Maintenance method and system for edge computing device
By preprocessing and predicting the system monitoring data of edge computing devices, combining self-attention mechanism and maintenance decision tree, the problem of lack of in-depth analysis and predictive maintenance in the existing technology is solved, and efficient equipment management and cost reduction are achieved .
Patent Information
- Application Number
- CN202411018464.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-05
AI Technical Summary
The existing edge computing device management system lacks in-depth data analysis and fault prediction capabilities, resulting in the inability to achieve predictive maintenance and high maintenance costs.
By obtaining the system monitoring data of edge computing devices for preprocessing, and using the self-attention mechanism in the Informer model for prediction, combined with the pre-built maintenance decision tree, equipment failures are judged and corresponding maintenance decisions are triggered to achieve predictive maintenance.
It improves the intelligence level of equipment management, reduces manual maintenance costs, and realizes efficient management and predictive maintenance of edge computing devices.
Smart Images

Figure CN118898030A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of edge computing and intelligent monitoring, and in particular to a maintenance method and system for edge computing equipment. Background Art
[0002] The focus of early edge computing research was on how to effectively distribute computing tasks and process data. As technology continues to develop, equipment maintenance and management issues have begun to receive attention. Traditional equipment management systems mainly rely on centralized monitoring solutions, which are inadequate when dealing with edge computing devices. In addition, early systems were unable to effectively predict equipment failures and often responded only after equipment problems occurred, resulting in system downtime and high maintenance costs.
[0003] After entering the intelligent era, the existing equipment management system began to integrate more automation tools and technologies, such as sensor technology and Internet of Things technology, so that equipment status data can be collected, transmitted and processed in real time. However, the level of intelligence and predictive maintenance capabilities are still limited. Most of the existing systems are still in the monitoring and reactive maintenance stage, lacking in-depth data analysis and fault prediction capabilities. In addition, these systems can collect equipment operation data, but lack sufficient algorithm support to analyze this data, making it impossible to achieve early prediction of faults and preventive maintenance. Therefore, it is particularly important to develop a highly intelligent management system that integrates the latest artificial intelligence algorithms to achieve predictive maintenance. Summary of the invention
[0004] In order to solve the above technical problems, the present invention provides a maintenance method and system for edge computing equipment, which determines whether a fault occurs and executes corresponding maintenance decisions by predicting the preprocessed system monitoring data, aiming to improve the intelligence level of equipment management and maintenance and reduce manual maintenance costs.
[0005] The present invention is achieved through the following technical solutions: In a first aspect, the present invention provides a method for maintaining an edge computing device, comprising: obtaining current system monitoring data related to maintenance of the edge computing device; Preprocessing the current system monitoring data; Input the preprocessed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; Matching the prediction system monitoring data with a pre-built maintenance decision tree to determine the fault type and trigger a corresponding maintenance decision; The prediction model is obtained by introducing a self-attention mechanism module into the encoder of the Informer model.
[0006] The present invention determines whether to query the fault type from the pre-established maintenance decision tree based on the prediction results of the system monitoring data. If a fault occurs or there is a risk of a possible fault, the corresponding maintenance decision is executed. It not only realizes the efficient management of edge computing devices, but also combines the deep learning model to provide predictive maintenance for edge devices, reduces the maintenance costs brought by traditional corrective maintenance and preventive maintenance, and improves the intelligence level of the equipment management system.
[0007] Optionally, after the preprocessed current system monitoring data is input into the pre-trained prediction model, the self-attention mechanism module is used to perform the following operations: Perform a dot product operation on the input query Q and key K, and scale it by the square root of dimension D to get the attention score matrix; Apply the mask to the attention score matrix and process it with the softmax function to get a new matrix; Based on the new matrix, for each query Q, select the top n keys K and values V with the highest attention scores and recalculate their attention scores; Finally, the value V is weighted and summed and output.
[0008] Optionally, the maintenance decision tree includes threshold conditions for possible failures of the edge computing device, corresponding failure types, and corresponding maintenance decisions.
[0009] Optionally, matching the prediction system monitoring data with a pre-built maintenance decision tree to determine the fault type and trigger a corresponding maintenance decision includes: The monitoring data of the prediction system is compared with the threshold conditions of possible failures of edge computing devices set in the maintenance decision tree, and the fault type is determined based on the comparison results to trigger the corresponding maintenance decision.
