Composite insulator temperature rise detection method, device and equipment and storage medium

Abstract

The invention relates to a composite insulator temperature rise detection method, device and equipment and a storage medium, and relates to the technical field of power equipment detection. The method comprises the following steps: acquiring an infrared image of a target composite insulator; the infrared image is input into a preset temperature rise detection model, temperature rise detection is carried out on the target composite insulator based on the temperature rise detection model, a temperature rise detection result of the target composite insulator is obtained, and the temperature rise detection result comprises a temperature rise area position of the target composite insulator and a temperature rise amplitude of the temperature rise area position; the temperature rise detection model is a model constructed based on a YOLOv12 model. Based on the attention mechanism of the YOLOv12 model, the multi-scale feature fusion technology, lightweight design and efficient reasoning ability, the temperature rise data is analyzed through the temperature rise detection model constructed based on the YOLOv12 model, the temperature rise features of the composite insulator are accurately captured, and the accuracy, adaptability and efficiency of temperature rise detection of the composite insulator are improved.

Description

Technical Field

[0001] This disclosure relates to the field of power equipment testing technology, and in particular to a method, apparatus, equipment and storage medium for detecting the temperature rise of composite insulators. Background Technology

[0002] Composite insulators, as core insulation components of transmission lines, are crucial to the safe operation of the power grid. Composite insulators can experience localized abnormal temperature rises due to electrical aging, mechanical stress, or pollution. These abnormal rises are characterized by their concealed spatial distribution and rapid evolution rate. If not detected in time, they may lead to insulation breakdown or mechanical degradation, potentially causing major transmission accidents. Therefore, accurate detection of temperature rise in composite insulators is of great significance for maintaining the reliability of power systems.

[0003] Temperature rise detection of composite insulators refers to assessing the operating status and potential defects of composite insulators by measuring the difference between the surface temperature of the composite insulator and the ambient temperature. While existing methods for detecting temperature rise in composite insulators can perform the task, they struggle to detect minute temperature rises, exhibit significant deviations under complex lighting conditions or with contamination, have low sensitivity to temperature anomalies, and rely heavily on manual inspection and judgment. This results in a high false alarm rate, poor environmental adaptability, and low efficiency in temperature rise detection for composite insulators. Summary of the Invention

[0004] To address the aforementioned technical problems, this disclosure provides a method, apparatus, equipment, and storage medium for detecting the temperature rise of composite insulators.

[0005] The first aspect of this disclosure provides a method for detecting the temperature rise of composite insulators, comprising:

[0006] Acquire infrared images of the target composite insulator;

[0007] The infrared image is input into a preset temperature rise detection model for composite insulators. The temperature rise of the target composite insulator is detected based on the temperature rise detection model to obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model built based on the YOLOv12 model.

[0008] A second aspect of this disclosure provides a composite insulator temperature rise detection device, comprising:

[0009] The first acquisition module is used to acquire infrared images of the target composite insulator;

[0010] The detection module is used to input infrared images into a preset temperature rise detection model for composite insulators, and to perform temperature rise detection on the target composite insulator based on the temperature rise detection model to obtain the temperature rise detection results of the target composite insulator. The temperature rise detection results include the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model built based on the YOLOv12 model.

[0011] A third aspect of this disclosure provides a computer device including a memory and a processor, wherein the memory stores a computer program that, when executed by the processor, can implement the composite insulator temperature rise detection method of the first aspect described above.

[0012] The fourth aspect of this disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the composite insulator temperature rise detection method of the first aspect described above.

[0013] The technical solution provided in this disclosure has the following advantages compared with the prior art:

[0014] This disclosure involves acquiring an infrared image of a target composite insulator; inputting the infrared image into a preset temperature rise detection model for composite insulators; performing temperature rise detection on the target composite insulator based on the temperature rise detection model; and obtaining the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise region of the target composite insulator and the temperature rise amplitude at the location of the temperature rise region. The temperature rise detection model is a model constructed based on the YOLOv12 model. This disclosure utilizes the attention mechanism and multi-scale feature fusion technology of the YOLOv12 model, along with lightweight design and efficient inference capabilities. By analyzing temperature rise data through a temperature rise detection model built on the YOLOv12 model, it accurately captures the temperature rise characteristics of composite insulators, determines the location and amplitude of temperature rise regions, improves the detection accuracy for small targets, obscured targets, and targets in complex backgrounds, reduces the probability of missed detections and false positives, significantly reduces reliance on manual intervention, and improves the accuracy, adaptability, and efficiency of composite insulator temperature rise detection. Furthermore, the temperature rise detection scheme of this disclosure can achieve real-time detection on embedded devices or edge computing platforms, meeting the needs of online monitoring. It can provide a scientific basis for the condition assessment and maintenance decisions of composite insulators, and enhance the intelligent level of power equipment operation and maintenance. Attached Figure Description

