Device operation data processing method and system for edge computing gateway

Abstract

The embodiment of the application provides a device operation data processing method and system for an edge computing gateway, first, a template device operation activity data sequence of a target edge computing gateway is acquired, wherein a plurality of template device operation activity data are included. Then, self-attention feature extraction is performed on each template device operation activity data, and according to a determined self-attention feature sequence, a device operation directed graph of each template device operation activity data is acquired. The device operation directed graph reflects an abnormal attention node of the template device operation activity data. Finally, the device operation directed graphs of the plurality of template device operation activity data are used to train a generated device anomaly diagnosis network, so that the device abnormality in the edge computing gateway can be effectively detected and diagnosed.

Description

Technical Field

[0001] This invention relates to the field of edge computing gateway technology, and more specifically, to a method and system for processing device operation data of an edge computing gateway. Background Technology

[0002] In edge computing networks, device anomaly detection presents a critical challenge due to the large number of devices and the complex operating environment. Current methods for device anomaly detection and diagnosis primarily rely on traditional approaches based on rules, statistical models, or machine learning. These methods often require extensive manual definition of rules or features and may lack accuracy or adaptability to the complex edge computing environment.

[0003] Therefore, a new method is needed that can automatically learn and effectively detect and diagnose device anomalies. Edge computing gateways, as key nodes connecting terminal devices and cloud platforms, play a crucial role in the entire edge computing architecture. Therefore, the detection and diagnosis of device anomalies in edge computing gateways has significant research value and practical application needs.

[0004] In recent years, techniques such as self-attention mechanisms and graph neural networks have made significant progress in the processing of sequential and graph data. Self-attention mechanisms can extract key features from sequential data, while graph neural networks can effectively model and analyze graph-structured data. However, a method that comprehensively utilizes self-attention mechanisms and graph neural networks is currently lacking to address the problem of anomaly detection and diagnosis in edge computing gateway devices. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method and system for processing device operation data of an edge computing gateway. First, a sequence of template device operation activity data for the target edge computing gateway is obtained, including multiple template device operation activity data sets. Then, self-attention features are extracted from each template device operation activity data set, and a directed graph of device operation activity data for each template device operation activity data set is obtained based on the determined self-attention feature sequence. This directed graph of device operation activity data reflects the abnormal attention nodes of the template device operation activity data. Finally, the directed graphs of device operation activity data from multiple template devices are used to train and generate a device anomaly diagnosis network, thereby effectively detecting and diagnosing device anomalies in the edge computing gateway.

[0006] According to one aspect of the present invention, a method and system for processing device operation data of an edge computing gateway are provided, the method comprising:

[0007] Obtain the template device operation activity data sequence of the target edge computing gateway, wherein the template device operation activity data sequence includes multiple template device operation activity data;

[0008] Self-attention features are extracted from the operational activity data of each template device, and a directed graph of device operation is obtained based on the determined self-attention feature sequence. The directed graph of device operation reflects the abnormal attention nodes of the operational activity data of the template device.

[0009] Based on the directed graph of device operation data from the multiple template devices, the device anomaly diagnosis network is trained and generated.

[0010] In an alternative implementation, the step of extracting self-attention features from the operational activity data of each template device and obtaining the directed graph of device operation data for each template device based on the determined self-attention feature sequence includes:

[0011] For each template device operation activity data, self-attention features are extracted from the template device operation activity data according to a preset self-attention network, the self-attention positioning operation link of the template device operation activity data is determined, and the self-attention positioning operation link is stored as self-attention positioning data.

[0012] The template device operation activity data is loaded into a preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the first encoded directed graph data of the template device operation activity data.

[0013] The self-attention localization data is loaded into the preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the second encoded directed graph data of the self-attention localization data.

[0014] By fusing the first coded directed graph data and the second coded directed graph data, a device operation directed graph of the template device operation activity data is generated.

[0015] In an alternative implementation, training the device anomaly diagnosis network based on the directed graph of device operation data from the plurality of template devices includes:

[0016] Based on the directed graph of device operation data of the multiple template devices, the weight parameters of the preset GAN model are updated to generate the device anomaly diagnosis network.

