Disconnector state determination method, device and disconnector state detection system

Abstract

The present application provides a method, device and system for determining the state of an isolating switch. The method includes: obtaining image data of the isolating switch; constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical image data obtained within a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the isolating switch; inputting the image data into the state analysis model to obtain the analysis results corresponding to the image data. This solution solves the problem that the opening and closing states of the isolating switch cannot be accurately detected in the prior art.

Description

Technical Field

[0001] The present application relates to the field of computer vision, and in particular to a method for determining the state of an isolating switch, a device for determining the state of an isolating switch, a computer program product, and a system for detecting the state of an isolating switch. Background Art

[0002] In recent years, as a key component of the power system, isolating switches have been used to effectively separate circuits and equipment, ensure the safety of workers, and facilitate equipment maintenance and repair. The accuracy of the isolating switch status is directly related to the safe operation of the power system.

[0003] However, in the current solution, it is impossible to accurately detect the open and closed status of the disconnector, resulting in potential power failures in the future, threatening the safety of power grid equipment and personnel. Summary of the invention

[0004] The main purpose of the present application is to provide a method for determining the state of an isolating switch, a device for determining the state of an isolating switch, a computer program product and a system for detecting the state of an isolating switch, so as to at least solve the problem that the opening and closing states of the isolating switch cannot be accurately detected in the prior art.

[0005] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for determining the state of an isolating switch is provided, comprising: acquiring image data of the isolating switch; constructing a state analysis model, wherein the state analysis model is trained by a YOLO algorithm and a DC algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical image data acquired within a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the isolating switch; inputting the image data into the state analysis model to obtain the analysis results corresponding to the image data.

[0006] Optionally, acquiring image data of the isolating switch includes: acquiring initial image data of the isolating switch; and enhancing the initial image data using an adaptive image enhancement algorithm to obtain the image data.

[0007] Optionally, before constructing the state analysis model, the method also includes: constructing an initial state analysis model, wherein the initial state analysis model is trained by a YOLOv8n algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes the historical image data acquired within the historical time period and the historical initial analysis results corresponding to the historical image data, wherein the historical initial analysis results are used to characterize the initial state of the historical opening and closing of the disconnector; optimizing the initial state analysis model to obtain the state analysis model, wherein the optimization processing method includes one or more of an embedded attention mechanism, feature fusion processing, and a loss function optimization processing.

[0008] Optionally, when the optimization processing method includes embedding an attention mechanism, the initial state analysis model is optimized, including: extracting the C2f module of the feature extraction network in the initial state analysis model; and replacing the C2f module with a Faster Block module embedded with a SimAM attention mechanism.

[0009] Optionally, when the optimization processing method includes feature fusion processing, the initial state analysis model is optimized, including: a first feature optimization step: compressing the external input features through a Conv convolution operation, processing the compressed features through a SimAM attention mechanism and a GSConv convolution operation, and outputting the first feature information after stacking and cascading fusion; a second feature optimization step: performing a Conv convolution operation on the external input features, stacking and cascading fusion the features subjected to the Conv convolution operation and the compressed features, and then outputting the second feature information after the Conv convolution operation.

[0010] Optionally, when the optimization processing method includes loss function optimization processing, the initial state analysis model is optimized, including: extracting at least part of the image data to obtain a training set; performing data enhancement processing on the image data in the training set to obtain a processed training set; inputting the processed training set into the initial state analysis model for training, and using the Inner-EIoU loss function to optimize the initial state analysis model.

[0011] Optionally, after optimizing the initial state analysis model to obtain the state analysis model, the method further includes: obtaining a first performance indicator of the initial state analysis model, wherein the first performance indicator includes one or more of a first analysis accuracy, a first analysis speed, a first calculation amount and a first parameter amount; obtaining a second performance indicator of the state analysis model, wherein the second performance indicator includes one or more of a second analysis accuracy, a second analysis speed, a second calculation amount and a second parameter amount; and displaying the first performance indicator and the second performance indicator in a display interface.

[0012] According to another aspect of the present application, a state determination device for an isolating switch is provided, comprising: a first acquisition unit, for acquiring image data of the isolating switch; a first construction unit, for constructing a state analysis model, wherein the state analysis model is trained by a YOLO algorithm and a DC algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical image data acquired within a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the isolating switch; and a first processing unit, for inputting the image data into the state analysis model to obtain the analysis results corresponding to the image data.

[0013] According to another aspect of the present application, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the steps of any one of the methods for determining the state of an isolating switch are implemented.

[0014] According to another aspect of the present application, a disconnector status detection system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the disconnector status determination methods.

[0015] By applying the technical solution of the present application, the YOLO algorithm and the DC algorithm can be used to construct a state analysis model, and the state analysis model can be used to analyze the opening and closing states of the disconnecting switch. The YOLO algorithm has a high accuracy rate for target detection and can accurately identify targets of various categories. The DC algorithm can compress the image, thereby retaining the important parts of the image, improving the image quality, and supporting more efficient image analysis. Therefore, in this solution, the YOLO algorithm and the DC algorithm are combined to accurately detect the opening and closing states of the disconnecting switch. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0017] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for determining the state of an isolating switch provided in an embodiment of the present application is shown;

[0018] Figure 2 A schematic flow chart of a method for determining a state of an isolating switch provided in accordance with an embodiment of the present application is shown;

[0019] Figure 3 Shown is a schematic diagram of the FBSAM module structure;

[0020] Figure 4 A schematic diagram of the structure of building a state analysis model is shown;

[0021] Figure 5 Shows a schematic diagram of the structure of the VoVGSCSPC-SAM module;

[0022] Figure 6 The first schematic diagram shows the calculation principle of the Inner-EIoU loss function;

[0023] Figure 7 The second schematic diagram shows the calculation principle of the Inner-EIoU loss function;

[0024] Figure 8 A structural block diagram of a state determination device for an isolating switch provided according to an embodiment of the present application is shown.