[0015] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a method for detecting the temperature rise of a composite insulator provided in an embodiment of this disclosure;

[0018] Figure 2 This is a flowchart of a method for training a temperature rise detection model for composite insulators according to an embodiment of this disclosure;

[0019] Figure 3 This is a schematic diagram of a temperature rise detection model provided in an embodiment of this disclosure;

[0020] Figure 4 This is a schematic diagram of the structure of a composite insulator temperature rise detection device provided in an embodiment of this disclosure;

[0021] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Detailed Implementation

[0022] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0023] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0024] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0025] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0026] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0027] Composite insulators, as core insulation components of transmission lines, are crucial to the safe operation of the power grid. Industry statistics show that in infrared spectroscopy testing of 1.92 million composite insulators with voltage levels of 500kV and above, localized heating defects accounted for 0.58 per ten thousand. These defects are mostly caused by electrical aging, mechanical stress, or pollution, initially manifesting as a slight temperature rise. They are characterized by their concealed spatial distribution and rapid evolution rate. If not identified in time, they may lead to insulation breakdown or mechanical degradation, ultimately causing major transmission accidents. Therefore, accurate detection of temperature rise defects in composite insulators is of great significance for the reliability maintenance of power systems.

[0028] Existing temperature rise detection methods rely on manual inspection, which depends on maintenance personnel climbing towers for close-range observation. Due to terrain and safety limitations, it is difficult to cover long-distance transmission lines and is difficult to identify minute temperature rise defects, resulting in low efficiency and high risk. Manual inspection combined with machine learning methods is difficult to effectively capture the mapping relationship between complex textures and temperature rise in the slender structure of composite insulators due to the multi-scale and nonlinear characteristics of the temperature field distribution. It is prone to detection deviations under non-ideal conditions such as complex lighting and dirt coverage, resulting in insufficient environmental robustness, increased false positive rate, and slow detection speed. It also has inherent defects in extracting local temperature rise features for targets with long aspect ratios and is not sensitive enough to temperature anomalies (often <0.5% pixel percentage) in key parts such as high-voltage and low-voltage ends, making it difficult to meet the real-time processing requirements of UAV inspection platforms.

[0029] To address the shortcomings of existing technologies in composite insulator temperature rise detection, this disclosure provides a method, apparatus, device, and storage medium for composite insulator temperature rise detection. Based on the attention mechanism and multi-scale feature fusion technology of the YOLOv12 model, along with lightweight design and efficient inference capabilities, it analyzes temperature rise data using a temperature rise detection model built on the YOLOv12 model. This accurately captures the temperature rise characteristics of composite insulators, determines the location and amplitude of temperature rise regions, improves the detection accuracy for small targets, obscured targets, and targets in complex backgrounds, reduces the probability of missed detections and false positives, significantly reduces reliance on manual intervention, and improves the accuracy, adaptability, and efficiency of composite insulator temperature rise detection. Furthermore, the temperature rise detection scheme of this disclosure can achieve real-time detection on embedded devices or edge computing platforms, meeting the needs of online monitoring. It can provide a scientific basis for composite insulator condition assessment and maintenance decisions, and enhance the intelligent level of power equipment operation and maintenance.

[0030] To better understand the inventive concept of the embodiments of this disclosure, the technical solutions of the embodiments of this disclosure will be described below in conjunction with exemplary embodiments.

[0031] Figure 1 This is a flowchart of a method for detecting the temperature rise of composite insulators according to an embodiment of this disclosure. This method can be executed by a computer device, which can be understood as any device with computing and processing capabilities. This computer device can include, but is not limited to, mobile terminals such as smartphones, laptops, personal digital assistants (PDAs), tablets (PADs), in-vehicle terminals, and wearable devices, as well as fixed electronic devices such as digital TVs and desktop computers. Figure 1 As shown, the composite insulator temperature rise detection method provided in this embodiment includes the following steps:

[0032] Step 110: Obtain an infrared image of the target composite insulator.

[0033] In this embodiment of the disclosure, a computer device can control a drone equipped with an infrared thermal imager to acquire infrared images of the target composite insulator.

