Network fault early warning method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a network fault early warning method and device, computer equipment and a storage medium. The network fault early warning method comprises the following steps: acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; performing prediction processing on the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by a pelican optimization algorithm; judging whether the fault prediction result meets a preset early warning condition or not; and if the fault prediction result meets a preset early warning condition, generating fault early warning information according to the fault prediction result, and determining an early warning processing scheme corresponding to the fault early warning information. The preset fault prediction model is easy to construct and train, the prediction effect of the network fault is ensured, the accuracy of fault prediction is improved, early warning is performed in time based on the fault prediction result, the processing scheme is determined, and the network fault processing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a network fault early warning method, device, computer equipment and storage medium. Background Art

[0002] In the information age, the Internet has become an indispensable part of modern society, supporting the daily operations and business development of all walks of life. In the financial industry, network failures will affect data transmission, information exchange, customer service and other businesses, causing serious economic losses and social impacts. In existing technologies, network operation and maintenance management mainly relies on manual monitoring and post-event response mechanisms, which have a certain lag and cannot timely warn of network failures. The fault handling mechanism is inefficient. Although some companies have tried to use AI technology to predict network failures, the causes of network failures are complex and changeable, which makes the construction and training of AI models time-consuming and labor-intensive, and the warning effect of network failures is not ideal, which cannot meet the actual application needs. Summary of the invention

[0003] Based on this, it is necessary to provide a network fault warning method, device, computer equipment and storage medium to address the above technical issues, so as to solve the problems of untimely discovery of existing network faults and poor warning effect.

[0004] A network fault early warning method, comprising: Acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; The operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; Determining whether the fault prediction result meets the preset warning condition; If the fault prediction result meets the preset warning condition, fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined.

[0005] A network fault early warning device, comprising: A preprocessing module, used to obtain network operation data, preprocess the network operation data, and obtain operation characteristic data; A model prediction module is used to predict the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; A fault judgment module is used to judge whether the fault prediction result meets the preset warning condition; The fault warning module is used to generate fault warning information according to the fault prediction result if the fault prediction result meets the preset warning condition, and determine the warning processing plan corresponding to the fault warning information.

[0006] A computer device comprises a memory, a processor and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the above-mentioned network fault early warning method when executing the computer-readable instructions.

[0007] A computer-readable storage medium stores computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the network fault early warning method as described above.

[0008] In the above-mentioned network fault warning method, device, computer equipment and storage medium, the network fault warning method obtains network operation data, pre-processes the network operation data, and obtains operation characteristic data; the operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result. The preset fault prediction model of the present invention is based on a multi-layer perceptron network as a basic model, and fully considers the complex factors affecting network failures. At the same time, the multi-layer perceptron network is optimized using the Pelican optimization algorithm, which reduces the difficulty of model construction and training, ensures the prediction effect of network failures, and improves the accuracy of fault prediction. Based on the obtained fault prediction results, the present invention determines whether the fault prediction results meet the preset warning conditions; if the fault prediction results meet the preset warning conditions, fault warning information is generated according to the fault prediction results, and a warning processing scheme corresponding to the fault warning information is determined, which helps to timely warn and handle network failures, maintain the high availability of the network, and improve the efficiency of network fault processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0010] Figure 1 This is a schematic diagram of an application environment of a network fault early warning method in one embodiment of the present invention; Figure 2 It is a flow chart of a network fault early warning method in one embodiment of the present invention; Figure 3 It is a schematic diagram of a fault handling process of a network fault early warning method in one embodiment of the present invention; Figure 4is a structural schematic diagram of a network fault early warning device in one embodiment of the present invention; Figure 5 is a schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The network fault early warning method provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In the application environment, the client establishes a communication connection with the server, and the network service architecture also includes a network operator and a network line (not shown in the figure). The network operator provides network services to the client through the network line, and the server is used to monitor whether the network service of the client fails. The client can be but is not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices. The server can be implemented with an independent server or a server cluster composed of multiple servers. The present invention is described in detail below through specific embodiments.

[0013] In the financial industry, network failures can affect data transmission, information exchange, and customer service. In terms of data transmission, the financial industry is highly dependent on data for business operations. Network failures can cause key information such as transaction data, market data, and customer information to be unable to be transmitted in a timely and accurate manner, and may even cause transaction delays, erroneous transactions, or transaction failures, leading to market chaos and customer dissatisfaction. In terms of information exchange, financial institutions and financial institutions and customers need to exchange information frequently. Network failures can cause information such as transaction confirmations, account change notifications, and market analysis reports to be unable to be transmitted in a timely manner, affecting market decisions. In terms of customer service, network failures can directly affect the customer satisfaction of financial institutions, resulting in the inability to use online transactions, account inquiries, customer service hotlines, and other functions normally, and even causing customer complaints and disputes. Among them, users in the financial industry attach great importance to online services, and the impact of network failures on customer service is particularly significant. When a user enjoys customer service through the client, the server can obtain the client's network operation data, pre-process the network operation data, and obtain operation characteristic data; predict and process the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; determine whether the fault prediction result meets the preset warning conditions; if the fault prediction result meets the preset warning conditions, generate fault warning information based on the fault prediction result, and determine the warning processing plan corresponding to the fault warning information to avoid customer service unavailability due to network failure.

