Wire temperature detection method, device, equipment, storage medium and program product

Abstract

The invention provides a wire temperature detection method and device, equipment, a storage medium and a program product, and relates to the technical field of electric power. According to the method, the infrared image of the to-be-detected wire is obtained, the infrared image is input into the wire temperature detection model for wire temperature detection, the detection information which is output by the wire temperature detection model and contains the wire temperature of the to-be-detected wire is obtained, and the accuracy of wire temperature detection is improved.

Description

Technical Field

[0001] This application relates to the field of power technologies, and in particular, to a method, device, equipment, storage medium, and program product for detecting the temperature of a wire. Background Art

[0002] In power transmission and distribution networks, wires are the core components for electric energy transmission, and their operating states are crucial for the stability and security of the entire power system. The wire temperature is one of the key indicators for evaluating the working condition of a wire. Excessive wire temperature may cause a decline in the physical properties of the wire, such as accelerating the aging of the insulation layer and reducing the metal conductivity, thereby shortening the service life of the wire. More seriously, excessive wire temperature may also trigger line failures, resulting in power outages, which have a significant impact on social production and life. Therefore, monitoring and managing the wire temperature is crucial for ensuring the safe operation of the power system.

[0003] In the related art, an infrared imager is used to detect the wire temperature. However, this method has the problem of low accuracy. Summary of the Invention

[0004] This application provides a method, device, equipment, storage medium, and program product for detecting the temperature of a wire, aiming to solve the problem of low accuracy in detecting the wire temperature in the related art.

[0005] In a first aspect, this application provides a method for detecting the temperature of a wire, including: obtaining an infrared image of the wire to be detected; inputting the infrared image into a wire temperature detection model for wire temperature detection to obtain detection information output by the wire temperature detection model, where the detection information includes the wire temperature of the wire to be detected.

[0006] In a possible implementation, the wire temperature detection model includes a backbone network, a bottleneck network, and a detection network. Inputting the infrared image into the wire temperature detection model for wire temperature detection to obtain detection information output by the wire temperature detection model includes: inputting the infrared image into the backbone network for convolutional processing to obtain a first feature map output by the backbone network, where the convolutional method of the backbone network is depthwise separable convolution, and the convolutional processing includes depth convolution processing and pointwise convolution processing; inputting the first feature map into the bottleneck network for multi-scale feature fusion processing to obtain a second feature map output by the bottleneck network, where the bottleneck network is a bidirectional feature pyramid network, and the multi-scale feature fusion processing includes upsampling processing, downsampling processing, and feature fusion processing; inputting the second feature map into the detection network for wire temperature detection to obtain detection information output by the detection network.

[0007] In a possible implementation, the detection information further includes target area location information of the wire to be detected.

[0008] In a possible implementation, the wire temperature detection model is trained as follows: Obtain an infrared image dataset; Based on a preset ratio, divide the infrared image dataset into a training set, a validation set, and a prediction set; Input the training set and the prediction set into the wire temperature detection model to be trained, and perform model training based on the training set and adjust the model parameters based on the prediction set to obtain a trained wire temperature detection model; Input the validation set into the trained wire temperature detection model for model validation to obtain the wire temperature detection model.

[0009] In a possible implementation, the infrared image dataset is obtained as follows: Collect infrared images of different wires under different simulated working conditions to obtain an original infrared image dataset; Preprocess each original infrared image in the original infrared image dataset to obtain a target infrared image dataset; For each target infrared image in the target infrared image dataset, label the recognition region corresponding to the wire included in the target infrared image and the temperature corresponding to the wire to obtain the infrared image dataset.

[0010] In a possible implementation, before inputting the training set and the prediction set into the wire temperature detection model to be trained, it further includes: For each infrared image in the training set, perform clustering analysis on the anchor box parameters of the wire temperature detection model to be trained according to the recognition region corresponding to the wire included in the infrared image and the temperature corresponding to the wire.

[0011] In a second aspect, the present application provides a wire temperature detection device, including:

[0012] An acquisition module, configured to acquire an infrared image of a wire to be detected;

[0013] A detection module, configured to input the infrared image into the wire temperature detection model to perform wire temperature detection, and obtain detection information output by the wire temperature detection model, where the detection information includes the wire temperature of the wire to be detected.

[0014] In a possible implementation, the wire temperature detection model includes a backbone network, a bottleneck network, and a detection network. The detection module is specifically configured to: Input the infrared image into the backbone network for convolutional processing to obtain a first feature map output by the backbone network. The convolutional method of the backbone network is depthwise separable convolution, and the convolutional processing includes depth convolution processing and pointwise convolution processing; Input the first feature map into the bottleneck network for multi-scale feature fusion processing to obtain a second feature map output by the bottleneck network. The bottleneck network is a bidirectional feature pyramid network, and the multi-scale feature fusion processing includes upsampling processing, downsampling processing, and feature fusion processing; Input the second feature map into the detection network for wire temperature detection to obtain the detection information output by the detection network.

