Intelligent edge server monitoring method and system based on artificial intelligence

Abstract

The invention relates to an edge server intelligent monitoring method and system based on artificial intelligence. The method comprises the following steps: acquiring various historical operation data and fault data to construct a historical data set; creating an intelligent monitoring model, cutting a historical data set into training data blocks with specified sizes, sequentially inputting the training data blocks into a data processing and analyzing module for training, and creating a reference model of a normal operation state of the edge server; various operation data are collected in real time and transmitted to the data processing and analyzing module for real-time feature extraction, and the data processing and analyzing module is optimized through real-time features; comparing the extracted features with features of a reference model, identifying abnormal data in the real-time operation data, and analyzing abnormal types and severity; classifying and grading the exception based on the exception type and severity of the exception data, and performing early warning processing; and the automatic processing module triggers a corresponding processing mechanism. Real-time monitoring and automatic early warning are achieved, the system stability is improved, and faults are reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent monitoring of servers, and particularly relates to an intelligent monitoring method and system for edge servers based on artificial intelligence. Background Art

[0002] Intelligent monitoring of edge servers is a technology that uses edge servers to perform real-time processing and analysis of monitoring data. Edge servers are usually deployed at the edge of the network, close to data sources and end users, and can process and respond to locally generated data more quickly, reducing data transmission latency and bandwidth consumption. With the rapid development of edge computing, as an important node for data processing and storage, the stability and availability of the operating state of edge servers are crucial to the overall performance of the system. However, traditional monitoring methods often rely on manual inspections or fixed threshold triggers, and there are problems such as slow response speed and inaccurate early warnings.

[0003] Therefore, how to improve the existing intelligent monitoring method for edge servers to achieve real-time monitoring and automatic early warning, so as to improve system stability and reduce the occurrence of faults, is a technical problem that needs to be solved urgently at present. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent monitoring method and system for edge servers based on artificial intelligence, so as to improve the existing intelligent monitoring method for edge servers, achieve real-time monitoring and automatic early warning, and improve system stability and reduce the occurrence of system faults.

[0005] To solve the above technical problems, the technical solution adopted by the present invention is as follows:

[0006] An intelligent monitoring method for edge servers based on artificial intelligence includes the following steps:

[0007] S1: Obtain various historical operation data and fault data of the edge server through a data acquisition module, construct a historical data set from the various historical operation data and fault data, and perform data preprocessing on the historical data set;

[0008] S2: Create an intelligent monitoring model, set a data processing and analysis module in the intelligent monitoring model, cut the preprocessed historical data set into training data blocks of a specified size, and sequentially input the training data blocks into the data processing and analysis module, and train the data processing and analysis module through the training data blocks to create a benchmark model for the normal operating state of the edge server;

[0009] S3: The data acquisition module collects various operation data of the edge server in real time, preprocesses the various operation data of the edge server, and then conveys it to the data processing and analysis module. The data processing and analysis module extracts real-time features from the various operation data of the edge server collected in real time, and optimizes the data processing and analysis module through the real-time features;

[0010] S4: The intelligent monitoring model compares the extracted features with the features of the benchmark model, identifies abnormal data in the real-time operation data based on the feature comparison result, and analyzes the abnormal type and severity of the server based on the abnormal data;

[0011] S5: The intelligent monitoring module classifies and grades the anomalies based on the abnormal type and severity of the abnormal data, generates corresponding warning information, and then performs warning processing on the warning information through the warning module. The warning information includes the abnormal type, the time when the anomaly occurs, the location where the anomaly occurs, and the scope affected by the anomaly;

[0012] S6: The automatic processing module automatically triggers corresponding processing mechanisms based on the warning information. The processing mechanisms include automatic restart operation, automatic isolation of faulty nodes, and automatic adjustment of resource allocation.

[0013] Preferably, the various operation data in step S1 include CPU usage rate, memory occupancy, disk space, and network bandwidth.