[0014] In one embodiment, if Figure 2 As shown, a network fault early warning method is provided, comprising the following steps S10-S40: S10: Acquire network operation data, and pre-process the network operation data to obtain operation characteristic data.

[0015] Understandably, in actual applications, the server can automatically trigger the process of the network fault warning method at a fixed periodic time, such as obtaining the client's network operation data every 5 minutes and performing data processing and predictive analysis. The client operation and maintenance personnel can also manually trigger the process of the network fault warning method as needed. Network operation data refers to various data and information generated during network communication to characterize the operating status of network lines, such as network delay, packet loss rate, bandwidth usage, etc. Network operation data can be used as a basis for evaluating the quality of the client's network connection, so as to determine whether there are potential network problems. As the raw data for model prediction, network operation data needs to be preprocessed to facilitate subsequent data analysis. Preprocessing is the process of processing network operation data to increase the readability of the model. Operation feature data is feature data obtained by extracting features from raw data to characterize the network operation status.

[0016] In one embodiment, in step S10, that is, preprocessing the network operation data to obtain operation characteristic data, includes: S101, performing data screening processing on the network operation data according to a preset characteristic index to obtain initial characteristic data, wherein the preset characteristic index is a characteristic index determined by similarity analysis of historical operation data; S102, normalizing the initial feature data to obtain normalized feature data; S103: Perform feature encoding processing on the normalized feature data to obtain the operation feature data.

[0017] Understandably, in the preprocessing process, the network operation data is firstly subjected to data screening processing according to the preset characteristic indicators to obtain the initial characteristic data. Among them, the preset characteristic indicators are one or more indicator types used to characterize the network operation status determined by the historical operation data through similarity analysis. Historical operation data refers to information that records the operation status of network lines and corresponding parameters in a specific historical time period, such as normal parameters of network lines and fault parameters of network lines in the past month. Data screening processing refers to selecting data from the network operation data according to the indicator type corresponding to the preset characteristic indicators. Initial characteristic data refers to one or more data corresponding to the preset characteristic indicators selected from the network operation data. Similarity analysis is used to measure the degree of correlation between different variables. Similarity analysis methods include similarity based on Pearson correlation, similarity based on Euclidean distance, cosine similarity, etc.

[0018] In one embodiment, the historical operation data includes ten data indicators, including whether the network is faulty, the time when the fault occurred, the type of fault (such as equipment or software), the duration of the fault, network traffic, bandwidth usage, device temperature, CPU usage, network delay rate and packet loss rate. In the similarity analysis, the Pearson correlation coefficient calculation method is used to calculate the Pearson similarity between "whether the network is faulty" and the other nine data, and finally determine the multiple data indicators with the highest correlation with "whether the network is faulty" as the preset feature indicators. Specifically, the seven data features with the highest correlation, including fault type, fault duration, bandwidth usage, CPU usage, network delay rate, packet loss rate and device temperature, are selected as the preset feature indicators.

[0019] Secondly, the initial feature data is normalized to obtain normalized feature data. The initial feature data under different evaluation indicators often have different dimensions and dimensional units, which will affect the results of data analysis. In order to eliminate the dimensional influence between indicators, normalization (data standardization) processing is required. Normalized feature data is the data obtained after the initial feature data is normalized. Each indicator is at the same order of magnitude, which is helpful for subsequent comprehensive comparative evaluation. Normalization methods include Min-Max normalization, Z-Score normalization, and nonlinear normalization methods. In one embodiment, the Z-Score normalization method is used to normalize the initial feature data to the same scale to obtain normalized feature data to avoid the influence of different scales on model training. The Z-Score normalization method standardizes the data based on the mean and standard deviation, and the processed data conforms to the standard normal distribution, that is, the mean is 0 and the standard deviation is 1.

[0020] Finally, the normalized feature data is feature encoded to obtain the operating feature data. Feature encoding is the process of converting non-numerical features into numerical form. For example, the network fault type is a non-numerical feature. After encoding using one-hot encoding, "0" is used to represent hardware failure, "1" is used to represent software failure, and "2" is used to represent others (such as normal network). In one embodiment, a one-hot encoding method is used to perform feature encoding on the normalized feature data. The specific method is to use a seven-bit status register to encode seven indicators such as fault type, fault duration, bandwidth utilization, CPU utilization, network delay rate, packet loss rate and device temperature. Each indicator has its own independent register bit, and only one bit is valid at any time. Operating feature data refers to data after feature encoding processing, which is completely expressed in numerical form and can be directly used for training or prediction of machine learning models.

