Power distribution cabinet air switch identification method and device, computer equipment, computer readable storage medium and computer program product

Abstract

The invention relates to a power distribution cabinet air switch identification method and device, computer equipment, a computer readable storage medium and a computer program product, and relates to the field of power distribution devices. The method comprises the following steps: performing target detection processing on an image of the power distribution cabinet to obtain a switch detection result of each air switch and a label detection result of each category label; the switch detection result comprises a switch on-off state and switch position information; the label detection result comprises a label image and label position information; performing character recognition processing on each label image to obtain character information in each category label; determining a category label corresponding to each air switch based on each piece of switch position information and each piece of label position information; and determining the on-off state of each air switch and the character information in the corresponding category label as an air switch identification result of the power distribution cabinet. By adopting the method, the monitoring efficiency of the on-off state of the air switch can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of power distribution devices, and in particular to a method, device, computer equipment, computer-readable storage medium and computer program product for identifying an air switch in a power distribution cabinet. Background Art

[0002] The power distribution cabinet is a power distribution device that connects the power supply and the load. The air switch configured in the power distribution cabinet plays a key role in circuit protection, fault isolation, and equipment maintenance. Therefore, it is necessary to monitor the switch status of the air switch configured in the power distribution cabinet in a timely manner.

[0003] In the related technology, monitoring personnel usually preliminarily judge the switch opening and closing status of the air switch based on the position information uploaded by the on-site equipment through the auxiliary contact of the secondary circuit, and then arrive at the equipment site to manually check the switch opening and closing status to obtain the actual switch opening and closing status of the air switch. Therefore, the efficiency of monitoring the switch opening and closing status of the air switch is low. Summary of the invention

[0004] Based on this, it is necessary to provide an identification method, device, computer equipment, computer-readable storage medium and computer program product for the air switch of a distribution cabinet that can improve the monitoring efficiency in order to address the technical problem of low efficiency in monitoring the opening and closing status of the above-mentioned switches.

[0005] In a first aspect, the present application provides a method for identifying an air switch of a power distribution cabinet, comprising:

[0006] Acquire an image of a power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch;

[0007] Performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label;

[0008] Performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels;

[0009] Based on the switch position information of each air switch and the label position information of each category label, matching processing is performed on each air switch and each category label to determine the category label corresponding to each air switch;

[0010] The switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet.

[0011] In one embodiment, before performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels, the method further includes:

[0012] Performing image super-resolution processing on the label image of each category label based on the pre-trained image super-resolution model to obtain the super-resolution image of each category label; the resolution of the super-resolution image is higher than the resolution of the label image;

[0013] The pre-trained text recognition model is used to perform text recognition processing on the label images of the respective category labels to obtain text information in the respective category labels, including:

[0014] Based on the text recognition model, text recognition processing is performed on the super-resolved images of each category label to obtain the text information in each category label.

[0015] In one embodiment, the image super-resolution model obtained by pre-training performs image super-resolution processing on the label image of each category label to obtain the super-resolution image of each category label, including:

[0016] For each category label, taking the label image of the category label as the current first label image;

[0017] Extracting image features of the first label image based on the deep image prior network in the image super-resolution model;

[0018] Based on the dynamic kernel prior network in the image super-resolution model and the image features, generating a dynamic convolution kernel corresponding to the first label image;

[0019] Based on the dynamic convolution kernel, the first label image is convolved to obtain a second label image, and the second label image is used as a new first label image, and the step of extracting image features of the first label image based on the deep image prior network in the image super-resolution model is returned until the obtained second label image meets a preset iteration condition, and the second label image that meets the preset iteration condition is determined as the super-resolution image of the category label.

[0020] In one embodiment, the matching processing of each air switch and each category label based on the switch position information of each air switch and the label position information of each category label to determine the category label corresponding to each air switch includes:

[0021] Obtaining the arrangement attribute of the power distribution cabinet; the arrangement attribute is used to characterize the arrangement correspondence between the air switches configured in the power distribution cabinet and the corresponding category labels;

[0022] For each air switch, according to the arrangement attribute, determining the matching degree between the switch position information of the air switch and the label position information of each category label;

[0023] Among the various category labels, the category label with the highest matching degree is determined as the category label corresponding to the air switch.

[0024] In one embodiment, after determining the switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the power distribution cabinet, it also includes:

[0025] When it is detected that the switch opening and closing state of any air switch is in the open state, based on the switch opening and closing state of the air switch and the text information in the category label corresponding to the air switch, generate warning information for the air switch;

[0026] The warning information is sent to the monitoring terminal corresponding to the power distribution cabinet.