[0015] In a possible implementation, the detection information further includes the target area positioning information of the wire to be detected.

[0016] In a possible implementation, the wire temperature detection model is trained as follows: obtaining an infrared image data set; dividing the infrared image data set into a training set, a validation set, and a prediction set based on a preset ratio; inputting the training set and the prediction set into the wire temperature detection model to be trained, training the model based on the training set, adjusting the model parameters based on the prediction set, and obtaining the trained wire temperature detection model; inputting the validation set into the trained wire temperature detection model for model validation to obtain the wire temperature detection model.

[0017] In a possible implementation, the infrared image data set is obtained as follows: collecting infrared images of different wires under different simulated working conditions to obtain an original infrared image data set; preprocessing each original infrared image in the original infrared image data set to obtain a target infrared image data set; for each target infrared image in the target infrared image data set, labeling the recognition area corresponding to the wire included in the target infrared image and the temperature corresponding to the wire to obtain the infrared image data set.

[0018] In a possible implementation, the wire temperature detection device further includes a clustering analysis module (not shown). The clustering analysis module is configured to, before inputting the training set and the prediction set into the wire temperature detection model to be trained, perform clustering analysis on the anchor box parameters of the wire temperature detection model to be trained according to the recognition area corresponding to the wire included in each infrared image in the training set and the temperature corresponding to the wire.

[0019] In a third aspect, the present application provides an electronic device, including a processor and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided in the first aspect as described above.

[0020] In a fourth aspect, the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the method provided in the first aspect as described above.

[0021] In a fifth aspect, the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the method provided in the first aspect as described above.

[0022] The wire temperature detection method, device, equipment, storage medium, and program product provided by this application obtain the infrared image of the wire to be detected, input the infrared image into the wire temperature detection model for wire temperature detection, and obtain the detection information including the wire temperature of the wire to be detected output by the wire temperature detection model, thereby improving the accuracy of wire temperature detection. Description of the Drawings

[0023] The drawings herein are incorporated into the specification and form a part of this specification, showing the embodiments consistent with this application, and are used together with the specification to explain the principles of this application.

[0024] Figure 1 Flow schematic of the wire temperature detection method provided by an embodiment of this application Figure 1 ;

[0025] Figure 2 Structural schematic of the wire temperature detection model provided by an embodiment of this application;

[0026] Figure 3 Flow schematic of the wire temperature detection method provided by an embodiment of this application Figure 2 ;

[0027] Figure 4 Flow schematic of the wire temperature detection method provided by an embodiment of this application Figure 3 ;

[0028] Figure 5 Structural schematic of the wire temperature detection system provided by an embodiment of this application;

[0029] Figure 6 Structural schematic of the wire temperature detection device provided by an embodiment of this application;

[0030] Figure 7 Structural schematic of the electronic device provided by an embodiment of this application.

[0031] Through the above drawings, the clear embodiments of this application have been shown, and there will be more detailed descriptions later. These drawings and textual descriptions are not intended to limit the scope of the concept of this application in any way, but to illustrate the concept of this application to those skilled in the art by referring to specific embodiments. Detailed Embodiments

[0032] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with this application. On the contrary, they are merely examples of the devices and methods consistent with some aspects of this application as detailed in the appended claims.

[0033] In the related art, an infrared imager is used to detect the temperature of a wire. Infrared imaging technology is a technology that, based on the characteristic of an object emitting infrared rays by itself, converts an invisible temperature distribution into a visual thermal image, so as to intuitively present the temperature condition on the surface of the wire. However, when operating in an outdoor environment, due to the widespread existence of sunlight refraction and infrared radiation reflection phenomena on metal wires, the infrared components in sunlight and the infrared components reflected in the environment interfere with the infrared rays emitted by the wire itself, and bright spots are very likely to form on the infrared photo. These bright spots not only change the original temperature distribution characteristics of the image, making the local temperature appear to rise falsely, but also seriously interfere with the subsequent temperature judgment process based on image analysis, resulting in difficulty in accurately extracting the true temperature information of the wire, and there is a problem of low accuracy in detecting the wire temperature.

[0034] Based on the problems existing in the related art, in the embodiments of the present application, by using a deep learning model such as a wire temperature detection model to detect the temperature of the infrared image of the wire to be detected, the accuracy of wire temperature detection can be improved.

[0035] The following will specifically describe the wire temperature detection method provided by the embodiments of the present application in conjunction with specific embodiments.