[0034] For example, the drone's flight altitude can be controlled between 5 and 10 meters, the shooting angle is ±15° perpendicular to the axis of the composite insulator, and the resolution is no less than 640×640 pixels.

[0035] In some embodiments, the infrared image of the composite insulator may include an infrared image of a heat-prone region of the composite insulator, which may include at least one of the high-voltage end, the low-voltage end, and the interface between the shed and the core rod. A computer device can acquire the infrared image of the heat-prone region of the target composite insulator. This allows the temperature rise detection model to quickly determine the location of the temperature rise region, improving the efficiency of temperature rise detection for composite insulators.

[0036] In some embodiments, after acquiring an infrared image of the target composite insulator, the computer device can preprocess the infrared image based on a preset first preprocessing method to obtain a preprocessed infrared image. The first preprocessing method may include at least one of grayscale processing, noise reduction processing, and normalization processing. This can unify the image size, enhance image quality, and reduce image noise.

[0037] Step 120: Input the infrared image into the preset temperature rise detection model for composite insulators, perform temperature rise detection on the target composite insulator based on the temperature rise detection model, and obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model built based on the YOLOv12 model.

[0038] In this embodiment of the present disclosure, after obtaining the infrared image of the target composite insulator, the computer device can input the infrared image into a preset temperature rise detection model for the composite insulator, and perform temperature rise detection on the target composite insulator based on the temperature rise detection model to obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result may include the location of the temperature rise region of the target composite insulator and the temperature rise amplitude at the location of the temperature rise region. The temperature rise detection model is a model constructed based on the YOLOv12 model.

[0039] The temperature rise amplitude can be understood as the difference between the surface temperature of the composite insulator and the ambient temperature.

[0040] The YOLOv12 model is a real-time object detection model with an attention mechanism at its core. It is a version of the YOLO (You Only Look Once) series of object detection models. It innovatively uses the attention mechanism as the core architecture, which significantly improves the accuracy of object detection while maintaining real-time detection speed.

[0041] Specifically, the temperature rise detection model mentioned above can include three parts: a backbone network, a neck network, and a head.

[0042] The above-mentioned temperature rise detection of composite insulators based on the temperature rise detection model, to obtain the temperature rise detection results of composite insulators, may include steps 1201-1204:

[0043] Step 1201: Based on the backbone network of the temperature rise detection model, perform multi-scale and multi-level feature extraction on the infrared image to obtain the multi-scale feature map of the infrared image.

[0044] Multi-scale feature maps can be understood as feature maps of multiple different scales. For example, an infrared image with a scale of 640×640 can be converted into a feature map with a scale of 320×320, a feature map with a scale of 160×160, and a feature map with a scale of 80×80.

[0045] Specifically, the backbone network can include Conv modules, C3k2 modules, and A2C2f modules:

[0046] The Conv module is a standard convolutional layer that can perform preliminary spatial feature extraction and dimensionality reduction on images; the C3k2 module improves feature representation and computational efficiency through a feature splitting-processing-fusion mechanism; the A2C2f module embeds AreaAttention into the C2f structure, dividing the feature map into vertical / horizontal regions (4 blocks by default), significantly reducing the computational complexity of attention while retaining a large receptive field.

[0047] Step 1202: The neck network based on the temperature rise detection model performs feature fusion on the multi-scale feature map to obtain the fused feature map of the infrared image.

[0048] The neck network can fuse features at different scales, increase the number of channels, and provide richer feature inputs for subsequent modules.

[0049] Specifically, the neck network can include Conv module, C3k2 module, A2C2f module, Concat module, Upsample module, and A2C2f module:

[0050] The Concat module can stitch together feature maps from different stages of the backbone network to achieve cross-level feature fusion; the Upsample module can upsample low-resolution feature maps and then align the low-resolution feature maps with the high-resolution features in terms of size; the A2C2f module further enhances the feature extraction capability of key regions.

[0051] Step 1203: The detection head based on the temperature rise detection model predicts the fused feature map to obtain the bounding box of the target composite insulator and the category probability of the bounding box.

[0052] The detection head can decode the bounding box of the target composite insulator from the fused feature map output by the neck network through the classification branch and the class probability of the bounding box from the fused feature map through the regression branch. The bounding box can represent the location of the temperature rise region of the target composite insulator, and the class probability of the bounding box can represent the temperature rise amplitude at the location of the temperature rise region.

