5G network data feature selection method and system based on graph immune network algorithm

Abstract

The invention discloses a 5G network data feature selection method and system based on a graph immune network algorithm, and the method comprises the following steps: carrying out the preprocessing of 5G network feature data, and obtaining a 5G network multi-feature time series data set; constructing a support vector machine model; performing target feature selection in the 5G network multi-feature time series data set based on a graph immune network algorithm and the support vector machine model to obtain a target feature group; training and testing the support vector machine model by using the target feature group; and detecting the 5G network by using the support vector machine model after the training test is completed. Therefore, the 5G network data dimension can be effectively reduced, redundant and irrelevant features are removed, and the real-time performance and accuracy of network anomaly detection are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of anomaly detection and intelligent computing, and particularly relates to a 5G network data feature selection method and system based on a graph immune network algorithm. BACKGROUND

[0002] In the process of large-scale deployment and intelligent evolution of 5G networks, anomaly detection of network equipment and communication links is a core link for guaranteeing ultra-low latency, high reliability transmission and network slice service quality. With the densification of 5G base stations, diversification of terminal equipment and distributed deployment of edge computing nodes, network systems collect multi-dimensional heterogeneous data such as signal strength, transmission delay, traffic load, signal-to-noise ratio and user connection number in real time through multi-source sensors and protocol interfaces, forming a high-dimensional spatio-temporal feature matrix. However, the problems of redundant features, dynamic noise and non-linear coupling between features hidden in the massive data seriously restrict the generalization ability and real-time inference efficiency of the anomaly detection model, and the existing technology faces many bottlenecks. A large number of irrelevant features (such as static configuration parameters and redundant protocol fields) and transient noise (such as channel burst interference and transient traffic peaks) in 5G network data cause feature space overload, and traditional methods (such as LASSO and random forest) cannot effectively select key features and cannot adapt to dynamic changes in network topology (such as base station load fluctuations and link switching), resulting in insufficient stability of the feature subset. The existing immune network algorithm needs to calculate the similarity between antibodies globally in the maintenance of antibody population diversity, and the computational complexity increases exponentially when facing a thousand-dimensional feature space, which is difficult to meet the millisecond-level anomaly detection requirements of 5G networks. For example, in the dense urban base station scenario, hourly data sampling leads to more than 500-dimensional feature dimensions, and traditional methods need tens of minutes to complete a single feature selection, which cannot adapt to real-time decision-making. 5G networks are essentially graph structure systems (such as base station-terminal connection graphs and core network topology graphs), but the existing feature selection methods (such as principal component analysis and support vector machines) do not effectively utilize the graph relationship of the network topology, resulting in insufficient generalization ability of the feature subset in dynamic environments. For example, in the mobile edge computing scenario, changes in user terminal positions cause local network topology reconstruction, and traditional methods frequently misjudge anomalies due to neglecting the relevance between nodes.

[0003] Early 5G network operation and maintenance relies on threshold alarm and rule engine (such as fixed threshold trigger based on traffic mutation), such method relies on artificial experience to set rules, it is difficult to quantify the abnormal boundary of complex network state, and it cannot adapt to the differentiated needs of multi-service scenarios. In order to improve the automation level, statistical learning based methods (such as isolation forest, single class SVM) are introduced, but their performance is limited by the steady state network assumption, and in the actual environment, due to user mobility and service burstiness, data distribution drift occurs, and the model is prone to false alarm or missed report. In recent years, although deep learning models (such as graph convolution network, spatio-temporal Transformer) can capture the spatio-temporal correlation of network topology, their black box characteristics lead to poor feature interpretability, and training relies on large-scale labeled data, which is limited in small and medium-sized edge node scenarios.

[0004] In summary, the existing methods face three major challenges of low feature redundancy elimination efficiency, real-time performance and poor topology adaptability in the high-dimensional, dynamic and graph structure data scenario of 5G network, and a new feature selection method is needed, which takes into account the computational efficiency and feature interpretability. SUMMARY

[0005] The present application provides a 5G network data feature selection method and system based on graph immune network algorithm, which can effectively reduce the dimension of 5G network data, remove redundant and irrelevant features, and significantly improve the real-time performance and accuracy of network anomaly detection. In a first aspect, a 5G network data feature selection method based on a graph immune network algorithm is provided, comprising the following steps: Preprocessing 5G network feature data to obtain a 5G network multi-feature time series data set; Constructing a support vector machine model; Performing target feature selection in the 5G network multi-feature time series data set based on a graph immune network algorithm and the support vector machine model to obtain a target feature group; Training and testing the support vector machine model using the target feature group; Detecting 5G network using the support vector machine model completed by training and testing.