[0021] This embodiment converts the original network operation data into operation feature data suitable for machine learning model processing through data screening, normalization and feature encoding processing, which helps to improve the accuracy and efficiency of subsequent model predictions.

[0022] S20. Predictively process the operating characteristic data using a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm.

[0023] Understandably, the preset fault prediction model is a pre-trained machine learning model used to predict the probability of network line failure based on network operation data. The preset fault prediction model is a multilayer perceptron (MLP) network model optimized by the Pelican Optimization Algorithm (POA). A multilayer perceptron is a neural network composed of fully connected layers with at least one hidden layer, and the output of each hidden layer is transformed by an activation function. The input of the preset fault prediction model is the preprocessed operation feature data, and the output is the probability prediction value of network line failure.

[0024] In one embodiment, step S20, that is, before predicting the operation characteristic data by using a preset fault prediction model, includes: S201, obtaining pre-collected historical operation data, performing sample preprocessing on the historical operation data, and obtaining sample feature data; S202, determining the constructed multi-layer perceptron network as an initial fault prediction model, wherein the initial fault prediction model includes an initial weight parameter and an initial threshold parameter, S203, performing iterative optimization processing on the initial weight parameter and the initial threshold parameter by using the Pelican optimization algorithm to obtain an optimized weight parameter and an optimized threshold parameter; S204, updating the parameters of the initial fault prediction model according to the optimization weight parameter and the optimization threshold parameter to obtain an optimized fault prediction model; S205: Training the optimized fault prediction model according to the sample feature data to obtain the preset fault prediction model.

[0025] Understandably, before predicting and processing the running feature data through the preset fault prediction model, it is necessary to generate the preset fault prediction model through model training. In the preset fault prediction model, the optimization of the multilayer perceptron network by the Pelican optimization algorithm is reflected in the optimization of weights and threshold parameters. Weights and thresholds are key parameters in the multilayer perceptron network, which determine how the model processes input data and generates outputs. In the multilayer perceptron network, weights are used to represent the importance of different input features. Each connection between an input node and a node in the hidden layer or output layer has a weight value. The larger the weight, the greater the influence of the input feature on the output result. The threshold, also known as bias, is used to control the activation conditions of neurons, determining which inputs will be processed and affect the final output. When the weighted input exceeds the threshold, the neuron will be activated and generate output. In the prior art, the search process of weights and thresholds in the multilayer perceptron network is complicated and needs to be adjusted through a large number of back-propagation algorithm processing processes. Therefore, the Pelican optimization algorithm is introduced to optimize the multilayer perceptron network.

[0026] First, obtain the pre-collected historical operation data, perform sample preprocessing on the historical operation data, and obtain sample feature data. As the original data for model training, the historical operation data also needs to be preprocessed to facilitate subsequent model training. Sample preprocessing is the process of processing historical operation data to increase the readability of the model. In addition to data screening, normalization and feature encoding processing, sample preprocessing also includes data cleaning between data screening and normalization processing, such as using mean filling for items with missing data in the data set, and directly eliminating items with outliers. Each sample feature data includes feature data corresponding to seven indicators and the actual fault state. Secondly, the constructed multi-layer perceptron network is determined as the initial fault prediction model, and the initial fault prediction model includes initial weight parameters and initial threshold parameters. The initial weight parameters refer to the initialization values ​​of the weight parameters that have not been optimized, and the initial threshold parameters refer to the initialization values ​​of the threshold parameters that have not been optimized. Thirdly, the initial weight parameters and initial threshold parameters are iteratively optimized by the Pelican optimization algorithm to obtain optimized weight parameters and optimized threshold parameters. The initial weight parameter refers to the optimized value of the weight parameter after iteration of the Pelican optimization algorithm, and the initial threshold parameter refers to the optimized value of the threshold parameter after iteration of the Pelican optimization algorithm. Again, the parameters of the initial fault prediction model are updated according to the optimized weight parameters and the optimized threshold parameters to obtain the optimized fault prediction model. The optimized fault prediction model refers to the model after the optimized weight parameters and the optimized threshold parameters are applied to the multi-layer perceptron network. Finally, the optimized fault prediction model is trained according to the sample feature data to continue to adjust the model parameters to obtain the preset fault prediction model.

[0027] This embodiment uses a multi-layer perceptron network as the basic model to build a fault prediction model, and introduces the Pelican optimization algorithm to optimize the multi-layer perceptron network, which can accelerate the iterative training of the model and obtain more accurate and appropriate weight thresholds. The optimized weight and threshold settings can improve the generalization ability and robustness of the model.

