer to obtain a first feature map, the first deconvolution unit is used to perform deconvolution processing on the first feature map output by the first convolution unit to obtain a second feature map, the second convolution unit is used to perform convolution processing on the second feature map output by the first deconvolution unit to obtain a third feature map, the second deconvolution unit is used to perform deconvolution processing on the third feature map output by the second convolution unit to obtain a semantic segmentation result map, and the output unit classifies and outputs the semantic segmentation result map to obtain a pixel-level classification result of the infrared image.

[0063] In this embodiment, in order to obtain a complete image recognition model, it is necessary to input the infrared image of the transformer into the image recognition model for training. The steps for training the image recognition model can be various steps known to those skilled in the art. In one example of the present invention, the specific training steps can be as follows: Figure 4 As shown. Figure 4 In the training step, the training step may include:

[0064] In step S31, the training set is injected into the image recognition model for training to update the parameters of the image recognition model.

[0065] In step S32, the verification set is input into the image recognition model for training to calculate the performance index of the image recognition model.

[0066] In step S33, it is determined whether the performance index is greater than a first set threshold value. If the performance index is greater than the first set threshold value, step S34 is executed, otherwise, step S31 is executed.

[0067] In step S34, the image recognition model is saved.

[0068] In this embodiment, further, the image size of the semantic segmentation result image is the same as the image size of the infrared image of the distribution network transformer to be identified.

[0069] In this implementation, in order to evaluate the performance of the image recognition model, it is also necessary to calculate its performance index. The types of the performance index may be various ones known to those skilled in the art. In one example of the present invention, the performance index may include:

[0070] Calculate the balanced F score of the image recognition model according to formula (1);

[0071]

[0072] Among them, F1 is the balanced F score, Precision is the precision, and Recall is the recall rate.

[0073] On the other hand, the present invention provides a semantic recognition method system for distribution network scheduling, and the semantic recognition system includes an image acquisition module, an image processing module, an image input module and a result display module. The image acquisition module is used to acquire an infrared image of the area where the distribution network transformer to be identified is located, the image processing module is used to preprocess the infrared image, the image input module is used to input the acquired infrared image into the image recognition model for processing, and the result display module is used to display the pixel-level classification result of the output infrared image, and the image of the distribution network transformer to be identified is obtained from the infrared image.

[0074] On the other hand, the computer-readable storage medium stores instructions, which are used to be read by a machine so that the machine executes any of the semantic recognition methods described above.

[0075] Through the above technical scheme, the present invention provides a semantic recognition method, system and storage medium for distribution network scheduling. First, an infrared image of the area where the target distribution network transformer is located is obtained, and then the infrared image is preprocessed. Finally, the infrared image is input into the image recognition model to obtain the pixel-level classification result of the infrared image. Compared with the prior art, the present invention uses infrared images to identify distribution network transformers, while traditional methods mainly rely on visible light images. Infrared images can provide richer information, especially in the case of insufficient light or blocked vision, and can still clearly display the location and status of the equipment, thereby improving the accuracy and robustness of recognition. Through the acquisition, preprocessing, construction of an image recognition model based on a fully convolutional network and the output of pixel-level classification results of infrared images, accurate identification of distribution network transformers is achieved. This method can be widely used in fields such as power inspection and equipment maintenance to improve work efficiency and safety.

[0076] 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.

[0077] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0078] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0079] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0080] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0081] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A semantic recognition method for distribution network scheduling, characterized in that: The semantic recognition method comprises: Obtain an infrared image of the area where the target distribution network transformer is located; Preprocessing the infrared image; Build an image recognition model based on a fully convolutional network; The infrared image is input into the image recognition model to obtain a pixel-level classification result of the infrared image.

2. The semantic recognition method according to claim 1, characterized in that: Acquire infrared images of target distribution transformers, including: Acquire an image of a target distribution network transformer; Acquire an image of the background area of ​​the target distribution transformer.

3. The semantic recognition method according to claim 2, characterized in that: Preprocessing the infrared image includes: Adjusting the size of the infrared image; Performing normalization processing on the infrared image; Using a median filter to remove noise from the infrared image; Enhance the contrast of infrared images by histogram equalization.

4. The semantic recognition method according to claim 3, characterized in that: The image recognition model includes: A first convolution unit, configured to perform convolution processing on the infrared image of the transformer to obtain a first feature map; A first deconvolution unit, configured to perform a deconvolution process on the first feature map output by the first convolution unit to obtain a second feature map; A second convolution unit, used for performing convolution processing on the second feature map output by the first deconvolution unit to obtain a third feature map; A second deconvolution unit, used to perform a deconvolution process on the third feature map output by the second convolution unit to obtain a semantic segmentation result map; An output unit classifies and outputs the semantic segmentation result map to obtain a pixel-level classification result of the infrared image.

5. The semantic recognition method according to claim 4, characterized in that: The image size of the semantic segmentation result image is the same as the image size of the infrared image of the distribution network transformer to be identified.

6. The semantic recognition method according to claim 5, characterized in that: Build an image recognition model based on a fully convolutional network, including: Dividing the infrared images of the transformer into a training set and a validation set; Injecting the training set into the image recognition model for training to update the parameters of the image recognition model; Inputting the verification set into the image recognition model for training to calculate the performance index of the image recognition model; Determining whether the performance indicator is greater than a first set threshold; Under the condition that it is determined that the performance indicator is greater than the first set threshold, the image recognition model is saved.

7. The semantic recognition method according to claim 6, characterized in that: Inputting the verification set into the image recognition model for training to calculate the performance index of the image recognition model includes: Calculate the balanced F score of the image recognition model according to formula (1); Among them, F1 is the balanced F score, Precision is the precision, and Recall is the recall rate.

8. A semantic recognition system for distribution network scheduling, characterized in that: The semantic recognition system comprises: An image acquisition module is used to acquire infrared images of the area where the distribution network transformer to be identified is located; An image processing module, used for preprocessing the infrared image; An image input module, used for inputting the collected infrared image into an image recognition model for processing; The result display module is used to display the pixel-level classification result of the output infrared image, and identify the image of the distribution network transformer to be identified from the infrared image.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and the instructions are used to be read by a machine so that the machine executes the semantic recognition method according to any one of claims 1 to 7.

Images