Network element health assessment method, device, equipment and computer readable storage medium

Abstract

The present disclosure relates to a network element health assessment method, device, equipment and computer readable storage medium. The present disclosure selects target features according to the contribution of network element instance samples to the prediction process, explains the influence of each feature data on the health degree of the network element instance, facilitates the positioning of health degree abnormality reasons, and then predicts the network element health probability based on the target features in the actual operation data. According to different network element categories or use scenarios, the corresponding target features can be flexibly selected, so that the network element health assessment method can be better applied to various network elements and scenes, and the flexibility and accuracy of the network element assessment method are improved.

Description

Technical Field

[0001] This disclosure relates to the field of network detection technology, and in particular to a method, apparatus, device, and computer-readable storage medium for assessing the health of network elements. Background Technology

[0002] Network elements play a crucial role in communication networks. They are devices with specific functions and independent operating capabilities, forming the basic units of the communication network and responsible for completing various communication tasks. To ensure the normal operation of the communication network, assessing the health of network elements and promptly detecting any abnormalities is a vital task.

[0003] Typically, staff will set corresponding deduction thresholds and deduction amounts for various operational metrics of network elements (such as equipment alarms, performance indicators, operation logs, cutover operations, dial-up testing, and service complaint volume). They will also set weighting factors for each dimension, and finally calculate and evaluate the network element's score through weighted summation. A lower score indicates a higher probability of the network element experiencing an anomaly, requiring timely maintenance. Another approach is to predict the alarm sequence of network elements based on a Hidden Markov Model, issuing an early warning when the prediction indicates that the network element is about to enter a fault state.

[0004] However, existing network element health assessment methods have relatively fixed scoring criteria and assumptions for hidden Markov prediction models, resulting in a lack of flexibility and accuracy in these methods. Summary of the Invention

[0005] To address the aforementioned technical problems, this disclosure provides a method, apparatus, device, and computer-readable storage medium for assessing network element health, thereby improving the flexibility and accuracy of network element assessment methods.

[0006] In a first aspect, embodiments of this disclosure provide a method for assessing the health of network elements, including:

[0007] Obtain a sample dataset of network element instances based on the historical operational data of network elements;

[0008] For each network element instance sample in the network element instance sample dataset, the network element instance sample is input into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features.

[0009] Calculate the contribution of each of the sample features in the prediction process;

[0010] Based on the contribution of each of the sample features, multiple target features are determined;

[0011] The target features are extracted based on the actual operating data of the network elements to obtain the network element target feature dataset;

[0012] The target feature dataset is input into the trained network element health assessment model so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

[0013] In some embodiments, obtaining the network element instance sample dataset based on the historical operational data of the network element includes:

[0014] An initial sample dataset is obtained based on historical operational data. The initial sample dataset includes normal network element samples and abnormal network element samples, with the number of abnormal network element samples being less than the number of normal network element samples.

[0015] For each of the abnormal network element samples, at least one synthetic network element sample is generated based on multiple neighboring samples of the abnormal network element sample, and the similarity between the synthetic network element sample and the abnormal network element sample is greater than a first preset threshold.

[0016] The normal network element samples, abnormal network element samples, and synthetic network element samples are used as the network element instance sample dataset, and the difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold.

[0017] In some embodiments, generating at least one synthetic network element sample based on neighboring samples of each anomalous network element sample includes:

[0018] Calculate the sample distance between the abnormal network element sample and every other abnormal network element sample;

[0019] The preset number of samples with the shortest sample distance are determined as the adjacent samples;

[0020] For each of the neighboring samples, calculate the feature distance between the neighboring sample and each feature of the abnormal network element sample;

[0021] Each of the feature distances is multiplied by a preset random factor to obtain multiple synthetic features;

[0022] The synthetic network element sample is composed of the multiple synthetic features.

[0023] In some embodiments, after obtaining the network element instance sample dataset based on the historical operational data of the network element, the method further includes:

[0024] The baseline prediction model is obtained by training the model based on the network element instance sample dataset.

[0025] In some embodiments, the prediction process includes a single prediction for each network element instance sample, and calculating the contribution of each sample feature in the prediction process includes:

[0026] For each of the sample features, calculate the marginal contribution of the sample feature in each of the single predictions;

[0027] Calculate the cumulative or average value of the marginal contribution in each of the single predictions to obtain the contribution of the sample feature in the prediction process.

[0028] In some embodiments, after determining multiple target features based on the contribution of each of the sample features, the method further includes:

[0029] Based on the historical operational data, the target features are extracted to obtain the network element feature sample dataset;

[0030] The baseline prediction model is updated and trained based on the network element feature sample dataset to obtain a trained network element health assessment model.