[0027] In one embodiment, the pre-trained target detection model is trained by the following method:

[0028] Collecting a plurality of sample images of the power distribution cabinet, and performing data enhancement processing on the plurality of sample images to obtain a plurality of sample images after data enhancement processing;

[0029] Annotating the switch opening and closing state and switch position information of the air switch in each sample image after data enhancement processing, annotating the label position information of the category label in each sample image after data enhancement processing, and dividing the multiple sample images after data enhancement processing into a training set and a verification set;

[0030] The target detection model to be trained is trained based on the training set and the verification set to obtain a target detection model that meets preset training conditions, which is used as the pre-trained target detection model.

[0031] In a second aspect, the present application also provides an identification device for an air switch of a distribution cabinet, comprising:

[0032] An image acquisition module is used to acquire an image of the power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch;

[0033] The target detection module is used to perform target detection processing on the image based on the pre-trained target detection model to obtain the switch detection results of each air switch and the label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label;

[0034] A text recognition module, used to perform text recognition processing on the label images of the respective category labels based on a pre-trained text recognition model to obtain text information in the respective category labels;

[0035] A position matching module, used to match each air switch with each category label based on the switch position information of each air switch and the label position information of each category label, and determine the category label corresponding to each air switch;

[0036] The status monitoring module is used to determine the switch opening and closing status of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the distribution cabinet, and obtain the air switch identification result of the distribution cabinet.

[0037] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0038] Acquire an image of a power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch;

[0039] Performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label;

[0040] Performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels;

[0041] Based on the switch position information of each air switch and the label position information of each category label, matching processing is performed on each air switch and each category label to determine the category label corresponding to each air switch;

[0042] The switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet.

[0043] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0044] Acquire an image of a power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch;

[0045] Performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label;

[0046] Performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels;

[0047] Based on the switch position information of each air switch and the label position information of each category label, matching processing is performed on each air switch and each category label to determine the category label corresponding to each air switch;

[0048] The switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet.

[0049] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0050] Acquire an image of a power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch;

[0051] Performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label;

[0052] Performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels;

[0053] Based on the switch position information of each air switch and the label position information of each category label, matching processing is performed on each air switch and each category label to determine the category label corresponding to each air switch;

[0054] The switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet.

[0055] The above-mentioned method, device, computer equipment, computer-readable storage medium and computer program product for identifying the air switch of the distribution cabinet first obtain an image of the distribution cabinet; the image contains multiple air switches configured in the distribution cabinet and the category label of each air switch; then, based on the pre-trained target detection model, the image is subjected to target detection processing to obtain the switch detection results of each air switch and the label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label; then, based on the pre-trained text recognition model, the label image of each category label is subjected to text recognition processing to obtain the text information in each category label; then, based on the switch position information of each air switch and the label position information of each category label, each air switch and each category label are matched to determine the category label corresponding to each air switch; finally, the switch opening and closing state of each air switch and the text information in the category label corresponding to each air switch are determined as the identification result of the air switch of the distribution cabinet. In this way, based on the target detection model, the switch opening and closing status and switch position information of each air switch, as well as the label image and label position information of each category label can be obtained. Based on the text recognition model, the text information in each category label can be obtained. Based on the switch position information of each air switch and the label position information of each category label, each air switch and each category label can be matched to determine the category label corresponding to each air switch, so that the air switch recognition result of the distribution cabinet can be obtained according to the switch opening and closing status of each air switch and the text information in the corresponding category label. The air switch recognition result obtained based on the above process can accurately monitor the switch opening and closing status of each air switch in the distribution cabinet remotely, thereby improving the monitoring efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0057] FIG1 is a schematic flow chart of a method for identifying an air switch of a power distribution cabinet in one embodiment;

[0058] FIG2 is a schematic flow chart of a method for identifying an air switch of a power distribution cabinet in another embodiment;

[0059] FIG3 is a flow chart showing steps of performing image super-resolution processing on a label image of each category label to obtain a super-resolution image of each category label in one embodiment;

[0060] FIG4 is a flowchart of the steps of matching each air switch with each category label to determine the category label corresponding to each air switch in one embodiment;

[0061] FIG5 is a schematic flow chart of the steps of providing an early warning to an air switch in one embodiment;

[0062] FIG6 is a schematic flow chart of a method for identifying an air switch of a power distribution cabinet in yet another embodiment;

[0063] FIG7 is a flow chart of a method for identifying the state and type of an air switch in a power distribution cabinet in one embodiment;

[0064] FIG8 is a structural block diagram of an identification device for an air switch of a power distribution cabinet in one embodiment;

[0065] FIG. 9 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0067] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0068] In one embodiment, as shown in FIG1 , a method for identifying an air switch in a power distribution cabinet is provided. This embodiment uses the method applied to a server as an example for illustration. It is understandable that the method can also be applied to a terminal, and can also be applied to a system including a server and a terminal, and is implemented through the interaction between the server and the terminal; wherein the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services; the terminal can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, etc. In this embodiment, the method includes the following steps:

[0069] Step S102, obtaining an image of the power distribution cabinet.

[0070] Among them, multiple air switches and a category label of each air switch are configured in the distribution cabinet, and the image contains multiple air switches configured in the distribution cabinet and a category label of each air switch.