[0028] In one embodiment, in step S203, the iterative optimization process of the initial weight parameter and the initial threshold parameter by the Pelican optimization algorithm to obtain the optimized weight parameter and the optimized threshold parameter includes: S2031. Determine the mean square error of the multilayer perceptron network as the fitness function in the Pelican optimization algorithm; S2032, performing an iterative optimization process on the initial weight parameter and the initial threshold parameter by using the Pelican optimization algorithm to obtain an updated weight parameter and an updated threshold parameter; S2033, determining a primary fitness value corresponding to the primary update weight parameter and the primary update threshold parameter according to the fitness function; S2034, determining a primary optimization weight parameter and a primary optimization threshold parameter according to the primary fitness value, so that the primary optimization weight parameter and the primary optimization threshold parameter are subjected to secondary iterative optimization processing by the Pelican optimization algorithm, and recording the number of iterative optimization times; S2035. When the number of iterative optimizations reaches a preset iteration number threshold, the optimization weight parameter and the optimization threshold parameter in the last iteration are obtained.

[0029] Understandably, the Pelican optimization algorithm requires an initialized population, where each individual in the population represents a set of weight parameters and threshold parameters of the multilayer perceptron network. The initialization formula for the Pelican population is: .in, Indicates The first pelican The location of the dimension; , represents the population size of pelicans; , Represents the dimensionality of the optimization problem; Represents a random number in the range [0,1]; and They represent the upper and lower boundaries of the jth dimension of the optimization problem. When evaluating the fitness, the mean square error of the multilayer perceptron network is determined as the fitness function in the Pelican optimization algorithm. The fitness function is .in, Indicates The fitness function value of each pelican individual; Indicates the number of training samples; Indicates The true value of samples; Indicates The predicted value of a sample.

[0030] In each iteration, the initial weight parameters and initial threshold parameters are optimized by the Pelican optimization algorithm in two stages. In the first stage, the Pelican first determines the location of the prey and then moves to the area where the prey is located. The mathematical formula for the Pelican to approach the prey is as follows: in, Indicates the first phase after the update The first pelican The location of the dimension; Represents a random number in the range [0,1]; A random number of 1 or 2; Indicates the prey The location of the dimension; represents the fitness function value of the prey, Indicates The fitness function value of the pelican individual. If the objective function value is The position of the pelican is improved, and the formula is: .in, Indicates The new position of the pelican individual, the values ​​of the group weight parameter and the threshold parameter; Indicates the first phase after the update The fitness function value of the new position of each individual pelican. In the second stage, after the pelican reaches the target area (water surface), it spreads its wings on the water surface. The hunting strategy of the pelican flying on the water surface allows them to capture more prey in the target area. The formula for the hunting strategy of the pelican flying on the water surface is .in, Indicates the second phase after the update The first pelican The location of the dimension; Represents a random number in the range [0,1]; A random number of 1 or 2; Indicates the current iteration number, that is, the number of iterative optimization; represents the maximum number of iterations, that is, the preset iteration threshold. The fitness function after the second stage update is .in, Indicates that after one iteration The new position of each pelican individual, that is, one update of the weight parameter and one update of the threshold parameter; Indicates that after one iteration The fitness function value of each pelican individual is the fitness value corresponding to the updated weight parameter and the updated threshold parameter. When the number of iterative optimizations reaches the preset iteration number threshold, the optimized weight parameter and the optimized threshold parameter in the last iteration are obtained.

[0031] This embodiment uses the Pelican optimization algorithm to optimize the weight parameters and threshold parameters of the multi-layer perceptron network, and uses the number of iterations as the termination condition to minimize the mean square error of the network, which helps to quickly iterate the network. This embodiment combines the powerful search ability of the swarm intelligence algorithm and the powerful learning ability of the multi-layer perceptron network, improves the efficiency of model training, and ensures the accuracy of subsequent fault prediction.

[0032] In one embodiment, in step S205, that is, training the optimized fault prediction model according to the sample feature data to obtain the preset fault prediction model, includes: S2051, inputting the sample feature data into the optimized fault prediction model to obtain the sample prediction value output by the optimized fault prediction model, S2052, obtaining a sample true value corresponding to the sample feature data, and determining a sample error according to the sample prediction value and the sample true value; S2053, performing back propagation training on the optimized fault prediction model based on the sample error, and stopping the training when the sample error reaches a preset accuracy threshold, to obtain a training weight parameter and a training threshold parameter; S2054: Update the parameters of the optimized fault prediction model according to the training weight parameters and the training threshold parameters to obtain the preset fault prediction model.

[0033] Understandably, after applying the optimized weight parameter and the optimized threshold parameter to the multi-layer perceptron network, if the obtained optimized fault prediction model is directly determined as the preset fault prediction model, there may be a prediction error. Therefore, the optimized fault prediction model can be continuously updated as needed to determine the preset fault prediction model. Input the sample feature data into the optimized fault prediction model to obtain the sample prediction value output by the optimized fault prediction model, that is, the probability prediction value of the network fault. Obtain the sample true value corresponding to the sample feature data, and determine the sample error, that is, the loss function value, based on the sample prediction value and the sample true value. Perform back propagation training on the optimized fault prediction model based on the sample error, and stop training when the sample error reaches the preset accuracy threshold to obtain the training weight parameter and the training threshold parameter. The preset accuracy threshold is a pre-set error critical value for terminating back propagation, such as stopping training when the sample error is less than the preset accuracy threshold. The training weight parameter refers to the weight parameter value after the back propagation algorithm is iterated, and the training threshold parameter refers to the threshold parameter value after the back propagation algorithm is iterated. Apply the training weight parameter and the training threshold parameter to the optimized fault prediction model for parameter update to obtain the preset fault prediction model.