[0014] Preferably, the specific process of data preprocessing in step S1 is as follows:

[0015] S11: Remove duplicate data in the historical data set, infer missing values in the historical data set after removing duplicate data based on the existing complete data set, and estimate the missing values by establishing a regression equation through the existing attribute values;

[0016] S12: Fill in the missing values in the historical data set based on the estimation result of the missing values, and perform regularization processing on the historical data set after filling in the missing values: calculate the P-norm of the historical data set, and obtain the components of the historical data set. Divide each component by the P-norm of the data set so that the P-norm of the processed historical data set is 1;

[0017] S13: Perform normalization processing on the historical data set after regularization processing: first determine the normalization range, and convert the regularized historical data set to the specified scale through the specified normalization formula.

[0018] Preferably, the data processing and analysis module in step S2 includes an input layer, a hidden layer, and an output layer. The input layer is connected to the hidden layer, and the hidden layer is connected to the output layer. The input layer is used to receive the data signal of the training data block and transmit it to the hidden layer; the hidden layer is used to perform a non-linear transformation on the data signal of the training data block. The hidden layer has multiple layers, each layer of the hidden layer has multiple neurons, and each neuron has different weights, biases, and activation functions. The data signal processed by the last layer of the hidden layer is output through the output layer.

[0019] Preferably, the specific process of training the data processing and analysis module with the training data block is as follows:

[0020] S21: Initialize the weights and biases of the data processing and analysis module. The input layer receives the current training data block, and the input layer transmits the received current training data block to the first hidden layer. The first hidden layer performs a weighted sum on the current training data block and transmits it to the second hidden layer after being processed by the activation function;

[0021] S22: The second hidden layer performs a weighted sum and the corresponding activation function processing on the data processed by the first hidden layer again, and then transmits it to the next hidden layer for processing until the last hidden layer finishes processing and transmits it to the output layer, and the output layer outputs the final processing result;

[0022] S23: Calculate the output error and output layer gradient of the output result of the output layer, calculate the gradient of the set loss function with respect to the weights of each layer, and adjust the weights and biases layer by layer from the output layer to the input layer.

[0023] Preferably, the specific process of the data processing and analysis module extracting real-time features from the real-time collected operation data is as follows:

[0024] The input layer of the trained data processing and analysis module receives the real-time collected operation data, and the input layer transmits it to the hidden layer. Each hidden layer automatically performs real-time selection and extraction of specified features on the real-time collected operation data layer by layer.

[0025] Preferably, the specific process of the intelligent monitoring model comparing the extracted features with the features of the reference model is as follows:

[0026] S41: Detect the extreme points of the features of the reference model and the real-time extracted features at different scales respectively, optimize the extreme points, obtain the feature points of the data based on the optimized extreme points, and assign one or more main directions to each feature point;

[0027] S42: Generate a descriptor for each feature point based on the main direction of each feature point and the position of the feature point, and perform descriptor matching processing on the descriptors of the reference model and the descriptors of the features extracted in real time;

[0028] S43: And output the descriptor matching result of the descriptor of the reference model and the descriptor of the features extracted in real time.

[0029] In a second aspect, an intelligent monitoring system for an edge server based on artificial intelligence is provided, which is used to implement any one of the intelligent monitoring methods for an edge server based on artificial intelligence, including a data acquisition module, a preprocessing module, a model creation module, an intelligent monitoring model, a data processing and analysis module, a data collection module, a feature comparison module, an anomaly classification and grading module, an early warning module, and an automatic processing module. The data acquisition module is connected to the preprocessing module, the preprocessing module is connected to the intelligent monitoring model, the model creation module is connected to the intelligent monitoring model, the intelligent monitoring model is connected to the data collection module, and the anomaly classification and grading module is connected to the early warning module and the automatic processing module;

[0030] The data acquisition module is used to acquire various historical operation data and fault data of the edge server, and construct a historical data set from the various historical operation data and fault data;

