Method, system and device for monitoring operating state of power switching equipment, computer device and readable storage medium

Abstract

The application relates to a method, system and device for monitoring the operation state of a power switch device, computer equipment and a readable storage medium. The method comprises the following steps: receiving initial operation data of the power switch device sent by a multi-modal sensor group in real time, screening the initial operation data based on a first lightweight multi-modal model to obtain target operation data; uploading the target operation data to the cloud and receiving model parameters returned by the cloud; wherein the model parameters are obtained by the cloud after knowledge distillation of a trained multi-modal large model, and the trained multi-modal large model is obtained by the cloud after incremental training of the multi-modal large model based on the target operation data; updating the first lightweight multi-modal model based on the model parameters, and monitoring the operation state of the power switch device based on the updated lightweight multi-modal model. The method provided by the application can improve the monitoring efficiency of the switch device.

Description

Technical Field

[0001] This application relates to the field of power grid monitoring technology, and in particular to a method, system, device, computer equipment, and computer-readable storage medium for monitoring the operating status of power switching equipment. Background Technology

[0002] With the continuous expansion of power system scale and the advancement of intelligent construction, the number of power grid equipment is growing exponentially. As a core component of the power system, the operating status of switchgear directly affects the reliability and security of power supply. Traditional switchgear operation and maintenance mainly relies on manual periodic inspections and preventive tests, but this manual maintenance method is inefficient. Summary of the Invention

[0003] Therefore, it is necessary to provide a method, system, device, computer equipment, and computer-readable storage medium for monitoring the operating status of power switchgear, which can improve the operation and maintenance efficiency of power switchgear, in response to the above-mentioned technical problems.

[0004] In a first aspect, this application provides a method for monitoring the operating status of power switching equipment, used in an edge computing terminal on the power grid side; the method includes:

[0005] The system receives initial operating data of the power switchgear from the multimodal sensor group in real time, and filters the initial operating data based on the first lightweight multimodal model to obtain target operating data.

[0006] The target running data is uploaded to the cloud, and the model parameters returned by the cloud are received; wherein, the model parameters are obtained by the cloud after performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after performing incremental training on the multimodal large model based on the target running data;

[0007] The first lightweight multimodal model is updated based on the model parameters, and the operating status of the power switchgear is monitored based on the updated lightweight multimodal model.

[0008] In one embodiment, the step of filtering the initial running data based on a first lightweight multimodal model to obtain target running data includes:

[0009] Input any initial running data into the first lightweight multimodal model and output the confidence level of the initial running data;

[0010] If the confidence level is less than the confidence level threshold, the initial running data is determined as the target running data.

[0011] In one embodiment, monitoring the operating status of the power switchgear based on the updated lightweight multimodal model includes:

[0012] Receive the current operating data of the power switchgear sent by the multimodal sensor group;

[0013] The current operating data is input into the updated lightweight multimodal model, and the current operating status of the power switchgear is output.

[0014] In one embodiment, the real-time reception of initial operating data of the power switching equipment sent by the multimodal sensor group includes:

[0015] Receive sensing data of the power switchgear sent by any one of the sensors in the multimodal sensor group, as well as the acquisition time of the sensing data;

[0016] The acquisition times of all sensor data are timestamped, and the initial running data is generated based on the timestamped sensor data.

[0017] In one embodiment, the model parameters are obtained by the cloud from a second lightweight multimodal model, which is generated by the cloud after performing knowledge distillation on the trained multimodal large model.

[0018] In one embodiment, all data transmitted between the edge computing terminal and the cloud is securely verified through an encrypted communication gateway.

[0019] Secondly, this application also provides an operating status monitoring system for power switchgear, the system comprising:

[0020] An edge computing terminal is used to receive initial operating data of power switchgear sent by a multimodal sensor group in real time, and filter the initial operating data based on a first lightweight multimodal model to obtain target operating data; upload the target operating data to the cloud and receive model parameters returned by the cloud; update the first lightweight multimodal model based on the model parameters, and monitor the operating status of the power switchgear based on the updated lightweight multimodal model;

[0021] The cloud platform receives target running data uploaded by the edge computing terminal; performs incremental training on the multimodal large model based on the target running data to obtain the trained multimodal large model; performs knowledge distillation on the trained multimodal large model to obtain model parameters; and sends the model parameters to the edge computing terminal.