[0053] Step 1204: Determine the location of the temperature rise region of the target composite insulator based on the bounding box of the target composite insulator, determine the temperature rise amplitude of the temperature rise region location based on the category probability of the bounding box, and determine the temperature rise region location and the temperature rise amplitude of the temperature rise region location of the target composite insulator as the temperature rise detection result of the target composite insulator.

[0054] For example, Figure 3 A schematic diagram of a temperature rise detection model is provided, such as... Figure 3 As shown, the temperature rise detection model can include three parts: a backbone network, a neck network, and a head. Steps 1201-1204 above can be performed using... Figure 3 The temperature rise detection model shown is executed, and the execution process may include the following steps:

[0055] The infrared image of the target composite insulator is input into the Backbone network. Initial downsampling is performed using the Conv module (convolution + BN + SiLU) to convert the input image (e.g., 640×640) into a 320×320 feature map for feature extraction. The Conv module calculation formula is as follows:

[0056] Y = SiLU(BN(Conv2d(X)));

[0057] Where Y is the output feature map with dimensions B×C. out ×H out ×W out (corresponding to batch size × number of output channels × height of input feature map × width of input feature map respectively), where X is the input feature map with dimensions B × C. in ×H in ×W in (These correspond to batch size × number of input channels × input feature map height × input feature map width, respectively).

[0058] Conv2d is a two-dimensional convolution, and its formula is:

[0059]

[0060] Where W is the convolution kernel with dimension C out ×C in ×k×k (corresponding to the number of output channels × the number of input channels × the core height × the core width, respectively), This is the bias term, with dimension C. out k is the kernel size, each output channel shares a bias value, and i and j represent the row and column indices of the kernel, ranging from 0 to k-1. The output feature map has dimensions B×C. out×H′×W′, where H′ and W′ represent the height and width of the output feature map, respectively, as shown in the following formula:

[0061]

[0062] Where s is the stride, which determines the spatial resolution of the output feature map, and P is the padding, which maintains the consistency of input and output sizes. BN normalizes the convolution output, as shown in the formula:

[0063]

[0064] Among them, Y b,c,h,w The output feature map has dimensions B×C. out ×H out ×W out X b,c,h,w The input feature map has dimensions B×C. in ×H in ×W in μ c The formula for calculating the batch mean is:

[0065]

[0066] The batch variance is calculated using the following formula:

[0067]

[0068] γ is a minimal constant (taken as 10⁻⁵). C β is a learnable scaling parameter (taken as 1). C To learn the offset parameter (set to 0).

[0069] SiLU is the activation function, and its formula is:

[0070]

[0071] Where X is the input feature map with dimensions B×C. in ×H in ×W in σ(x) is the sigmoid function.

[0072] The C3k2 module improves feature representation capabilities and computational efficiency through a feature splitting-processing-fusion mechanism. The splitting calculation formula is as follows:

[0073] X1,X2=Split(X,[c,c2-c]);

[0074] X1 is the main branch: c is the number of channels, processed by Bottleneck; X2 is the residual branch, with the number of channels being c2-c, directly connected to the output; Split means splitting the input channels by a ratio of e, c = c2·e, where e is the expansion coefficient (taken as 0.5), and c2 is the number of output channels, controlling the output feature dimension of the module.

[0075] The formula for calculating Bottleneck processing is as follows:

[0076] Bottleneck(X1)=X1+Conv(X1);

[0077] The processed main branch Bottleneck(X1) is concatenated and merged with the residual branch X2. The fusion calculation formula is as follows:

[0078] Y=Conv(Concat([Bottleneck(X1),X2]));

[0079] Where Y is the output feature map.

[0080] The A2C2f module introduces region attention, dividing the feature map into A×A regions, each with a size of H / A×W / A. For each region, the query, key, and value are calculated:

[0081] Q,K,V=Linear(X),Linear(X),Linear(X);

[0082] Here, Linear represents matrix multiplication, using three different weight matrices W. q W k W v Multiply by X, i.e., Q = XW q K = XW k V = XW v , Take d q =d k =d v .

[0083] Calculate the attention score matrix:

[0084]

[0085] Where d k It is the dimension of the k matrix.