[0025] The above drawings include the following reference numerals:

[0026] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] As introduced in the background technology, the prior art is unable to accurately detect the open and closed states of an isolating switch. To solve the above problem, an embodiment of the present application provides a method for determining the state of an isolating switch, a device for determining the state of an isolating switch, a computer program product and an isolating switch state detection system.

[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0032] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for determining the state of an isolating switch according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0033] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the display method of device information in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The above-mentioned specific examples of the network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] In this embodiment, a method for determining the state of an isolating switch running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0035] Figure 2 FIG. 1 is a flow chart of a method for determining the state of an isolating switch according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0036] Step S201, acquiring image data of the isolating switch;

[0037] Specifically, the image data of the isolating switch may be obtained, that is, the image data may be obtained from a camera or other device connected to the isolating switch.

[0038] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0039] Specifically, a state analysis model is constructed by training multiple sets of training data using the YOLO algorithm and the DC algorithm. These training data include historical image data acquired in a historical time period and historical analysis results corresponding to these image data. These historical analysis results are used to describe the historical opening and closing states of the disconnector. By analyzing these historical data, a model can be established to predict the state of the disconnector in the future, thereby helping to monitor and manage the operation of the power system.

[0040] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0041] Specifically, the image data is input into a state analysis model, which then analyzes the image data and generates corresponding results.

[0042] Through this embodiment, the YOLO algorithm and the DC algorithm can be used to construct a state analysis model, and the state analysis model can be used to analyze the opening and closing states of the disconnecting switch. The YOLO algorithm has a high accuracy rate for target detection and can accurately identify targets of various categories. The DC algorithm can compress the image, thereby retaining the important parts of the image, improving the image quality, and supporting more efficient image analysis. Therefore, in this solution, the YOLO algorithm and the DC algorithm are combined to accurately detect the opening and closing states of the disconnecting switch.

[0043] Specifically, in the current solution, by accurately detecting the opening and closing status of the disconnector, potential power failures can be avoided and the safety of power grid equipment and personnel can be guaranteed. In addition, accurate detection of the disconnector status can also help improve the reliability and stability of the power system and provide important support for system operation. Therefore, the importance of disconnector opening and closing detection is self-evident. It is not only related to the normal operation of the power system, but also directly related to the safety of the staff and the overall efficiency of the power system.

[0044] At present, based on the widespread and successful application of target detection technology in computer vision, this technology can also be used to achieve rapid and accurate recognition of the disconnector opening and closing states, helping operation and maintenance personnel to promptly determine whether the disconnector is in the open or closed state, and avoiding safety accidents caused by misjudgment. By utilizing the powerful extraction capability of the target detection algorithm model for video image feature information, the automatic detection of the disconnector state is achieved, the accuracy and efficiency of the detection are improved, the automation level of the disconnector state detection in the power system is improved, the dependence on manual operation is reduced, and the errors caused by human factors are reduced, thereby ensuring the safe operation of the power system.

[0045] Specifically, this solution is aimed at the deployment of algorithm models for edge devices such as video surveillance or drones. A disconnector opening and closing detection model (i.e., a state analysis model) is designed with both high-precision detection performance and low complexity to improve its response speed and feedback speed in engineering applications and reduce the economic loss of electricity caused by time delays.

[0046] Specifically, the relevant data of the isolating switch can also be obtained, and the relevant data includes one or more of distance data, mechanical data, image data, auxiliary contact data and temperature data. The distance data of the isolating switch refers to the change in the distance between the switches when the switch is opened or closed. Mechanical data refers to the performance of the switch under the application of force, such as the wear resistance and bearing capacity of the switch. Image data refers to data on the appearance, structure and other aspects of the switch. Auxiliary contact data refers to the performance data of the auxiliary contacts of the switch, such as contact resistance, contact life, etc. Temperature data refers to the performance change of the switch at different temperatures. YOLO algorithm and DC algorithm can also be used to train with relevant data to obtain a state analysis model. Specifically, multiple sets of training data can be used to train with YOLO algorithm and DC algorithm, and each set of training data in the above multiple sets of training data includes historical relevant data obtained in the historical time period and historical analysis results corresponding to the above historical relevant data; the above relevant data is input into the above state analysis model to obtain the analysis results corresponding to the above relevant data.

[0047] Since the image samples of disconnectors taken by video surveillance, cameras, drones, etc. are affected by factors such as multi-scale, complex environment, and uneven lighting, the detection accuracy of the disconnector opening and closing states in small targets and at night drops sharply, making it difficult to meet the detection performance requirements in engineering applications. In the specific implementation process, the image data of the disconnector can be obtained through the following steps: obtaining the initial image data of the above-mentioned disconnector; and using an adaptive image enhancement algorithm to enhance the above-mentioned initial image data to obtain the above-mentioned image data.

[0048] In this solution, an adaptive image enhancement algorithm can be used to perform image enhancement processing on the acquired initial image data to obtain enhanced image data. The adaptive image enhancement algorithm can automatically adjust parameters according to the characteristics and content of the image to make the enhancement effect more accurate and natural. This can improve the clarity and contrast of the image, repair the noise and distortion in the image, and ensure that the image data is more accurate.

[0049] Specifically, this scheme proposes an adaptive histogram correction enhancement network (AHCENet) to dynamically enhance the edge feature information of the isolation switch and the contrast of the night image to meet the high-precision detection performance of small targets and low brightness conditions.