[0071] Among them, the category label contains text content, and the text content of the category label is used to characterize the power type of the power supply to which the corresponding air switch belongs, such as AC power supply, battery power supply, communication power supply, protection device power supply, operation power supply and motor power supply, etc.

[0072] Furthermore, if the power distribution cabinet is a distribution panel cabinet, the image of the power distribution cabinet may be an image obtained by an image acquisition device installed outside the distribution panel cabinet through the cabinet door of the distribution panel cabinet, photographing the air switch and the category label. If the power distribution cabinet is not a distribution panel cabinet, the image of the power distribution cabinet may be an image obtained by an image acquisition device installed inside the distribution box directly photographing the air switch and the category label.

[0073] Specifically, the server obtains an image of the power distribution cabinet taken by an image acquisition device, wherein the image contains multiple air switches configured in the power distribution cabinet and a category label of each air switch.

[0074] Step S104, performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label.

[0075] The pre-trained target detection model uses the open / closed state and position of the air switch, as well as the image and position of the category label as detection targets. In a specific application, the pre-trained target detection model is a target detection model based on the YOLO algorithm (You Only Look Once, a real-time target detection algorithm). Preferably, the pre-trained target detection model uses the YOLOv8 model.

[0076] The switch detection result of each air switch includes at least the switch opening and closing state and switch position information of the air switch. The switch opening and closing state is used to characterize whether the air switch is in a closed state or an open state. The switch position information is used to characterize the position of the air switch in the image. In practical applications, the coordinates of the air switch output by the target detection model can be used to represent the switch position information.

[0077] The label detection result of each category label includes at least the label image and label position information of the category label. The label image is the image of the area where the category label is located in the image of the distribution cabinet. The label position information is used to characterize the position of the category label in the image. In practical applications, the coordinates of the category label output by the target detection model can be used to represent the label position information.

[0078] Specifically, the server inputs the image of the distribution cabinet into the pre-trained YOLOv8 target detection model. The YOLOv8 target detection model outputs the switch opening and closing status and switch position information of each air switch as the switch detection result of each air switch through target detection of the image, and outputs the label image and label position information of each category label as the label detection result of each category label.

[0079] In this step, the server can identify the opening and closing status of the air switch through the switch opening and closing status output by the target detection model.

[0080] Step S106, performing text recognition processing on the label images of each category label based on the pre-trained text recognition model to obtain the text information in each category label.

[0081] The pre-trained text recognition model is used to recognize text content in the input image. In a specific application, the pre-trained text recognition model is a target detection model based on an OCR algorithm (Optical Character Recognition). Preferably, the pre-trained text recognition model uses a PP-OCRv4 model.

[0082] Specifically, the server inputs the label images of each category label into the pre-trained PP-OCRv4 text recognition model, and the PP-OCRv4 text recognition model outputs the text information in each category label by recognizing the text in each label image.

[0083] Step S108, based on the switch position information of each air switch and the tag position information of each category tag, matching processing is performed on each air switch and each category tag to determine the category tag corresponding to each air switch.

[0084] Specifically, since there are multiple air switches and multiple category labels in the image, in order to determine which air switch a certain category label corresponds to, the server also needs to match each air switch and each category label according to the switch position information of each air switch and the label position information of each category label, so as to determine the correspondence between the air switch and each category label, and then determine the category label corresponding to each air switch; based on the category label corresponding to each air switch, the server can determine the power type of the power supply to which each air switch belongs.

[0085] For example, assuming that after matching the air switch and the category tag, the server determines that air switch A1 matches category tag B1, then the server determines that there is a corresponding relationship between air switch A1 and category tag B1, that is, the category tag corresponding to air switch A1 is category tag B1; if the text information in category tag B1 is "battery power supply", then the server can determine that the power supply corresponding to air switch A1 is a battery power supply.

[0086] Step S110, determining the switch opening and closing status of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the distribution cabinet.

[0087] The air switch identification result includes the switch opening and closing state of each air switch and the power type corresponding to the air switch.

[0088] Specifically, the server obtains the air switch identification result of the distribution cabinet based on the switch opening and closing status of each air switch and the text information in the class label corresponding to the air switch.

[0089] In the above-mentioned state monitoring method of the air switch of the distribution cabinet, first, an image of the distribution cabinet is obtained; the image contains multiple air switches configured in the distribution cabinet and the category label of each air switch; then, based on the pre-trained target detection model, the image is subjected to target detection processing to obtain the switch detection results of each air switch and the label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and the switch position information of the air switch; the label detection result of each category label at least includes the label image and the label position information of the category label; then, based on the pre-trained text recognition model, the label image of each category label is subjected to text recognition processing to obtain the text information in each category label; then, based on the switch position information of each air switch and the label position information of each category label, each air switch and each category label are matched to determine the category label corresponding to each air switch; finally, the switch opening and closing state of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch recognition result of the distribution cabinet. In this way, based on the target detection model, the switch opening and closing status and switch position information of each air switch, as well as the label image and label position information of each category label can be obtained. Based on the text recognition model, the text information in each category label can be obtained. Based on the switch position information of each air switch and the label position information of each category label, each air switch and each category label can be matched to determine the category label corresponding to each air switch, so that the air switch identification result of the distribution cabinet can be obtained according to the switch opening and closing status of each air switch and the text information in the corresponding category label. Based on the above process, the switch opening and closing status of the air switch of the distribution cabinet can be accurately monitored remotely, thereby improving the monitoring efficiency.