[0036] Figure 1 It is a schematic flow of the wire temperature detection method provided by the embodiments of the present application Figure 1 As Figure 1 shown, the specific implementation manner of this wire temperature detection method may include the following steps:

[0037] S101, obtain an infrared image of the wire to be detected.

[0038] Exemplarily, the wire to be detected may be a wire used for transmitting electric energy in an actual application scenario.

[0039] Exemplarily, the infrared image of the wire to be detected can be collected by an infrared imager, or can be collected by an infrared spectral camera carried on a drone.

[0040] The present application does not limit the acquisition method of the infrared image of the wire to be detected, and it can be specifically determined according to actual application requirements.

[0041] S102, input the infrared image into the wire temperature detection model to perform wire temperature detection, and obtain the detection information output by the wire temperature detection model. This detection information includes the wire temperature of the wire to be detected.

[0042] Exemplarily, the wire temperature detection model may be an improved YOLOV5 network model.

[0043] Exemplarily, the wire temperature of the wire to be detected can be the highest temperature of the wire to be detected.

[0044] In the embodiment of the present application, by acquiring the infrared image of the wire to be detected, inputting the infrared image into the wire temperature detection model for wire temperature detection, and obtaining the detection information including the wire temperature of the wire to be detected output by the wire temperature detection model, the accuracy of wire temperature detection is improved.

[0045] Next, in conjunction with Figure 2 the model structure of the wire temperature detection model provided by the embodiment of the present application will be described in detail.

[0046] Figure 2 It is a schematic structural diagram of the wire temperature detection model provided by the embodiment of the present application. As Figure 2 shown, the wire temperature detection model includes a backbone network, a bottleneck network, and a detection network.

[0047] Among them, compared with the backbone network in the YOLOV5 network model, the convolution architecture in the backbone network provided by the embodiment of the present application can be a hybrid convolution architecture formed by combining a traditional convolution kernel and a depthwise separable convolution, and the convolution method of the backbone network can be a depthwise separable convolution. Specifically, when performing multi-scale basic feature extraction on the input infrared image through the backbone network, first perform convolution operations on each input channel separately through depth convolution, that is, channel-separated convolution, to reduce the amount of calculation, and then fuse the feature information between different channels through pointwise convolution to achieve the fusion of feature information between different channels. The backbone network provided by the embodiment of the present application can maintain the feature extraction ability while reducing parameters.

[0048] Exemplarily, the number of input channels can be 3. The present application does not limit the number of input channels, and it can be determined specifically according to actual application requirements.

[0049] Compared with the bottleneck network in the YOLOV5 network model, in the bottleneck network provided in the embodiments of the present application, the concat layer of the YOLOV5 network model is improved to a bidirectional feature pyramid network fusion layer (concat-bifpn). That is, the neck network provided in the embodiments of the present application can be a bidirectional feature pyramid network. The bidirectional feature pyramid network includes top-down and bottom-up bidirectional feature fusion paths. At the same time, a multi-scale feature fusion module is constructed in the neck network. The multi-scale feature fusion module is used to build a residual (shortcut) connection between feature layers of different depths to realize cross-layer flow and reuse of features. For example, through the multi-scale feature fusion module, the detailed information rich in the shallow feature map can be organically fused with the semantic information carried by the deep feature map, so that the wire temperature detection model can not only sensitively capture fine features such as wire edges and textures, but also understand the overall temperature distribution pattern of the wire, further improving the effectiveness of feature extraction. Specifically, when the neck network performs cross-scale feature correction fusion processing on feature maps of different scales output by the backbone network, based on the top-down path, the deep feature maps in the feature maps of different scales are gradually fused with the shallow feature maps in the feature maps of different scales through upsampling to transfer the semantic information in the feature maps; based on the bottom-up path, the shallow feature maps in the feature maps of different scales are gradually fused with the deep feature maps in the feature maps of different scales through downsampling to supplement the detailed information in the feature maps. After multiple repeated cross-scale feature correction fusion processes, multi-level and all-round fusion of features is achieved.

[0050] Exemplarily, the bidirectional feature pyramid network adopts a Bi-FPN network.

[0051] It should be noted that the number of Bi-FPN layers in the bidirectional feature pyramid network in the embodiments of the present application is not limited, and can be specifically determined according to actual application requirements.

[0052] Next, in conjunction with Figure 3 a specific implementation manner of inputting the infrared image into the wire temperature detection model to perform wire temperature detection and obtaining the detection information output by the wire temperature detection model in step S102 will be described in detail.