[0034] Based on the optimization of the multi-layer perceptron network using the Pelican optimization algorithm, this embodiment obtains a fully trained and optimized model through back propagation. Back propagation continuously adjusts the parameters of the model to reduce errors, which helps to improve the prediction accuracy of the preset fault prediction model.

[0035] S30: Determine whether the fault prediction result meets a preset warning condition.

[0036] It can be understood that the fault prediction result is a probability prediction value used to characterize the current network operation failure. After obtaining the fault prediction result, it is necessary to further determine whether the fault prediction result meets the preset warning condition. The preset warning condition is a pre-set standard for evaluating whether a warning needs to be triggered.

[0037] In one embodiment, step S30, i.e., determining whether the fault prediction result meets a preset warning condition, includes: S301, determining whether the fault prediction result is greater than a first preset warning threshold; S302: If the fault prediction result is greater than a first preset warning threshold, determining that the fault prediction result meets a preset warning condition; S303: If the fault prediction result is less than or equal to a first preset warning threshold, determine that the fault prediction result does not meet a preset warning condition.

[0038] Understandably, the probability prediction value in the fault prediction result is compared with the first preset warning threshold, and whether the preset warning condition is met is determined based on the comparison result. The first preset warning threshold is a preset minimum critical value of the probability for triggering a warning. The first preset warning threshold can be set with a default value, or it can be adjusted as needed, such as setting the default value of the first preset warning threshold to 50%. When the fault prediction result is greater than the first preset warning threshold, it indicates that the probability prediction value of the network fault has reached the minimum critical value, then it is determined that the fault prediction result meets the preset warning condition, and it is necessary to trigger a warning at this time. When the fault prediction result is less than or equal to the first preset warning threshold, it indicates that the probability prediction value of the network fault has not yet reached the minimum critical value, then it is determined that the fault prediction result does not meet the preset warning condition, and it is not necessary to trigger a warning at this time.

[0039] This embodiment distinguishes whether a preset warning condition is met based on the first preset warning threshold, and can more accurately understand the state of network operation, so that a timely warning can be issued when a fault is predicted.

[0040] S40: If the fault prediction result meets the preset warning condition, generate fault warning information according to the fault prediction result, and determine a warning processing plan corresponding to the fault warning information.

[0041] Understandably, on the basis that the fault prediction result meets the preset warning conditions, the server will generate fault warning information according to the fault prediction result and determine the warning processing plan corresponding to the fault warning information. Fault warning information is information used to prompt the client operation and maintenance personnel that the network has failed. The warning processing plan refers to the instruction or request to adjust the network line based on the fault warning information.

[0042] This embodiment obtains network operation data, pre-processes the network operation data, and obtains operation characteristic data; the operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result. The preset fault prediction model of this embodiment is based on a multi-layer perceptron network as a basic model, and fully considers the complex factors that affect network failures. At the same time, the multi-layer perceptron network is optimized using the Pelican optimization algorithm, which reduces the difficulty of model construction and training, ensures the prediction effect of network failures, and improves the accuracy of fault prediction. Based on the obtained fault prediction results, this embodiment determines whether the fault prediction results meet the preset warning conditions; if the fault prediction results meet the preset warning conditions, fault warning information is generated according to the fault prediction results, and a warning processing plan corresponding to the fault warning information is determined, which helps to handle network failures in a timely manner, maintain the high availability of the network, and improve the efficiency of network fault processing.

[0043] In one embodiment, the fault warning information includes first-level warning information and second-level warning information, and the first-level warning is higher than the second-level warning; in step S40, if the fault prediction result meets the preset warning condition, the fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined, including: S401, determining whether the fault prediction result is greater than a second preset warning threshold; S402: if the fault prediction result is greater than a second preset warning threshold, generate first level warning information, and send a line switching instruction to the client, so that the client performs a line switching operation; S403. If the fault prediction result is less than or equal to the second preset warning threshold, a second level warning information is generated, and a line switching request is sent to the preset management party. After receiving the line switching confirmation information corresponding to the line switching request sent by the preset management party, a line switching instruction is sent to the client, so that the client performs a line switching operation.