[0050] Specifically, adaptive histogram correction and enhancement are performed on the isolation switch image samples with multi-scale, complex background and night conditions taken by video surveillance, cameras, drones, etc., in order to improve the robustness and generalization ability of the detection algorithm in engineering applications.

[0051] Specifically, before the state analysis model is constructed, an adaptive histogram correction enhancement network module can also be constructed. The adaptive histogram correction enhancement network (AHCENet) is an adaptive image enhancement algorithm based on histogram equalization. Its working principle is to adjust the histogram equalization parameters in combination with local contrast information to achieve more accurate histogram correction technology, and to achieve adaptive enhancement of the image by dynamically adjusting the histogram equalization parameters.

[0052] Assume that the input disconnector opening and closing images are as follows: , the histogram equalization function is , then the enhancement function expression of the adaptive histogram correction enhancement network AHCENet is: , where Is the histogram equalization function, used to calculate the corresponding pixel The pixel value after histogram equalization, To consider the local contrast information, the position The enhancement function parameters at .

[0053] Specifically, local contrast information is generally obtained by calculating the grayscale variance of pixels in the local area, but this scheme proposes to use a 3×3 convolution kernel as a window function to extract local area contrast information, and convolve each pixel center point (x, y) in the disconnector opening and closing image, and use the convolution result as the local contrast information. Combined with the original pixel value, the adjusted enhancement function is:

[0054] ,

[0055] Specifically, in the formula, represents a 3×3 convolution operation, It is an adjustable coefficient, usually between 0 and 1. Indicates that the image is at the center of the pixel The mean value of Indicates that the image is at the center of the pixel The mean value of .

[0056] In the specific implementation process, before constructing the state analysis model, the method also includes the following steps: constructing an initial state analysis model, wherein the initial state analysis model is obtained by training with the YOLOv8n algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes the historical image data obtained within the historical time period and the historical initial analysis results corresponding to the historical image data, wherein the historical initial analysis results are used to characterize the initial state of the historical opening and closing of the above-mentioned disconnector; optimizing the initial state analysis model to obtain the above-mentioned state analysis model, wherein the optimization processing method includes one or more of embedded attention mechanism, feature fusion processing, and loss function optimization processing.

[0057] In this solution, the YOLOv8n algorithm can be used for machine learning to build an initial state analysis model, and then the initial state analysis model can be optimized to obtain a more accurate state analysis model. Optimization methods such as embedding attention mechanism and feature fusion processing can improve the performance of the model, making the model perform state analysis faster and more stably, thereby more accurately judging the state of the disconnector.

[0058] During the specific implementation process, when the optimization processing method includes embedding the attention mechanism, the above-mentioned initial state analysis model is optimized, which can be achieved through the following steps: extracting the C2f module of the feature extraction network in the above-mentioned initial state analysis model; using the Faster Block module embedded with the SimAM attention mechanism to replace the above-mentioned C2f module.

[0059] In this scheme, in the field of deep learning, the initial state analysis model is optimized by introducing the attention mechanism. The attention mechanism can automatically learn and enhance the weights of important features while suppressing the influence of unimportant features, effectively improving the discriminability of features, while reducing information redundancy and improving model efficiency, further improving the performance and efficiency of the model.

[0060] Specifically, a Faster Block module (FBSAM module) embedded with the SimAM attention mechanism can be constructed. This solution designs a Faster Block module embedded with the SimAM attention mechanism, namely the FBSAM module, to replace the C2f module in the original YOLOv8n model feature extraction network (Backbone). The FBSAM module structure is as follows: Figure 3 shown.

[0061] Specifically, a 3×3 PConv layer is connected in series with two 1×1 Conv layers to process the input features. After processing and superimposing with the features processed by the SAM attention mechanism, the output features of the FBSAM module can be obtained. The calculation expression is as follows: , where Represents the SimAM attention mechanism processing function, is the Conv layer processing function with a convolution kernel of 1×1, is the PConv layer processing function with a convolution kernel of 3×3, is the superposition processing function of feature stacking; the SAM attention mechanism processing function expression is:

[0062] ⊙ ,

[0063] In the formula, Group all feature vectors of the disconnector opening and closing images in the channel and space dimensions, ⊙ represents the same or same operator symbol, is a common activation function, and its expression is 1 / (1+ex). Assuming that the dimension of the input disconnector opening and closing image is H×W×C, if a 3×3 Conv ordinary convolution is used for processing, its parameter volume is H×W×2C; and after the PConv convolution operation, its parameter volume is H×W×2C p , which is equivalent to C of ordinary convolution p / C times. So by setting C p The parameter amount of the FBSAM module is reduced by adjusting the size of the value, and then the C2f module is replaced by this module to build a parameter reduction enhanced feature extraction network based on the FBSAM module. The introduction of the SimAM attention mechanism can improve the performance of feature extraction without increasing the number of parameters.

[0064] In the specific implementation process, when the optimization processing method includes feature fusion processing, the optimization processing of the above-mentioned initial state analysis model can be achieved through the following steps: the first feature optimization step: compress the external input features through Conv convolution operation, process the compressed features through SimAM attention mechanism and GSConv convolution operation, and output the first feature information after stacking and cascading fusion; the second feature optimization step: perform Conv convolution operation on the external input features, stack and cascade fusion the features subjected to Conv convolution operation and the compressed features, and then output the second feature information after Conv convolution operation.