[0053] Figure 3 is a schematic flow chart of the wire temperature detection method provided in the embodiments of the present application Figure 2 . As Figure 3 shown, a specific implementation manner of inputting the infrared image into the wire temperature detection model to perform wire temperature detection and obtaining the detection information output by the wire temperature detection model may include the following steps:

[0054] S301. Input the infrared image into the backbone network for convolutional processing to obtain the first feature map output by the backbone network. The convolutional method of the backbone network is depthwise separable convolution, and the convolutional processing includes depthwise convolution processing and pointwise convolution processing.

[0055] In a possible implementation, input the infrared image into the backbone network. First, perform convolutional operations on the infrared image in each input channel respectively through depthwise convolution processing to reduce the computational amount. Then, fuse the feature information between different channels through pointwise convolution processing to achieve the fusion of feature information between different channels, and obtain the first feature map output by the backbone network.

[0056] Exemplarily, the first feature map may include multiple feature maps of different scales.

[0057] It can be understood that by performing convolutional processing on the input infrared image through the backbone network, the extraction of the basic features of the infrared image can be achieved.

[0058] S302. Input the first feature map into the bottleneck network for multi-scale feature fusion processing to obtain the second feature map output by the bottleneck network. The bottleneck network is a bidirectional feature pyramid network, and the multi-scale feature fusion processing includes upsampling processing, downsampling processing, and feature fusion processing.

[0059] In a possible implementation, based on the top-down fusion path, starting from the deep feature map in the input first feature map, perform upsampling layer by layer, and perform feature fusion processing with the shallow feature map in the first feature map. Then, based on the bottom-up fusion path, starting from the shallow feature map after feature fusion processing, perform downsampling layer by layer, and perform feature fusion processing with the deep feature map. Then, continue based on the top-down fusion path, starting from the deep feature map in the input first feature map, perform upsampling layer by layer, and perform feature fusion processing with the shallow feature map in the first feature map, and perform iterative loops in sequence.

[0060] Exemplarily, the number of iterative loops can be 3 times. The embodiments of the present application do not limit the number of iterative loops, and it can be determined specifically according to actual application requirements.

[0061] It can be understood that by based on the top-down fusion path, starting from the deep feature map in the input first feature map, performing upsampling layer by layer, and performing feature fusion with the shallow feature map in the first feature map, the transmission of semantic information in the first feature map can be achieved, such as the overall temperature distribution pattern of the wire to be detected, etc.; based on the bottom-up fusion path, starting from the shallow feature map after feature fusion processing, performing downsampling layer by layer, and performing feature fusion processing with the deep feature map, the supplement of detailed information in the first feature map can be achieved, such as the edges, textures and other fine features of the wire to be detected.

[0062] It is understandable that through feature fusion processing, the detailed information rich in the shallow feature map can be organically fused with the semantic information carried by the deep feature map, enabling the wire temperature detection model to not only keenly capture the subtle features such as the wire edge and texture of the wire to be detected, but also understand the overall temperature distribution pattern of the wire to be detected, further improving the effectiveness of feature extraction.

[0063] S303, input the second feature map into the detection network for wire temperature detection to obtain the detection information output by the detection network.

[0064] In the embodiment of the present application, by inputting the infrared image into the backbone network for depth convolution processing and pointwise convolution processing, the first feature map output by the backbone network is obtained. The first feature map is input into the bottleneck network for upsampling processing, downsampling processing, and feature fusion processing to obtain the second feature map output by the bottleneck network. Further, the second feature map is input into the detection network for wire temperature detection to obtain the detection information output by the detection network, improving the accuracy of wire temperature detection.

[0065] It is understandable that in the wire temperature detection method provided by the embodiment of the present application, by comprehensively and deeply improving the YOLOV5 network model, the accuracy of wire temperature detection is improved. On the one hand, in the wire temperature detection model provided by the embodiment of the present application, by adopting a hybrid convolution architecture and a multi-scale feature fusion module, the wire temperature detection model can accurately capture the key features of the wire temperature, such as the subtle temperature changes on the wire surface and the overall temperature distribution trend of the wire, thereby improving the detection accuracy of the wire temperature detection model and effectively reducing the false positive rate of the wire temperature, such as false high-temperature regions.

[0066] Optionally, the detection information further includes the target area positioning information of the wire to be detected.

[0067] Exemplarily, the target area positioning information may be the position information corresponding to the area where the wire temperature of the wire to be detected is located obtained by detection.

[0068] It is understandable that the temperature of the wire to be detected in this target area is relatively high, that is, by outputting the target area positioning information, it can provide accurate wire abnormal positions for power maintenance personnel and improve the processing efficiency of wire abnormalities.

[0069] The following combines Figure 4 to elaborate in detail on the training method of the wire temperature detection model provided by the embodiment of the present application.