[0044] Understandably, if Figure 3 The fault handling flowchart shown shows the network service architecture, including the client, the server, and the operator of the network line, where the network line includes the main line and the backup line, and there can be one or more backup lines. The operator is the main entity that provides network services and is responsible for building and maintaining the network infrastructure, including network lines, data centers, etc. The network line is the physical or logical channel used by the operator to connect to the client. The client is the device or software that the user uses to access the network service, such as a personal computer, a smart phone, an application, etc. The server includes a preset fault prediction model and an early warning information center. The server obtains network operation data from the client and predicts whether the network service fails through the preset fault prediction model. The early warning information center of the server is used to send early warning information to the client and the operator when a fault is predicted, so that the client can switch from the main line to the backup line according to the early warning information, and the operator can repair the network line according to the early warning information.

[0045] Fault warning information is divided into first-level warning information and second-level warning information according to the degree of impact of the fault on the network, and the first-level warning is higher than the second-level warning. The first-level warning information refers to the prompt information when the network fault causes the network to be unavailable, such as the warning information when the network connection fails. The second-level warning information refers to the prompt information when the network fault does not cause the network to be unavailable but affects the client business, such as the warning information when the network fluctuates.

[0046] On the premise that the fault prediction result is greater than the first preset warning threshold, it is necessary to further determine whether the fault prediction result is greater than the second preset warning threshold. The second preset warning threshold is a pre-set probability critical value used to distinguish the degree of fault impact. The second preset warning threshold can be set with a default value, or it can be adjusted as needed, such as setting the default value of the second preset warning threshold to 85%. When the fault prediction result is greater than the second preset warning threshold, it indicates that the impact of the network failure has reached the level that causes the network to be unavailable, and a first-level warning information is generated. At the same time, the server sends a line switching instruction to the client, so that the client performs a line switching operation. The line switching instruction refers to an action instruction sent by the server to the client to control the client to switch the network line to be used, such as switching from the main line to the backup line.

[0047] When the fault prediction result is less than or equal to the second preset warning threshold, it indicates that the probability impact of the network fault has not yet reached the level that causes the network to be unavailable, and the second level warning information is generated. At the same time, the server sends a line switching request to the preset management party, and after receiving the line switching confirmation information corresponding to the line switching request sent by the preset management party, sends a line switching instruction to the client, so that the client performs the line switching operation. The preset management party refers to the client operation and maintenance personnel pre-set to receive the line switching request. The line switching request is a request information for the operation and maintenance personnel to determine whether the line switching is required. After receiving the line switching request, the operation and maintenance personnel can manually select "switch" or "no switch". When the switch is manually selected, the preset management party will send the line switching confirmation information corresponding to the line switching request to the server. The line switching confirmation information is used to characterize the information that the network line is switched by manual selection. After receiving the line switching confirmation information, the server sends a line switching instruction to the client, so that the client performs the line switching operation.

[0048] In one embodiment, the network fault warning method is applied in the financial customer service business. When the fault prediction result (network fault probability prediction value) output by the preset fault prediction model is less than or equal to 50% (the first preset warning threshold), there is no need to perform warning and line switching processing. When the fault prediction result (network fault probability prediction value) output by the preset fault prediction model is greater than 50% (the first preset warning threshold), it is determined that the fault prediction result meets the preset warning condition. Further, when the fault prediction result (network fault probability prediction value) is greater than 85% (the second preset warning threshold), a first-level warning information is generated, and a line switching instruction is sent to the client to automatically switch the line to avoid customer service interruption due to network failure. When the fault prediction result (network fault probability prediction value) is less than or equal to 85% (the second preset warning threshold), a second-level warning information is generated, and a line switching request is sent to the preset management party to manually switch the line to ensure the stability of customer service.

[0049] This embodiment distinguishes the first level warning information from the second level warning information based on the second preset warning threshold, and adopts different processing solutions for different situations, switching lines to ensure high availability of network services. The hierarchical warning and response mechanism of this embodiment helps to handle different types of fault prediction results more flexibly, ensuring that lines can be switched in time in emergency situations, while maintaining reasonable allocation of network lines and service stability in non-emergency situations.

[0050] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0051] In one embodiment, a network fault early warning device is provided, which corresponds one-to-one to the network fault early warning method in the above embodiment. Figure 4 As shown, the network fault warning device includes a preprocessing module 10, a model prediction module 20, a fault judgment module 30 and a fault warning module 40. The functional modules are described in detail as follows: The preprocessing module 10 is used to obtain network operation data, preprocess the network operation data, and obtain operation characteristic data; A model prediction module 20 is used to predict the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; A fault judgment module 30 is used to judge whether the fault prediction result meets the preset warning condition; The fault warning module 40 is used to generate fault warning information according to the fault prediction result if the fault prediction result meets the preset warning condition, and determine a warning processing scheme corresponding to the fault warning information.

[0052] In one embodiment, the pre-processing module 10 includes: A data screening and processing unit, used to perform data screening and processing on the network operation data according to a preset characteristic index to obtain initial characteristic data, wherein the preset characteristic index is a characteristic index determined by similarity analysis of historical operation data; A normalization processing unit, used for performing normalization processing on the initial feature data to obtain normalized feature data; The feature coding processing unit is used to perform feature coding processing on the normalized feature data to obtain the operation feature data.