[0031] The preprocessing module is used to perform data preprocessing on the historical data set and the various operation data of the edge server collected in real time;

[0032] The model creation module is used to create an intelligent monitoring model, and set a data processing and analysis module and a feature comparison module in the intelligent monitoring model;

[0033] The intelligent monitoring model is used to perform real-time feature extraction on the various operation data of the edge server collected in real time, optimize the data processing and analysis module through the real-time features, compare the extracted features with the features of the reference model, identify abnormal data in the real-time operation data based on the feature comparison result, and analyze the abnormal type and severity of the server based on the abnormal data;

[0034] The data processing and analysis module performs real-time feature extraction on the various operation data of the edge server collected in real time, and optimizes the data processing and analysis module through the real-time features;

[0035] The data collection module is used to collect various operation data of the edge server in real time;

[0036] The feature comparison module is used to compare the extracted features with the features of the benchmark model, identify abnormal data in the real-time operation data based on the feature comparison result, and analyze the abnormal type and severity of the server based on the abnormal data;

[0037] The anomaly classification and grading module is used to classify and grade anomalies based on the anomaly type and severity of the abnormal data;

[0038] The early warning module is used to perform early warning processing on the early warning information;

[0039] The automatic processing module is used to automatically trigger corresponding processing mechanisms based on the early warning information, and the processing mechanisms include automatic restart operation, automatic isolation of faulty nodes, and automatic adjustment of resource allocation.

[0040] The beneficial effects of the present invention include:

[0041] The intelligent monitoring method and system for edge servers based on artificial intelligence provided by the present invention obtain various historical operation data and fault data to construct a historical data set; create an intelligent monitoring model, cut the historical data set into training data blocks of a specified size and input them into the data processing and analysis module for training in sequence, and create a benchmark model for the normal operation state of the edge server; collect various operation data in real time and send them to the data processing and analysis module for real-time feature extraction, and optimize the data processing and analysis module through real-time features; compare the extracted features with the features of the benchmark model, identify abnormal data in the real-time operation data, analyze the abnormal type and severity; classify and grade the anomalies based on the anomaly type and severity of the abnormal data, and perform early warning processing; the automatic processing module triggers corresponding processing mechanisms. It realizes real-time monitoring and automatic early warning, improves system stability and reduces the occurrence of faults.

[0042] First, by removing duplicate data in the historical data set, inferring missing values for the historical data set after removing duplicate data, estimating the missing values by establishing a regression equation based on existing attribute values, filling the missing values in the historical data set, and performing regularization processing on the historical data set after filling the missing values; the process of normalizing the regularized historical data set realizes the processing of the historical data set or real-time collected data, avoids the impact of duplicate data, missing data or data from different data sources on the subsequent data processing and analysis results, and improves the accuracy of the data analysis and processing results.

[0043] Secondly, by initializing the weights and biases of the data processing and analysis module, the input layer receives the current training data block, transmits the received current training data block to the hidden layer, and processes it layer by layer through multiple processing layers; calculates the output error and output layer gradient of the output result of the output layer, calculates the gradient of the set loss function with respect to the weights of each layer, and adjusts the weights and biases layer by layer from the output layer to the input layer, improving the performance of the data processing and analysis module and enhancing the accuracy of the subsequent processing and analysis results of real-time data.

[0044] Finally, by optimizing the extreme points detected at different scales for the features of the benchmark model and the features extracted in real time respectively, and assigning one or more main directions to the feature points of the data obtained from the optimized extreme points, generating a descriptor for each feature point based on the main direction and position of each feature point, and performing descriptor matching processing based on the descriptors of the benchmark model and the descriptors of the features extracted in real time; outputting the descriptor matching result of the descriptors of the benchmark model and the descriptors of the features extracted in real time, realizing the analysis of real-time data, and being able to efficiently discover abnormal data in real-time data, so as to timely detect abnormalities in the server for timely processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic flow chart of the intelligent monitoring method for an edge server based on artificial intelligence according to the present invention.