[0070] Figure 4 is the flow schematic of the wire temperature detection method provided by the embodiment of the present application Figure 3 As Figure 4As shown in the figure, the training method of the wire temperature detection model in the wire temperature detection method includes the following steps:

[0071] S401, obtain an infrared image dataset.

[0072] Exemplarily, the infrared images included in the infrared image dataset can be infrared images of different wires under different simulated working conditions.

[0073] In a possible implementation manner, the infrared image dataset is obtained through the following method: collect infrared images of different wires under different simulated working conditions to obtain an original infrared image dataset; preprocess each original infrared image in the original infrared image dataset to obtain a target infrared image dataset; for each target infrared image in the target infrared image dataset, label the recognition area corresponding to the wire included in the target infrared image and the temperature corresponding to the wire to obtain the infrared image dataset.

[0074] In a possible implementation manner, the original infrared image dataset is obtained through the following method: simulate various environmental conditions close to the actual wire conditions of transmission lines in the laboratory, arrange wires of different specifications and materials, and set different temperature values of the wires through temperature control equipment. Use a high-resolution and high-sensitivity infrared imager to take pictures of the wires in different temperature states from multiple directions, angles, and distances to obtain a large number of original infrared images, and thus construct the original infrared image dataset.

[0075] Exemplarily, in the process of constructing the original infrared image dataset, it also includes diversified operations such as rotation, flipping, scaling, and adding noise to the original infrared images to simulate various possible actual working conditions of the wires.

[0076] Exemplarily, the preprocessing includes filtering processing, histogram equalization processing, and screening processing, etc.

[0077] It can be understood that through filtering processing, the noise in the infrared image can be removed; through histogram equalization processing, the contrast of the infrared image can be enhanced; through screening processing, the blurred and defective infrared images caused by factors such as equipment jitter and electromagnetic interference can be eliminated.

[0078] A possible implementation manner of labeling the recognition area corresponding to the wire included in the target infrared image and the temperature corresponding to the wire can be: based on a labeling tool, label the recognition area corresponding to the wire and the temperature corresponding to the wire on the infrared image.

[0079] Exemplarily, the temperature corresponding to the labeled wire can be the highest temperature of the dot line.

[0080] Exemplarily, the recognition area corresponding to the wire can be geometric information such as the size and shape of the area where the wire is located.

[0081] S402. Divide the infrared image dataset into a training set, a validation set, and a prediction set based on a preset ratio.

[0082] Exemplarily, the preset ratio can be 7:2:1.

[0083] The embodiments of the present application do not limit the preset ratio, and it can be specifically determined according to actual application requirements.

[0084] S403. Input the training set and the prediction set into the wire temperature detection model to be trained, perform model training based on the training set, and adjust the model parameters based on the prediction set to obtain a trained wire temperature detection model.

[0085] Exemplarily, before training the wire temperature detection model to be trained, first set the initial training parameters of the wire temperature detection model to be trained, such as parameters like the initial learning rate decay strategy, batch size dynamic adjustment mechanism, and number of training rounds.

[0086] In a possible implementation, input the training set and the prediction set into the wire temperature detection model to be trained at the same time, perform model training on the wire temperature detection model to be trained based on the training set, and at the same time, adjust the parameters of the wire temperature detection model during the training process based on the prediction set, loss function value, and evaluation metrics to obtain a trained wire temperature detection model.

[0087] Exemplarily, the loss function can be the hybrid loss function of the YOLOV5 network model or the mean square error loss function. The present application does not limit the type of the loss function.

[0088] S404. Input the validation set into the trained wire temperature detection model for model validation to obtain a wire temperature detection model.

[0089] In a possible implementation, input the validation set into the trained wire temperature detection model to obtain the detection results output by the trained wire temperature detection model, such as wire temperature prediction values and target area positioning information corresponding to the wire, and based on metrics such as Precision and Recall, evaluate the recognition ability of the wire temperature detection model for wire targets from the classification perspective and accuracy according to the detection results. Further, according to the evaluation results, deeply analyze the defects and deficiencies existing in the wire temperature detection model, such as overfitting, underfitting, insufficient feature extraction, etc., and adjust the model parameters, optimize the network structure, or supplement training data accordingly, and iterate and optimize repeatedly until the model performance metrics meet the requirements to obtain a wire temperature detection model.

[0090] It should be noted that in the wire temperature detection model provided by the embodiments of the present application, the detection network includes two different convolutions, which are respectively used to output the wire temperature category and the wire temperature accuracy. Correspondingly, the loss function of the wire temperature detection model includes a category loss function and an accuracy loss function to adjust the parameters of the wire temperature detection model.