[0017] The GAN model includes a feature extraction network and a prediction network, and the feature extraction network includes a feature extraction unit and a feature restoration unit.

[0018] The process of updating the weight parameters of the preset GAN model based on the directed graph of device operation data from the multiple template devices includes:

[0019] Based on the preset reconstruction loss function, the weight parameters of the directed graph of the running activity data of the multiple template devices are updated, the first weight information of the feature extraction unit and the first weight information of the feature restoration unit are determined, and the depth-coded directed graph data generated by the feature extraction unit after extracting features from the directed graph of the running activity data of the multiple template devices based on its first weight information is obtained.

[0020] Based on the preset training adversarial supervision cost function, the weight parameters of the deep encoded directed graph data are updated to determine the second weight information of the prediction network, the second weight information of the feature extraction unit, and the second weight information of the feature restoration unit.

[0021] In an alternative implementation, the plurality of template device operation activity data includes active device operation activity data and passive device operation activity data;

[0022] The step of training and generating the device anomaly diagnosis network based on the directed graph of device operation data from the multiple template devices includes:

[0023] When the weight of negative device operation activity data in the template device operation activity data sequence is greater than the weight of positive device operation activity data, during the current training phase of training the device anomaly diagnosis network, the negative device operation activity data in the template device operation activity data sequence is traversed to generate multiple traversed negative device operation activity data, and the number of traversed negative device operation activity data is the same as the number of positive device operation activity data.

[0024] The current training phase is conducted based on the directed graph of device operation data of the positive device operation activity data and the directed graph of device operation data of the negative device operation activity data.

[0025] In the next training phase of training the device anomaly diagnosis network, the remaining negative device operation activity data in the template device operation activity data sequence, excluding the traversed negative device operation activity data, are traversed to generate multiple new traversed negative device operation activity data. The number of the new traversed negative device operation activity data is the same as the number of the positive device operation activity data.

[0026] Based on the directed graph of device operation data of the positive device operation activity data and the new directed graph of device operation data of the negative device operation activity data, the next training phase is carried out.

[0027] In an alternative implementation, the method further includes:

[0028] Obtain operational activity data of candidate devices;

[0029] Self-attention features are extracted from the candidate device operation activity data, and a device operation directed graph of the candidate device operation activity data is obtained based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data.

[0030] Based on the directed graph of the candidate device operation activity data and the preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data to determine the anomaly diagnosis result of the candidate device operation activity data.

[0031] In an alternative implementation, the preset device anomaly diagnosis network includes a feature extraction network and a prediction network; the feature extraction network includes a feature extraction unit and a feature restoration unit.

[0032] Based on the directed graph of the candidate device operation activity data and a preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data, and the anomaly diagnosis results of the candidate device operation activity data include:

[0033] The device operation directed graph of the candidate device operation activity data is loaded into the feature extraction network, and the depth-coded directed graph data generated by the feature extraction unit of the feature extraction network after performing feature extraction on the device operation directed graph is obtained.

[0034] The deep-encoded directed graph data is loaded into the prediction network to determine the abnormal diagnostic results of the candidate device's operational activity data.

[0035] According to another aspect of the present invention, a method and system for processing device operation data of an edge computing gateway are provided, the system comprising:

[0036] The first acquisition module is used to acquire a sequence of template device operation activity data, wherein the sequence of template device operation activity data includes multiple template device operation activity data.

[0037] The first extraction module is used to extract self-attention features from the operational activity data of each template device, and to obtain the directed graph of device operation corresponding to each operational activity data of the template device based on the determined self-attention feature sequence. The directed graph of device operation reflects the abnormal attention nodes of the operational activity data of the template device.

[0038] The training module is used to train and generate the device anomaly diagnosis network based on the directed graph of device operation data of the multiple template devices.

[0039] In an alternative implementation, the system further includes:

[0040] The second acquisition module is used to acquire operational activity data of candidate devices;

[0041] The second extraction module is used to extract self-attention features from the candidate device operation activity data, and obtain the device operation directed graph of the candidate device operation activity data based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data.