[0010] Optionally, obtaining the system monitoring data of the edge computing device includes: parsing a system file through a python script to obtain the system monitoring data.
[0011] Optionally, the training method of the prediction model includes: Obtain samples of system monitoring data related to maintenance of edge computing devices; Divide the system monitoring data samples into training set samples and test set samples; Preprocessing the system monitoring data in the training set samples and the test set samples; Input the system monitoring data in the preprocessed training set samples into the prediction model obtained based on the improvement of the Informer model to train the prediction model; The prediction accuracy of the trained prediction model is tested through the test set samples. If the prediction accuracy reaches the preset accuracy requirement, the training is stopped; otherwise, the prediction model is continued to be trained through the training set samples.
[0012] Optionally, the preprocessing includes: The isolation forest outlier detection algorithm is used to filter the acquired system monitoring data and delete sudden abnormal data; The filtered system monitoring data is normalized.
[0013] In a second aspect, the present invention provides an intelligent maintenance unit for an edge computing device, comprising: A storage module, used to store edge computing device information and obtain and store system monitoring data related to the maintenance of the edge computing device; A preprocessing module, used for preprocessing the system monitoring data, wherein the system monitoring data includes current system monitoring data; The abnormal prediction module is used to input the pre-processed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; A maintenance decision generation module, used to match the prediction system monitoring data with a pre-built maintenance decision tree, determine the fault type and generate a corresponding maintenance decision; The maintenance instruction issuing module is used to trigger maintenance decisions.
[0014] In a third aspect, the present invention provides an intelligent maintenance system for an edge computing device, comprising the intelligent maintenance unit as described in the second aspect, The intelligent maintenance unit and the intelligent management unit are connected in series, and the intelligent maintenance unit receives system monitoring data of the edge device unit and transmits it to the intelligent management unit; The intelligent maintenance unit further comprises an alarm module, the alarm module performs threshold detection on the system monitoring data in the storage module and issues an alarm for abnormal data; The intelligent management unit includes a user login / registration module, a device management module, a monitoring module, an abnormal alarm list generation module and a user instruction upload module; After the user registers and logs in through the user login / registration module, the device management module requests to obtain the edge computing device information stored in the storage module; the monitoring module requests to obtain the system monitoring data stored in the storage module through the edge computing device information obtained by the device management module; the abnormal alarm list generation module obtains the edge computing device information and system monitoring data in the device management module and the monitoring module, filters the alarm information, and displays it on the page; the maintenance decision generation module analyzes the alarm information displayed on the page, outputs the maintenance decision to the maintenance instruction issuing module, and the maintenance instruction issuing module converts it into a shell command and sends it to a specific edge computing device; the user instruction upload module sends the shell command entered by the user to the maintenance instruction issuing module, and the maintenance instruction issuing module sends the shell command to the corresponding edge computing device In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program / instructions stored thereon, wherein when the computer program is executed by a processor, the steps of the edge computing device maintenance method as described in the first aspect are implemented.
[0015] Compared with the prior art, the edge computing device maintenance method and system provided by the present invention can at least achieve the following beneficial effects: The present invention improves the Informer model by introducing a self-attention mechanism module to achieve accurate prediction of system monitoring data; establishes a maintenance decision tree through the possible fault types and maintenance decisions of edge computing devices; and then matches the data prediction results with the data in the maintenance decision tree to determine whether the equipment has failed and whether there is a risk of failure; if a failure occurs or there is a risk of failure, the corresponding maintenance decision is selected for maintenance, thereby achieving the effect of intelligent management and maintenance, which can improve the intelligence level of equipment management and maintenance and reduce manual maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 Shown is a flow chart of an embodiment of a method for maintaining an edge computing device provided by the present invention; Figure 2 The figure is a structural block diagram of an embodiment of a maintenance system for edge computing devices provided by the present invention; Figure 3 Shown is a flow chart of a maintenance method for edge computing equipment of the present invention; Figure 4 The figure shows the flow chart of data processing by the self-attention mechanism; Figure 5 The figure shows the prediction effect diagram of the Informer model of the present invention; Figure 6The figure shows the prediction effect diagram of the improved Informer model of the present invention; Figure 7 Shown is a schematic diagram of a maintenance function of an embodiment of the present invention; Figure 8 Shown is a partial user function schematic diagram of an embodiment of the present invention; Fig. 9 Shown is a schematic diagram of part of the historical data exported according to an embodiment of the present invention. DETAILED DESCRIPTION
[0017] The following is a further description in conjunction with the accompanying drawings and specific embodiments. In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first", "second", etc. may explicitly or implicitly include one or more of the features.