[0053] In one embodiment, the model prediction module 20 includes: A sample preprocessing unit, used to obtain pre-collected historical operation data, perform sample preprocessing on the historical operation data, and obtain sample feature data; The initial fault prediction model determination unit is used to determine the constructed multi-layer perceptron network as the initial fault prediction model, wherein the initial fault prediction model includes an initial weight parameter and an initial threshold parameter. An iterative optimization processing unit, used for iteratively optimizing the initial weight parameters and the initial threshold parameters by using a Pelican optimization algorithm to obtain optimized weight parameters and optimized threshold parameters; A parameter updating unit, configured to update the parameters of the initial fault prediction model according to the optimization weight parameter and the optimization threshold parameter to obtain an optimized fault prediction model; A sample training unit is used to train the optimized fault prediction model according to the sample feature data to obtain the preset fault prediction model.

[0054] In one embodiment, the model prediction module 20 further includes: A fitness function determination unit, used for determining the mean square error of the multi-layer perceptron network as a fitness function in the Pelican optimization algorithm; A primary iterative processing unit, used for performing a primary iterative optimization process on the initial weight parameter and the initial threshold parameter by using the Pelican optimization algorithm to obtain a primary updated weight parameter and a primary updated threshold parameter; A primary fitness value determining unit, used to determine a primary fitness value corresponding to the primary update weight parameter and the primary update threshold parameter according to the fitness function; A primary optimization parameter determination unit, used to determine a primary optimization weight parameter and a primary optimization threshold parameter according to the primary fitness value, so that the primary optimization weight parameter and the primary optimization threshold parameter are subjected to secondary iterative optimization processing by the Pelican optimization algorithm, and the number of iterative optimization times is recorded; The iterative optimization stopping unit is used to obtain the optimization weight parameter and the optimization threshold parameter in the last iteration when the iterative optimization times reaches a preset iteration times threshold.

[0055] In one embodiment, the model prediction module 20 further includes: a sample prediction value determination unit, configured to input the sample feature data into the optimized fault prediction model to obtain a sample prediction value output by the optimized fault prediction model, A sample error determination unit, used to obtain a sample true value corresponding to the sample feature data, and determine a sample error according to the sample prediction value and the sample true value; A back propagation training unit, used to perform back propagation training on the optimized fault prediction model based on the sample error, and stop the training when the sample error reaches a preset accuracy threshold, to obtain a training weight parameter and a training threshold parameter; The preset fault prediction model determination unit is used to update the parameters of the optimized fault prediction model according to the training weight parameters and the training threshold parameters to obtain the preset fault prediction model.

[0056] In one embodiment, the fault judgment module 30 includes: A first threshold comparison unit, used to determine whether the fault prediction result is greater than a first preset warning threshold; a warning condition satisfaction determination unit, configured to determine that the fault prediction result satisfies a preset warning condition if the fault prediction result is greater than a first preset warning threshold; The warning condition not satisfied determination unit is used to determine that the fault prediction result does not satisfy the preset warning condition if the fault prediction result is less than or equal to a first preset warning threshold.

[0057] In one embodiment, the fault warning module 40 includes: A second threshold comparison unit, used to determine whether the fault prediction result is greater than a second preset warning threshold; A first warning processing unit, configured to generate first-level warning information if the fault prediction result is greater than a second preset warning threshold, and send a line switching instruction to the client, so that the client performs a line switching operation; The second warning processing unit is used to generate second-level warning information if the fault prediction result is less than or equal to a second preset warning threshold, and send a line switching request to a preset management party, and after receiving the line switching confirmation information corresponding to the line switching request sent by the preset management party, send a line switching instruction to the client, so that the client performs a line switching operation.

[0058] For the specific definition of the network fault warning device, please refer to the definition of the network fault warning method above, which will not be repeated here. Each module in the above-mentioned network fault warning device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0059] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium and an internal memory. The readable storage medium stores an operating system, computer-readable instructions and a database. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The database of the computer device is used to store data involved in the network fault early warning method. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, a network fault early warning method is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0060] In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor executes the computer-readable instructions to implement the following steps: Acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; The operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; Determining whether the fault prediction result meets the preset warning condition; If the fault prediction result meets the preset warning condition, fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined.

[0061] In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided. The readable storage media provided in this embodiment include non-volatile readable storage media and volatile readable storage media. The readable storage media store computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the following steps are implemented: Acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; The operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; Determining whether the fault prediction result meets the preset warning condition; If the fault prediction result meets the preset warning condition, fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined.

[0062] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0063] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0064] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A network failure early warning method, characterized in that: include: Acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; The operation characteristic data is predicted and processed by a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; Determining whether the fault prediction result meets the preset warning condition; If the fault prediction result meets the preset warning condition, fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined.

