A method and system for power equipment image recognition based on cloud-edge collaborative technology

Abstract

The present invention is applicable to the field of electrical engineering technology, and provides a method and system for image recognition of electric power equipment based on cloud-edge collaborative technology, the method comprising the following steps: obtaining sample image data, preprocessing the sample image data to obtain processed image data; using a convolutional neural network to extract features from the processed image data to obtain sample feature information; acquiring electric power equipment images through an image acquisition device; processing and analyzing the electric power equipment images through an edge processing module, and when the feature information in the electric power equipment image does not match the sample feature information, sending the electric power equipment image to a cloud platform; determining non-conforming electric power equipment and non-conforming feature information; determining possible abnormal electric power equipment through an electric power equipment related library, and obtaining electric power equipment data that may have abnormal electric power equipment. By analyzing the electric power equipment data, it is possible to learn the impact of faulty electric power equipment on associated electric power equipment.

Description

Technical Field

[0001] The present invention relates to the field of electrical engineering technology, and specifically to a method and system for image recognition of electric power equipment based on cloud-edge collaborative technology. Background Art

[0002] With the development of image processing and computer vision technology, the use of computer vision to identify power equipment and analyze the daily operating status of power equipment plays an increasingly important role in power system detection. Real-time monitoring of the operating status of power equipment is particularly important. However, a large amount of image data will be generated through image processing, and it will continue to increase at an exponential rate. If all this data is gathered in the cloud for intelligent analysis, it will bring unlimited pressure to the bandwidth requirements and real-time requirements of communication. This requires us to provide edge intelligent services nearby on the device node side or the network edge side close to the data source, that is, to reduce the pressure on the cloud through cloud-edge collaborative technology. Although cloud-edge collaborative technology can ensure that the image of power equipment is successfully identified and abnormal equipment is found, power equipment is basically not independently operated, and existing means cannot determine the impact of abnormal equipment on related power equipment. Therefore, it is necessary to provide a method and system for power equipment image recognition based on cloud-edge collaborative technology to solve the above problems. Summary of the invention

[0003] In view of the shortcomings of the prior art, the purpose of the present invention is to provide an image recognition method and system for power equipment based on cloud-edge collaborative technology to solve the problems existing in the above-mentioned background technology.

[0004] The present invention is implemented as follows: a method for image recognition of electric power equipment based on cloud-edge collaborative technology, the method comprising the following steps:

[0005] Acquire sample image data through an image acquisition device, and pre-process the sample image data to obtain processed image data;

[0006] Using a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, region features, boundary features, and power equipment data features;

[0007] Collecting images of power equipment by using an image acquisition device;

[0008] The power equipment image is processed and analyzed by an edge processing module, wherein the edge processing module stores sample feature information. When the feature information in the power equipment image does not match the sample feature information, the non-matching feature information is marked, and the power equipment image is sent to the cloud platform;

[0009] The power equipment images are deeply processed through the cloud platform, specifically: non-compliant power equipment and non-compliant feature information are determined to generate abnormal information; power equipment that may have abnormalities is determined through the power equipment related library, power equipment images that may have abnormal power equipment are retrieved, and power equipment data that may have abnormal power equipment is obtained.

[0010] As a further solution of the present invention: the step of preprocessing the sample image data to obtain processed image data specifically includes:

[0011] Remove noise from sample image data;

[0012] Enhance the details of the sample image data;

[0013] After improving the signal-to-noise ratio of the sample image data, processed image data is obtained.

[0014] As a further solution of the present invention: the step of using a convolutional neural network to extract features from processed image data to obtain sample feature information specifically includes:

[0015] A convolutional neural network consisting of convolutional layers and fully connected layers is established, where the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized;

[0016] Normalizing the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features;

[0017] The feature vector of the power equipment image is established for the output of the fully connected layer to obtain the sample feature information;

[0018] Receive normal range values ​​of power equipment data and complete sample feature information.

[0019] As a further solution of the present invention: the step of determining the power equipment that may have abnormalities through the power equipment related library, retrieving the power equipment image that may have abnormal power equipment, and obtaining the power equipment data that may have abnormal power equipment specifically includes:

[0020] Establishing a power equipment related library, wherein the power equipment related library records mutually related equipment, and when one power equipment fails, the power equipment mutually related to the power equipment is likely to fail;

[0021] Input the abnormal information into the power equipment related library, and output the power equipment that may have abnormalities;

[0022] Sending an image acquisition instruction to a corresponding image acquisition device to obtain an image of a power device where abnormal power devices may exist;

[0023] The power equipment data features of the power equipment image are identified to obtain the power equipment data of the power equipment that may contain abnormal power equipment.