[0006] In some embodiments, the step of preprocessing 5G network feature data to obtain a 5G network multi-feature time series data set comprises: Aligning each feature data and data label in the 5G network feature data according to the timestamp, deleting data with missing features or labels, and obtaining a 5G network multi-feature time series data set.

[0007] In some embodiments, the support vector machine model is as follows: ; Wherein, the kernel function ; wherein, is a support vector machine model; is a Lagrange multiplier; is a bias; , are different sample points, subscript , denotes the index of different samples; is a kernel function parameter; is a sample label.

[0008] In some embodiments, the graph-based immune network algorithm and the support vector machine model perform target feature selection on the 5G network multi-feature time series dataset to obtain a target feature group, including: randomly initializing the 5G network multi-feature time series dataset to generate a target antibody group, each individual including a plurality of encoded features, and performing feature selection on each individual based on a graph-based immune network algorithm; detecting the antibody group based on the support vector machine model; performing fitness evaluation on each individual according to the feature selection result and the detection result; performing a first redundancy subgroup screening on the antibody group according to the fitness evaluation result of each individual; performing cloning and mutation on the remaining antibody group according to the first screening result, and performing a second redundancy subgroup screening on the antibody group after cloning and mutation; performing iterative cloning and mutation on the second screening result until a condition is met to terminate to obtain a target feature group.

[0009] In some embodiments, the method of performing fitness evaluation on each individual according to the feature selection result and the detection result is as shown in the following formula: ; wherein, is a fitness value; and are weights; Accuracy is the detection result of the support vector machine model; F is the number of selected features; P is the total number of all features of each individual.

[0010] In some embodiments, the first redundancy subgroup screening on the antibody group according to the fitness evaluation result of each individual includes: dividing the antibody group into a plurality of subgroups using a graph generation technique, calculating the sub-similarity between each subgroup and the average fitness value of each subgroup; deleting the corresponding subgroup when the sub-similarity is less than a preset threshold, and deleting the corresponding subgroup when the average similarity is less than a preset threshold.

[0011] In some embodiments, the step of obtaining the target feature group according to the first screening result and performing clone mutation on the remaining antibody group, and performing the second screening on the redundant subgroups of the antibody group, comprises: performing clone mutation and binary adaptive mutation on the remaining subgroups in sequence; merging the mutated individuals and the individuals before cloning to obtain a merged group, calculating the similarity between each individual in the merged group, and deleting the corresponding individual when the similarity value is less than a preset threshold; selecting a target individual with the highest individual fitness from the remaining individuals to obtain the target feature group.

[0012] In some embodiments, the step of performing clone mutation and binary adaptive mutation on the remaining subgroups in sequence comprises:

[0013] The method of performing binary adaptive mutation on the cloned subset is as follows:

[0014] In the formula, is the number of clones; is the clone factor; is the mutation rate; is the mutation factor; and are the maximum fitness value and the minimum fitness value in the remaining subgroup, respectively; is the fitness value of the individual ; is a positive number; is a random number between 0 and 1; is one of the characteristics of the individual ; is the mutated individual .

[0015] In a second aspect, a 5G network data feature selection system based on a graph immune network algorithm is provided, comprising: a preprocessing module for preprocessing 5G network feature data to obtain a 5G network multi-feature time series data set; a model construction module for constructing a support vector machine model; a feature selection module in communication connection with the preprocessing module and the model construction module, for performing target feature selection in the 5G network multi-feature time series data set based on a graph immune network algorithm and the support vector machine model to obtain a target feature group; a training test module, in communication connection with the model construction module and the feature selection module, configured to train and test the support vector machine model by using the target feature group; and a network detection module, in communication connection with the training test module, configured to detect the 5G network by using the support vector machine model trained and tested.

[0016] In a third aspect, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the 5G network data feature selection method based on the graph immune network algorithm as described above.