[0042] The diagnostic module is used to perform anomaly diagnosis on the candidate device operation activity data based on the device operation directed graph and a preset device anomaly diagnosis network, and to determine the anomaly diagnosis result of the candidate device operation activity data.

[0043] According to another aspect of the present invention, a server is provided, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory is used to store a computer program; and the processor is used to execute the computer program to implement the device operation data processing method steps for an edge computing gateway as described above.

[0044] According to another aspect of the present invention, a readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, can perform the steps of the above-described device operation data processing method for an edge computing gateway.

[0045] To make the above-mentioned objects, features and advantages of the embodiments of the present invention more apparent and understandable, a detailed description will be given below in conjunction with the embodiments and the accompanying drawings. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A schematic diagram of the components of a server provided in an embodiment of the present invention is shown;

[0048] Figure 2 A flowchart illustrating the device operation data processing method for an edge computing gateway provided in an embodiment of the present invention is shown.

[0049] Figure 3 This invention provides a functional block diagram of a device operation data processing system for an edge computing gateway, based on an embodiment of the present invention. Detailed Implementation

[0050] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] The terms “first,” “second,” “third,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0052] Figure 1An exemplary component diagram of server 100 is shown. Server 100 may include one or more processors 104, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. Server 100 may also include any storage medium 106 for storing any kind of information such as code, settings, data, etc. Without limitation, for example, storage medium 106 may include any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any storage medium can use any technology to store information. Further, any storage medium may provide volatile or non-volatile retention of information. Further, any storage medium may represent a fixed or removable component of server 100. In one case, server 100 may perform any operation of the associated instructions when processor 104 executes associated instructions stored in any storage medium or combination of storage media. Server 100 also includes one or more drive units 108 for interacting with any storage medium, such as hard disk drive units, optical disk drive units, etc.

[0053] Server 100 also includes input / output 110 (I / O) for receiving various inputs (via input unit 112) and providing various outputs (via output unit 114). A specific output mechanism may include a presentation device 116 and an associated graphical user interface (GUI) 118. Server 100 may also include one or more network interfaces 120 for exchanging data with other devices via one or more communication units 122. One or more communication buses 124 couple the components described above together.

[0054] The communication unit 122 can be implemented in any manner, such as via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, or any combination thereof. The communication unit 122 may include any combination of hardwired links, wireless links, routers, gateway functions, name server 100, etc., governed by any protocol or combination of protocols.

[0055] Figure 2 This diagram illustrates a flowchart of a device operation data processing method and system for an edge computing gateway provided by an embodiment of the present invention. This device operation data processing method and system for an edge computing gateway can be derived from... Figure 1 The detailed steps of the data processing method executed by the server 100 shown are described below for the device running the edge computing gateway.

[0056] Step S110: Obtain the template device operation activity data sequence of the target edge computing gateway, wherein the template device operation activity data sequence includes multiple template device operation activity data;

[0057] Step S120: Self-attention features are extracted for each template device operation activity data, and a device operation directed graph is obtained for each template device operation activity data based on the determined self-attention feature sequence. The device operation directed graph reflects the abnormal attention nodes of the template device operation activity data.

[0058] Step S130: Based on the directed graph of device operation data of the multiple template devices, train and generate the device anomaly diagnosis network.

[0059] In an alternative implementation, the method further includes:

[0060] Step S140: Obtain candidate device operation activity data;

[0061] Step S150: Extract self-attention features from the candidate device operation activity data, and obtain the device operation directed graph of the candidate device operation activity data based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data.

[0062] Step S160: Based on the directed graph of the candidate device operation activity data and the preset device anomaly diagnosis network, perform anomaly diagnosis on the candidate device operation activity data to determine the anomaly diagnosis result of the candidate device operation activity data.

[0063] Based on the above steps, this embodiment first obtains the template device operation activity data sequence of the target edge computing gateway, including multiple template device operation activity data. Then, self-attention features are extracted from each template device operation activity data, and a directed graph of device operation is obtained for each template device operation activity data based on the determined self-attention feature sequence. This directed graph of device operation reflects the abnormal attention nodes of the template device operation activity data. Finally, the directed graph of device operation of multiple template device operation activity data is used to train and generate a device anomaly diagnosis network, thereby effectively detecting and diagnosing device anomalies in the edge computing gateway.