[0024] As a further solution of the present invention: the method also includes outputting comprehensive abnormal information, and the output comprehensive abnormal information includes images of non-compliant power equipment, non-compliant feature information, and power equipment data where abnormal power equipment may exist.

[0025] As a further solution of the present invention: each image acquisition device corresponds to an edge processing module, all edge processing modules are communicatively connected to the cloud platform, and an image acquisition device is used to acquire one or more specific power equipment images.

[0026] Another object of the present invention is to provide a power equipment image recognition system based on cloud-edge collaboration technology, the system comprising:

[0027] An image acquisition device is used to acquire sample image data, pre-process the sample image data to obtain processed image data, and is used to acquire images of power equipment;

[0028] A sample feature information determination module; using a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, regional features, boundary features, and power equipment data features;

[0029] An edge processing module, used for processing and analyzing the power equipment image, wherein the edge processing module stores sample feature information, and when the feature information in the power equipment image does not match the sample feature information, the inconsistent feature information is marked, and the power equipment image is sent to the cloud platform; and

[0030] The cloud platform is used to perform in-depth processing on the power equipment images, specifically: determine the non-compliant power equipment and non-compliant feature information, and generate abnormal information; determine the power equipment that may have abnormalities through the power equipment related library, retrieve the power equipment images that may have abnormal power equipment, and obtain the power equipment data that may have abnormal power equipment.

[0031] As a further solution of the present invention: the image acquisition device comprises:

[0032] A noise removal unit, used to remove noise from sample image data;

[0033] A detail enhancement unit, used for enhancing details of sample image data;

[0034] The signal-to-noise ratio improving unit is used to improve the signal-to-noise ratio of the sample image data to obtain processed image data.

[0035] As a further solution of the present invention: the sample feature information determination module includes:

[0036] A convolutional neural network building unit is used to build a convolutional neural network consisting of a convolutional layer and a fully connected layer, wherein the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized;

[0037] A normalization processing unit, used to perform normalization processing on the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features;

[0038] A sample feature information determination unit, used to establish a feature vector of the power equipment image for the output of the fully connected layer to obtain sample feature information;

[0039] The sample characteristic information completion unit is used to receive the normal range value of the power equipment data and complete the sample characteristic information.

[0040] Compared with the prior art, the beneficial effects of the present invention are: the present invention can generate abnormal information based on non-compliant power equipment and non-compliant characteristic information; determine the power equipment that may have abnormalities through the power equipment related library, and retrieve the power equipment image of the power equipment that may have abnormalities, and obtain the power equipment data of the power equipment that may have abnormalities. In this way, by analyzing the power equipment data, the staff can know the impact of the faulty power equipment on the related power equipment, and then it is convenient to determine the specific cause of the fault and formulate a maintenance plan as soon as possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of a method for image recognition of power equipment based on cloud-edge collaborative technology.

[0042] Figure 2 The present invention is a flowchart for preprocessing sample image data to obtain processed image data in an image recognition method for electric power equipment based on cloud-edge collaborative technology.

[0043] Figure 3 This is a flow chart of a method for image recognition of electric power equipment based on cloud-edge collaborative technology, which uses a convolutional neural network to extract features from processed image data to obtain sample feature information.

[0044] Figure 4 This is a flow chart of determining power equipment that may have abnormalities through a power equipment related library in a power equipment image recognition method based on cloud-edge collaborative technology.

[0045] Figure 5This is a structural diagram of an electric power equipment image recognition system based on cloud-edge collaborative technology.

[0046] Figure 6 This is a structural schematic diagram of the image acquisition device in an electric power equipment image recognition system based on cloud-edge collaborative technology.