[0017] Compared with the prior art, the present application has the following advantages: 5G network feature data is collected, and the 5G network feature data is preprocessed to obtain a 5G network multi-feature time series data set. Then, target feature selection is performed in the 5G network multi-feature time series data set based on a graph immune network algorithm and a support vector machine model to obtain a target feature group. Finally, the data selected by the feature is used for training and testing of the support vector machine. Therefore, the present application can effectively reduce the dimension of 5G network data, remove redundant and irrelevant features, and significantly improve the real-time performance and accuracy of network anomaly detection. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a flowchart of an embodiment of the 5G network data feature selection method based on the graph immune network algorithm of the present application; Figure 2 is a flowchart of another embodiment of the 5G network data feature selection method based on the graph immune network algorithm of the present application; Figure 3 is a comparison result graph of the present application and other algorithms in the accuracy of 5G network anomaly detection; Figure 4 is a structural schematic diagram of a 5G network data feature selection system based on the graph immune network algorithm of the present application. DETAILED DESCRIPTION

[0019] Reference will now be made in detail to the present application, examples of which are illustrated in the accompanying drawings. While the present application will be described in conjunction with the specific embodiments, it will be understood that the present application is not intended to be limited to the described embodiments. On the contrary, the present application is intended to cover modifications, alternatives and equivalents, which are within the spirit and scope of the present application, as defined by the appended claims. It should be noted that the method steps described herein can all be implemented by any functional block or functional arrangement, and any functional block or functional arrangement can be implemented as a physical entity or a logical entity, or a combination of both.

[0020] In order for those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0021] Note: the examples to be introduced next are only specific examples and not as limiting embodiments of the present application must be as follows specific steps, values, conditions, data, order, etc. Those skilled in the art can use the concept of the present application by reading the present specification to construct more embodiments not mentioned in the present specification.

[0022] Referring to Figure 1 and Figure 2 As shown in the present application, a 5G network data feature selection method based on a graph immune network algorithm is provided, comprising the following steps: S100, pre-processing 5G network feature data to obtain a 5G network multi-feature time series data set; S200, constructing a support vector machine model; S300, performing target feature selection in the 5G network multi-feature time series data set based on a graph immune network algorithm and the support vector machine model to obtain a target feature group; S400, training and testing the support vector machine model using the target feature group; S500, detecting a 5G network using the support vector machine model completed by training and testing.

[0023] Specifically, in the present application, a graph immune network algorithm (GINA) is proposed for 5G network data feature selection, and combined with a support vector machine (SVM) for anomaly detection. The specific steps include: collecting industrial equipment feature data-5G network feature data, such as signal strength, transmission delay, bandwidth utilization, traffic load, etc., and pre-processing the 5G network feature data to obtain a 5G network multi-feature time series data set. Then, based on the graph immune network algorithm and the support vector machine model, target feature selection is performed in the 5G network multi-feature time series data set to obtain a target feature group. Finally, the data selected by the feature is used for training and testing of the support vector machine, i.e. the data set can be divided into 70% training set and 30% test set to ensure the data independence of the model training and performance evaluation. Therefore, the present application can effectively reduce the dimension of 5G network data, remove redundant and irrelevant features, and significantly improve the real-time performance and accuracy of network anomaly detection.

[0024] Preferably, in another embodiment of the present application, the S100, pre-processing 5G network feature data to obtain a 5G network multi-feature time series data set step comprises: Align each feature data in the 5G network feature data with a data label according to a timestamp, delete data with missing features or labels, and obtain a 5G network multi-feature time series data set.

[0025] Specifically, in the embodiments of the present application, 5G network feature data and labels are collected, aligned according to timestamps, and processed to ensure that all feature data at the same time point are consistent, forming a unified time series data set. Then, data with a large number of missing features or labels in the aligned 5G network feature data set is deleted to obtain a data set suitable for model training and detection-5G network multi-feature time series data set.

[0026] Preferably, in another embodiment of the present application, for SVM model construction, the goal of SVM is to find a hyperplane that can separate two classes of data by maximizing the margin, and the convex quadratic programming problem model is described as:

[0027] wherein represents a slack variable for allowing certain classification errors, is a penalty factor, is a mapping function, is a sample point, is a normal vector of the separating hyperplane, is a bias, and by introducing a Lagrange multiplier , the above optimization problem is converted into its dual form:

[0028] Here is a Lagrange multiplier, is a kernel function, and a radial basis function is selected, described as follows:

[0029] Therefore, the final support vector machine model is as follows:

[0030] According to the positive and negative of the output value, the normal and abnormal of the sample are described, is a sign function, and if the value in the parentheses is ≥0, +1 is output, otherwise -1 is output.