[0018] Embodiment 1, this embodiment introduces a maintenance method of an edge computing device, which can be performed by an intelligent maintenance unit, and the intelligent maintenance unit can be configured as an independent server, such as Figure 1 As shown, the method specifically comprises the following steps:
[0019] obtaining current system monitoring data related to maintenance of the edge computing device; Preprocessing the current system monitoring data; Input the preprocessed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; Matching the prediction system monitoring data with a pre-built maintenance decision tree to determine the fault type and trigger a corresponding maintenance decision; The prediction model is obtained by introducing a self-attention mechanism module into the encoder of the Informer model.
[0020] The present invention improves the Informer model by introducing a self-attention mechanism module to achieve accurate prediction of system monitoring data; establishes a maintenance decision tree through the possible fault types and maintenance decisions of edge computing devices; and then matches the data prediction results with the data in the maintenance decision tree to determine whether the equipment has failed and whether there is a risk of failure; if a failure occurs or there is a risk of failure, the corresponding maintenance decision is selected for maintenance, thereby achieving the effect of intelligent management and maintenance, which can improve the intelligence level of equipment management and maintenance and reduce manual maintenance costs.
[0021] The specific implementation of each step in the method of this embodiment is described as follows: 1. Dataset Acquisition and Processing By deploying a python script on the edge computing device, it is used to collect system monitoring data of the edge computing device, such as CPU utilization, GPU utilization, network latency, network bandwidth, hardware temperature, fan speed, etc. There are two ways to obtain edge data: (1) Use python scripts to parse system files such as / proc / cpuinfo, / proc / meminfo, / sys / class / hwmon / hwmon* / device / fan* and other files to obtain the required data. (2) Use system commands such as python's re module to process the command output obtained by subprocess, and extract the required data through corresponding regular expressions. This embodiment adopts the first method to obtain the required data. After the data is acquired, the results are temporarily stored in the SQLite lightweight database using the python list structure, waiting for upload. In order to effectively reduce the pressure of data uploaded to the server, the acquired data is directly distributed and processed at the edge to reduce the delay and network load of data upload. For example, the threshold judgment method is used to judge the data generated by the edge computing device, and the alarm is directly processed locally; the feature selection algorithm is used to delete the default data, compress all data into the [0,1] interval, and then reduce the high-dimensional data to one dimension, and dynamically adjust the upload frequency according to the monitored network bandwidth and network delay to reduce the network load. Convert the data into JSON format before uploading, and then use the function of the urllib.request library to send the data to the server through HTTP POST request.
[0022] The prediction model should be trained with a data set before it is put into use. When constructing a data set, default values and outliers may be encountered for the data uploaded to the server, and the amount of data is large. Therefore, before the model is trained, the embodiment of the present invention adopts an isolation forest outlier detection algorithm, which is trained by inputting abnormal data samples, filtering the collected data, and deleting sudden abnormal data, thereby reducing noise and improving data quality. In order to optimize the accuracy of model prediction, the filtered data is normalized, a sliding window mechanism is introduced to extract features, the data is converted into a series of fixed-length data segments, and the data set is reasonably divided, and the data set is divided into a training set and a test set in a ratio of 6:4.
[0023] When the prediction model is actually used, the collected current system monitoring data can also be pre-processed using the above-mentioned processing method before being input into the prediction model. Specifically, it may include: using the isolation forest outlier detection algorithm to filter the acquired system monitoring data and delete sudden abnormal data; normalizing the filtered system monitoring data.
[0024] 2. Construction and training of prediction models 2.1 Model Construction The embodiment of the present invention improves the Informer model for long time series prediction to make it suitable for the prediction task of edge computing device system monitoring data. The improved Informer model includes an input embedding layer, an encoder, a decoder and an output layer arranged in sequence; wherein a self-attention mechanism module is introduced in the encoder.