[0046] Figure 2 It is a schematic flow chart of feature comparison according to the present invention.

[0047] Figure 3 It is a schematic architecture diagram of the intelligent monitoring system for an edge server based on artificial intelligence according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0048] The following further elaborates on the present invention in conjunction with the attached Figures 1 to 3 to make a more detailed description:

[0049] Embodiment 1

[0050] Referring to the attached Figure 1 As shown, an intelligent monitoring method for an edge server based on artificial intelligence includes the following steps:

[0051] S1: Obtain various historical operation data and fault data of the edge server through the data acquisition module, construct a historical data set from the various historical operation data and fault data, and perform data preprocessing on the historical data set. By removing duplicate data, filling in missing values, identifying and processing outliers through data preprocessing, and converting data from different sources into a unified data format and converting data types, the quality of the data is improved, ensuring the reliability of the analysis results.

[0052] S2: Create an intelligent monitoring model, set up a data processing and analysis module in the intelligent monitoring model, cut the preprocessed historical data set into training data blocks of a specified size, and sequentially input the training data blocks into the data processing and analysis module. Train the data processing and analysis module with the training data blocks to create a benchmark model for the normal operation state of the edge server. By cutting the historical data set into training data blocks of a specified size, the number of samples to be processed during each training can be reduced, thereby reducing memory occupancy and computational volume and improving training efficiency. Processing only one data block at a time can complete the training process faster. By using multi-threading or multi-processing to process different data blocks in parallel, the training process can be accelerated, and it also helps to better understand and debug the model.

[0053] S3: The data acquisition module collects various operation data of the edge server in real time, and after preprocessing the various operation data of the edge server, conveys it to the data processing and analysis module. The data processing and analysis module extracts real-time features from the various operation data of the edge server collected in real time, and optimizes the data processing and analysis module through the real-time features. Through the process of feature extraction, representative information can be extracted from the historical data set or the data collected in real time, reducing the dimension of the data, reducing noise and redundant information, thereby improving the learning efficiency and accuracy of the model, enabling the model to better understand and process data, avoiding overfitting phenomena, and enhancing the generalization ability of the model.

[0054] S4: The intelligent monitoring model compares the extracted features with the features of the benchmark model, identifies abnormal data in the real-time operation data based on the feature comparison results, and analyzes the abnormal type and severity of the server based on the abnormal data.

[0055] S5: The intelligent monitoring module classifies and grades the anomalies based on the abnormal type and severity of the abnormal data, generates corresponding warning information, and performs warning processing on the warning information through the warning module. The warning information includes the abnormal type, the time when the anomaly occurred, the location where the anomaly occurred, and the scope affected by the anomaly;

[0056] S6: The automatic processing module automatically triggers corresponding processing mechanisms based on the warning information. The processing mechanisms include automatic restart operations, automatic isolation of faulty nodes, and automatic adjustment of resource allocation.

[0057] In this embodiment, the various operation data in step S1 include CPU usage rate, memory occupancy, disk space, and network bandwidth.

[0058] Embodiment 2

[0059] Based on Embodiment 1, the specific process of data preprocessing in step S1 is as follows:

[0060] S11: Remove the duplicate data in the historical dataset, infer the missing values for the historical dataset after removing duplicate data based on the existing complete dataset, and estimate the missing values by establishing a regression equation based on the existing attribute values;

[0061] S12: Fill the missing values in the historical dataset based on the estimation results of the missing values, and perform regularization processing on the historical dataset after filling the missing values: calculate the P-norm of the historical dataset, and obtain the components of the historical dataset, divide each component by the P-norm of the dataset, so that the P-norm of the processed historical dataset is 1;

[0062] S13: Perform normalization processing on the historical dataset after regularization processing: first determine the normalization range, and convert the regularized historical dataset to the specified scale by specifying the normalization formula.