[0031] , is a different sample point, and subscript , represents the index of different samples; is a kernel function parameter; is a sample label.

[0032] Preferably, in the additional embodiments of the application, the S300, based on the graph immune network algorithm and the support vector machine model, performs target feature selection in the 5G network multi-feature time series data set to obtain a target feature group step, comprising: S310, randomly initializing the 5G network multi-feature time series data set to generate a target individual antibody group, each individual including a plurality of encoded features, and performing feature selection on each individual based on the graph immune network algorithm; S320, detecting the antibody group based on the support vector machine model; S330, performing fitness evaluation on each individual according to the feature selection result and the detection result; S340, performing a first redundancy subgroup screening on the antibody group according to the fitness evaluation result of each individual; S350, performing cloning and mutation on the remaining antibody group according to the first screening result, and performing a second redundancy subgroup screening on the antibody group after cloning and mutation; S360, iteratively performing cloning and mutation on the second screening result until a condition is met to terminate and obtain a target feature group.

[0033] Specifically, in the embodiments of the application, the immune network algorithm feature selection, specifically, in the initialization process, a random initial antibody group is generated, that is, an antibody group of individuals is generated, each individual representing a possible feature combination, using to represent, each individual including a plurality of features, using binary coding, that is, 0 and 1 to represent feature states, if the parameter value is 1, it represents that the feature is selected, if it is 0, it represents that the feature is not selected.

[0034] Detecting the antibody group based on the support vector machine model, and performing fitness evaluation on each individual according to the feature selection result and the detection result, specifically as follows:

[0035] In the formula, is the fitness value; and are weights, respectively considering the detection accuracy Accuracy of the SVM and the feature redundancy of the feature subset; F is the number of selected features; P is the total number of all features of each individual, including the number of selected and unselected.

[0036] Preferably, in the additional embodiments of the application, the S340, according to the fitness evaluation result of each individual, performs a first redundancy subgroup screening on the antibody group step, comprising: dividing the antibody population into a plurality of sub-populations by using a sub-graph generation technique, calculating a sub-similarity between each of the sub-populations and an average fitness value of each of the sub-populations; deleting the sub-population corresponding to the sub-similarity less than a preset threshold value, and deleting the sub-population corresponding to the average similarity less than a preset threshold value.

[0037] Specifically, in the embodiment of the present application, the encoded antibodies are divided into a plurality of sub-graph communities by using a sub-graph generation technique, the average value of each feature dimension of each sub-graph antibody is calculated to obtain a center point, the Euclidean distance of different sub-graph center points is calculated to judge the similarity of the sub-graphs and calculate the average fitness value of each sub-graph antibody, and the redundant sub-population screening is performed to delete the sub-population with low fitness and similarity.

[0038] Preferably, in another embodiment of the present application, the S350, according to the first screening result, performs clone variation on the remaining antibody population, performs second redundant sub-population screening on the antibody population after clone variation, and obtains a target feature population step, comprising: sequentially performing clone and binary adaptive variation on the remaining sub-population; merging the individuals after variation and the individuals before clone to obtain a merged population, calculating the similarity between each individual in the merged population, and deleting the individual corresponding to the similarity value less than a preset threshold value; selecting a target individual with the highest individual fitness from the remaining individuals to obtain a target feature population.

[0039] Specifically, in the embodiment of the present application, clone variation is performed in the remaining sub-population, the parameter combination with high fitness in each sub-population is cloned to generate a clone subset, the clone number is proportional to the fitness value, and then adaptive variation is performed on the subset, that is, the mutation rate is inversely proportional to the fitness value, and the variation amplitude is inversely proportional to the fitness value.

[0040] The method of performing clone in the remaining sub-population is as follows:

[0041] In order to make the individuals after variation meet the binary coding, a binary adaptive variation operation is designed, and the method of performing binary adaptive variation on the cloned subset is as follows:

[0042] In the formula, is the clone number; is the clone factor; is the mutation rate; is the mutation factor; and are the maximum fitness value and the minimum fitness value in the remaining sub-population, respectively. fitness value of the individual fitness value of the individual a very small positive number to avoid division by zero.

[0043] using to dynamically adjust the mutation probability, a random number between 0 and 1, each feature of each individual by probability, ensuring that high fitness individuals have a small mutation probability, and low fitness individuals have a large mutation probability, fitness value of the individual one of the features; is the individual after mutation.