[0064] In an alternative implementation, the step of extracting self-attention features from the operational activity data of each template device and obtaining the directed graph of device operation data for each template device based on the determined self-attention feature sequence includes:

[0065] For each template device operation activity data, self-attention features are extracted from the template device operation activity data according to a preset self-attention network, the self-attention positioning operation link of the template device operation activity data is determined, and the self-attention positioning operation link is stored as self-attention positioning data.

[0066] The template device operation activity data is loaded into a preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the first encoded directed graph data of the template device operation activity data.

[0067] The self-attention localization data is loaded into the preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the second encoded directed graph data of the self-attention localization data.

[0068] By fusing the first coded directed graph data and the second coded directed graph data, a device operation directed graph of the template device operation activity data is generated.

[0069] In an alternative implementation, training the device anomaly diagnosis network based on the directed graph of device operation data from the plurality of template devices includes:

[0070] Based on the directed graph of device operation data of the multiple template devices, the weight parameters of the preset GAN model are updated to generate the device anomaly diagnosis network.

[0071] The GAN model includes a feature extraction network and a prediction network, and the feature extraction network includes a feature extraction unit and a feature restoration unit.

[0072] The process of updating the weight parameters of the preset GAN model based on the directed graph of device operation data from the multiple template devices includes:

[0073] Based on the preset reconstruction loss function, the weight parameters of the directed graph of the running activity data of the multiple template devices are updated, the first weight information of the feature extraction unit and the first weight information of the feature restoration unit are determined, and the depth-coded directed graph data generated by the feature extraction unit after extracting features from the directed graph of the running activity data of the multiple template devices based on its first weight information is obtained.

[0074] Based on the preset training adversarial supervision cost function, the weight parameters of the deep encoded directed graph data are updated to determine the second weight information of the prediction network, the second weight information of the feature extraction unit, and the second weight information of the feature restoration unit.

[0075] In an alternative implementation, the plurality of template device operation activity data includes active device operation activity data and passive device operation activity data;

[0076] The step of training and generating the device anomaly diagnosis network based on the directed graph of device operation data from the multiple template devices includes:

[0077] When the weight of negative device operation activity data in the template device operation activity data sequence is greater than the weight of positive device operation activity data, during the current training phase of training the device anomaly diagnosis network, the negative device operation activity data in the template device operation activity data sequence is traversed to generate multiple traversed negative device operation activity data, and the number of traversed negative device operation activity data is the same as the number of positive device operation activity data.

[0078] The current training phase is conducted based on the directed graph of device operation data of the positive device operation activity data and the directed graph of device operation data of the negative device operation activity data.

[0079] In the next training phase of training the device anomaly diagnosis network, the remaining negative device operation activity data in the template device operation activity data sequence, excluding the traversed negative device operation activity data, are traversed to generate multiple new traversed negative device operation activity data. The number of the new traversed negative device operation activity data is the same as the number of the positive device operation activity data.

[0080] Based on the directed graph of device operation data of the positive device operation activity data and the new directed graph of device operation data of the negative device operation activity data, the next training phase is carried out.

[0081] In an alternative implementation, the method further includes:

[0082] Obtain operational activity data of candidate devices;

[0083] Self-attention features are extracted from the candidate device operation activity data, and a device operation directed graph of the candidate device operation activity data is obtained based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data.

[0084] Based on the directed graph of the candidate device operation activity data and the preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data to determine the anomaly diagnosis result of the candidate device operation activity data.

[0085] In an alternative implementation, the preset device anomaly diagnosis network includes a feature extraction network and a prediction network; the feature extraction network includes a feature extraction unit and a feature restoration unit.

[0086] Based on the directed graph of the candidate device operation activity data and a preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data, and the anomaly diagnosis results of the candidate device operation activity data include:

[0087] The device operation directed graph of the candidate device operation activity data is loaded into the feature extraction network, and the depth-coded directed graph data generated by the feature extraction unit of the feature extraction network after performing feature extraction on the device operation directed graph is obtained.