[0031] In some embodiments, the method further includes:

[0032] During the real-time prediction of the health probability of the network element by the network element health assessment model, the contribution of each target feature in the real-time prediction process is calculated.

[0033] The target features are updated based on the contribution of each of the target features;

[0034] The health probability of the network element is predicted based on the updated target characteristics.

[0035] Secondly, embodiments of this disclosure provide a network element health assessment device, comprising:

[0036] The first acquisition module is used to acquire a sample dataset of network element instances based on the historical operational data of the network elements.

[0037] The first prediction module is used to input each network element instance sample in the network element instance sample dataset into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features.

[0038] A calculation module is used to calculate the contribution of each of the sample features in the prediction process;

[0039] A determination module is used to determine multiple target features based on the contribution of each of the sample features;

[0040] The second acquisition module is used to extract the target features based on the actual operating data of the network element to obtain the network element target feature dataset;

[0041] The second prediction module is used to input the target feature dataset into the trained network element health assessment model, so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

[0042] Thirdly, embodiments of this disclosure provide an electronic device, including:

[0043] Memory;

[0044] Processor; and

[0045] Computer programs;

[0046] The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in the first aspect.

[0047] Fourthly, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the method described in the first aspect.

[0048] Fifthly, embodiments of this disclosure also provide a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the network element health assessment method as described above.

[0049] The network element health assessment method, apparatus, device, and computer-readable storage medium provided in this disclosure select target features based on the contribution of network element instance samples to the prediction process, explain the impact of each feature data on the health of the network element instance, facilitate the location of the cause of health anomalies, and then predict the network element health probability based on the target features in the actual operating data. It can flexibly select the corresponding target features according to different network element categories or usage scenarios, so that the network element health assessment method can be better applied to various network elements and scenarios, improving the flexibility and accuracy of the network element assessment method. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0051] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of the network element health assessment method provided in this embodiment of the disclosure;

[0053] Figure 2 A flowchart of a network element health assessment method provided in another embodiment of this disclosure;

[0054] Figure 3 This is a schematic diagram of the structure of the network element health assessment device provided in the embodiments of this disclosure;

[0055] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0056] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0057] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0058] Traditional rule-based scoring methods set six dimensions: equipment alarms (A), performance indicators (B), operation logs (C), cutover operations (D), testing dimensions (E), and business complaint volume (F). Based on different indicators within each dimension, corresponding scoring thresholds and scoring amounts are set, along with weighting factors α, β, ε for A, B, E, and F. Finally, the health score W of the network element is evaluated by weighted summation, and the calculation method is as follows:

[0059]

[0060] However, the aforementioned indicator deduction values, thresholds, and dimensional weights all rely on expert experience, introducing a degree of subjectivity. Individual differences exist between different network elements, production environments, and user groups, resulting in a lack of universal interpretability for this method. Furthermore, the threshold-based interval deduction method can only roughly quantify health deductions. For example, if the maximum concurrent online users of the PGW (PDNGateway) fluctuate by 5% for two consecutive periods, the health loss factor is increased by 5 points from its initial value. Therefore, when the indicator change does not exceed the deduction threshold, the health deduction is not reflected. That is, if the maximum concurrent online users of the PGW fluctuate by 4.9% and 5.1% for two consecutive periods respectively, failing to reach a fluctuation of more than 5% for two consecutive periods, the health loss factor is increased by 0 points from its initial value. Additionally, there are discrepancies in expert experience across different operational environments, leading to situations where the method is unusable, significantly increasing user costs and reducing its overall usability.

[0061] Another approach is to predict alarm sequences of network elements based on Hidden Markov Models. This method only depends on the state at each time step, such as whether an alarm is triggered, and the corresponding observed object, such as a certain AMF (Access and Mobility Management Function) network element. At a certain time t, different instances of the same type of AMF network element express completely different states. For example, AMF network element 1 generates an alarm, while AMF network element 2 does not generate an alarm. However, at several times after time t, the states of the two network element instances are consistent. Therefore, the prediction results of this method are also consistent, that is, different input alarm sequences yield the same prediction output. This inconsistency causes the method to fail.

[0062] To address the aforementioned issues, this disclosure provides a method for assessing the health of network elements. The method will be described below with reference to specific embodiments.

[0063] Figure 1 This is a flowchart illustrating the network element health assessment method provided in this embodiment. This method can be applied to terminal devices with data processing capabilities, such as smartphones, PDAs, tablets, wearable devices with displays, desktop computers, laptops, all-in-one computers, and smart home devices. It is understood that the network element health assessment method provided in this embodiment can also be applied in other scenarios.