[0022] Thirdly, this application also provides an operating status monitoring device for power switchgear, used in an edge computing terminal on the power grid side; the device includes:

[0023] The receiving module is used to receive the initial operating data of the power switchgear sent by the multimodal sensor group in real time, and to filter the initial operating data based on the first lightweight multimodal model to obtain the target operating data.

[0024] The upload module is used to upload the target running data to the cloud and receive the model parameters returned by the cloud; wherein the model parameters are obtained by the cloud after performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after performing incremental training on the multimodal large model based on the target running data;

[0025] The update module is used to update the first lightweight multimodal model based on the model parameters, and to monitor the operating status of the power switchgear based on the updated lightweight multimodal model.

[0026] Fourthly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the methods in any of the above embodiments.

[0027] Fifthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the methods in any of the above embodiments.

[0028] The aforementioned method, system, device, computer equipment, and computer-readable storage medium for monitoring the operating status of power switchgear receive initial operating data of the power switchgear sent by a multimodal sensor group in real time, and filter the initial operating data based on a first lightweight multimodal model to obtain target operating data; upload the target operating data to the cloud, and receive model parameters returned by the cloud; wherein, the model parameters are obtained by the cloud performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud performing incremental training on the multimodal large model based on the target operating data; update the first lightweight multimodal model based on the model parameters, and monitor the operating status of the power switchgear based on the updated lightweight multimodal model. The method provided in this application sends the target operating data of the power switchgear to the cloud, and then the cloud performs knowledge distillation on the multimodal large model based on the target operating data to obtain model parameters. Finally, the first lightweight multimodal model is updated based on the model parameters. In this way, the updated lightweight multimodal model is used to monitor the operating status of the power switchgear, which can improve both the monitoring efficiency and the accuracy of the determined operating status of the power switchgear. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is an application environment diagram of the power switchgear operation status monitoring method in one embodiment;

[0031] Figure 2 This is a flowchart illustrating a method for monitoring the operating status of power switching equipment in one embodiment;

[0032] Figure 3 This is a flowchart illustrating the steps for determining target runtime data in one embodiment;

[0033] Figure 4 This is a flowchart illustrating a method for monitoring the operating status of power switching equipment in another embodiment;

[0034] Figure 5 This is a structural block diagram of a power switchgear operation status monitoring device in one embodiment;

[0035] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0037] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0038] The method for monitoring the operating status of power switchgear provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, Figure 1 The system includes a multimodal sensor array 101, an edge computing terminal 102 on the power grid side, and a cloud platform 103. The edge computing terminal 102 communicates with the cloud platform 103 via a network. The edge computing terminal 102 is installed at the substation side of the power grid, and the cloud platform 103 is deployed at the power system master station or a private cloud server. The multimodal sensor array 101 may include, but is not limited to, visible light cameras, acoustic sensors, partial discharge detectors, and electrical quantity sensors. The multimodal sensor array 101 sends the collected operating data of the power switching equipment to the edge computing terminal 102; the edge computing terminal 102 sends the filtered operating data to the cloud platform 103; the cloud platform 103 performs knowledge distillation on the multimodal large model based on the received operating data to obtain model parameters, and sends the model parameters to the edge computing terminal 102; the edge computing terminal 102 updates the first lightweight multimodal model according to the model parameters.

[0039] In one exemplary embodiment, such as Figure 2 As shown, a method for monitoring the operating status of power switchgear is provided, which can be applied to... Figure 1 Taking the edge computing terminal 102 as an example, the explanation includes the following steps 202 to 206. Wherein:

[0040] S202. Receive the initial operating data of the power switchgear sent by the multimodal sensor group in real time, and filter the initial operating data based on the first lightweight multimodal model to obtain the target operating data.

[0041] Among them, the lightweight multimodal model is a miniaturized AI model that adapts to edge computing power, can integrate multiple types of monitoring data, and is used for real-time fault diagnosis of power switching equipment.

[0042] Optionally, if the fault corresponding to the initial operating data is a complex fault that the first lightweight multimodal model has never seen before, thus reducing the accuracy of the first lightweight multimodal model's monitoring results of the power switchgear's status, then the operating data corresponding to this complex fault needs to be filtered out for subsequent model training. The complex fault may, but is not limited to, novel foreign object connections or special discharge patterns.

[0043] S204. Upload the target running data to the cloud and receive the model parameters returned by the cloud; wherein, the model parameters are obtained by the cloud after knowledge distillation of the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after incremental training of the multimodal large model based on the target running data.