[0063] By removing the duplicate data in the historical dataset, inferring the missing values for the historical dataset after removing duplicate data, estimating the missing values by establishing a regression equation based on the existing attribute values, filling the missing values in the historical dataset, and performing regularization processing on the historical dataset after filling the missing values; the process of performing normalization processing on the historical dataset after regularization processing realizes the processing of the historical dataset or the real-time collected data, avoids the influence of duplicate data, missing data or data from different data sources on the subsequent data processing and analysis results, and improves the accuracy of the data analysis and processing results.

[0064] In this embodiment, the data processing and analysis module in step S2 includes an input layer, a hidden layer and an output layer. The input layer is connected to the hidden layer, and the hidden layer is connected to the output layer. The input layer is used to receive the data signal of the training data block and transmit it to the hidden layer; the hidden layer is used to perform non-linear transformation on the data signal of the training data block. There are multiple layers of hidden layers, each layer of hidden layer has multiple neurons, and each neuron has different weights, biases and activation functions. The data signal processed by the last layer of hidden layer is output through the output layer.

[0065] Embodiment 3

[0066] On the basis of Embodiment 1 or Embodiment 2, the specific process of training the data processing and analysis module by using the training data block is as follows:

[0067] S21: Initialize the weights and biases of the data processing and analysis module. The input layer receives the current training data block, and the input layer transmits the received current training data block to the first hidden layer. The first hidden layer performs weighted summation on the current training data block and processes it through an activation function before transmitting it to the second hidden layer;

[0068] S22: The second hidden layer performs weighted summation again on the data processed by the first hidden layer and processes it through the corresponding activation function before transmitting it to the next hidden layer for processing. This process continues until the last hidden layer finishes processing and then transmits the result to the output layer. The output layer outputs the final processing result;

[0069] S23: Calculate the output error and output layer gradient of the output result of the output layer, calculate the gradient of the set loss function with respect to the weights of each layer, and adjust the weights and biases layer by layer from the output layer to the input layer.

[0070] By initializing the weights and biases of the data processing and analysis module, the input layer receives the current training data block, the input layer transmits the received current training data block to the hidden layer, and processes it layer by layer through multiple processing layers; calculate the output error and output layer gradient of the output result of the output layer, calculate the gradient of the set loss function with respect to the weights of each layer, and adjust the weights and biases layer by layer from the output layer to the input layer. This process improves the performance of the data processing and analysis module and enhances the accuracy of the subsequent processing and analysis results of real-time data.

[0071] In this embodiment, the specific process of the data processing and analysis module for real-time feature extraction of various operation data collected in real time is as follows: The input layer of the trained data processing and analysis module receives various operation data collected in real time, and the input layer transmits it to the hidden layer. Each hidden layer automatically selects and extracts specified features layer by layer from the various operation data collected in real time.

[0072] The specific process of the intelligent monitoring model for comparing the extracted features with the features of the reference model is as follows:

[0073] S41: Detect extreme points of the features of the reference model and the features extracted in real time at different scales, optimize the extreme points, obtain the feature points of the data based on the optimized extreme points, and assign one or more main directions to each feature point;

[0074] S42: Generate a descriptor for each feature point according to the main direction and position of each feature point, and perform matching processing of the descriptors based on the descriptors of the reference model and the descriptors of the features extracted in real time;

[0075] S43: Output the descriptor matching result of the descriptors of the reference model and the descriptors of the features extracted in real time.

[0076] By optimizing the detection of extreme points of the features of the reference model and the features extracted in real time at different scales respectively, and assigning one or more main directions to the feature points of the data obtained from the optimized extreme points, generating a descriptor for each feature point based on the main direction and the position of each feature point, and performing a matching process on the descriptors based on the descriptors of the reference model and the descriptors of the features extracted in real time; the matching result of the descriptors of the reference model and the descriptors of the features extracted in real time is output, realizing the analysis of real-time data, and being able to efficiently discover abnormal data in real-time data so as to timely discover the abnormality of the server for timely processing.