[0090] In an exemplary embodiment, in the above step S106, before text recognition processing is performed on the label image of each category label based on the pre-trained text recognition model to obtain the text information in each category label, the following content is specifically included: image super-resolution processing is performed on the label image of each category label based on the pre-trained image super-resolution model to obtain a super-resolution image of each category label; the resolution of the super-resolution image is higher than the resolution of the label image.

[0091] The pre-trained image super-resolution model is used to reconstruct the input image and improve the resolution of the input image. In a specific application, the pre-trained image super-resolution model is an image super-resolution model based on deep learning. Preferably, the pre-trained image super-resolution model adopts a DIP (Deep Image Prior)-DKP (dynamic kernel prior) image super-resolution model.

[0092] Among them, image super-resolution processing refers to the process of reconstructing the input image to improve the resolution of the input image.

[0093] It should be noted that in a distribution cabinet, the category label is usually small, and the image of the distribution cabinet is obtained by taking pictures of the image acquisition device. In particular, for a distribution panel cabinet, the image of the distribution cabinet is taken through the cabinet door of the distribution panel cabinet, which results in the label image quality being usually low, that is, the resolution of the label image is usually low. Therefore, in order to improve the text recognition effect of the label image, it is necessary to first improve the resolution of the label image to increase the clarity of the text content in the label image before performing text recognition.

[0094] Specifically, for each category label, the server inputs the label image of the category label into the pre-trained DIP-DKP image super-resolution model. The DIP-DKP image super-resolution model outputs an image with higher resolution by super-resolution processing of the label image, thereby obtaining a super-resolution image of the category label.

[0095] The above step S108, based on the pre-trained text recognition model, performs text recognition processing on the label image of each category label to obtain the text information in each category label, specifically includes the following contents: based on the text recognition model, performs text recognition processing on the super-resolution image of each category label to obtain the text information in each category label.

[0096] Specifically, after performing image super-resolution processing on the label images of the category labels, the server inputs the super-resolution images of each category label into the pre-trained PP-OCRv4 text recognition model. The PP-OCRv4 text recognition model outputs the text information in each category label by recognizing the text in each super-resolution image.

[0097] Therefore, as shown in FIG. 2 , this embodiment further provides a method for identifying an air switch in a power distribution cabinet, the method comprising the following steps:

[0098] Step S202, obtaining an image of the power distribution cabinet.

[0099] Step S204, performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label.

[0100] Step S206, performing image super-resolution processing on the label image of each category label based on the pre-trained image super-resolution model to obtain a super-resolution image of each category label.

[0101] Step S208, performing text recognition processing on the super-resolved images of each category label based on the text recognition model to obtain the text information in each category label.

[0102] Step S210, based on the switch position information of each air switch and the tag position information of each category tag, matching processing is performed on each air switch and each category tag to determine the category tag corresponding to each air switch.

[0103] Step S212, determining the switch opening and closing status of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the distribution cabinet.

[0104] In this embodiment, the server can improve the resolution of the label image by super-resolution processing of the label image, and can improve the accuracy of text recognition by performing text recognition on the super-resolution image with higher resolution, thereby improving the accuracy of air switch recognition and the efficiency of monitoring the switch opening and closing status.

[0105] In an exemplary embodiment, as shown in FIG3 , the above step S206, performing image super-resolution processing on the label image of each category label based on the pre-trained image super-resolution model to obtain the super-resolution image of each category label, specifically includes the following steps:

[0106] Step S302: for each category label, use the label image of the category label as the current first label image.

[0107] Step S304: extracting image features of the first label image based on the deep image prior network in the image super-resolution model.

[0108] Step S306, generating a dynamic convolution kernel corresponding to the first label image based on the dynamic kernel prior network and image features in the image super-resolution model.

[0109] Step S308, based on the dynamic convolution kernel, convolution processing is performed on the first label image to obtain a second label image, and the second label image is used as the new first label image, and the step of extracting image features of the first label image based on the deep image prior network in the image super-resolution model is returned until the obtained second label image meets the preset iteration conditions, and the second label image that meets the preset iteration conditions is determined as the super-resolution image of the category label.

[0110] Among them, the image super-resolution model at least includes a deep image prior network, a dynamic kernel prior network and a backbone network.

[0111] Among them, the deep image prior network is constructed based on the DIP deep image prior algorithm and is used to extract the deep features of the input image.