[0025] The input embedding layer is used to receive and process the current system monitoring data, convert it into a numerical form that the model can understand, and provide a suitable input representation for the subsequent encoder part; The encoder is used to receive data from the input embedding layer, extract features from the data through transformation operations, and use the introduced self-attention mechanism module to enable the encoder to consider other time steps in the time series when processing the data of the current time step, so as to better capture the long-term dependencies between data. The self-attention mechanism module can be integrated with the original Informer model through a specific interface. In this embodiment, the self-attention mechanism can be inserted at the beginning of the encoder layer of the encoder to ensure that the data can be processed by self-attention when flowing through the encoder; The decoder is used to receive the output of the encoder and predict future system monitoring data based on the extracted features; The output layer is used to convert the system monitoring data output by the decoder, i.e., the predicted system monitoring data, into the same format or dimension as the current system monitoring data so as to be directly compared with the actual situation or used for further decision making.
[0026] The flowchart of the self-attention mechanism module processing data is as follows Figure 4 As shown, perform the following specific operations: Perform a dot product operation on the input query Q and key K, and scale it by the square root of dimension D to obtain an attention score matrix; apply a mask to the attention score matrix and process it with a softmax function to obtain a new matrix; based on the new matrix, for each query Q, select the top n keys K and values V with the highest attention scores, and recalculate their attention scores; based on the recalculated attention scores, perform a weighted sum on the values V and output them.
[0027] The comparison of the evaluation indicators of the improved Informer model and the original Informer model is shown in Table 1 below: Mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE) and mean absolute percentage error (MAPE) are selected as evaluation indicators. MAE represents the average of the absolute value of the difference between the actual value and the predicted value, MSE calculates the average of the square of the difference between the actual value and the predicted value, RMSE is the square root of MSE and is more sensitive to larger errors, and MAPE represents the ratio of the error size to the actual value. The smaller the value of the above metrics, the more accurate the model prediction.
[0028]
[0029] 2.2 Model Training The divided training set is input into the improved Informer model for training. After each round of training, the test set is input into the model for evaluation. Repeat the above process to continuously optimize the parameters of the improved Informer model. For example, the hidden layer feature dimension is set to 512, the number of attention heads is 10, the number of stacked layers of encoder and decoder is 2, the dimension of the fully connected layer is 2048, the number of sampling factors is 5, the data padding is 0, the Dropout rate is 0.05, and the learning rate is 0.0001. During the training process, the encoder input length is 96, the decoder input length is 48, the prediction sequence length is 12, and the epoch is set to 10. If the loss does not improve for 3 consecutive epochs, the training is stopped to prevent overfitting. Adam is selected as the optimizer and MAE as the loss function. It can adaptively adjust the learning strategy, improve the training efficiency, and finally obtain the trained model.
[0030] The training effect is Figure 5 and Figure 6 The prediction effect of the Informer model is as follows Figure 5 As shown in Figure 2, the prediction effect of the improved Informer model is as follows: Figure 6 shown. Figure 5 Figure 6 The blue curve represents the real data, and the red curve represents the data predicted by the Informer model. It can be seen from the comparison that the data predicted by the improved Informer model is closer to the real data and has a higher accuracy rate.
[0031] The system monitoring data to be predicted is input into the trained model, the input system monitoring data is predicted, and the system monitoring data that may appear in the future is output.
[0032] 3. Intelligent maintenance and management 3.1 Establishing a maintenance decision tree Pre-build a maintenance decision tree based on the threshold conditions, fault types, and corresponding maintenance decisions of possible failures of edge computing devices. First, obtain the historical data of the system monitoring data of the edge computing device, which can include characteristic values such as CPU utilization, GPU utilization, network latency, network bandwidth, hardware temperature, and fan speed. Define the thresholds of each characteristic value based on these historical data and environmental characteristics in advance and save them. Preset the thresholds of various performance indicators in the Spring Boot service class and integrate them into the Spring Boot controller. The fault type can be network abnormality, high temperature, insufficient memory, etc.; the corresponding maintenance decision can be reconnecting to the network, increasing the fan speed, shutting down unnecessary processes, etc.
[0033] 3.2 Intelligent Management According to the prediction results of the improved Informer model, it is evaluated whether it is necessary to retrieve the corresponding maintenance decision from the maintenance decision tree: the prediction results are matched with the conditions in the decision tree. If a certain feature value exceeds the preset threshold, the corresponding alarm mechanism and maintenance decision are triggered. The Spring Boot controller parses the output data of the model, extracts key performance indicators, and activates the corresponding alarm mechanism and maintenance decision through logical judgment. Finally, the server converts the maintenance decision into a shell command, establishes a connection with the edge computing device through the JSch library, and executes the shell command.