[0088] The deep-encoded directed graph data is loaded into the prediction network to determine the abnormal diagnostic results of the candidate device's operational activity data.

[0089] Figure 3 A functional block diagram of a device operation data processing system 200 for an edge computing gateway, according to an embodiment of the present invention, is shown. The functions implemented by this device operation data processing system 200 for an edge computing gateway correspond to the steps performed by the above-described method. This device operation data processing system 200 for an edge computing gateway can be understood as the aforementioned server 100, or the processor of server 100, or it can be understood as a component that implements the functions of the present invention under the control of server 100, independent of server 100 or the processor. Figure 3 As shown below, the functions of each functional module of the device operation data processing system 200 based on the edge computing gateway will be described in detail.

[0090] The first acquisition module 210 is used to acquire a sequence of template device operation activity data, wherein the sequence of template device operation activity data includes multiple template device operation activity data.

[0091] The first extraction module 220 is used to extract self-attention features from the operation activity data of each template device, and obtain the directed graph of device operation corresponding to each template device operation activity data based on the determined self-attention feature sequence. The directed graph of device operation reflects the abnormal attention nodes of the operation activity data of the template device.

[0092] Training module 230 is used to train and generate the device anomaly diagnosis network based on the directed graph of device operation data of the multiple template devices.

[0093] In an alternative implementation, the system further includes:

[0094] The second acquisition module 240 is used to acquire candidate device operation activity data;

[0095] The second extraction module 250 is used to extract self-attention features from the candidate device operation activity data and obtain the device operation directed graph of the candidate device operation activity data based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data.

[0096] The diagnostic module 260 is used to perform anomaly diagnosis on the candidate device operation activity data based on the device operation directed graph of the candidate device operation activity data and a preset device anomaly diagnosis network, and to determine the anomaly diagnosis result of the candidate device operation activity data.

[0097] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0098] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

Claims

1. A method for processing device operation data of an edge computing gateway, characterized in that, The method includes: Obtain the template device operation activity data sequence of the target edge computing gateway, wherein the template device operation activity data sequence includes multiple template device operation activity data; Self-attention features are extracted from the operational activity data of each template device, and a directed graph of device operation is obtained based on the determined self-attention feature sequence. The directed graph of device operation reflects the abnormal attention nodes of the operational activity data of the template device. Based on the directed graph of device operation data from the multiple template devices, a device anomaly diagnosis network is trained and generated.

2. The device operation data processing method for an edge computing gateway according to claim 1, characterized in that, The step of extracting self-attention features from the operational activity data of each template device and obtaining the directed graph of device operation data for each template device based on the determined self-attention feature sequence includes: For each template device operation activity data, self-attention features are extracted from the template device operation activity data according to a preset self-attention network, the self-attention positioning operation link of the template device operation activity data is determined, and the self-attention positioning operation link is stored as self-attention positioning data. The template device operation activity data is loaded into a preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the first encoded directed graph data of the template device operation activity data; The self-attention localization data is loaded into the preset graph autoencoder to generate the encoded data of the preset graph autoencoder, and the encoded data is used as the second encoded directed graph data of the self-attention localization data. By fusing the first coded directed graph data and the second coded directed graph data, a device operation directed graph of the template device operation activity data is generated.

3. The device operation data processing method for an edge computing gateway according to claim 1 or 2, characterized in that, The step of training and generating the device anomaly diagnosis network based on the directed graph of device operation data from the multiple template devices includes: Based on the directed graph of device operation data of the multiple template devices, the weight parameters of the preset GAN model are updated to generate the device anomaly diagnosis network. The GAN model includes a feature extraction network and a prediction network, and the feature extraction network includes a feature extraction unit and a feature restoration unit. The process of updating the weight parameters of the preset GAN model based on the directed graph of device operation data from the multiple template devices includes: Based on the preset reconstruction loss function, the weight parameters of the directed graph of the running activity data of the multiple template devices are updated, the first weight information of the feature extraction unit and the first weight information of the feature restoration unit are determined, and the depth-coded directed graph data generated by the feature extraction unit after extracting features from the directed graph of the running activity data of the multiple template devices based on its first weight information is obtained. Based on the preset training adversarial supervision cost function, the weight parameters of the deep encoded directed graph data are updated to determine the second weight information of the prediction network, the second weight information of the feature extraction unit, and the second weight information of the feature restoration unit.