[0112] Among them, the dynamic kernel prior network is constructed based on the DKP dynamic kernel prior algorithm, which is used to dynamically generate corresponding dynamic convolution kernels based on deep features.

[0113] The backbone network is constructed based on the degradation model of the blind super-resolution task. In specific applications, the goal of the blind super-resolution task in the backbone network is to restore the corresponding high-resolution image based only on the given low-resolution image without knowing the specific parameters of the image degradation process (including blur kernel, downsampling processing and additional noise, etc.).

[0114] Specifically, for each category label, the server uses the label image of the category label as the current first label image, extracts the deep features of the first label image through a deep image prior network, and obtains the image features of the first label image; then, the server generates a dynamic convolution kernel of the first label image according to the image features of the first label image through a dynamic kernel prior network; then, the server performs convolution processing on the first label image according to the dynamic convolution kernel through a backbone network to obtain a second label image, and returns the second label image as a new first label image to step S304, and iteratively convolves the label image based on back propagation until the obtained second label image meets a preset iteration condition, and determines the second label image that meets the preset iteration condition as the super-resolution image of the category label.

[0115] In specific applications, the dynamic kernel prior network specifically includes a random kernel sampling module (RKS, random kernel sampling) and a prior kernel estimation module (PKE, prior kernel estimation). Among them, the random kernel sampling module is used to generate a prior kernel in the fuzzy kernel distribution through Markov chain Monte Carlo sampling, and the prior kernel estimation module is used to guide the update of kernel estimation network parameters based on the prior kernel generated by the random kernel sampling module.

[0116] The server replaces the kernel prior network in the existing DIP framework with the above-mentioned dynamic kernel prior network to obtain the DIP-DKP image super-resolution model.

[0117] In this embodiment, since the DIP-DKP image super-resolution model does not require the cost of data labeling and supervised training, the server can restore high-resolution images in a plug-and-play manner based on the DIP-DKP image super-resolution model, reducing the cost of image super-resolution processing.

[0118] In an exemplary embodiment, as shown in FIG. 4 , the above step S108, based on the switch position information of each air switch and the label position information of each category label, performs matching processing on each air switch and each category label to determine the category label corresponding to each air switch, specifically includes the following steps:

[0119] Step S402, obtaining the layout attributes of the power distribution cabinet.

[0120] Step S404: for each air switch, according to the arrangement attribute, determine the matching degree between the switch position information of the air switch and the label position information of each category label.

[0121] Step S406: Determine the category label with the highest matching degree among the category labels as the category label corresponding to the air switch.

[0122] The power distribution cabinet is preset with an arrangement attribute, which is used to characterize the arrangement correspondence between the air switches configured in the power distribution cabinet and the corresponding category labels.

[0123] Among them, the arrangement correspondence between the air switch and the corresponding category label refers to whether the category label corresponding to the air switch is arranged above, below, on the left or on the right of the air switch in the distribution cabinet.

[0124] Specifically, the server obtains the preset layout attributes of the distribution cabinet; and for each air switch, according to the layout attributes, determines the matching degree between the switch position information of the air switch and the label position information of each category label, and determines the category label with the highest matching degree as the category label corresponding to the air switch.

[0125] For example, assuming that in the distribution cabinet, the category label corresponding to the air switch is arranged below the air switch; then, for each air switch, the server determines the probability that each category label is located below the air switch based on the switch position information of the air switch and the label position information of each category label, as the matching degree between the category label and the air switch, and determines the category label with the highest probability of being located below the air switch as the category label corresponding to the air switch.

[0126] In this embodiment, the server can match each air switch with each category tag based on the switch position information of each air switch and the label position information of each category tag, determine the category label corresponding to each air switch, thereby realizing the category identification of the air switch.

[0127] In an exemplary embodiment, as shown in FIG5 , in the above step S110, after the switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet, the following steps are specifically included to realize early warning of the air switch:

[0128] Step S502, when it is detected that the switch opening and closing state of any air switch is in the open state, based on the switch opening and closing state of the air switch and the text information in the category label corresponding to the air switch, generate warning information for the air switch.

[0129] Step S504: Send the warning information to the monitoring terminal corresponding to the power distribution cabinet.

[0130] Among them, the monitoring terminal is the terminal used by the monitoring personnel of the distribution cabinet.

[0131] Specifically, after the server obtains the identification result of the air switch of the distribution cabinet, if it monitors that the switch opening and closing state of any air switch is in the disconnected state, it will immediately generate an early warning message for the air switch based on the switch opening and closing state of the air switch and the text information in the corresponding category label, and send the early warning message to the monitoring terminal corresponding to the distribution cabinet, so that the monitoring personnel of the distribution cabinet can know that the air switch has been disconnected, and then perform corresponding operation and maintenance processing on the air switch and its corresponding power supply in time.