[0034] Taking Jeston Nano as an example, if the hardware temperature is detected to be high, the fan speed is increased. The corresponding shell command is: "sudo sh -c 'echo 255> / sys / devices / pwm-fan / target_pwm'". If the device network status is detected to be abnormal, the network connection is checked and reconnected to the network. The corresponding shell command is "sudo nmcli dev wificonnect 'wifi_name' password 'wifi_password' ifname wlan0, where wifi_name and wifi_password are the preset wifi name and password.
[0035] Embodiment 2 is based on the same inventive concept as Embodiment 1. This embodiment introduces an intelligent maintenance unit of an edge computing device, including:
[0036] A storage module, used to store edge computing device information and obtain and store system monitoring data related to the maintenance of the edge computing device; A preprocessing module, used for preprocessing the system monitoring data, wherein the system monitoring data includes current system monitoring data; The abnormal prediction module is used to input the pre-processed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; A maintenance decision generation module, used to match the prediction system monitoring data with a pre-built maintenance decision tree, determine the fault type and generate a corresponding maintenance decision; The maintenance instruction issuing module is used to trigger the maintenance decision. The implementation steps of the method in this embodiment can refer to Figure 3 .
[0037] Example 3, based on the same inventive concept as Example 2, this example introduces an intelligent maintenance system for edge computing devices, such as Figure 2 As shown, it includes an intelligent maintenance unit and an intelligent management unit, the intelligent maintenance unit and the intelligent management unit are connected in series, and the intelligent maintenance unit receives the system monitoring data of the edge device unit and transmits it to the intelligent management unit;
[0038] The intelligent maintenance unit is generated by a storage module, an alarm module, an abnormality prediction module, a maintenance decision production module and a maintenance instruction delivery module; the storage module receives system monitoring data from an edge computing device, the alarm module performs threshold detection on the system monitoring data in the storage module and issues an alarm for abnormal data, the abnormality prediction module inputs a model to the data in the storage module and outputs a prediction result, the maintenance decision generation module analyzes the result output by the abnormality prediction module and outputs a maintenance decision to the maintenance instruction delivery module, the maintenance instruction delivery module converts it into a shell command and sends it to a specific edge computing device; The intelligent management unit is composed of a user login / registration module, a device management module, a monitoring module, an abnormal alarm list generation module and a user instruction upload module; after the user registers and logs in through the user login / registration module, the backup management module, the monitoring module and the abnormal alarm list generation module request to obtain the system monitoring data in the storage module of the intelligent maintenance unit and display it on the page; the user enters a shell command in the page, sends it to the server, and the server forwards it to the corresponding edge computing device.
[0039] The intelligent maintenance unit and intelligent management unit of the present invention are presented in the form of a web page, and some screenshots are shown as follows: Figure 7-9 As shown in the figure, it provides users with an intuitive and easy-to-operate interface to facilitate monitoring and management of edge devices. Through a well-designed user interface, including clear interface elements, menus, and toolbars, the system ensures that users can quickly and intuitively find and use the required functions. At the same time, using data visualization technology, such as charts and reports, users can intuitively understand the operating trends and performance indicators of the equipment, improving the efficiency of equipment monitoring.
[0040] like Figure 7 The monitoring module and abnormal alarm list generation module of the intelligent management unit shown in the figure present the CPU utilization, memory utilization, fan speed, hardware temperature change trend and network status of the device in intuitive forms such as pie charts and curve charts. The alarm module includes an alarm list and a specific alarm information pop-up window. This design effectively integrates real-time data monitoring and abnormal alarm functions, which can help users quickly understand the status of the device.
[0041] like Figure 8 The user interface of the system of the present invention shown in the figure, after registration and login, the user can manage the edge computing device in the console, add, delete, rename, view real-time data, export historical data and other operations. The user can choose to export the historical data of the device in the form of an Excel spreadsheet for the past day, week, month, half a year or year. Some of the exported historical data are as follows Fig. 9 shown.
[0042] Embodiment 4: This embodiment introduces a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the edge computing device maintenance method introduced in Embodiment 1 are implemented.
[0043] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take 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 code.
[0044] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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.
[0045] 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.
[0046] 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 A step that specifies a function in one or more boxes.
[0047] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.