4. The device operation data processing method for an edge computing gateway according to claim 1, characterized in that, The operational activity data of the multiple template devices includes active equipment operational activity data and passive equipment operational activity data; The step of training and generating the device anomaly diagnosis network based on the directed graph of device operation data from the multiple template devices includes: When the weight of negative device operation activity data in the template device operation activity data sequence is greater than the weight of positive device operation activity data, during the current training phase of training the device anomaly diagnosis network, the negative device operation activity data in the template device operation activity data sequence is traversed to generate multiple traversed negative device operation activity data, and the number of traversed negative device operation activity data is the same as the number of positive device operation activity data. The current training phase is conducted based on the directed graph of device operation data of the positive device operation activity data and the directed graph of device operation data of the negative device operation activity data. In the next training phase of training the device anomaly diagnosis network, the remaining negative device operation activity data in the template device operation activity data sequence, excluding the traversed negative device operation activity data, are traversed to generate multiple new traversed negative device operation activity data. The number of the new traversed negative device operation activity data is the same as the number of the positive device operation activity data. Based on the directed graph of device operation data of the positive device operation activity data and the new directed graph of device operation data of the negative device operation activity data, the next training phase is carried out.

5. The device operation data processing method for an edge computing gateway according to claim 1, characterized in that, The method further includes: Obtain operational activity data of candidate devices; Self-attention features are extracted from the candidate device operation activity data, and a device operation directed graph of the candidate device operation activity data is obtained based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data. Based on the directed graph of the candidate device operation activity data and the preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data to determine the anomaly diagnosis result of the candidate device operation activity data.

6. The device operation data processing method for an edge computing gateway according to claim 5, characterized in that, The preset equipment anomaly diagnosis network includes a feature extraction network and a prediction network; the feature extraction network includes a feature extraction unit and a feature restoration unit. Based on the directed graph of the candidate device operation activity data and a preset device anomaly diagnosis network, anomaly diagnosis is performed on the candidate device operation activity data, and the anomaly diagnosis results of the candidate device operation activity data include: The device operation directed graph of the candidate device operation activity data is loaded into the feature extraction network to obtain the depth-coded directed graph data generated by the feature extraction unit of the feature extraction network after performing feature extraction on the device operation directed graph; The deep-encoded directed graph data is loaded into the prediction network to determine the abnormal diagnostic results of the candidate device's operational activity data.

7. A device operation data processing system for an edge computing gateway, characterized in that, include: The first acquisition module is used to acquire a sequence of template device operation activity data, wherein the sequence of template device operation activity data includes multiple template device operation activity data. The first extraction module is used to extract self-attention features from the operational activity data of each template device, and to obtain the directed graph of device operation corresponding to each operational activity data of the template device based on the determined self-attention feature sequence. The directed graph of device operation reflects the abnormal attention nodes of the operational activity data of the template device. The training module is used to train and generate a device anomaly diagnosis network based on the directed graph of device operation data from the multiple template devices.

8. The device operation data processing system for edge computing gateways according to claim 7, characterized in that, Also includes: The second acquisition module is used to acquire operational activity data of candidate devices; The second extraction module is used to extract self-attention features from the candidate device operation activity data, and obtain the device operation directed graph of the candidate device operation activity data based on the determined self-attention feature sequence. The device operation directed graph of the candidate device operation activity data reflects the abnormal attention nodes of the candidate device operation activity data. The diagnostic module is used to perform anomaly diagnosis on the candidate device operation activity data based on the device operation directed graph and a preset device anomaly diagnosis network, and to determine the anomaly diagnosis result of the candidate device operation activity data.

9. A server, characterized in that, include: The processor, communication interface, memory, and communication bus are provided, wherein the processor, communication interface, and memory communicate with each other via the communication bus. The memory is used to store computer programs; the processor is used to execute the computer programs to implement the device operation data processing method steps for an edge computing gateway as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the device operation data processing method for an edge computing gateway as described in any one of claims 1-6.

Images