[0077] An intelligent monitoring system for edge servers based on artificial intelligence, which is used to implement any one of the intelligent monitoring methods for edge servers based on artificial intelligence, including a data acquisition module, a preprocessing module, a model creation module, an intelligent monitoring model, a data processing and analysis module, a data collection module, a feature comparison module, an anomaly classification and grading module, a warning module and an automatic processing module. The data acquisition module is connected to the preprocessing module, the preprocessing module is connected to the intelligent monitoring model, the model creation module is connected to the intelligent monitoring model, the intelligent monitoring model is connected to the data collection module, and the anomaly classification and grading module is connected to the warning module and the automatic processing module.

[0078] The data acquisition module is used to acquire various historical operation data and fault data of the edge server, and construct a historical data set from the various historical operation data and fault data; the preprocessing module is used to perform data preprocessing on the historical data set and the real-time collected operation data of the edge server; the model creation module is used to create an intelligent monitoring model, and set a data processing and analysis module and a feature comparison module in the intelligent monitoring model; the intelligent monitoring model is used to perform real-time feature extraction on the real-time collected operation data of the edge server, optimize the data processing and analysis module through the real-time features, and compare the extracted features with the features of the benchmark model, and identify abnormal data in the real-time operation data based on the feature comparison result, and analyze the abnormal type and severity of the server based on the abnormal data; the data processing and analysis module performs real-time feature extraction on the real-time collected operation data of the edge server, and optimizes the data processing and analysis module through the real-time features; the data acquisition module is used to collect the real-time operation data of the edge server; the feature comparison module is used to compare the extracted features with the features of the benchmark model, and identify abnormal data in the real-time operation data based on the feature comparison result, and analyze the abnormal type and severity of the server based on the abnormal data; the abnormal classification and grading module is used to classify and grade the abnormal based on the abnormal type and severity of the abnormal data; the warning module is used to perform warning processing on the warning information; the automatic processing module is used to automatically trigger a corresponding processing mechanism based on the warning information, and the processing mechanism includes automatic restart operation, automatic isolation of faulty nodes, and automatic adjustment of resource allocation.

[0079] In summary, the intelligent monitoring method and system for edge servers based on artificial intelligence provided by the present invention acquire various historical operation data and fault data to construct a historical data set; create an intelligent monitoring model, cut the historical data set into training data blocks of a specified size and input them into the data processing and analysis module for training in sequence, and create a benchmark model for the normal operation state of the edge server; collect various operation data in real time and send them to the data processing and analysis module for real-time feature extraction, and optimize the data processing and analysis module through the real-time features; compare the extracted features with the features of the benchmark model, identify abnormal data in the real-time operation data, and analyze the abnormal type and severity; classify and grade the abnormal based on the abnormal type and severity of the abnormal data, and perform warning processing; the automatic processing module triggers a corresponding processing mechanism. It realizes real-time monitoring and automatic warning, improves system stability and reduces the occurrence of faults.

[0080] By removing duplicate data from the historical dataset, imputing missing values in the historical dataset after removing duplicate data, estimating the missing values by establishing a regression equation based on existing attribute values, filling in the missing values in the historical dataset, and performing regularization and normalization processing on the historical dataset after filling in the missing values, the processing of the historical dataset or real-time collected data is realized, avoiding the impact of duplicate data, missing data, or data from different data sources on the subsequent data processing and analysis results, and improving the accuracy of the data analysis and processing results. Initialize the weights and biases of the data processing and analysis module. The input layer receives the current training data block, and the input layer transmits the received current training data block to the hidden layer and processes it layer by layer through multiple processing layers; calculate the output error and output layer gradient of the output result of the output layer, calculate the gradient of the set loss function with respect to the weights of each layer, and adjust the weights and biases layer by layer from the output layer to the input layer to improve the performance of the data processing and analysis module and the accuracy of the subsequent processing and analysis results of real-time data.