[0091] In the embodiments of the present application, based on a preset ratio, the obtained infrared image dataset is divided into a training set, a validation set, and a prediction set. The training set and the prediction set are input into the wire temperature detection model to be trained. The model is trained based on the training set, and the model parameters are adjusted based on the prediction set to obtain a trained wire temperature detection model. Further, the validation set is input into the trained wire temperature detection model for model validation to obtain the wire temperature detection model, which can improve the accuracy and robustness of the model.

[0092] It can be understood that compared with the conventional deep learning models in the related art, for an object like a wire with a specific shape, size, and relatively complex temperature distribution characteristics, they lack the ability of targeted feature extraction and processing, and cannot fully mine the key information closely related to the wire temperature in the image. Therefore, when facing different lighting conditions, shooting angles, shooting distances, and wire working conditions changes, they show poor adaptability and robustness problems. On the one hand, the wire temperature detection model provided by the embodiments of the present application can accurately capture the key features of the wire temperature by using a hybrid convolution architecture and a multi-scale feature fusion module in the wire temperature detection model, improving the detection accuracy of the wire temperature; on the other hand, by obtaining infrared images under simulated different wire working conditions to train the wire temperature detection model, the accuracy and robustness of the model are improved.

[0093] Optionally, in the wire temperature detection method provided by the embodiments of the present application, before inputting the training set and the prediction set into the wire temperature detection model to be trained, it further includes: for each infrared image in the training set, clustering analysis is performed on the anchor box parameters of the wire temperature detection model to be trained according to the recognition area corresponding to the wire included in the infrared image and the temperature corresponding to the wire.

[0094] In a possible implementation, based on the k-means++ algorithm, clustering analysis is performed on the initial anchor box parameters of the wire temperature detection model to be trained according to the geometric information such as the size and shape of the wire region marked in the infrared images of the training set to determine the key attributes such as the size and aspect ratio of the anchor box, so that the design of the anchor box is more in line with the actual shape of the wire, improving the accuracy and efficiency of the wire temperature detection model to be trained in the target localization stage.

[0095] In summary, Figure 5This is a schematic structural diagram of the wire temperature detection system provided by the embodiments of the present application. As Figure 5 shown, the wire temperature detection system includes a data acquisition module, a model training module, and a wire temperature detection module.

[0096] Among them, in the data acquisition module, a high-precision and high-stability infrared imager is equipped to clearly capture the tiny temperature differences on the wire surface, ensuring that accurate infrared images can be obtained immediately under different ambient temperatures. By installing the infrared imager on a robotic arm or a pan-tilt head that can be flexibly adjusted in angle and orientation, it is convenient to take pictures of the wire from various directions according to experimental or actual detection requirements. The infrared imager closely cooperates with the temperature control device and the data transmission unit to realize setting different wire temperatures while synchronously collecting the corresponding infrared images and quickly and stably transmitting the data to the subsequent processing module.

[0097] The model training module, by integrating mainstream deep learning frameworks such as PyTorch or TensorFlow, provides powerful computing power support and a rich toolset for model training. Receiving the training set data from the data preprocessing module, combined with preset training parameters such as the learning rate decay strategy and the batch size dynamic adjustment mechanism, conducts multi-round and high-intensity training on the wire temperature detection model to be trained. During the training process, the change of the loss function value and the evaluation index of the model is monitored in real time, and the training strategy is adjusted in time according to the feedback information to ensure the accuracy of the model.

[0098] The wire temperature detection module, by collecting the infrared image of the wire to be detected, quickly inputs it into the wire temperature detection model that has been fully trained and optimized, obtains the wire temperature detection result output by the wire temperature detection model, and further feeds back the detection result to the power operation and maintenance personnel, facilitating the power operation and maintenance personnel to quickly and intuitively understand the temperature state of the wire and make operation and maintenance decisions in time.

[0099] In summary, the wire temperature detection method provided by the embodiments of the present application has the following beneficial effects:

[0100] 1) By comprehensively and deeply improving the YOLOv5 network model, the accuracy of wire temperature detection is improved. In terms of feature layer extraction optimization, the wire temperature detection model adopts a hybrid convolution architecture and a multi-scale feature fusion module, enabling the wire temperature detection model to accurately capture the key features of the wire temperature. Whether it is the subtle temperature changes on the wire surface or the overall temperature distribution trend of the wire, they are effectively extracted and utilized. Further, it can help power workers accurately distinguish the true heating area and the false high-temperature area of the wire, improve the detection accuracy of the wire temperature, and effectively reduce the misjudgment rate of the wire temperature.