2. The network failure early warning method according to claim 1, characterized in that: Preprocessing the network operation data to obtain operation characteristic data includes: Performing data screening processing on the network operation data according to preset characteristic indicators to obtain initial characteristic data, wherein the preset characteristic indicators are characteristic indicators determined by similarity analysis of historical operation data; Normalizing the initial feature data to obtain normalized feature data; Perform feature encoding processing on the normalized feature data to obtain the operation feature data.

3. The network failure early warning method according to claim 1, characterized in that: Before predicting the operation characteristic data by using a preset fault prediction model, the method includes: Acquire pre-collected historical operation data, perform sample preprocessing on the historical operation data, and obtain sample feature data; Determine the constructed multi-layer perceptron network as an initial fault prediction model, wherein the initial fault prediction model includes an initial weight parameter and an initial threshold parameter; Iteratively optimizing the initial weight parameters and the initial threshold parameters by using the Pelican optimization algorithm to obtain optimized weight parameters and optimized threshold parameters; Update the parameters of the initial fault prediction model according to the optimization weight parameter and the optimization threshold parameter to obtain an optimized fault prediction model; The optimized fault prediction model is trained according to the sample feature data to obtain the preset fault prediction model.

4. The network failure early warning method according to claim 3, characterized in that: The iterative optimization process of the initial weight parameter and the initial threshold parameter by the Pelican optimization algorithm to obtain the optimized weight parameter and the optimized threshold parameter includes: The mean square error of the multilayer perceptron network is determined as the fitness function in the Pelican optimization algorithm; Performing an iterative optimization process on the initial weight parameter and the initial threshold parameter by using the Pelican optimization algorithm to obtain an updated weight parameter and an updated threshold parameter; Determine a primary fitness value corresponding to the primary update weight parameter and the primary update threshold parameter according to the fitness function; Determine a primary optimization weight parameter and a primary optimization threshold parameter according to the primary fitness value, so that the primary optimization weight parameter and the primary optimization threshold parameter are subjected to secondary iterative optimization processing by the Pelican optimization algorithm, and record the number of iterative optimization times; When the number of iterative optimizations reaches a preset iteration number threshold, the optimization weight parameter and the optimization threshold parameter in the last iteration are obtained.

5. The network failure early warning method according to claim 3, characterized in that: The step of training the optimized fault prediction model according to the sample feature data to obtain the preset fault prediction model includes: Inputting the sample feature data into the optimized fault prediction model to obtain a sample prediction value output by the optimized fault prediction model; Obtaining a sample true value corresponding to the sample feature data, and determining a sample error according to the sample predicted value and the sample true value; Performing back propagation training on the optimized fault prediction model based on the sample error, and stopping the training when the sample error reaches a preset accuracy threshold, to obtain a training weight parameter and a training threshold parameter; The parameters of the optimized fault prediction model are updated according to the training weight parameters and the training threshold parameters to obtain the preset fault prediction model.

6. The network failure early warning method according to claim 1, characterized in that: The determining whether the fault prediction result meets the preset warning condition includes: Determining whether the fault prediction result is greater than a first preset warning threshold; If the fault prediction result is greater than a first preset warning threshold, determining that the fault prediction result meets a preset warning condition; If the fault prediction result is less than or equal to the first preset warning threshold, it is determined that the fault prediction result does not meet the preset warning condition.

7. The network failure early warning method according to claim 6, characterized in that: The fault warning information includes first-level warning information and second-level warning information, and the first-level warning is higher than the second-level warning; If the fault prediction result meets the preset warning condition, fault warning information is generated according to the fault prediction result, and a warning processing scheme corresponding to the fault warning information is determined, including: Determining whether the fault prediction result is greater than a second preset warning threshold; If the fault prediction result is greater than a second preset warning threshold, generating first level warning information, and sending a line switching instruction to the client, so that the client performs a line switching operation; If the fault prediction result is less than or equal to the second preset warning threshold, a second level warning information is generated, and a line switching request is sent to the preset management party. After receiving the line switching confirmation information corresponding to the line switching request sent by the preset management party, a line switching instruction is sent to the client, so that the client performs the line switching operation.

8. A network fault early warning device, characterized in that: include: A preprocessing module, used to obtain network operation data, preprocess the network operation data, and obtain operation characteristic data; A model prediction module is used to predict the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by the Pelican optimization algorithm; A fault judgment module is used to judge whether the fault prediction result meets the preset warning condition; The fault warning module is used to generate fault warning information according to the fault prediction result if the fault prediction result meets the preset warning condition, and determine the warning processing plan corresponding to the fault warning information.

9. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that: When the processor executes the computer-readable instructions, the network failure early warning method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing computer-readable instructions, characterized in that: When the computer-readable instructions are executed by one or more processors, the one or more processors execute the network fault early warning method according to any one of claims 1 to 7.

Images