[0065] In this scheme, the steps of feature optimization are formulated, which include different processing operations on the features of external input to improve the representation ability of the features. In the first feature optimization step, the input features are first compressed by the Conv convolution operation, and then the compressed features are processed by the SimAM attention mechanism and the GSConv convolution operation, and finally the first feature information is obtained through stacking and cascading fusion. In the second feature optimization step, the Conv convolution operation is first performed on the features of the external input, and then the features that have undergone the Conv convolution operation are stacked and cascaded with the features compressed in the first step, and finally the second feature information is output through the Conv convolution operation. In this way, different operations can be fully utilized to process the features in different aspects, thereby further improving the representation ability and discrimination of the features. Through operations such as compression, attention mechanism, stacking and cascading fusion, the features can be made richer and more effective, which helps to improve the performance and generalization ability of the model. At the same time, combining different processing methods together can make feature optimization more comprehensive and diversified, which helps to improve the model's ability to handle complex data.

[0066] Specifically, a YOLO-DC-based disconnector opening and closing state detection model (i.e., state analysis model) is constructed. The YOLO-DC disconnector opening and closing state detection model proposed in this scheme has the following structure: Figure 4 As shown in the figure, it mainly uses YOLOv8n as the benchmark model, combines the aforementioned designed FBSAM module to improve the feature extraction network, and improves the feature fusion network based on the designed Slim-Neck-SAM module.

[0067] Specifically, the disconnector opening and closing images are input into the feature extraction network, and first pass through two CBR modules (Conv layer, BN layer, ReLU activation function) for feature compression, and then pass through an FBSAM module designed as mentioned above for feature enhancement extraction, and then after processing through a CBR module and an FBSAM module, the large-scale feature T1 is obtained; it is further passed through a CBS module and an FBSAM module to obtain the medium-scale feature T2; finally, the small-scale feature T3 is extracted through the CBR module, FBSAM and SPPF modules respectively.

[0068] Specifically, in the innovative Slim-Neck-SAM feature fusion network proposed in the present invention, the extracted small-scale feature T3 is cascaded and fused with the medium-scale feature T2 after upsampling to obtain the fused feature T4, and then the VoVGSCSPC-SAM module is used to perform feature enhancement processing to obtain the enhanced feature T5; the enhanced feature T5 is cascaded and fused with the large-scale feature T1 after upsampling to obtain the fused feature T6, and then the enhanced feature T7 (i.e., the feature T7 to be decoded) is obtained after feature enhancement processing by the VoVGSCSPC-SAM module; the enhanced feature T7 is cascaded and fused with the enhanced feature T5 through a GBR module (Conv layer, BN layer, ReLU activation function) to obtain the fused feature T8, and then the feature T9 to be decoded is obtained after enhancement by the VoVGSCSPC-SAM module; at the same time, the feature T9 to be decoded will be further cascaded and fused with the small-scale feature T3 through a GBR module to obtain the fused feature T 10 , and then strengthened by the VoVGSCSPC-SAM module to obtain the feature T to be decoded 11 .

[0069] Specifically, in the above feature fusion process, the main innovation is the Slim-Neck-SAM structural paradigm designed based on the GBR module and the VoVGSCSPC-SAM module. Figure 5 As shown in the figure. The VoVGSCSPC-SAM module compresses the number of channels c1 of the input feature into half after ordinary convolution (Conv) processing. One branch is input to GSBottleneck-SAM, processed with two GSConvs through the SimAM attention mechanism, and then stacked and cascaded as output. The other branch is processed by Conv ordinary convolution and then stacked and cascaded with the GSBottleneck-SAM output features, and finally processed by Conv to become the output feature of the final VoVGSCSPC-SAM module. The VoVGSCSPC-SAM module can improve the extraction and utilization of spatial feature information while being lightweight.

[0070] Specifically, the features T7, T9, T7 and T9 of large, medium and small scales are obtained through feature fusion network enhancement. 11 All are input into the YOLO Head detection head in the feature decoding network for decoding prediction. First, the three scale features containing the semantic information of the disconnector's opening and closing states are normalized to 128-dimensional channels through a 1×1 convolution layer to reduce the amount of calculation. Then, a 3×3 convolution layer with two independent branches is used for parallel processing to predict the opening and closing state type and the switch position in the disconnector image.

[0071] In the specific implementation process, when the optimization processing method includes loss function optimization processing, the optimization processing of the above-mentioned initial state analysis model can be achieved through the following steps: extracting at least part of the above-mentioned image data to obtain a training set; performing data enhancement processing on the above-mentioned image data in the above-mentioned training set to obtain a processed training set; inputting the above-mentioned processed training set into the above-mentioned initial state analysis model for training, and using the Inner-EIoU loss function to optimize the above-mentioned initial state analysis model.

[0072] In this scheme, part of the data is extracted from the image data as a training set, and the image data in the training set is data augmented to increase the diversity and quantity of the training set. The processed training set is input into the initial state analysis model for training, and the Inner-EIoU loss function is used to optimize the parameters of the model. Data augmentation can improve the generalization and robustness of the model, so that the model performs well in a wider range of scenarios. The use of the Inner-EIoU loss function can better optimize the model, so that the model fits the training data more accurately, thereby further improving the performance of the model. By inputting the processed training set into the model for training, the convergence speed of the model can be accelerated, further improving the training efficiency.

[0073] Specifically, since the existing technology does not take into account the high-precision and low-complexity model performance requirements when deploying edge devices such as video surveillance or drones, this solution explicitly proposes a disconnector opening and closing state detection model based on YOLO-DC (i.e., the state analysis model mentioned in this solution). YOLO-DC is mainly constructed based on the designed FBSAM module and Slim-Neck-SAM module to improve the feature extraction capability of the YOLOv8n benchmark model, and the Inner-EIoU loss function is used during the training process to improve the regression precision and accuracy of the detection frame. Combining the above improvement methods, the model parameters and calculation amount are reduced together, while still maintaining high-precision disconnector opening and closing detection accuracy.