[0044] Merge the individual after mutation with the original cloned individual, and calculate the Euclidean distance between the individuals and , described as:

[0045] If two individuals are considered similar, and individuals with relatively low fitness values are removed, is the similarity threshold set. The suppression operation is to remove similar individuals and maintain the diversity of the population. The individuals are sorted according to fitness, and the individuals with the highest fitness values form the new population and proceed to the next generation of iteration. This cycle is repeated until the maximum number of iterations is reached, and the optimal solution is output, at which time the target feature group is obtained.

[0046] Referring to Figure 3 , it is shown that the 5G network anomaly detection accuracy of the method of the present application is significantly higher than that of other algorithms.

[0047] Referring to Figure 4 , the 5G network data feature selection system based on the graph immune network algorithm provided by the embodiment of the present application comprises: a preprocessing module for preprocessing 5G network feature data to obtain 5G network multi-feature time series data set; a model construction module for constructing a support vector machine model; a feature selection module in communication connection with the preprocessing module and the model construction module, for selecting target features in the 5G network multi-feature time series data set based on the graph immune network algorithm and the support vector machine model, and obtaining a target feature group; The training test module is connected in communication with the model construction module and the feature selection module, and is configured to train and test the support vector machine model by using the target feature group. The network detection module is connected in communication with the training test module, and is configured to detect the 5G network by using the support vector machine model after the training and test.

[0048] To sum up, the beneficial effects of the present application are: collecting industrial equipment feature data such as signal strength, transmission delay, bandwidth utilization, traffic load, etc., and performing time sequence alignment on the data, and deleting a large amount of missing data of features and labels; immune network algorithm feature selection, specifically, randomly initializing an antibody group in the initialization process, encoding the features as antibodies, and using binary 0 and 1 to represent whether the features are selected; constructing an antibody group graph structure, where the nodes represent feature subset antibodies, and the edge weights reflect the similarity between antibodies; using subgraph generation technology to divide the antibody group into subgraph communities, cloning and mutating the antibodies in each subgraph, dynamically adjusting the cloning scale through the fitness of the antibodies, and introducing binary adaptive mutation to increase the mutation probability of low fitness antibodies to help the algorithm jump out of the local optimum and improve the global search ability of feature selection; calculating the subgraph center and selecting redundant subgroups based on the center similarity, calculating the similarity between antibodies in the retained subgroups, eliminating duplicate feature subsets, and selecting the optimal feature subset through multi-generation evolution; finally, using the data after feature selection for training and testing of the support vector machine. The present application can effectively reduce the dimensionality of 5G network data, remove redundant and irrelevant features, and significantly improve the real-time performance and accuracy of network anomaly detection.

[0049] Specifically, the present embodiment corresponds to the above-mentioned method embodiment one by one, and the functions of each module have been described in detail in the corresponding method embodiment, so they will not be repeated here.

[0050] Based on the same inventive concept, the embodiments of the present application also provide a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement all method steps or part of the method steps of the above method.

[0051] The present application implements all or part of the processes in the above method, and can also be completed by a computer program instructing related hardware. The computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0052] Based on the same inventive concept, the embodiments of the present application also provide an electronic device, including a memory and a processor, the memory stores a computer program running on the processor, and the processor implements all or part of the method steps of the above method when executing the computer program.

[0053] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and connects all parts of the computer device through various interfaces and lines.

[0054] The memory can be used to store computer programs and / or modules, and the processor realizes various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and at least one application program required by a function (for example, a sound playing function, an image playing function, etc.); and the data storage area can store data created according to use of the mobile phone (for example, audio data, video data, etc.). In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0055] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, a server or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer usable program codes.

[0056] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems), servers and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be realized by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows or blocks.

[0057] These computer program instructions can also be stored in a computer readable memory capable of directing the computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction apparatus, which realizes the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that carries out the functions specified in one or more flows or blocks.

[0058] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable devices provide processes for implementing the flow Figure 1 one or more flows and / or blocks Figure 1 one or more blocks or steps of the functions specified in the flow

[0059] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

Claims

1. A 5G network data feature selection method based on graph immune network algorithm, characterized in that: The following steps are involved: Preprocess the 5G network feature data to obtain a 5G network multi-feature time series dataset; Build a support vector machine model; Based on the graph immune network algorithm and the support vector machine model, target features are selected in the 5G network multi-feature time series dataset to obtain a target feature group; Using the target feature group to train and test the support vector machine model; The trained and tested support vector machine model is used to detect the 5G network.