[0086] To perform residual connection calculation, the original input X is added to the output Attn of the attention mechanism:

[0087] Xattn =X + Attn;

[0088] Connection calculation with feedforward network:

[0089] X ffn =X attn +MLP(X attn );

[0090] The MLP (Multilayer Perceptron) is a multilayer perceptron that further performs a nonlinear transformation on the output Xattn after residual connections. The calculation formula is as follows:

[0091]

[0092] Among them, w (L) Let b be the weight matrix of layer l. (L) The bias vector of layer l is σ. (L) The activation function for the l-th layer is ReLU, and its calculation formula is as follows:

[0093] σ (L) =ReLU(w (L) X attn +b (L) ) = max(0, w (L) X attn +b (L) );

[0094] The Neck network employs Conv, C3k2, A2C2f, Upsample, and Contact modules to fuse feature maps of different scales output from the Backbone, enhancing the model's ability to detect targets of various sizes. The Concat module fuses features of different scales / meanings, increasing the number of channels and providing richer feature inputs for subsequent modules. The calculation formula is as follows:

[0095] Y = Concat([X1,X2]);

[0096] The Upsample module improves resolution through nearest neighbor interpolation, calculated using the following formula:

[0097] Y b,c,ih,iw =X b,c,[ih / s],[iw / s] ;

[0098] Where s is the amplification factor;

[0099] The Head network, based on the multi-scale feature map output by the Neck network, decodes bounding boxes and class probabilities from the multi-scale feature map through a classification branch and a regression branch. The bounding boxes represent the locations of temperature-rising regions in the composite insulator, and the class probabilities of the bounding boxes represent the temperature rise magnitude at those locations. The classification branch predicts the class probability of each detection box using a three-layer convolution operation, calculated as follows:

[0100] Y = Conv(Conv(Conv2d(X)));

[0101] The regression branch is used to output the distribution parameters of the bounding box, and the calculation formula is as follows:

[0102] Y=Conv2d(Conv(DWConv(Conv(DWConv(X)))));

[0103] Where DWConv is a depthwise separable convolution, calculated using the following formula:

[0104]

[0105] in, For pointwise convolution kernels, Here, k is the kernel size, i and j represent the row and column indices of the kernel, ranging from 0 to k-1, and b... c′ This is a bias term.

[0106] Therefore, based on the attention mechanism and multi-scale feature fusion technology of the YOLOv12 model, as well as its lightweight design and efficient inference capabilities, a temperature rise detection model built on the YOLOv12 model can be used to analyze temperature rise data, accurately capture the temperature rise characteristics of composite insulators, determine the location and amplitude of temperature rise regions, improve the detection accuracy of small targets, obscured targets, and targets in complex backgrounds, reduce the probability of missed detections and false judgments, greatly reduce reliance on manual labor, improve the accuracy and adaptability of composite insulator temperature rise detection, and improve the efficiency of composite insulator temperature rise detection. Furthermore, the temperature rise detection scheme disclosed herein can achieve real-time detection on embedded devices or edge computing platforms, meeting the needs of online monitoring, providing a scientific basis for composite insulator condition assessment and maintenance decisions, and improving the intelligent level of power equipment operation and maintenance.

[0107] In some embodiments of this disclosure, before inputting the infrared image into a preset temperature rise detection model for composite insulators, the computer device may execute... Figure 2 A flowchart is provided for a training method of a temperature rise detection model for composite insulators. (For example...) Figure 2 As shown, the temperature rise detection model training method for composite insulators provided in this embodiment includes the following steps:

[0108] Step 210: Obtain infrared images of multiple composite insulators and temperature rise labeling data for each infrared image. The temperature rise labeling data includes the location of the temperature rise area of ​​the composite insulator and the temperature rise amplitude at the location of the temperature rise area.

[0109] In this embodiment of the disclosure, the computer device can acquire infrared images of multiple composite insulators and temperature rise labeling data for each infrared image. The temperature rise labeling data may include the location of the temperature rise region of the composite insulator and the temperature rise amplitude at that location. For example, the Labelme labeling tool can be used for labeling.

[0110] In some embodiments, the computer device can acquire infrared images of the heat-prone areas of the composite insulator, which may include at least one of the high-voltage end, the low-voltage end, and the interface between the shed and the core rod. This allows the temperature rise detection model to quickly determine the location of the temperature rise area, improving the training efficiency of the model.

[0111] Step 220: Construct an infrared image dataset based on infrared images of multiple composite insulators and temperature rise annotation data for each infrared image.

[0112] In this embodiment of the disclosure, a computer device can construct an infrared image dataset based on infrared images of multiple composite insulators and temperature rise annotation data of each infrared image.

[0113] In some embodiments, after obtaining infrared images of multiple composite insulators, the computer device can preprocess each infrared image based on a preset second preprocessing method to obtain a preprocessed infrared image; and construct an infrared image dataset based on the preprocessed infrared images and the temperature rise annotation data of each preprocessed infrared image.