[0132] In this embodiment, the server can promptly notify the monitoring personnel by generating and sending warning information so that the monitoring personnel can promptly perform corresponding operation and maintenance processing on the air switch and its corresponding power supply, thereby ensuring the stable operation of the distribution system where the distribution cabinet is located.

[0133] In an exemplary embodiment, the pre-trained target detection model in any of the above embodiments is trained by the following method:

[0134] A plurality of sample images of the power distribution cabinet are collected, and data enhancement processing is performed on the plurality of sample images to obtain a plurality of sample images after data enhancement processing.

[0135] Specifically, the server pre-collects multiple sample images of the power distribution cabinet through an image acquisition device, and expands the multiple sample images through data enhancement processing to obtain more sample images. The specific data enhancement processing process is as follows:

[0136] On the one hand, the server randomly selects 4 sample images from multiple sample images each time, randomly crops them, and splices them into a new sample image; on the other hand, the server randomly selects 2 sample images from multiple sample images each time, weightedly sums the pixel values ​​of the 2 sample images according to a certain weight, and weighted sums the labels corresponding to the 2 sample images in the same proportion to obtain a new sample image; through the above process, the server expands the distribution space of sample images to a certain extent, and obtains multiple sample images after data enhancement processing.

[0137] The switch opening and closing state and switch position information of the air switch in each sample image after data enhancement processing are marked, the label position information of the category label in each sample image after data enhancement processing is marked, and multiple sample images after data enhancement processing are divided into a training set and a verification set.

[0138] Specifically, the server performs labeling on each sample image after data enhancement processing, annotates the switch opening and closing status and switch position information of the air switch in each image, and annotates the label position information of the category label in each image, and at the same time divides 80% of the sample images after multiple data enhancement processing into a training set and the remaining 20% ​​into a verification set.

[0139] The target detection model to be trained is trained based on the training set and the validation set to obtain a target detection model that meets the preset training conditions as the pre-trained target detection model.

[0140] Specifically, the server iteratively trains the target detection model to be trained based on the training set and the validation set until the trained target detection model meets the preset training conditions. The server uses the target detection model that meets the preset training conditions as the pre-trained target detection model.

[0141] In this embodiment, the server can obtain a target detection model that can output the switch opening and closing status and switch position information of the air switch in the input image, as well as the label image and label position information of the category label in the input image by training the target detection model to be trained, thereby being able to remotely recognize the opening and closing status of the air switch.

[0142] In an exemplary embodiment, as shown in FIG6 , another method for identifying an air switch in a power distribution cabinet is provided, and the method is described by taking the application of the method to a server as an example, and the method includes the following steps:

[0143] Step S602: Acquire an image of the power distribution cabinet.

[0144] Step S604, performing target detection processing on the image based on the pre-trained target detection model to obtain the switch opening and closing state and switch position information of each air switch in the image, as well as the label image and label position information of each category label in the image.

[0145] Step S606: Perform image super-resolution processing on the label image of each category label based on the pre-trained image super-resolution model to obtain a super-resolution image of each category label.

[0146] Step S608, performing text recognition processing on the super-resolved images of each category label based on the text recognition model to obtain the text information in each category label.

[0147] Step S610, obtaining the layout attributes of the power distribution cabinet.

[0148] Step S612, for each air switch, determine the matching degree between the switch position information of the air switch and the label position information of each category label according to the arrangement attribute, and determine the category label with the highest matching degree among the category labels as the category label corresponding to the air switch.

[0149] Step S614, when it is detected that the switch opening and closing state of any air switch is in the disconnected state, an early warning message for the air switch is generated based on the switch opening and closing state of the air switch and the text information in the category label corresponding to the air switch, and the early warning message is sent to the monitoring terminal corresponding to the distribution cabinet.

[0150] In this embodiment, based on the target detection model, the server can obtain the switch opening and closing status and switch position information of each air switch, as well as the label image and label position information of each category label. Based on the text recognition model, the server can obtain the text information in each category label. Based on the switch position information of each air switch and the label position information of each category label, the server can match each air switch with each category label and determine the category label corresponding to each air switch, so as to obtain the air switch identification result of the distribution cabinet according to the switch opening and closing status of each air switch and the text information in the corresponding category label. Based on the above process, the switch opening and closing status of the air switch of the distribution cabinet can be accurately monitored remotely, thereby improving the monitoring efficiency.

[0151] In order to more clearly illustrate the identification method of the air switch of the power distribution cabinet provided in the embodiment of the present application, the identification method of the air switch of the power distribution cabinet is specifically described below with a specific embodiment, but it should be understood that the embodiment of the present application is not limited to this. As shown in Figure 7, in one of the exemplary embodiments, the present application also provides a state and category identification method of the air switch of the power distribution cabinet, which specifically includes the following steps:

[0152] 1. Input the image of the power distribution cabinet.

[0153] 2. The server uses the YOLOv8 target detection model to detect the air switch and text labels in the image. Specifically, the YOLOv8 target detection model needs to output the opening and closing state and position of the air switch, as well as the image and position of the text label.