[0047] Figure 7 This is a structural schematic diagram of a sample feature information determination module in an electric power equipment image recognition system based on cloud-edge collaborative technology. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0050] like Figure 1 As shown, an embodiment of the present invention provides a method for image recognition of electric power equipment based on cloud-edge collaboration technology, and the method includes the following steps:

[0051] S100, acquiring sample image data through an image acquisition device, and preprocessing the sample image data to obtain processed image data;

[0052] S200, using a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, region features, boundary features, and power equipment data features;

[0053] S300, collecting images of power equipment by using an image collection device;

[0054] S400, processing and analyzing the power equipment image through an edge processing module, wherein the edge processing module stores sample feature information, and when feature information in the power equipment image does not match the sample feature information, marking the non-matching feature information, and sending the power equipment image to a cloud platform;

[0055] S500, deep processing of the power equipment image is performed through the cloud platform, specifically: determining non-compliant power equipment and non-compliant feature information, generating abnormal information; determining the power equipment that may have abnormalities through the power equipment related library, retrieving the power equipment image that may have abnormal power equipment, and obtaining the power equipment data that may have abnormal power equipment.

[0056] It should be noted that with the development of image processing and computer vision technology, the use of computer vision to identify power equipment and analyze the daily operating status of power equipment plays an increasingly important role in power system detection, and real-time monitoring of the operating status of power equipment is particularly important. However, image processing will generate a large amount of image data, which is increasing at an exponential rate. If all this data is gathered in the cloud for intelligent analysis, it will bring unlimited pressure to the bandwidth requirements and real-time requirements of communication. This requires us to provide edge intelligent services nearby on the device node side or the network edge side close to the data source, that is, to reduce the pressure on the cloud through cloud-edge collaborative technology. Although cloud-edge collaborative technology can ensure that the image of power equipment is successfully identified and abnormal equipment is found, power equipment basically does not operate independently, and existing means cannot determine the impact of abnormal equipment on related power equipment.

[0057] In an embodiment of the present invention, sample image data is acquired through an image acquisition device, and the sample image data is preprocessed to obtain processed image data. The sample image data shows a picture of the power equipment under normal operation. The preprocessing is performed to make the feature information of the sample image data more prominent and facilitate the subsequent feature extraction work. Then, a convolutional neural network is used to extract features from the processed image data to obtain sample feature information. Training and learning the image data through a convolutional neural network to obtain feature information is a conventional means of image recognition technology, which will not be repeated here. The sample feature information includes image shape features, regional features, boundary features and power equipment data features; then, the power equipment image is acquired through the image acquisition device, and the power equipment image is processed and analyzed through an edge processing module. The edge processing module stores the sample feature information, and the feature information in the power equipment image is compared with the image shape features, regional features, boundary features and power equipment data features in the sample feature information. When the feature information is inconsistent, it means that the power equipment image does not conform to It is required that the power equipment is likely to have abnormalities, for example, it is inconsistent with the image shape features, regional features and boundary features, indicating that the power equipment is damaged; it is inconsistent with the power equipment data features, indicating that the voltage, current, power, etc. in the power equipment are abnormal. When the feature information in the power equipment image is inconsistent with the sample feature information, the inconsistent feature information is marked, and the power equipment image is sent to the cloud platform. The cloud platform generates abnormal information based on the non-compliant power equipment and the non-compliant feature information; determine the power equipment that may have abnormalities through the power equipment related library, which records interrelated devices. If one power equipment fails, the power equipment that is mutually related to the power equipment is likely to fail, and retrieve the power equipment image that may have abnormal power equipment to obtain the power equipment data that may have abnormal power equipment. In this way, the staff can understand the impact of the faulty power equipment on the related power equipment by analyzing the power equipment data, and then determine the specific cause of the failure and formulate a maintenance plan as soon as possible.

[0058] like Figure 2 As shown, as a preferred embodiment of the present invention, the step of preprocessing the sample image data to obtain processed image data specifically includes:

[0059] S101, removing noise from sample image data;

[0060] S102, enhancing details of sample image data;

[0061] S103, improving the signal-to-noise ratio of the sample image data to obtain processed image data.

[0062] In the embodiment of the present invention, the preprocessing includes image noise reduction, detail enhancement and signal-to-noise ratio improvement, the purpose of which is to remove interference, noise and differences from the sample image data of the power equipment and convert the sample image into a form suitable for subsequent feature extraction by a computer.