[0081] Optimize the detection of extreme points for the features of the benchmark model and the features extracted in real time at different scales, and assign one or more main directions to the feature points that obtain data for the optimized extreme points. Generate descriptors for each feature point based on the main direction and position of each feature point, and perform descriptor matching processing based on the descriptors of the benchmark model and the descriptors of the features extracted in real time; output the descriptor matching results of the descriptors of the benchmark model and the descriptors of the features extracted in real time, realizing the analysis of real-time data and being able to efficiently detect abnormal data in real-time data so as to timely detect server anomalies for timely processing.

Claims

1. An intelligent monitoring method for an edge server based on artificial intelligence, characterized in that, It includes the following steps: S1: Obtain various historical operation data and fault data of the edge server through the data acquisition module, construct a historical data set from the various historical operation data and fault data, and perform data preprocessing on the historical data set; S2: Create an intelligent monitoring model, set a data processing and analysis module in the intelligent monitoring model, cut the preprocessed historical data set into training data blocks of a specified size, and sequentially input the training data blocks into the data processing and analysis module. Train the data processing and analysis module through the training data blocks to create a benchmark model for the normal operation state of the edge server; S3: Real-time collect various operation data of the edge server through the data collection module, perform data preprocessing on the various operation data of the edge server, and then transport it to the data processing and analysis module. The data processing and analysis module performs real-time feature extraction on the various operation data of the edge server collected in real time, and optimizes the data processing and analysis module through the real-time features; S4: The intelligent monitoring model compares the extracted features with the features of the benchmark model, identifies abnormal data in the real-time operation data based on the feature comparison result, and analyzes the abnormal type and severity of the server based on the abnormal data; S5: The intelligent monitoring module classifies and grades the anomalies based on the abnormal type and severity of the abnormal data, generates corresponding warning information, and then performs warning processing on the warning information through the warning module. The warning information includes the abnormal type, the time when the anomaly occurred, the location where the anomaly occurred, and the scope affected by the anomaly; S6: The automatic processing module automatically triggers corresponding processing mechanisms based on the warning information. The processing mechanisms include automatic restart operation, automatic isolation of faulty nodes, and automatic adjustment of resource allocation.

2. The intelligent monitoring method for an edge server based on artificial intelligence according to claim 1, wherein The various operation data in step S1 include CPU usage rate, memory occupancy, disk space, and network bandwidth.

3. The intelligent monitoring method for an edge server based on artificial intelligence according to claim 1, characterized in that, The specific process of data preprocessing in step S1 is as follows: S11: Remove duplicate data in the historical data set, infer missing values in the historical data set after removing duplicate data based on the existing complete data set, and estimate the missing values by establishing a regression equation with the existing attribute values based on the inference result; S12: Fill in the missing values in the historical data set based on the estimation result of the missing values, and perform regularization processing on the historical data set after filling in the missing values: calculate the P-norm of the historical data set, and obtain the components of the historical data set. Divide each component by the P-norm of the data set so that the P-norm of the processed historical data set is 1; S13: Perform normalization processing on the historical data set after regularization processing: first determine the normalization range, and convert the regularized historical data set to a specified scale through a specified normalization formula.

4. A method for intelligent monitoring of an edge server based on artificial intelligence according to claim 1, characterized in that, The data processing and analysis module in step S2 includes an input layer, a hidden layer, and an output layer. The input layer is connected to the hidden layer, and the hidden layer is connected to the output layer. The input layer is used to receive the data signal of the training data block and transmit it to the hidden layer; the hidden layer is used to perform a non-linear transformation on the data signal of the training data block. There are multiple layers of hidden layers, and each layer of the hidden layer has multiple neurons. Each neuron has different weights, biases, and activation functions. The data signal processed by the last layer of the hidden layer is output through the output layer.