[0074] Specifically, the YOLO-DC model (i.e., state analysis model) can be trained and tested based on the Inner-EIoU improved loss function.

[0075] Specifically, based on the opening and closing videos of disconnectors shot by the Internet and power grid companies, including day and night, various shooting angles, and various meteorological conditions, the videos are converted into images at a time interval of 1 frame every 10 frames, and an image dataset of disconnector images in both open and closed states is constructed, with a total of more than 1,200 images. Labels are created based on the LabelImg tool, and the label names are "Fen" and "He".

[0076] Specifically, the disconnector image dataset can be divided into training set, validation set and test set in a ratio of 8:1:1. The training set is first enhanced by randomly combining data, i.e., 50% probability of Mosaic data enhancement and 50% probability of mixup data enhancement. The enhanced data is input into the YOLO-DC model for training. The training process uses the Inner-EIoU loss function proposed in this scheme to measure the position and category in the training process, and combines the validation set for real-time detection to verify the quality of model training, so as to further adjust the learning rate. After 100 rounds of training and verification, the optimization model is adjusted to obtain the YOLO-DC disconnector opening and closing state analysis model with the best performance.

[0077] Specifically, the calculation diagram of the Inner-EIoU loss function in the above steps is as follows: Figure 6 and Figure 7 As shown, the expression is as follows:

[0078] ,

[0079] ,

[0080] ,

[0081] ,

[0082] ,

[0083] ,

[0084] In the formula, Indicates the Inner-EIoU loss value, represents the EIoU loss value, It represents the intersection-over-union ratio of the prior frame and the real frame. It represents the intersection-over-union ratio between the prior frame and the real frame after embedding the inner. represents the Euclidean distance metric function, and are the predicted box and the true box respectively. Indicates the width of the prediction box, Represents the height of the prediction box, represents the width of the prior box, Indicates the width of the real box, Represents the height of the prior box, Indicates the height of the real frame, is the diagonal length of the minimum circumscribed rectangle of the real box and the predicted box, is the scale factor, are the coordinates of the real box and its internal center point, and is the coordinate of the prior box and its internal center point, and Respectively represent the positive and negative deviations of the true frame width, and Respectively represent the positive and negative deviations of the real frame height, and Respectively represent the positive and negative deviations of the prediction box width, and Indicates the positive and negative deviations of the predicted box height, represents the intersection area, Represents the area of ​​the union.

[0085] In the specific implementation process, after optimizing the above-mentioned initial state analysis model and obtaining the above-mentioned state analysis model, the above-mentioned method also includes the following steps: obtaining the first performance indicator of the above-mentioned initial state analysis model, wherein the above-mentioned first performance indicator includes one or more of a first analysis accuracy, a first analysis speed, a first calculation amount and a first parameter amount; obtaining the second performance indicator of the above-mentioned state analysis model, wherein the above-mentioned second performance indicator includes one or more of a second analysis accuracy, a second analysis speed, a second calculation amount and a second parameter amount; displaying the above-mentioned first performance indicator and the above-mentioned second performance indicator in a display interface.

[0086] In this solution, the first performance indicator and the second performance indicator of the analysis model in the initial state are obtained and displayed in the display interface. Displaying these indicators in the display interface can allow the user to intuitively understand the performance of the model. The user can directly see the performance of the model. The user can easily compare the performance of different models and choose the model that best suits their needs. Providing intuitive information display can improve the user experience.

[0087] Specifically, the disconnector opening and closing state images in the test set can be input into the best YOLO-DC model after training for batch detection, and the model detection accuracy (mAP), detection speed (FPS), calculation amount (FLOPs), and parameter amount (Params) performance evaluation indicators are calculated based on the detection results. Table 1 shows the statistical results of the performance evaluation indicators of the model calculated based on the detection results.

[0088] Table 1

[0089]

[0090] In summary, the above-mentioned scheme of the present application has strong adaptability and can still accurately detect small targets in complex environments and special situations such as at night; it has superior performance and can achieve high-precision detection of the opening and closing status of the disconnecting switch; it saves computing resources, and the disconnecting switch opening and closing status detection model with low parameter quantity and computational complexity has lower computing performance requirements for the inference device; it is easy to promote and apply, and the detection algorithm model can be embedded in the edge devices of various application scenarios in combination with different work task requirements, and it is highly practical and promotable.

[0091] The embodiment of the present application also provides a state determination device for an isolating switch. It should be noted that the state determination device for an isolating switch in the embodiment of the present application can be used to execute the state determination method for an isolating switch provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0092] The following is an introduction to the state determination device of the isolating switch provided in an embodiment of the present application.

[0093] Figure 8 is a structural block diagram of a state determination device for an isolating switch according to an embodiment of the present application. Figure 8 As shown, the device comprises:

[0094] A first acquisition unit 10 is used to acquire image data of the disconnector;

[0095] A first construction unit 20 is used to construct a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0096] The first processing unit 30 is used to input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0097] Through this embodiment, the YOLO algorithm and the DC algorithm can be used to construct a state analysis model, and the state analysis model can be used to analyze the opening and closing states of the disconnecting switch. The YOLO algorithm has a high accuracy rate for target detection and can accurately identify targets of various categories. The DC algorithm can compress the image, thereby retaining the important parts of the image, improving the image quality, and supporting more efficient image analysis. Therefore, in this solution, the YOLO algorithm and the DC algorithm are combined to accurately detect the opening and closing states of the disconnecting switch.