[0044] Among them, the multimodal large model refers to a large-scale AI model deployed in the cloud, with hundreds of billions of parameters, which can integrate image, voiceprint, partial discharge and electrical quantity data for fault understanding and annotation.

[0045] Alternatively, knowledge distillation can be used to migrate the capabilities of large cloud models to lightweight edge models, making the lightweight edge models more powerful while remaining compact.

[0046] Optionally, incremental training of the multimodal large model based on the target operating data can enable the trained multimodal large model to identify fault characteristics corresponding to the target operating data.

[0047] S206. Update the first lightweight multimodal model based on the model parameters, and monitor the operating status of the power switchgear based on the updated lightweight multimodal model.

[0048] Optionally, the model parameters can be determined as new model parameters for the first lightweight multimodal model, thereby updating the first lightweight multimodal model.

[0049] In the aforementioned method for monitoring the operating status of power switchgear, initial operating data of the power switchgear sent by a multimodal sensor group is received in real time. The initial operating data is then filtered based on a first lightweight multimodal model to obtain target operating data. This target operating data is uploaded to the cloud, and model parameters returned from the cloud are received. The model parameters are obtained by the cloud through knowledge distillation of a trained multimodal large model, which is then incrementally trained based on the target operating data. The first lightweight multimodal model is updated based on the model parameters, and the operating status of the power switchgear is monitored based on the updated lightweight multimodal model. The method provided in this application sends the target operating data of the power switchgear to the cloud, whereby the cloud performs knowledge distillation on the multimodal large model based on the target operating data to obtain model parameters. Finally, the first lightweight multimodal model is updated based on these model parameters. This method uses the updated lightweight multimodal model to monitor the operating status of the power switchgear, which improves both monitoring efficiency and the accuracy of the determined operating status.

[0050] In some embodiments, such as Figure 3 As shown, the initial running data is filtered based on the first lightweight multimodal model to obtain the target running data, including:

[0051] S302. Input any initial running data into the first lightweight multimodal model and output the confidence level of the initial running data.

[0052] S304. If the confidence level is less than the confidence level threshold, the initial running data shall be determined as the target running data.

[0053] Optionally, when the diagnostic confidence is detected to be lower than a preset threshold (e.g., 85%) or an undefined feature pattern is detected, the corresponding initial running data is labeled, and the labeled running data is used as samples for subsequent training of the model.

[0054] In this embodiment, only the initial running data with insufficient confidence and difficult for the model to identify is uploaded as target data to filter out a large amount of low-value data and reduce invalid transmission; the target data is unknown or new fault feature data, which is used for incremental training in the cloud to improve the model's ability to identify new faults and enhance the system's adaptive evolution capability.

[0055] In some embodiments, monitoring the operating status of power switchgear based on an updated lightweight multimodal model includes: receiving current operating data of the power switchgear sent by a multimodal sensor group; inputting the current operating data into the updated lightweight multimodal model; and outputting the current operating status of the power switchgear.

[0056] Optionally, the multimodal sensor group is installed on the switchgear body or the inspection robot. The multimodal sensor group may include, but is not limited to, a visible light camera, an acoustic sensor, a partial discharge detector, and an electrical quantity sensor. Among them, the high-definition visible light camera is used to collect the appearance image and meter readings of the switchgear, the acoustic sensor is used to collect the mechanical vibration audio of the equipment during operation, the partial discharge detector is used to collect ultra-high frequency or ultrasonic signals, and the electrical quantity sensor is used to collect voltage, current, and contact temperature data.

[0057] In this embodiment, relying on the updated model, it is possible to accurately identify known faults and newly added unknown fault characteristics, thereby improving the accuracy of equipment operation status monitoring and fault diagnosis; real-time inference monitoring can be completed locally at the edge, with low response latency, and timely detection of equipment anomalies.

[0058] In some embodiments, receiving initial operating data of the power switchgear sent by the multimodal sensor group in real time includes: receiving sensing data of the power switchgear sent by any one of the sensors in the multimodal sensor group, as well as the acquisition time of the sensing data; aligning the acquisition times of all sensing data with timestamps, and generating initial operating data based on the timestamp-aligned sensing data.

[0059] Optionally, the data from each sensor are timestamped using a synchronous clock and aggregated to the edge computing terminal, facilitating the analysis of the multidimensional characteristics of the device at the same time and the discovery of hidden faults.

[0060] In this embodiment, timestamps are aligned for different types of sensor data to ensure time synchronization of multi-source data; analysis errors caused by acquisition time deviations are eliminated, and the accuracy of multimodal model fusion analysis and fault diagnosis is improved.