[0114] The second preprocessing method may include at least one of image enhancement processing, image cropping processing, grayscale processing, noise reduction processing, and image normalization processing.

[0115] Image enhancement processing can include at least one of rotation, scaling, flipping, translation, brightness adjustment, and contrast adjustment. This allows for the rapid acquisition of multiple infrared images and the rapid construction of datasets.

[0116] Image cropping, grayscale conversion, denoising, and image normalization can reduce image noise, improve image quality, and thus enhance the stability of the model's gradient descent and improve the model's generalization ability.

[0117] Step 230: Iteratively train the YOLOv12 model based on the infrared image dataset until the accuracy of the temperature rise detection results output by the YOLOv12 model meets the preset accuracy requirements, thus completing the training of the YOLOv12 model.

[0118] Specifically, the infrared image dataset can be divided into training, validation, and test sets at a ratio of 60%, 20%, and 20%, respectively. The YOLOv12 model is trained based on the training and validation sets, and the temperature rise detection results output by the YOLOv12 model are tested based on the test set until the accuracy of the temperature rise detection results output by the YOLOv12 model is greater than or equal to a preset accuracy threshold, thus completing the training of the YOLOv12 model.

[0119] The preset accuracy threshold can be set as needed; there are no restrictions here.

[0120] Step 240: Determine the trained YOLOv12 model as the temperature rise detection model for composite insulators.

[0121] Therefore, a YOLOv12 model can be trained using an infrared image dataset constructed from multiple infrared images of composite insulators and temperature rise annotations for each infrared image. This yields a temperature rise detection model for composite insulators. Based on the attention mechanism and multi-scale feature fusion technology of the YOLOv12 model, as well as its lightweight design and efficient inference capabilities, the temperature rise detection model built on the YOLOv12 model can analyze the temperature rise data, accurately capture the temperature rise characteristics of composite insulators, determine the location and amplitude of the temperature rise region, improve the detection accuracy for small targets, occluded targets, and targets in complex backgrounds, reduce the probability of missed detections and false positives, significantly reduce reliance on manual intervention, improve the accuracy and adaptability of composite insulator temperature rise detection, and increase the efficiency of composite insulator temperature rise detection.

[0122] In some embodiments of this disclosure, after obtaining the temperature rise detection results of the target composite insulator, the computer device can compare the temperature rise amplitude at the temperature rise area of ​​the target composite insulator in the temperature rise detection results with a preset temperature rise threshold. When the temperature rise amplitude at the temperature rise area of ​​the target composite insulator is greater than the preset temperature rise threshold, it can be determined that the target composite insulator is in an abnormal temperature rise state. Then, the computer device can send an alarm message to the target personnel that the target composite insulator is in an abnormal temperature rise state, so that the target personnel can repair or replace the target composite insulator.

[0123] The preset temperature rise threshold can be set as needed, such as 10K, but this is not limited here.

[0124] Therefore, when an abnormal temperature rise is detected in the composite insulator, relevant personnel can be notified in a timely manner so that they can repair or replace the composite insulator in a timely manner, ensuring the stable operation of the composite insulator and improving the stability of the power system.

[0125] Figure 4This is a schematic diagram of a composite insulator temperature rise detection device provided in an embodiment of this disclosure. This device can be understood as the aforementioned computer equipment or a functional module within the aforementioned computer equipment. Figure 4 As shown, the composite insulator temperature rise detection device 400 includes:

[0126] The first acquisition module 410 is used to acquire an infrared image of the target composite insulator;

[0127] The detection module 420 is used to input infrared images into a preset temperature rise detection model for composite insulators, and to perform temperature rise detection on the target composite insulator based on the temperature rise detection model to obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model built based on the YOLOv12 model.

[0128] Optionally, the above-mentioned composite insulator temperature rise detection device includes:

[0129] The first preprocessing module is used to preprocess the infrared image based on a preset first preprocessing method to obtain a preprocessed infrared image. The first preprocessing method includes at least one of grayscale processing, noise reduction processing, and normalization processing.

[0130] Optionally, the above temperature rise detection model includes a backbone network, a neck network, and a detection head;

[0131] The above detection module includes:

[0132] The feature extraction submodule is used to perform multi-scale and multi-level feature extraction on infrared images based on the backbone network of the temperature rise detection model, so as to obtain multi-scale feature maps of infrared images.