[0154] 3. The server performs image super-resolution on the text label based on the DIP-DKP image super-resolution model. Specifically, the DIP-DKP image super-resolution model needs to output a text label image with a higher resolution to make the text content in the text label clearer.

[0155] 4. The server uses the PP-OCRv4 text recognition model to recognize the content of the text label. Specifically, the content of the text label is the power type of the power supply to which the air switch corresponding to the text label belongs.

[0156] 5. The server matches the air switch and the text label based on the location relationship.

[0157] 6. The power type and opening and closing status corresponding to the server output air switch.

[0158] In this embodiment, the server also compared the recognition accuracy and recognition recall rate before and after the introduction of the DIP-DKP image super-resolution model. After comparison, it was found that after the introduction of the DIP-DKP image super-resolution model, the recognition accuracy rate increased by about 2.2% and the recognition recall rate increased by about 1.9%.

[0159] In this embodiment, by introducing the DIP-DKP image super-resolution model, the accuracy of text recognition is improved, thereby improving the recognition accuracy of air switches. It has strong generalization and meets the task requirements of air switch recognition in distribution cabinets.

[0160] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0161] Based on the same inventive concept, the embodiment of the present application also provides a device for identifying an air switch in a power distribution cabinet for implementing the above-mentioned method for identifying an air switch in a power distribution cabinet. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the embodiments of the device for identifying one or more air switches in a power distribution cabinet provided below can refer to the limitations of the method for identifying an air switch in a power distribution cabinet above, and will not be repeated here.

[0162] In an exemplary embodiment, as shown in FIG8 , a device for identifying an air switch in a power distribution cabinet is provided, comprising: an image acquisition module 802 , a target detection module 804 , a text recognition module 806 , a position matching module 808 and a result determination module 810 , wherein:

[0163] The image acquisition module 802 is used to acquire an image of the power distribution cabinet; the image contains multiple air switches configured in the power distribution cabinet and a category label of each air switch.

[0164] The target detection module 804 is used to perform target detection processing on the image based on the pre-trained target detection model to obtain the switch detection results of each air switch and the label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label.

[0165] The text recognition module 806 is used to perform text recognition processing on the label images of each category label based on the pre-trained text recognition model to obtain the text information in each category label.

[0166] The position matching module 808 is used to match each air switch with each category tag based on the switch position information of each air switch and the label position information of each category tag, so as to determine the category label corresponding to each air switch.

[0167] The result determination module 810 is used to determine the switch opening and closing status of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the distribution cabinet.

[0168] In an exemplary embodiment, the recognition device of the air switch of the distribution cabinet also includes an image super-resolution module, which is used to perform image super-resolution processing on the label image of each category label based on a pre-trained image super-resolution model to obtain a super-resolution image of each category label; the resolution of the super-resolution image is higher than the resolution of the label image.

[0169] The text recognition module 806 is further used to perform text recognition processing on the super-resolved images of each category label based on the text recognition model to obtain the text information in each category label.

[0170] In an exemplary embodiment, the image super-resolution module is also used to use the label image of the category label as the current first label image for each category label; extract image features of the first label image based on the deep image prior network in the image super-resolution model; generate a dynamic convolution kernel corresponding to the first label image based on the dynamic kernel prior network and image features in the image super-resolution model; perform convolution processing on the first label image based on the dynamic convolution kernel to obtain a second label image, and use the second label image as the new first label image, return to the step of extracting image features of the first label image based on the deep image prior network in the image super-resolution model, until the obtained second label image meets the preset iteration condition, and the second label image that meets the preset iteration condition is determined as the super-resolution image of the category label.

[0171] In an exemplary embodiment, the position matching module 808 is further used to match each air switch with each category label based on the switch position information of each air switch and the label position information of each category label, and determine the category label corresponding to each air switch, including:

[0172] Obtain the layout attributes of the distribution cabinet; the layout attributes are used to characterize the layout correspondence between the air switches configured in the distribution cabinet and the corresponding category labels; for each air switch, determine the matching degree between the switch position information of the air switch and the label position information of each category label according to the layout attributes; among each category label, the category label with the highest matching degree is determined as the category label corresponding to the air switch.

[0173] In an exemplary embodiment, the identification device of the air switch of the distribution cabinet also includes a switch warning module, which is used to generate warning information for the air switch based on the switch opening and closing state of the air switch and the text information in the category label corresponding to the air switch when it is detected that the switch opening and closing state of any air switch is in the disconnected state; and send the warning information to the monitoring terminal corresponding to the distribution cabinet.

[0174] In an exemplary embodiment, the identification device of the air switch of the distribution cabinet also includes a model training module, which is used to collect multiple sample images of the distribution cabinet, and perform data enhancement processing on the multiple sample images to obtain multiple sample images after data enhancement processing; annotate the switch opening and closing status and switch position information of the air switch in each sample image after data enhancement processing, annotate the label position information of the category label in each sample image after data enhancement processing, and divide the multiple sample images after data enhancement processing into a training set and a verification set; train the target detection model to be trained based on the training set and the verification set to obtain a target detection model that meets preset training conditions as the pre-trained target detection model.