[0098] Since the image samples of disconnectors taken by video surveillance, cameras, drones, etc. are affected by factors such as multi-scale, complex environment, and uneven lighting, the detection accuracy of the disconnector opening and closing states under small targets and at night drops sharply, making it difficult to adapt to the detection performance requirements in engineering applications. In the specific implementation process, the first acquisition unit includes an acquisition module and a first processing module, the acquisition module is used to acquire the initial image data of the above-mentioned disconnector; the first processing module is used to enhance the above-mentioned initial image data using an adaptive image enhancement algorithm to obtain the above-mentioned image data.

[0099] In this solution, an adaptive image enhancement algorithm can be used to perform image enhancement processing on the acquired initial image data to obtain enhanced image data. The adaptive image enhancement algorithm can automatically adjust parameters according to the characteristics and content of the image to make the enhancement effect more accurate and natural. This can improve the clarity and contrast of the image, repair the noise and distortion in the image, and ensure that the image data is more accurate.

[0100] During the specific implementation process, the above-mentioned device also includes a second construction unit and a second processing unit. The second construction unit is used to construct an initial state analysis model before constructing the state analysis model, wherein the above-mentioned initial state analysis model is obtained by training with the YOLOv8n algorithm using multiple sets of training data, and each set of training data in the above-mentioned multiple sets of training data includes the above-mentioned historical image data obtained within the above-mentioned historical time period and the historical initial analysis results corresponding to the above-mentioned historical image data, wherein the above-mentioned historical initial analysis results are used to characterize the initial state of the historical opening and closing of the above-mentioned disconnector; the second processing unit is used to optimize the above-mentioned initial state analysis model to obtain the above-mentioned state analysis model, wherein the optimization processing method includes one or more of embedded attention mechanism, feature fusion processing, and loss function optimization processing.

[0101] In this solution, the YOLOv8n algorithm can be used for machine learning to build an initial state analysis model, and then the initial state analysis model can be optimized to obtain a more accurate state analysis model. Optimization methods such as embedding attention mechanism and feature fusion processing can improve the performance of the model, making the model perform state analysis faster and more stably, thereby more accurately judging the state of the disconnector.

[0102] During the specific implementation process, when the optimization processing method includes embedding an attention mechanism, the second processing unit includes a first extraction module and a replacement module. The first extraction module is used to extract the C2f module of the feature extraction network in the above-mentioned initial state analysis model; the replacement module is used to replace the above-mentioned C2f module with a Faster Block module embedded in the SimAM attention mechanism.

[0103] In this scheme, in the field of deep learning, the initial state analysis model is optimized by introducing the attention mechanism. The attention mechanism can automatically learn and enhance the weights of important features while suppressing the influence of unimportant features, effectively improving the discriminability of features, while reducing information redundancy and improving model efficiency, further improving the performance and efficiency of the model.

[0104] During the specific implementation process, when the optimization processing method includes feature fusion processing, the second processing unit includes a first optimization module and a second optimization module. The first optimization module is used to execute the first feature optimization step: compress the external input features through a Conv convolution operation, process the compressed features through a SimAM attention mechanism and a GSConv convolution operation, and output the first feature information after stacking and cascading fusion; the second optimization module is used for the second feature optimization step: perform a Conv convolution operation on the external input features, stack and cascade fusion the features subjected to the Conv convolution operation and the compressed features, and then output the second feature information after the Conv convolution operation.

[0105] In this scheme, the steps of feature optimization are formulated, which include different processing operations on the features of external input to improve the representation ability of the features. In the first feature optimization step, the input features are first compressed by the Conv convolution operation, and then the compressed features are processed by the SimAM attention mechanism and the GSConv convolution operation, and finally the first feature information is obtained through stacking and cascading fusion. In the second feature optimization step, the Conv convolution operation is first performed on the features of the external input, and then the features that have undergone the Conv convolution operation are stacked and cascaded with the features compressed in the first step, and finally the second feature information is output through the Conv convolution operation. In this way, different operations can be fully utilized to process the features in different aspects, thereby further improving the representation ability and discrimination of the features. Through operations such as compression, attention mechanism, stacking and cascading fusion, the features can be made richer and more effective, which helps to improve the performance and generalization ability of the model. At the same time, combining different processing methods together can make feature optimization more comprehensive and diversified, which helps to improve the model's ability to handle complex data.

[0106] During the specific implementation process, when the optimization processing method includes loss function optimization processing, the second processing unit includes a second extraction module, a second processing module and a third processing module. The second extraction module is used to extract at least part of the data in the above-mentioned image data to obtain a training set; the second processing module is used to perform data enhancement processing on the above-mentioned image data in the above-mentioned training set to obtain a processed training set; the third processing module is used to input the above-mentioned processed training set into the above-mentioned initial state analysis model for training, and use the Inner-EIoU loss function to optimize the above-mentioned initial state analysis model.

[0107] In this scheme, part of the data is extracted from the image data as a training set, and the image data in the training set is data augmented to increase the diversity and quantity of the training set. The processed training set is input into the initial state analysis model for training, and the Inner-EIoU loss function is used to optimize the parameters of the model. Data augmentation can improve the generalization and robustness of the model, so that the model performs well in a wider range of scenarios. The use of the Inner-EIoU loss function can better optimize the model, so that the model fits the training data more accurately, thereby further improving the performance of the model. By inputting the processed training set into the model for training, the convergence speed of the model can be accelerated, further improving the training efficiency.