[0133] The feature fusion submodule is used to fuse multi-scale feature maps using a neck network based on a temperature rise detection model to obtain a fused feature map of the infrared image.

[0134] The prediction submodule is used to predict the fused feature map based on the detection head of the temperature rise detection model to obtain the bounding box of the target composite insulator and the class probability of the bounding box.

[0135] The determination submodule is used to determine the location of the temperature rise region of the target composite insulator based on the bounding box of the target composite insulator, determine the temperature rise amplitude of the temperature rise region location based on the category probability of the bounding box, and determine the temperature rise region location and the temperature rise amplitude of the temperature rise region location of the target composite insulator as the temperature rise detection result of the target composite insulator.

[0136] Optionally, the above-mentioned composite insulator temperature rise detection device includes:

[0137] The second acquisition module is used to acquire infrared images of multiple composite insulators and temperature rise labeling data for each infrared image. The temperature rise labeling data includes the location of the temperature rise area of ​​the composite insulator and the temperature rise amplitude at the location of the temperature rise area.

[0138] The module is used to build an infrared image dataset based on infrared images of multiple composite insulators and temperature rise annotation data of each infrared image;

[0139] The training module is used to iteratively train the YOLOv12 model based on the infrared image dataset until the accuracy of the temperature rise detection result output by the YOLOv12 model is greater than or equal to the preset accuracy threshold, thus completing the training of the YOLOv12 model.

[0140] The determination module is used to determine the trained YOLOv12 model as a temperature rise detection model for composite insulators.

[0141] Optionally, the above-mentioned building modules include:

[0142] The preprocessing submodule is used to preprocess each infrared image based on a preset second preprocessing method to obtain a preprocessed infrared image.

[0143] A submodule is built to construct an infrared image dataset based on the preprocessed infrared images and the temperature rise annotation data of each preprocessed infrared image;

[0144] The second preprocessing method includes at least one of image enhancement processing, image cropping processing, grayscale processing, noise reduction processing, and image normalization processing;

[0145] Image enhancement processing includes at least one of rotation, scaling, flipping, translation, brightness adjustment, and contrast adjustment.

[0146] Optionally, the infrared image mentioned above includes an infrared image of the heat-prone area of ​​the composite insulator, and the heat-prone area includes at least one of the high-voltage end, the low-voltage end, and the interface of the shed core rod.

[0147] Optionally, the above-mentioned composite insulator temperature rise detection device includes:

[0148] The status determination module is used to determine that the target composite insulator is in an abnormal temperature rise state when the temperature rise amplitude at the temperature rise area of ​​the target composite insulator is greater than the preset temperature rise threshold in the temperature rise detection results.

[0149] The alarm module is used to send alarm information to the target personnel that the target composite insulator is in an abnormal temperature rise state, so that the target personnel can repair or replace the target composite insulator.

[0150] The apparatus provided in this disclosure can implement the methods of any of the above embodiments, and its execution and beneficial effects are similar, so they will not be described again here.

[0151] This disclosure also provides a computer device, which includes a processor and a memory, wherein the memory stores a computer program. When the computer program is executed by the processor, it can implement the methods of any of the above embodiments. The execution method and beneficial effects are similar, and will not be described again here.

[0152] The computer device in this disclosure can be understood as any device with processing and computing capabilities. This device may include, but is not limited to, mobile terminals such as smartphones, laptops, personal digital assistants (PDAs), tablet computers (PADs), in-vehicle terminals, and wearable devices, as well as fixed electronic devices such as digital TVs and desktop computers.

[0153] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure, such as... Figure 5 As shown, the computer device 500 may include a processor 510 and a memory 520. The memory 520 stores a computer program 521. When the computer program 521 is executed by the processor 510, it can implement the method provided in any of the above embodiments. Its execution mode and beneficial effects are similar and will not be described again here.

[0154] Of course, for the sake of simplicity, Figure 5 Only some of the components of the computer device 500 relevant to the present invention are shown in this illustration; components such as buses, input / output interfaces, input devices, and output devices are omitted. In addition, the computer device 500 may include any other suitable components depending on the specific application.

[0155] This disclosure provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it can implement the methods of any of the above embodiments. The execution method and beneficial effects are similar, and will not be described again here.

[0156] The aforementioned computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0157] The computer program described above can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this disclosure. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer device, partially on the user's device, as a standalone software package, partially on the user's computer device and partially on a remote computer device, or entirely on a remote computer device or server.