[0101] 2) Based on data augmentation techniques, through diverse operations such as rotating, flipping, scaling, and adding noise to the original image, various actual working conditions of possible wires are simulated, enabling the wire temperature detection model to train on more types of infrared image scenarios during the training process. As a result, the wire temperature detection model can be applicable to different infrared image scenarios such as different lighting conditions, shooting angles, and changes in wire working conditions, and then stably output reliable detection results, enhancing the generalization ability and robustness of the wire temperature detection model. At the same time, through the classification loss function and precision loss function, it is ensured that the model parameters of the wire temperature detection model are adjusted to the optimal in the two tasks of temperature prediction and target classification, and the optimization of the anchor box parameters can make the wire temperature detection model more accurate and efficient in locating the wire target, further improving the reliability of the entire system.

[0102] 3) This method and system can achieve rapid and accurate detection of wire temperature, providing strong technical support for the safe operation of the power system. In practical applications, it can monitor the wire temperature in real time, promptly discover potential overheating hazards of the wire, provide accurate temperature status information and decision-making basis for power operation and maintenance personnel, help take preventive measures in advance, and avoid line failures and power outages caused by wire overheating, having high practical value and broad promotion prospects.

[0103] The following is the device embodiment of the present application, which can be used to execute the method embodiment of the present application. For the details not disclosed in the device embodiment of the present application, please refer to the method embodiment of the present application.

[0104] Figure 6 It is a schematic structural diagram of the wire temperature detection device provided by the embodiment of the present application. As Figure 6 shown, the wire temperature detection device 60 includes: an acquisition module 610 and a detection module 620.

[0105] Among them, the acquisition module 610 is used to acquire the infrared image of the wire to be detected;

[0106] The detection module 620 is used to input the infrared image into the wire temperature detection model for wire temperature detection, and obtain the detection information output by the wire temperature detection model, and the detection information includes the wire temperature of the wire to be detected.

[0107] In a possible implementation, the wire temperature detection model includes a backbone network, a bottleneck network, and a detection network. The detection module 620 is specifically configured to: input an infrared image into the backbone network for convolutional processing to obtain a first feature map output by the backbone network. The convolutional method of the backbone network is depthwise separable convolution, and the convolutional processing includes depth convolution processing and pointwise convolution processing; input the first feature map into the bottleneck network for multi-scale feature fusion processing to obtain a second feature map output by the bottleneck network. The bottleneck network is a bidirectional feature pyramid network, and the multi-scale feature fusion processing includes upsampling processing, downsampling processing, and feature fusion processing; input the second feature map into the detection network for wire temperature detection to obtain detection information output by the detection network.

[0108] In a possible implementation, the detection information further includes target area localization information of the wire to be detected.

[0109] In a possible implementation, the wire temperature detection model is trained in the following manner: obtain an infrared image dataset; based on a preset ratio, divide the infrared image dataset into a training set, a validation set, and a prediction set; input the training set and the prediction set into the wire temperature detection model to be trained, and perform model training based on the training set and adjust model parameters based on the prediction set to obtain a trained wire temperature detection model; input the validation set into the trained wire temperature detection model for model validation to obtain the wire temperature detection model.

[0110] In a possible implementation, the infrared image dataset is obtained in the following manner: collect infrared images of different wires under different simulated working conditions to obtain an original infrared image dataset; preprocess each original infrared image in the original infrared image dataset to obtain a target infrared image dataset; for each target infrared image in the target infrared image dataset, label the recognition area corresponding to the wire included in the target infrared image and the temperature corresponding to the wire to obtain the infrared image dataset.

[0111] In a possible implementation, the wire temperature detection device further includes a clustering analysis module (not shown). The clustering analysis module is configured to, before inputting the training set and the prediction set into the wire temperature detection model to be trained, perform clustering analysis on the anchor box parameters of the wire temperature detection model to be trained according to the recognition area corresponding to the wire included in each infrared image in the training set and the temperature corresponding to the wire.

[0112] The wire temperature detection device provided in this embodiment can execute the method provided in the above method embodiment, and its implementation principle and technical effects are similar, which will not be elaborated here in this embodiment.

[0113] Figure 7 This is a schematic structural diagram of an electronic device provided in an embodiment of the present application. AsFigure 7 As shown in the figure, the electronic device 70 provided in this embodiment includes at least one processor 701 and a memory 702. Optionally, the device 70 further includes a communication component 703. Among them, the processor 701, the memory 702, and the communication component 703 are connected through a bus 704.

[0114] In a specific implementation process, at least one processor 701 executes computer-executable instructions stored in the memory 702, so that at least one processor 701 executes the above-mentioned method.

[0115] For the specific implementation process of the processor 701, reference can be made to the above method embodiment. The implementation principle and technical effect are similar, and will not be elaborated here in this embodiment.