[0175] Each module in the above-mentioned identification device of the air switch of the power distribution cabinet can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0176] In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in FIG9. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for identifying an air switch of a distribution cabinet is implemented.

[0177] Those skilled in the art will understand that the structure shown in FIG. 9 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0178] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0179] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0180] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0181] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0182] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0183] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for identifying an air switch in a power distribution cabinet, characterized in that: The method comprises: Acquire an image of a power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch; Performing target detection processing on the image based on the pre-trained target detection model to obtain switch detection results of each air switch and label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label; Performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels; Based on the switch position information of each air switch and the label position information of each category label, matching processing is performed on each air switch and each category label to determine the category label corresponding to each air switch; The switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch are determined as the air switch identification result of the distribution cabinet.

2. The method according to claim 1, characterized in that Before performing text recognition processing on the label images of the respective category labels based on the pre-trained text recognition model to obtain the text information in the respective category labels, the method further includes: Performing image super-resolution processing on the label image of each category label based on the pre-trained image super-resolution model to obtain the super-resolution image of each category label; the resolution of the super-resolution image is higher than the resolution of the label image; The pre-trained text recognition model is used to perform text recognition processing on the label images of the respective category labels to obtain text information in the respective category labels, including: Based on the text recognition model, text recognition processing is performed on the super-resolved images of each category label to obtain the text information in each category label.

3. The method according to claim 2, characterized in that The image super-resolution model obtained by pre-training performs image super-resolution processing on the label image of each category label to obtain the super-resolution image of each category label, including: For each category label, taking the label image of the category label as the current first label image; Extracting image features of the first label image based on the deep image prior network in the image super-resolution model; Based on the dynamic kernel prior network in the image super-resolution model and the image features, generating a dynamic convolution kernel corresponding to the first label image; Based on the dynamic convolution kernel, the first label image is convolved to obtain a second label image, and the second label image is used as a new first label image, and the step of extracting image features of the first label image based on the deep image prior network in the image super-resolution model is returned until the obtained second label image meets a preset iteration condition, and the second label image that meets the preset iteration condition is determined as the super-resolution image of the category label.

4. The method according to claim 1, characterized in that: The matching process of each air switch and each category label based on the switch position information of each air switch and the label position information of each category label to determine the category label corresponding to each air switch includes: Obtaining the arrangement attribute of the power distribution cabinet; the arrangement attribute is used to characterize the arrangement correspondence between the air switches configured in the power distribution cabinet and the corresponding category labels; For each air switch, according to the arrangement attribute, determining the matching degree between the switch position information of the air switch and the label position information of each category label; Among the various category labels, the category label with the highest matching degree is determined as the category label corresponding to the air switch.

5. The method according to claim 1, characterized in that After determining the switch opening and closing states of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the power distribution cabinet, the method further includes: When it is detected that the switch opening and closing state of any air switch is in the open state, based on the switch opening and closing state of the air switch and the text information in the category label corresponding to the air switch, generate warning information for the air switch; The warning information is sent to the monitoring terminal corresponding to the power distribution cabinet.

6. The method according to any one of claims 1 to 5, characterized in that: The pre-trained target detection model is trained by the following method: Collecting a plurality of sample images of the power distribution cabinet, and performing data enhancement processing on the plurality of sample images to obtain a plurality of sample images after data enhancement processing; Annotating the switch opening and closing state and switch position information of the air switch in each sample image after data enhancement processing, annotating the label position information of the category label in each sample image after data enhancement processing, and dividing the multiple sample images after data enhancement processing into a training set and a verification set; The target detection model to be trained is trained based on the training set and the verification set to obtain a target detection model that meets preset training conditions, which is used as the pre-trained target detection model.

7. An identification device for an air switch of a power distribution cabinet, characterized in that: The device comprises: An image acquisition module is used to acquire an image of the power distribution cabinet; the image contains a plurality of air switches configured in the power distribution cabinet and a category label of each air switch; The target detection module is used to perform target detection processing on the image based on the pre-trained target detection model to obtain the switch detection results of each air switch and the label detection results of each category label; the switch detection result of each air switch at least includes the switch opening and closing state and switch position information of the air switch; the label detection result of each category label at least includes the label image and label position information of the category label; A text recognition module, used to perform text recognition processing on the label images of the respective category labels based on a pre-trained text recognition model to obtain text information in the respective category labels; A position matching module, used to match each air switch with each category label based on the switch position information of each air switch and the label position information of each category label, and determine the category label corresponding to each air switch; The result determination module is used to determine the switch opening and closing status of each air switch and the text information in the category label corresponding to each air switch as the air switch identification result of the distribution cabinet.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images