[0108] During the specific implementation process, the above-mentioned device also includes a second acquisition unit, a third acquisition unit and a display unit. The second acquisition unit is used to obtain the first performance indicator of the above-mentioned initial state analysis model after optimizing the above-mentioned initial state analysis model to obtain the above-mentioned state analysis model, wherein the above-mentioned first performance indicator includes one or more of a first analysis accuracy, a first analysis speed, a first calculation amount and a first parameter amount; the third acquisition unit is used to obtain the second performance indicator of the above-mentioned state analysis model, wherein the above-mentioned second performance indicator includes one or more of a second analysis accuracy, a second analysis speed, a second calculation amount and a second parameter amount; the display unit is used to display the above-mentioned first performance indicator and the above-mentioned second performance indicator in a display interface.

[0109] In this solution, the first performance indicator and the second performance indicator of the analysis model in the initial state are obtained and displayed in the display interface. Displaying these indicators in the display interface can allow the user to intuitively understand the performance of the model. The user can directly see the performance of the model. The user can easily compare the performance of different models and choose the model that best suits their needs. Providing intuitive information display can improve the user experience.

[0110] The state determination device of the disconnector includes a processor and a memory, and the first acquisition unit, the first construction unit, the first processing unit, etc. are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions. The modules are all located in the same processor; or, the modules are located in different processors in any combination.

[0111] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be provided, and the problem that the opening and closing states of the isolating switch cannot be accurately detected in the prior art can be solved by adjusting the kernel parameters.

[0112] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0113] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the state determination method of the isolating switch.

[0114] Specifically, the method for determining the state of the isolating switch includes:

[0115] Step S201, acquiring image data of the isolating switch;

[0116] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0117] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0118] An embodiment of the present invention provides a processor, and the processor is used to run a program, wherein the state determination method of the isolation switch is executed when the program is run.

[0119] Specifically, the method for determining the state of the isolating switch includes:

[0120] Step S201, acquiring image data of the isolating switch;

[0121] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0122] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0123] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the steps of the method for determining the state of an isolating switch are implemented.

[0124] Specifically, the method for determining the state of the isolating switch includes:

[0125] Step S201, acquiring image data of the isolating switch;

[0126] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0127] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0128] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0129] A computer program product includes a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the state determination method of the isolation switch in each embodiment of the present application are implemented.

[0130] Specifically, the method for determining the state of the isolating switch includes:

[0131] Step S201, acquiring image data of the isolating switch;

[0132] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0133] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0134] The present application provides an isolating switch state detection system, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the isolating switch state determination methods.

[0135] Specifically, the method for determining the state of the isolating switch includes:

[0136] Step S201, acquiring image data of the isolating switch;

[0137] Step S202, constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired in a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector;

[0138] Step S203: input the image data into the state analysis model to obtain analysis results corresponding to the image data.

[0139] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0140] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0141] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0142] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0143] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0144] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0145] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0146] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0147] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0148] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0149] 1) A state determination method for an isolating switch of the present application can use the YOLO algorithm and the DC algorithm to construct a state analysis model, and use the state analysis model to analyze the opening and closing states of the isolating switch. The YOLO algorithm has a high accuracy rate for target detection and can accurately identify targets of various categories. The DC algorithm can compress the image, thereby retaining the important parts of the image, improving the image quality, and supporting more efficient image analysis. Therefore, in this solution, the YOLO algorithm and the DC algorithm are combined to accurately detect the opening and closing states of the isolating switch.

[0150] 2) A state determination device for an isolating switch of the present application can use the YOLO algorithm and the DC algorithm to construct a state analysis model, and use the state analysis model to analyze the opening and closing states of the isolating switch. The YOLO algorithm has a high accuracy rate for target detection and can accurately identify targets of various categories. The DC algorithm can compress the image, thereby retaining the important parts of the image, improving the image quality, and supporting more efficient image analysis. Therefore, in this solution, the YOLO algorithm and the DC algorithm are combined to accurately detect the opening and closing states of the isolating switch.

[0151] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining the state of an isolating switch, characterized in that: include: Acquire image data of the disconnector; Constructing a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired within a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector; Inputting the image data into the state analysis model to obtain analysis results corresponding to the image data; The structure of the state analysis model is as follows: the disconnector opening and closing images are input into the feature extraction network, and firstly, the features are compressed through two CBR modules, and the CBR modules include a Conv layer, a BN layer, and a ReLU activation function. Then, the features are enhanced and extracted through an FBSAM module, and the FBSAM module is a FasterBlock module that uses an embedded SimAM attention mechanism. After being processed through a CBR module and an FBSAM module, a large-scale feature T1 is obtained; the medium-scale feature T2 is obtained after it is further processed through a CBR module and an FBSAM module; finally, the small-scale feature T3 is extracted through the CBR module, the FBSAM and the SPPF module respectively; in the Slim-Neck-SAM feature fusion network, the extracted small-scale feature T3 is up-sampled and cascaded with the medium-scale feature T2 to obtain the fused feature T4, and then the fused feature T4 is obtained through the VoVGSCSPC- The SAM module performs feature enhancement processing to obtain enhanced feature T5; the enhanced feature T5 is then upsampled and cascaded with the large-scale feature T1 to obtain fused feature T6, and then enhanced by the VoVGSCSPC-SAM module to obtain enhanced feature T7, that is, feature T7 to be decoded; the enhanced feature T7 is cascaded with the enhanced feature T5 through a GBR module, and the GBR module includes a Conv layer, a BN layer, and a ReLU activation function to obtain fused feature T8, and then enhanced by the VoVGSCSPC-SAM module to obtain feature T9 to be decoded; at the same time, the feature T9 to be decoded will be further cascaded with the small-scale feature T3 through a GBR module to obtain fused feature T 10 , and then strengthened by the VoVGSCSPC-SAM module to obtain the feature T to be decoded 11 The VoVGSCSPC-SAM module compresses the number of channels c1 of the input feature to half after ordinary convolution processing. One branch is input to GSBottleneck-SAM, processed with two GSConvs through the SimAM attention mechanism, and then stacked and cascaded as the output. The other branch is processed by Conv ordinary convolution and then stacked and cascaded with the GSBottleneck-SAM output features. Finally, it is processed by Conv to become the output feature of the final VoVGSCSPC-SAM module.