[0158] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0159] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0160] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting the temperature rise of composite insulators, characterized in that, include: Acquire infrared images of the target composite insulator; The infrared image is input into a preset temperature rise detection model for composite insulators. The temperature rise of the target composite insulator is detected based on the temperature rise detection model to obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model constructed based on the YOLOv12 model.

2. The method according to claim 1, characterized in that, Before inputting the infrared image into a preset temperature rise detection model for composite insulators, the method further includes: The infrared image is preprocessed based on a preset first preprocessing method to obtain a preprocessed infrared image. The first preprocessing method includes at least one of grayscale processing, noise reduction processing, and normalization processing.

3. The method according to claim 1, characterized in that, The temperature rise detection model includes a backbone network, a neck network, and a detection head; The temperature rise detection of the target composite insulator based on the temperature rise detection model, to obtain the temperature rise detection result of the target composite insulator, includes: Based on the backbone network of the temperature rise detection model, multi-scale and multi-level feature extraction is performed on the infrared image to obtain a multi-scale feature map of the infrared image. The neck network based on the temperature rise detection model performs feature fusion on the multi-scale feature map to obtain the fused feature map of the infrared image; The detection head based on the temperature rise detection model predicts the fused feature map to obtain the bounding box of the target composite insulator and the class probability of the bounding box; The location of the temperature rise region of the target composite insulator is determined based on the bounding box of the target composite insulator. The temperature rise amplitude of the temperature rise region is determined based on the category probability of the bounding box. The location of the temperature rise region and the temperature rise amplitude of the temperature rise region are determined as the temperature rise detection result of the target composite insulator.

4. The method according to claim 1, characterized in that, Before inputting the infrared image into a preset temperature rise detection model for composite insulators, the method further includes: Acquire infrared images of multiple composite insulators and temperature rise labeling data for each infrared image, wherein the temperature rise labeling data includes the location of the temperature rise region of the composite insulator and the temperature rise amplitude at the location of the temperature rise region; An infrared image dataset is constructed based on infrared images of multiple composite insulators and temperature rise annotation data for each infrared image; The YOLOv12 model is iteratively trained based on the infrared image dataset until the accuracy of the temperature rise detection result output by the YOLOv12 model is greater than or equal to a preset accuracy threshold, thus completing the training of the YOLOv12 model. The trained YOLOv12 model was selected as the temperature rise detection model for composite insulators.

5. The method according to claim 4, characterized in that, The infrared image dataset is constructed based on infrared images of multiple composite insulators and temperature rise annotation data of each infrared image, including: Each infrared image is preprocessed based on a preset second preprocessing method to obtain a preprocessed infrared image; An infrared image dataset is constructed based on the preprocessed infrared images and the temperature rise annotation data of each preprocessed infrared image. The second preprocessing method includes at least one of image enhancement processing, image cropping processing, grayscale processing, noise reduction processing, and image normalization processing; The image enhancement processing includes at least one of rotation, scaling, flipping, translation, brightness adjustment, and contrast adjustment.

6. The method according to claim 1 or 4, characterized in that, The infrared image includes an infrared image of the heat-prone area of ​​the composite insulator, and the heat-prone area includes at least one of the high-voltage end, the low-voltage end, and the interface of the shed core rod.

7. The method according to claim 1, characterized in that, After obtaining the temperature rise detection result of the target composite insulator, the method further includes: If the temperature rise amplitude at the temperature rise region of the target composite insulator is greater than the preset temperature rise threshold in the temperature rise detection results, the target composite insulator is determined to be in an abnormal temperature rise state. An alarm message is sent to the target personnel indicating that the target composite insulator is in an abnormal temperature rise state, so that the target personnel can repair or replace the target composite insulator.

8. A composite insulator temperature rise detection device, characterized in that, include: The first acquisition module is used to acquire infrared images of the target composite insulator; The detection module is used to input the infrared image into a preset temperature rise detection model for composite insulators, and to perform temperature rise detection on the target composite insulator based on the temperature rise detection model to obtain the temperature rise detection result of the target composite insulator. The temperature rise detection result includes the location of the temperature rise area of ​​the target composite insulator and the temperature rise amplitude at the location of the temperature rise area. The temperature rise detection model is a model constructed based on the YOLOv12 model.

9. A computer device, characterized in that, include: A memory and a processor, wherein the memory stores a computer program that, when executed by the processor, implements the composite insulator temperature rise detection method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the composite insulator temperature rise detection method as described in any one of claims 1-7.

Images