[0063] like Figure 3 As shown, as a preferred embodiment of the present invention, the step of using a convolutional neural network to extract features from processed image data to obtain sample feature information specifically includes:

[0064] S201, establishing a convolutional neural network consisting of a convolutional layer and a fully connected layer, wherein the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized;

[0065] S202, normalizing the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features;

[0066] S203, establishing a feature vector of the power equipment image for the output of the fully connected layer to obtain sample feature information;

[0067] S204, receiving the normal range value of the power equipment data, and completing the sample characteristic information.

[0068] In the embodiment of the present invention, the existing power equipment image recognition method requires manual annotation of a large amount of sample data, which is time-consuming and labor-intensive, has high economic costs, and the feature extraction of the image is too single, which cannot better describe the global features of the image, resulting in the problem that the recognition accuracy of the final image is not high. In order to better utilize the convolutional neural network algorithm to automatically learn to describe the global multi-dimensional features of the image, the present invention uses a convolutional neural network to extract features that comprehensively reflect the image feature information of the power equipment, and establishes a convolutional neural network with 4 convolutional layers and 2 fully connected layers, wherein the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized. , and normalize the extracted 4-dimensional features, which include image shape features, regional features, boundary features and power equipment data features. The feature vector of the power equipment image is established for the output of the first of the two fully connected layers for subsequent use. In addition, it is easy to understand that the image shape features, regional features and boundary features of each power equipment are unique, and the correct feature information can be determined by identifying enough pictures. However, the correct feature information of the power equipment data features may be within a range, which can be inferred by relevant staff of the data range. Therefore, it is sufficient to directly upload the correct data range to supplement the power equipment data features.

[0069] like Figure 4As shown, as a preferred embodiment of the present invention, the steps of determining the power equipment that may have abnormalities through the power equipment related library, retrieving the power equipment image that may have abnormal power equipment, and obtaining the power equipment data that may have abnormal power equipment specifically include:

[0070] S501, establishing a power equipment related library, wherein the power equipment related library records mutually related equipment, and when one power equipment fails, the power equipment mutually related to the power equipment is likely to fail;

[0071] S502, inputting the abnormal information into the power equipment related library, and outputting the power equipment that may have abnormalities;

[0072] S503, sending an image acquisition instruction to a corresponding image acquisition device to obtain an image of a power device where abnormal power devices may exist;

[0073] S504, identifying the electric power equipment data features of the electric power equipment image to obtain the electric power equipment data of the electric power equipment that may contain abnormal electric power equipment.

[0074] In an embodiment of the present invention, each image acquisition device corresponds to an edge processing module, and all edge processing modules are communicatively connected to the cloud platform. An image acquisition device is used to acquire one or more specific power equipment images. That is to say, by obtaining a power equipment image acquired by a certain image acquisition device, the specific power equipment information in the power equipment image can be known. The present invention requires the establishment of a power equipment related library in advance, and the interrelated equipment information needs to be uploaded by the staff. Then, the abnormal information is input into the power equipment related library, and the power equipment that may have abnormalities is automatically output. An image retrieval instruction is sent to the corresponding image acquisition device to obtain a power equipment image that may have abnormal power equipment, and then the power equipment data features of the power equipment image are identified to obtain the power equipment data of the power equipment that may have abnormal power equipment.

[0075] As a preferred embodiment of the present invention, the method also includes outputting comprehensive abnormal information, which includes images of non-compliant power equipment, non-compliant feature information, and power equipment data of possible abnormal power equipment, so that the staff can first determine the faulty power equipment, and by analyzing the power equipment data, they can know the impact of the faulty power equipment on related power equipment, and then facilitate the determination of the specific cause of the fault and formulate a maintenance plan as soon as possible.

[0076] like Figure 5 As shown, an embodiment of the present invention further provides a power equipment image recognition system based on cloud-edge collaboration technology, the system comprising:

[0077] The image acquisition device 100 is used to acquire sample image data, pre-process the sample image data to obtain processed image data, and is used to acquire images of power equipment;

[0078] Sample feature information determination module 200: using a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, region features, boundary features, and power equipment data features;

[0079] The edge processing module 300 is used to process and analyze the power equipment image. The edge processing module 300 stores sample feature information. When the feature information in the power equipment image does not match the sample feature information, the inconsistent feature information is marked and the power equipment image is sent to the cloud platform 400.

[0080] The cloud platform 400 is used to perform in-depth processing on the power equipment image, specifically: determine the non-compliant power equipment and non-compliant feature information, and generate abnormal information; determine the power equipment that may have abnormalities through the power equipment related library, retrieve the power equipment image that may have abnormal power equipment, and obtain the power equipment data that may have abnormal power equipment.