2. The method according to claim 1, characterized in that Obtain image data of the disconnector, including: Acquiring initial image data of the isolating switch; The initial image data is enhanced by using an adaptive image enhancement algorithm to obtain the image data.

3. The method according to claim 1, characterized in that Before building the state analysis model, the method also includes: Constructing an initial state analysis model, wherein the initial state analysis model is obtained by training with a YOLOv8n algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes the historical image data acquired within the historical time period and the historical initial analysis results corresponding to the historical image data, wherein the historical initial analysis results are used to characterize the initial state of the historical opening and closing of the disconnector; The initial state analysis model is optimized to obtain the state analysis model, wherein the optimization processing method includes one or more of embedded attention mechanism, feature fusion processing, and loss function optimization processing.

4. The method according to claim 3, characterized in that In the case where the optimization processing method includes embedding the attention mechanism, the initial state analysis model is optimized, including: Extracting a C2f module of a feature extraction network in the initial state analysis model; The Faster Block module embedded with the SimAM attention mechanism is used to replace the C2f module.

5. The method according to claim 3, characterized in that: When the optimization processing method includes feature fusion processing, the initial state analysis model is optimized, including: The first feature optimization step: compress the external input features through the Conv convolution operation, process the compressed features through the SimAM attention mechanism and the GSConv convolution operation, and output the first feature information after stacking and cascading fusion; The second feature optimization step: perform Conv convolution operation on the external input features, stack and cascade the Conv convolution operation features and the compressed features, and then output the second feature information after Conv convolution operation.

6. The method according to claim 3, characterized in that When the optimization processing method includes loss function optimization processing, the initial state analysis model is optimized, including: Extracting at least part of the image data to obtain a training set; Performing data enhancement processing on the image data in the training set to obtain a processed training set; The processed training set is input into the initial state analysis model for training, and the initial state analysis model is optimized using the Inner-EIoU loss function.

7. The method according to claim 3, characterized in that After optimizing the initial state analysis model to obtain the state analysis model, the method further includes: Acquire a first performance indicator of the initial state analysis model, wherein the first performance indicator includes one or more of a first analysis accuracy, a first analysis speed, a first calculation amount, and a first parameter amount; Acquire a second performance indicator of the state analysis model, wherein the second performance indicator comprises one or more of a second analysis accuracy, a second analysis speed, a second calculation amount, and a second parameter amount; The first performance indicator and the second performance indicator are displayed in a display interface.

8. A state determination device for an isolating switch, characterized in that: include: A first acquisition unit, used to acquire image data of the disconnector; A first construction unit is used to construct a state analysis model, wherein the state analysis model is obtained by training with a YOLO algorithm and a DC algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical image data acquired within a historical time period and historical analysis results corresponding to the historical image data, wherein the historical analysis results are used to characterize the historical opening and closing states of the disconnector; A first processing unit, configured to input the image data into the state analysis model to obtain an analysis result corresponding to the image data; The structure of the state analysis model is as follows: the disconnector opening and closing images are input into the feature extraction network, and firstly, the features are compressed through two CBR modules, and the CBR modules include a Conv layer, a BN layer, and a ReLU activation function. Then, the features are enhanced and extracted through an FBSAM module, and the FBSAM module is a FasterBlock module that uses an embedded SimAM attention mechanism. After being processed through a CBR module and an FBSAM module, a large-scale feature T1 is obtained; the medium-scale feature T2 is obtained after it is further processed through a CBR module and an FBSAM module; finally, the small-scale feature T3 is extracted through the CBR module, the FBSAM and the SPPF module respectively; in the Slim-Neck-SAM feature fusion network, the extracted small-scale feature T3 is up-sampled and cascaded with the medium-scale feature T2 to obtain the fused feature T4, and then the fused feature T4 is obtained through the VoVGSCSPC- The SAM module performs feature enhancement processing to obtain enhanced feature T5; the enhanced feature T5 is then upsampled and cascaded with the large-scale feature T1 to obtain fused feature T6, and then enhanced by the VoVGSCSPC-SAM module to obtain enhanced feature T7, that is, feature T7 to be decoded; the enhanced feature T7 is cascaded with the enhanced feature T5 through a GBR module, and the GBR module includes a Conv layer, a BN layer, and a ReLU activation function to obtain fused feature T8, and then enhanced by the VoVGSCSPC-SAM module to obtain feature T9 to be decoded; at the same time, the feature T9 to be decoded will be further cascaded with the small-scale feature T3 through a GBR module to obtain fused feature T 10 , and then strengthened by the VoVGSCSPC-SAM module to obtain the feature T to be decoded 11 The VoVGSCSPC-SAM module compresses the number of channels c1 of the input feature to half after ordinary convolution processing. One branch is input to GSBottleneck-SAM, processed with two GSConvs through the SimAM attention mechanism, and then stacked and cascaded as the output. The other branch is processed by Conv ordinary convolution and then stacked and cascaded with the GSBottleneck-SAM output features. Finally, it is processed by Conv to become the output feature of the final VoVGSCSPC-SAM module.

9. 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 for determining the state of the disconnector according to any one of claims 1 to 7 are implemented.

10. A disconnector status detection system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the state determination method of the disconnector according to any one of claims 1 to 7.

Images