[0061] In some embodiments, the model parameters are obtained from a second lightweight multimodal model in the cloud, which is generated by the cloud after performing knowledge distillation on the trained multimodal large model.

[0062] Optionally, after receiving the target running data, the cloud uses the powerful understanding capabilities of the multimodal large model to perform automatic annotation or assisted expert annotation, and then triggers an incremental training task to update the weights of the large model, and generates a new version of the lightweight model adapted to edge hardware through knowledge distillation technology.

[0063] In this embodiment, by performing knowledge distillation on the trained multimodal large model, the fault identification capability of the large model is transferred to the second lightweight multimodal model, taking into account both the model's recognition accuracy and lightweight characteristics. Distillation generates a lightweight model adapted to edge computing power, and the edge model can be updated simply by issuing model parameters, adapting to scenarios where edge computing terminal hardware resources are limited.

[0064] In some embodiments, all data transmitted between the edge computing terminal and the cloud is securely verified through an encrypted communication gateway.

[0065] In some embodiments, such as Figure 4 As shown, a power switchgear operation status monitoring system is provided, characterized in that the system includes:

[0066] The edge computing terminal is used to receive initial operating data of the power switchgear sent by the multimodal sensor group in real time, and filter the initial operating data based on the first lightweight multimodal model to obtain target operating data; upload the target operating data to the cloud and receive the model parameters returned by the cloud; update the first lightweight multimodal model based on the model parameters, and monitor the operating status of the power switchgear based on the updated lightweight multimodal model.

[0067] In the cloud, it receives target running data uploaded by edge computing terminals; it incrementally trains a multimodal large model based on the target running data to obtain a trained multimodal large model; it performs knowledge distillation on the trained multimodal large model to obtain model parameters; and it sends the model parameters to the edge computing terminal.

[0068] Optionally, the system in this embodiment includes a multimodal sensor array, an edge computing terminal, a cloud training platform, and an encrypted communication gateway. During operation, the edge computing terminal uses a lightweight model to monitor the switching equipment in real time. When encountering abnormal data that cannot be confirmed, it automatically uploads it to the cloud. The cloud uses a large model for incremental learning and knowledge distillation to generate a new lightweight model and sends it back to the edge. Because this invention adopts a co-evolutionary approach of "large cloud model teaching and small edge model application," coupled with multimodal data fusion technology, the system architecture is flexible, the iteration efficiency is high, and it is convenient to achieve adaptive diagnosis of new faults in substations with limited computing resources.

[0069] Optionally, the system in this embodiment connects the cloud and the edge via a dedicated power grid or wireless network. It is characterized by including a multimodal sensor array, an edge computing terminal, an encrypted communication gateway, and a cloud training platform. The input end of the edge computing terminal is connected to the multimodal sensor array, and its output end is bidirectionally connected to the cloud training platform via the encrypted communication gateway. This controls the edge computing terminal to collect data, perform local inference, and interact with the cloud for model parameters and sample data.

[0070] Optionally, a multimodal sensor array is mounted on the switchgear itself or an inspection robot, including a high-definition visible light camera, an acoustic fingerprint sensor, a partial discharge detector, and electrical quantity sensors. The high-definition visible light camera is used to acquire external images and meter readings of the switchgear; the acoustic fingerprint sensor is used to acquire mechanical vibration audio during equipment operation; the partial discharge detector is used to acquire ultra-high frequency or ultrasonic signals; and the electrical quantity sensors are used to acquire voltage, current, and contact temperature data. The data from each sensor are timestamped using a synchronous clock and aggregated at an edge computing terminal, facilitating the analysis of the equipment's multidimensional characteristics at any given time and the discovery of hidden faults.

[0071] Optionally, the edge computing terminal is installed on the substation side and includes a data preprocessing module, a lightweight inference engine, and a sample screening module. The edge computing terminal is built on a domestically produced AI chip and features low power consumption and real-time performance. The lightweight inference engine internally deploys a compressed lightweight multimodal model (student model) for real-time fault diagnosis of the collected data. The sample screening module monitors the confidence level of the inference results. When the diagnostic confidence level is detected to be lower than a preset threshold (e.g., 85%) or an undefined feature pattern is detected, the multimodal data set is marked as a "high-value difficult example sample" and uploaded to the cloud via an encrypted communication gateway, enabling the accurate extraction of key learning materials from massive amounts of data.