Claims
1. A method for maintaining an edge computing device, characterized in that: include: obtaining current system monitoring data related to maintenance of the edge computing device; Preprocessing the current system monitoring data; Input the preprocessed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; Matching the prediction system monitoring data with a pre-built maintenance decision tree to determine the fault type and trigger a corresponding maintenance decision; The prediction model is obtained by introducing a self-attention mechanism module into the encoder of the Informer model.
2. The method according to claim 1, characterized in that: After the preprocessed current system monitoring data is input into the pre-trained prediction model, the self-attention mechanism module is used to perform the following operations: Perform a dot product operation on the input query Q and key K, and scale it by the square root of dimension D to get the attention score matrix; Apply the mask to the attention score matrix and process it with the softmax function to get a new matrix; Based on the new matrix, for each query Q, select the top n keys K and values V with the highest attention scores and recalculate their attention scores; Finally, the value V is weighted and summed and output.
3. The method according to claim 1, characterized in that: The maintenance decision tree includes threshold conditions for possible failures of edge computing devices, corresponding failure types, and corresponding maintenance decisions.
4. The method according to claim 1, characterized in that: The predictive system monitoring data is matched with a pre-built maintenance decision tree to determine the fault type and trigger a corresponding maintenance decision, including: The monitoring data of the prediction system is compared with the threshold conditions of possible failures of edge computing devices set in the maintenance decision tree, and the fault type is determined based on the comparison results to trigger the corresponding maintenance decision.
5. The method according to claim 1, characterized in that: The obtaining of the system monitoring data of the edge computing device includes: obtaining the system monitoring data by parsing a system file through a python script.
6. The method according to claim 1, characterized in that: The training method of the prediction model comprises: Obtain samples of system monitoring data related to maintenance of edge computing devices; Divide the system monitoring data samples into training set samples and test set samples; Preprocessing the system monitoring data in the training set samples and the test set samples; Input the system monitoring data in the preprocessed training set samples into the prediction model obtained based on the improvement of the Informer model to train the prediction model; The prediction accuracy of the trained prediction model is tested through the test set samples. If the prediction accuracy reaches the preset accuracy requirement, the training is stopped; otherwise, the prediction model is continued to be trained through the training set samples.
7. The method according to any one of claims 1 or 6, characterized in that: The preprocessing comprises: The isolation forest outlier detection algorithm is used to filter the acquired system monitoring data and delete sudden abnormal data; The filtered system monitoring data is normalized.
8. An intelligent maintenance unit for an edge computing device, characterized in that: include A storage module, used to store edge computing device information and obtain and store system monitoring data related to the maintenance of the edge computing device; A preprocessing module, used for preprocessing the system monitoring data, wherein the system monitoring data includes current system monitoring data; The abnormal prediction module is used to input the pre-processed current system monitoring data into the pre-trained prediction model to obtain the predicted system monitoring data; A maintenance decision generation module, used to match the prediction system monitoring data with a pre-built maintenance decision tree, determine the fault type and generate a corresponding maintenance decision; The maintenance instruction issuing module is used to trigger maintenance decisions.
9. An intelligent maintenance system for edge computing equipment, characterized in that: comprising an intelligent management unit and the intelligent maintenance unit as claimed in claim 8; The intelligent maintenance unit and the intelligent management unit are connected in series, and the intelligent maintenance unit receives system monitoring data of the edge device unit and transmits it to the intelligent management unit; The intelligent management unit includes a user login / registration module, a device management module, a monitoring module, an abnormal alarm list generation module and a user instruction upload module; After the user registers and logs in through the user login / registration module, the device management module requests to obtain the edge computing device information stored in the storage module; the monitoring module requests to obtain the system monitoring data stored in the storage module through the edge computing device information obtained by the device management module; the abnormal alarm list generation module obtains the edge computing device information and system monitoring data in the device management module and the monitoring module, filters the alarm information, and displays it on the page; the user instruction upload module sends the shell command input by the user corresponding to the maintenance decision to the maintenance instruction issuance module, and the maintenance instruction issuance module sends the shell command to the corresponding edge computing device.
10. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the edge computing device maintenance method as described in any one of claims 1 to 7 are implemented.
Citation Information
Patent Citations
Predictive industrial equipment maintenance method and maintenance system based on edge computing
CN109635958A
Water quality early warning method based on Informer model
CN114385614A
Hydropower station transformer multi-parameter temperature prediction and risk early warning method and system
CN118378016A