2. The 5G network data feature selection method based on the graph immune network algorithm according to claim 1 is characterized in that: The step of preprocessing the 5G network feature data to obtain a 5G network multi-feature time series data set includes: Each feature data in the 5G network feature data is aligned with the data label according to the timestamp, and the data with missing features or labels is deleted to obtain a 5G network multi-feature time series data set.

3. The 5G network data feature selection method based on the graph immune network algorithm according to claim 1 is characterized in that: The support vector machine model is shown below: Among them, the kernel function ; Where, is the support vector machine model; is the Lagrange multiplier; is bias; 、 For different sample points, subscript , Indicates the index of different samples; is the kernel function parameter; is the sample label.

4. The 5G network data feature selection method based on the graph immune network algorithm according to claim 1, characterized in that: The step of selecting target features in the 5G network multi-feature time series dataset based on the graph immune network algorithm and the support vector machine model to obtain a target feature group includes: The 5G network multi-feature time series dataset is randomly initialized to generate a target individual antibody group, each individual includes multiple encoded features, and feature selection is performed on each individual based on a graph immune network algorithm; Detecting the antibody group based on the support vector machine model; The fitness of each individual is evaluated based on the feature selection results and test results; Performing a first redundant subgroup screening on the antibody group according to the fitness evaluation results of each individual; Based on the results of the first screening, the remaining antibody population is cloned and mutated, and the cloned antibody population is subjected to a second redundant subgroup screening; The second screening results are subjected to iterative cloning and mutation until the target feature group is obtained when the termination conditions are met.

5. The 5G network data feature selection method based on the graph immune network algorithm according to claim 4 is characterized in that: The method for evaluating the fitness of each individual based on the feature selection results and the test results is shown in the following formula: Where, is the fitness value; and is the weight; Accuracy is the detection result of the support vector machine model; F is the number of selected features; P is the total number of all features of each individual.

6. The 5G network data feature selection method based on the graph immune network algorithm according to claim 4 is characterized in that: The step of performing a first redundant subgroup screening on the antibody group according to the fitness evaluation result of each individual comprises: Dividing the antibody group into multiple subgroups using a subgraph generation technology, and calculating the sub-similarity between each of the subgroups and the average fitness value of each of the subgroups; The subgroup corresponding to when the sub-similarity is less than a preset threshold is deleted, and the subgroup corresponding to when the average similarity is less than a preset threshold is deleted.

7. The 5G network data feature selection method based on the graph immune network algorithm according to claim 6, characterized in that: The step of performing clonal mutation on the remaining antibody population according to the first screening results, and performing a second redundant subgroup screening on the antibody population after clonal mutation to obtain the target feature population includes: Perform cloning and binary adaptive mutation in the remaining subgroups in turn; Merging the mutated individuals with the individuals before cloning to obtain a merged group, calculating the similarity between each individual in the merged group, and deleting individuals whose similarity values ​​are less than a preset threshold; The target individual with the highest individual fitness is selected from the remaining individuals to obtain the target feature group.

8. The 5G network data feature selection method based on the graph immune network algorithm according to claim 7, characterized in that: The steps of sequentially performing cloning and binary adaptive mutation in the remaining subgroups include: The method for cloning in the remaining subgroups is as follows: The method for performing binary adaptive mutation on the cloned subset is as follows: Where, is the number of clones; is a cloning factor; is the mutation rate; is a mutation factor; and are the maximum fitness value and the minimum fitness value in the remaining subgroups respectively; For individuals The fitness value of is a positive number; A random number between 0 and 1; For individuals One of the features; for The mutated individual.

9. A 5G network data feature selection system based on graph immune network algorithm, characterized in that: include: A preprocessing module is used to preprocess the 5G network feature data to obtain a 5G network multi-feature time series data set; Model building module, used to build support vector machine models; a feature selection module, communicatively connected to the preprocessing module and the model building module, for selecting target features in the 5G network multi-feature time series dataset based on a graph immune network algorithm and the support vector machine model to obtain a target feature group; A training and testing module, in communication with the model building module and the feature selection module, for training and testing the support vector machine model using the target feature group; as well as, The network detection module is in communication with the training and testing module and is used to detect the 5G network using the support vector machine model completed by the training and testing.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for selecting 5G network data features based on a graph immune network algorithm as described in any one of claims 1 to 8 is implemented.

Images