[0072] Optionally, the edge training platform is deployed at the power system master station or a private cloud server, including a multimodal large model base (teacher model), an incremental learning module, and a knowledge distillation module. The multimodal large model base has a massive parameter scale (tens of billions or more), possessing powerful generalization and multi-task processing capabilities. The incremental learning module uses received "high-value difficult example samples" to fine-tune the large model, enabling it to grasp newly emerging fault characteristics. The knowledge distillation module employs a distillation technique based on feature relationship preservation to migrate the updated knowledge of the large model to a lightweight model architecture, generating a smaller but performance-closer updated student model. The cloud training platform distributes the updated model parameter package to the edge computing terminal through an encrypted communication gateway, completing the system's online upgrade.

[0073] In one exemplary embodiment, such as Figure 4 As shown, another method for monitoring the operating status of power switchgear is provided, including the following:

[0074] A cloud-edge collaborative architecture is adopted, and a multi-modal sensor group is deployed at the edge of the substation. Among them, high-definition visible light cameras, acoustic sensors, partial discharge detectors and electrical quantity sensors collect appearance images and meter readings of switchgear, mechanical vibration audio, ultra-high frequency or ultrasonic partial discharge signals, as well as status data such as voltage, current and contact temperature.

[0075] The data preprocessing module of the edge computing terminal receives the sensing data and corresponding acquisition time sent by each sensor, performs timestamp alignment processing on the multi-source heterogeneous sensing data, eliminates the acquisition timing deviation between different sensors, and then generates standardized initial operation data.

[0076] Subsequently, the lightweight inference engine of the edge computing terminal inputs the preprocessed initial operating data into the lightweight multimodal model, and the model infers and outputs the current operating status of the power switchgear and the corresponding diagnostic confidence level. At the same time, the sample screening module, based on the preset confidence level threshold, identifies and marks unknown anomalies and new fault feature data with confidence levels below the threshold and which are difficult for the model to reliably identify as high-value difficult case samples.

[0077] The selected high-value difficult examples are securely transmitted via an encrypted communication gateway and uploaded to the cloud training platform on the main cloud site. Leveraging its powerful multimodal feature understanding capabilities, the multimodal large model on the cloud training platform automatically labels or assists experts in labeling the received unlabeled difficult examples, constructing a training sample set with fault category labels.

[0078] Based on the completed training sample set, the cloud initiates incremental training tasks and updates model weights for the multimodal large model, enabling the multimodal large model to learn and master new fault characteristics while retaining its original fault identification capabilities. Then, through knowledge distillation technology, the fault identification knowledge learned by the iteratively optimized multimodal large model is transferred and distributed to generate a second lightweight multimodal model adapted to the computing power resources of edge terminal hardware, and the corresponding model parameter package is extracted from it.

[0079] The system achieves bidirectional secure transmission of data and model parameters between the cloud and the edge through edge-side data acquisition, time-series alignment preprocessing, real-time model inference, and hard example sample screening, in conjunction with an encrypted communication gateway. Combined with the complete process of cloud-based multimodal large model sample annotation, incremental training, and knowledge distillation, a closed-loop iterative mechanism is formed, ultimately enabling the entire system to continuously self-evolve and iteratively optimize its ability to identify faults in power switchgear.

[0080] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0081] Based on the same inventive concept, this application also provides an operating status monitoring device for power switchgear, used to implement the above-described method for monitoring the operating status of power switchgear. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the operating status monitoring device for power switchgear provided below can be found in the limitations of the operating status monitoring method for power switchgear described above, and will not be repeated here.

[0082] In one exemplary embodiment, such as Figure 5 As shown, a power switchgear operation status monitoring device 500 is provided, comprising: a receiving module 501, an uploading module 502, and an updating module 503, wherein:

[0083] The receiving module 501 is used to receive the initial operating data of the power switchgear sent by the multimodal sensor group in real time, and to filter the initial operating data based on the first lightweight multimodal model to obtain the target operating data.

[0084] The upload module 502 is used to upload the target running data to the cloud and receive the model parameters returned by the cloud; wherein the model parameters are obtained by the cloud after performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after performing incremental training on the multimodal large model based on the target running data.

[0085] The update module 503 is used to update the first lightweight multimodal model based on the model parameters, and to monitor the operating status of the power switchgear based on the updated lightweight multimodal model.

[0086] In some embodiments, the receiving module 501 is further configured to input any initial running data into the first lightweight multimodal model, output the confidence level of the initial running data, and determine the initial running data as the target running data if the confidence level is less than a confidence level threshold.