[0081] In an embodiment of the present invention, sample image data is acquired through an image acquisition device 100, and the sample image data is preprocessed to obtain processed image data. The sample image data shows a picture of the power equipment under normal operation. The preprocessing is performed to make the feature information of the sample image data more prominent and facilitate the subsequent feature extraction work. Then, a convolutional neural network is used to extract features from the processed image data to obtain sample feature information. Training and learning the image data through a convolutional neural network and then obtaining feature information is a conventional means of image recognition technology, which will not be repeated here. The sample feature information includes image shape features, regional features, boundary features and power equipment data features; then, the image of the power equipment is acquired through the image acquisition device 100, and the image of the power equipment is processed and analyzed through the edge processing module 300. The edge processing module 300 stores the sample feature information, and the feature information in the power equipment image is compared with the image shape features, regional features, boundary features and power equipment data features in the sample feature information. When the feature information is inconsistent, it indicates that the power equipment image If it does not meet the requirements, the power equipment is likely to have an abnormality, for example, it is inconsistent with the image shape features, regional features and boundary features, indicating that the power equipment is damaged; it is inconsistent with the power equipment data features, indicating that the voltage, current, power, etc. in the power equipment are abnormal. When the feature information in the power equipment image is inconsistent with the sample feature information, the inconsistent feature information is marked, and the power equipment image is sent to the cloud platform 400. The cloud platform 400 generates abnormal information based on the non-compliant power equipment and the non-compliant feature information; the power equipment that may have abnormalities is determined through the power equipment related library, and the power equipment related library records interrelated devices. If one power equipment fails, the power equipment that is interrelated with the power equipment is likely to fail, and the power equipment image that may have abnormal power equipment is retrieved to obtain the power equipment data that may have abnormal power equipment. In this way, the staff can understand the impact of the faulty power equipment on the related power equipment by analyzing the power equipment data, and then it is convenient to determine the specific cause of the failure and formulate a maintenance plan as soon as possible.

[0082] like Figure 6 As shown, as a preferred embodiment of the present invention, the image acquisition device 100 includes:

[0083] A noise removal unit 101, used to remove noise from sample image data;

[0084] A detail enhancement unit 102, used to enhance the details of the sample image data;

[0085] The signal-to-noise ratio improving unit 103 is used to improve the signal-to-noise ratio of the sample image data to obtain processed image data.

[0086] like Figure 7 As shown, as a preferred embodiment of the present invention, the sample feature information determination module 200 includes:

[0087] A convolutional neural network establishing unit 201 is used to establish a convolutional neural network consisting of a convolutional layer and a fully connected layer, wherein the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized;

[0088] A normalization processing unit 202, used to perform normalization processing on the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features;

[0089] A sample feature information determination unit 203 is used to establish a feature vector of the power equipment image for the output of the fully connected layer to obtain sample feature information; and

[0090] The sample characteristic information completion unit 203 is used to receive the normal range value of the power equipment data and complete the sample characteristic information.

[0091] The above only describes in detail the preferred embodiments of the present invention, which is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0092] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. 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 performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0093] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0094] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A method for image recognition of power equipment based on cloud-edge collaborative technology, It is characterized in that The method comprises the following steps: Acquire sample image data through an image acquisition device, and pre-process the sample image data to obtain processed image data; Using a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, region features, boundary features, and power equipment data features; Collecting images of power equipment by using an image acquisition device; The power equipment image is processed and analyzed by an edge processing module, wherein the edge processing module stores sample feature information. When the feature information in the power equipment image does not match the sample feature information, the non-matching feature information is marked, and the power equipment image is sent to the cloud platform; The power equipment image is deeply processed through the cloud platform, specifically: determining the non-compliant power equipment and non-compliant feature information, generating abnormal information; determining the power equipment that may have abnormalities through the power equipment related library, retrieving the power equipment image that may have abnormal power equipment, and obtaining the power equipment data that may have abnormal power equipment; The step of using a convolutional neural network to extract features from processed image data to obtain sample feature information specifically includes: A convolutional neural network consisting of convolutional layers and fully connected layers is established, where the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized; Normalizing the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features; The feature vector of the power equipment image is established for the output of the fully connected layer to obtain the sample feature information; Receive normal range values ​​of power equipment data and complete sample feature information; The step of determining the power equipment that may have abnormalities through the power equipment related library, retrieving the power equipment image that may have abnormal power equipment, and obtaining the power equipment data that may have abnormal power equipment specifically includes: Establishing a power equipment related library, wherein the power equipment related library records mutually related equipment, and when one power equipment fails, the power equipment mutually related to the power equipment is likely to fail; Input the abnormal information into the power equipment related library, and output the power equipment that may have abnormalities; Sending an image acquisition instruction to a corresponding image acquisition device to obtain an image of a power device where abnormal power devices may exist; The power equipment data features of the power equipment image are identified to obtain the power equipment data of the power equipment that may contain abnormal power equipment.