5. The intelligent monitoring method of an edge server based on artificial intelligence according to claim 4, characterized in that, The specific process of training the data processing and analysis module with the training data block is as follows: S21: Initialize the weights and biases of the data processing and analysis module. The input layer receives the current training data block, and the input layer transmits the received current training data block to the first hidden layer. The first hidden layer performs a weighted sum on the current training data block and processes it through the activation function and then transmits it to the second hidden layer; S22: The second hidden layer performs a weighted sum and the corresponding activation function processing on the data processed by the first hidden layer again and then transmits it to the next hidden layer for processing until the last hidden layer finishes processing and then transmits it to the output layer, and the output layer outputs the final processing result; S23: Calculate the output error and output layer gradient of the output result of the output layer, calculate the gradient of the set loss function with respect to the weights of each layer, and adjust the weights and biases layer by layer from the output layer to the input layer.

6. The intelligent monitoring method for an edge server based on artificial intelligence according to claim 5, characterized in that, The specific process of the data processing and analysis module for real-time feature extraction of the real-time collected operation data is as follows: The input layer of the trained data processing and analysis module receives the real-time collected operation data, and the input layer transmits it to the hidden layer. Each hidden layer automatically performs real-time selection and extraction of specified features on the real-time collected operation data layer by layer.

7. An intelligent monitoring method for an edge server based on artificial intelligence according to claim 6, characterized in that, The specific process of the intelligent monitoring model for comparing the extracted features with the features of the reference model is as follows: S41: Detect the extreme points of the features of the reference model and the real-time extracted features at different scales respectively, optimize the extreme points, obtain the feature points of the data based on the optimized extreme points, and assign one or more main directions to each feature point; S42: Generate a descriptor for each feature point according to the main direction and position of each feature point, and perform matching processing on the descriptors based on the descriptors of the reference model and the descriptors of the real-time extracted features; S43: Output the descriptor matching result of the descriptor of the reference model and the descriptor of the real-time extracted features.

8. An intelligent monitoring system for an edge server based on artificial intelligence, which is used to implement an intelligent monitoring method for an edge server based on artificial intelligence according to any one of claims 1-7, characterized in that, It includes a data acquisition module, a preprocessing module, a model creation module, an intelligent monitoring model, a data processing and analysis module, a data collection module, a feature comparison module, an anomaly classification and grading module, a warning module, and an automatic processing module. The data acquisition module is connected to the preprocessing module, the preprocessing module is connected to the intelligent monitoring model, the model creation module is connected to the intelligent monitoring model, the intelligent monitoring model is connected to the data collection module, and the anomaly classification and grading module is connected to the warning module and the automatic processing module; The data acquisition module is used to acquire various historical operation data and fault data of the edge server, and construct a historical data set from the various historical operation data and fault data; The preprocessing module is used to perform data preprocessing on the historical data set and the real-time collected operation data of the edge server; The model creation module is used to create an intelligent monitoring model, and set a data processing and analysis module and a feature comparison module in the intelligent monitoring model; The intelligent monitoring model is used to perform real-time feature extraction on the real-time collected operation data of the edge server, optimize the data processing and analysis module through the real-time features, compare the extracted features with the features of the reference model, identify abnormal data in the real-time operation data based on the feature comparison results, and analyze the abnormal type and severity of the server based on the abnormal data analysis; The data processing and analysis module performs real-time feature extraction on the real-time collected operation data of the edge server, and optimizes the data processing and analysis module through the real-time features; The data collection module is used to collect the real-time operation data of the edge server; The feature comparison module is used to compare the extracted features with the features of the reference model, identify abnormal data in the real-time operation data based on the feature comparison results, and analyze the abnormal type and severity of the server based on the abnormal data analysis; The abnormal classification and grading module is used to classify and grade the abnormal based on the abnormal type and severity of the abnormal data; The warning module is used to perform warning processing on the warning information; The automatic processing module is used to automatically trigger corresponding processing mechanisms based on the warning information, and the processing mechanisms include automatic restart operation, automatic isolation of fault nodes, and automatic adjustment of resource allocation.

Images