[0087] In some embodiments, the update module 503 is further configured to receive the current operating data of the power switchgear sent by the multimodal sensor group; input the current operating data into the updated lightweight multimodal model; and output the current operating status of the power switchgear.

[0088] In some embodiments, the receiving module 501 is further configured to receive sensing data of the power switchgear sent by any one of the sensors in the multimodal sensor group, and the acquisition time of the sensing data; to timestamp-align the acquisition times of all sensing data, and to generate the initial operating data based on the timestamp-aligned sensing data.

[0089] In some embodiments, the power switchgear operation status monitoring device 500 is specifically used for the model parameters being obtained by the cloud from a second lightweight multimodal model, which is generated by the cloud after performing knowledge distillation on the trained multimodal large model.

[0090] In some embodiments, the power switchgear operation status monitoring device 500 is also used to ensure that all data transmitted between the edge computing terminal and the cloud is securely verified through an encrypted communication gateway.

[0091] Each module in the aforementioned power switchgear operation status monitoring device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0092] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores monitoring data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for monitoring the operating status of power switching equipment.

[0093] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0094] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0095] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0096] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0097] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0098] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0099] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method of operating state monitoring of a power switching device, characterized by, An edge computing terminal for the power grid side; the method includes: The system receives initial operating data of the power switchgear from the multimodal sensor group in real time, and filters the initial operating data based on the first lightweight multimodal model to obtain target operating data. The target running data is uploaded to the cloud, and the model parameters returned by the cloud are received; wherein, the model parameters are obtained by the cloud after performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after performing incremental training on the multimodal large model based on the target running data; The first lightweight multimodal model is updated based on the model parameters, and the operating status of the power switchgear is monitored based on the updated lightweight multimodal model.

2. The method according to claim 1, characterized in that, The process of filtering the initial running data based on the first lightweight multimodal model to obtain the target running data includes: Input any initial running data into the first lightweight multimodal model and output the confidence level of the initial running data; If the confidence level is less than the confidence level threshold, the initial running data is determined as the target running data.

3. The method according to claim 1, characterized in that, The monitoring of the operating status of the power switchgear based on the updated lightweight multimodal model includes: Receive the current operating data of the power switchgear sent by the multimodal sensor group; The current operating data is input into the updated lightweight multimodal model, and the current operating status of the power switchgear is output.

4. The method according to claim 1, characterized in that, The real-time reception of initial operating data of the power switchgear sent by the multimodal sensor group includes: Receive sensing data of the power switchgear sent by any one of the sensors in the multimodal sensor group, as well as the acquisition time of the sensing data; The acquisition times of all sensor data are timestamped, and the initial running data is generated based on the timestamped sensor data.

5. The method according to claim 1, characterized in that, The model parameters are obtained by the cloud from the second lightweight multimodal model, which is generated by the cloud after performing knowledge distillation on the trained multimodal large model.

6. The method according to claim 1, characterized in that, All data transmitted between the edge computing terminal and the cloud is securely verified through an encrypted communication gateway.

7. A monitoring system for the operating status of power switchgear, characterized in that, The system includes: An edge computing terminal is used to receive initial operating data of power switchgear sent by a multimodal sensor group in real time, and filter the initial operating data based on a first lightweight multimodal model to obtain target operating data; upload the target operating data to the cloud and receive model parameters returned by the cloud; update the first lightweight multimodal model based on the model parameters, and monitor the operating status of the power switchgear based on the updated lightweight multimodal model; The cloud platform receives target running data uploaded by the edge computing terminal; performs incremental training on the multimodal large model based on the target running data to obtain the trained multimodal large model; performs knowledge distillation on the trained multimodal large model to obtain model parameters; and sends the model parameters to the edge computing terminal.

8. A device for monitoring the operating status of power switchgear, characterized in that, An edge computing terminal for the power grid side; the device includes: The receiving module is used to receive the initial operating data of the power switchgear sent by the multimodal sensor group in real time, and to filter the initial operating data based on the first lightweight multimodal model to obtain the target operating data. The upload module is used to upload the target running data to the cloud and receive the model parameters returned by the cloud; wherein the model parameters are obtained by the cloud after performing knowledge distillation on the trained multimodal large model, and the trained multimodal large model is obtained by the cloud after performing incremental training on the multimodal large model based on the target running data; The update module is used to update the first lightweight multimodal model based on the model parameters, and to monitor the operating status of the power switchgear based on the updated lightweight multimodal model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

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

Images