[0116] In the above embodiment, it should be understood that the processor may be a central processing unit (Central Processing Unit, abbreviated as CPU), or other general-purpose processors, digital signal processors (Digital Signal Processor, abbreviated as DSP), application-specific integrated circuits (Application Specific Integrated Circuit, abbreviated as ASIC), etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the invention can be directly embodied as being executed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0117] The memory may include a high-speed memory (Random Access Memory, abbreviated as RAM), and may also include a non-volatile memory (Non-volatile Memory, abbreviated as NVM), such as at least one disk memory.

[0118] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of this application is not limited to only one bus or one type of bus.

[0119] This embodiment of the present application also provides a computer program product, including a computer program, which implements the above-mentioned method when executed by a processor.

[0120] An embodiment of the present application further provides a computer-readable storage medium, in which computer-executable instructions are stored. When the processor executes the computer-executable instructions, the above-mentioned method is implemented.

[0121] The above-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk or an optical disc. The readable storage medium can be any available medium accessible by a general-purpose or special-purpose computer.

[0122] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in a device.

[0123] The division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces, and the indirect coupling or communication connection of devices or units can be in an electrical, mechanical or other form.

[0124] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0125] In addition, in each embodiment of the present invention, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0126] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0127] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When this program is executed, it executes the steps including the above method embodiments; and the aforementioned storage medium includes: ROMs, RAMs, magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0128] Finally, it should be noted that: After considering the specification and practicing the invention disclosed herein, those skilled in the art will easily think of other implementation manners of the present invention. The present invention is intended to cover any variations, uses, or adaptive changes of the present invention. These variations, uses, or adaptive changes follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field of the present invention that are not disclosed in the present invention. It is not limited to the exact structures described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present invention is only limited by the appended claims.

Claims

1. A method for detecting wire temperature, characterized in that: include: Acquire an infrared image of the wire to be inspected; The infrared image is input into a wire temperature detection model to detect the wire temperature, and detection information output by the wire temperature detection model is obtained, wherein the detection information includes the wire temperature of the wire to be detected.

2. The wire temperature detection method according to claim 1, characterized in that: The wire temperature detection model comprises a backbone network, a bottleneck network and a detection network. The infrared image is input into the wire temperature detection model to detect the wire temperature, and the detection information output by the wire temperature detection model is obtained, including: Inputting the infrared image into the backbone network for convolution processing to obtain a first feature map output by the backbone network, wherein the convolution mode of the backbone network is depthwise separable convolution, and the convolution processing includes depthwise convolution processing and pointwise convolution processing; Inputting the first feature map into the bottleneck network for multi-scale feature fusion processing to obtain a second feature map output by the bottleneck network, wherein the bottleneck network is a bidirectional feature pyramid network, and the multi-scale feature fusion processing includes upsampling processing, downsampling processing and feature fusion processing; The second characteristic graph is input into the detection network to detect the temperature of the conductor, and the detection information output by the detection network is obtained.

3. The wire temperature detection method according to claim 1, characterized in that: The detection information also includes target area positioning information of the wire to be detected.

4. The wire temperature detection method according to any one of claims 1 to 3, characterized in that: The wire temperature detection model is trained in the following way: Obtain infrared image dataset; Based on a preset ratio, the infrared image data set is divided into a training set, a validation set and a prediction set; Inputting the training set and the prediction set into the wire temperature detection model to be trained, performing model training based on the training set, adjusting model parameters based on the prediction set, and obtaining a trained wire temperature detection model; The verification set is input into the trained wire temperature detection model for model verification to obtain the wire temperature detection model.

5. The wire temperature detection method according to claim 4, characterized in that: The infrared image dataset is obtained in the following way: Collect infrared images of different conductors under different simulated working conditions to obtain original infrared image data sets; Preprocessing each original infrared image in the original infrared image dataset to obtain a target infrared image dataset; For each target infrared image in the target infrared image data set, an identification area corresponding to the wire contained in the target infrared image and a temperature corresponding to the wire are marked to obtain the infrared image data set.

6. The wire temperature detection method according to claim 5, characterized in that: Before inputting the training set and the prediction set into the wire temperature detection model to be trained, the method further includes: For each infrared image in the training set, cluster analysis is performed on anchor frame parameters of the wire temperature detection model to be trained according to the identification area corresponding to the wire contained in the infrared image and the temperature corresponding to the wire.

7. A conductor temperature detection device, characterized in that: include: An acquisition module, used for acquiring an infrared image of the wire to be detected; The detection module is used to input the infrared image into a wire temperature detection model to perform wire temperature detection, and obtain detection information output by the wire temperature detection model, wherein the detection information includes the wire temperature of the wire to be detected.

8. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 6 when executed by a processor.

10. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 6 when being executed by a processor.

Images