2. According to the method for power equipment image recognition based on cloud-edge collaborative technology described in claim 1, It is characterized in that The step of preprocessing the sample image data to obtain processed image data specifically includes: Remove noise from sample image data; Enhance the details of the sample image data; After improving the signal-to-noise ratio of the sample image data, processed image data is obtained.

3. According to the method for power equipment image recognition based on cloud-edge collaborative technology described in claim 1, It is characterized in that The method further includes outputting abnormal comprehensive information, wherein the output abnormal comprehensive information includes an image of non-compliant electric equipment, non-compliant feature information, and electric equipment data where abnormal electric equipment may exist.

4. According to the method for power equipment image recognition based on cloud-edge collaborative technology described in claim 1, It is characterized in that Each image acquisition device corresponds to an edge processing module, and all edge processing modules are connected to the cloud platform for communication. An image acquisition device is used to acquire one or more specific images of power equipment.

5. A power equipment image recognition system based on cloud-edge collaborative technology, It is characterized in that The system comprises: An image acquisition device is used to acquire sample image data, pre-process the sample image data to obtain processed image data, and is used to acquire images of power equipment; Sample feature information determination module: uses a convolutional neural network to extract features from processed image data to obtain sample feature information, wherein the sample feature information includes image shape features, area features, boundary features, and power equipment data features; An edge processing module, used for processing and analyzing the power equipment image, wherein the edge processing module stores sample feature information, and when the feature information in the power equipment image does not match the sample feature information, the inconsistent feature information is marked, and the power equipment image is sent to the cloud platform; and The cloud platform is used to perform in-depth processing on the power equipment image, specifically: determine the non-compliant power equipment and non-compliant feature information, and generate abnormal information; determine the power equipment that may have abnormalities through the power equipment related library, retrieve the power equipment image that may have abnormal power equipment, and obtain the power equipment data that may have abnormal power equipment; The sample feature information determination module includes: A convolutional neural network building unit is used to build a convolutional neural network consisting of a convolutional layer and a fully connected layer, wherein the last fully connected layer has the same dimension as the input processed image data, and all weights of the network are randomly initialized; A normalization processing unit, used to perform normalization processing on the multiple dimensional features extracted in the previous step, wherein the multiple dimensional features include image shape features, region features, boundary features, and power equipment data features; A sample feature information determination unit, used to establish a feature vector of the power equipment image for the output of the fully connected layer to obtain sample feature information; A sample characteristic information completion unit, used to receive a normal range value of the power equipment data and complete the sample characteristic information; The determining of the possible abnormal power equipment through the power equipment related library, retrieving the power equipment image of the possible abnormal power equipment, and obtaining the power equipment data of the possible abnormal power equipment specifically includes: Establishing a power equipment related library, wherein the power equipment related library records mutually related equipment, and when one power equipment fails, the power equipment mutually related to the power equipment is likely to fail; Input the abnormal information into the power equipment related library, and output the power equipment that may have abnormalities; Sending an image acquisition instruction to a corresponding image acquisition device to obtain an image of a power device where abnormal power devices may exist; The power equipment data features of the power equipment image are identified to obtain the power equipment data of the power equipment that may contain abnormal power equipment.

6. According to claim 5, a power equipment image recognition system based on cloud-edge collaborative technology, It is characterized in that The image acquisition device comprises: A noise removal unit, used to remove noise from sample image data; A detail enhancement unit, used for enhancing details of sample image data; The signal-to-noise ratio improving unit is used to improve the signal-to-noise ratio of the sample image data to obtain processed image data.

Images