[0064] The following is about Figure 1 The network element health assessment method shown is introduced below, and the specific steps of this method are as follows:

[0065] S101. Obtain the network element instance sample dataset based on the historical operation data of the network element.

[0066] Historical operational data of a network element refers to the data such as the operating status and performance parameters recorded and saved by the network element device over a period of time.

[0067] Specifically, the historical operating data of network elements includes, but is not limited to: performance data, such as central processing unit (CPU) utilization, memory usage, interface traffic, packet loss rate, latency, etc.; alarm data, such as the occurrence time, type, level, and possible causes of alarms; and configuration data, such as hardware configuration, software version, interface configuration, routing configuration, etc.

[0068] The network element instance sample dataset is extracted from the historical operating data of the network element over a period of time. It includes the operating index values ​​of various operating indicators of the network element at different time points (i.e., the sample feature values ​​of each sample feature), as well as the network element status label, which is used to indicate whether the network element status is normal at that time point.

[0069] That is, a single network element instance sample includes the network element feature value at a certain point in time, and the network element status label at that point in time.

[0070] S102. For each network element instance sample in the network element instance sample dataset, the network element instance sample is input into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features.

[0071] Optionally, a benchmark prediction model can be obtained by training the model based on the network element instance sample dataset.

[0072] The baseline prediction model is used to predict the probability that a network element is in a normal state based on the network element feature values ​​in the aforementioned network element instance samples. This model is trained based on network element instance samples; specifically, it is trained based on the training set samples within the network element instance samples.

[0073] The model is trained based on the network element instance sample dataset to obtain a benchmark prediction model, including: for each network element instance sample, inputting the network element feature value therein into the benchmark prediction model to be trained to obtain the prediction result output by the benchmark prediction model to be trained; updating the parameters of the benchmark prediction model to be trained to guide the prediction result to approach the network element state label.

[0074] Finally, when the error between the prediction result and the network element state label is controlled within the preset error range, the trained benchmark prediction model is obtained.

[0075] For example, the baseline prediction model outputs the probability that a network element is in a normal state, with a value ranging from 0 to 1. The closer the value is to 1, the greater the probability that the network element is in a normal state. The network element state label of a network element instance sample is 0 or 1, where 0 indicates an abnormal network element and 1 indicates a normal network element. During the training of the baseline prediction model, for abnormal network element samples with a network element state label of 0, the prediction result output by the baseline prediction model is guided to be closer to 0; for normal network element samples with a network element state label of 1, the prediction result output by the baseline prediction model is guided to be closer to 1.

[0076] S103. Calculate the contribution of each of the sample features in the prediction process.

[0077] In the process of making predictions, each sample feature has its own contribution. The higher the contribution, the greater the influence of that sample feature on the prediction result; the lower the contribution, the smaller the influence of that sample feature on the prediction result.

[0078] S104. Determine multiple target features based on the contribution of each of the sample features.

[0079] The sample features are ranked according to their contribution, and the top-ranked sample features are selected as target features.

[0080] Optionally, a preset number of sample features with the highest ranking results can be selected as target features; or, sample features with a contribution greater than a preset contribution threshold can be selected as target features; or, a preset number of sample features can be selected as target features based on expert experience.

[0081] S105. Extract the target features based on the actual operation data of the network elements to obtain the network element target feature dataset.

[0082] Health assessments of network elements are actually performed on network elements that are currently in operation, therefore it is necessary to obtain the actual operational data of the network elements.

[0083] Similar to the historical operational data of network elements, the actual operational data of network elements refers to the operational data of the network element at the current point in time. Further selection of the operational indicators and their values ​​corresponding to the target features yields the network element target feature dataset.

[0084] S106. Input the target feature dataset into the trained network element health assessment model so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

[0085] This embodiment of the disclosure obtains a network element instance sample dataset based on the historical operational data of network elements. For each network element instance sample in the dataset, the sample is input into a benchmark prediction model to predict its health probability. The sample includes multiple sample features. The contribution of each feature in the prediction process is calculated. Multiple target features are determined based on their contribution. The target features are extracted from the actual operational data of the network element to obtain a target feature dataset. The target feature dataset is input into a trained network element health assessment model to predict the health probability of the network element in real time. By selecting target features based on their contribution to the prediction process, the influence of each feature on the health of the network element instance is explained, facilitating the identification of health anomalies. Furthermore, the network element health probability is predicted based on the target features in the actual operational data. This allows for flexible selection of appropriate target features according to different network element categories or usage scenarios, enabling the network element health assessment method to be applied to various network elements and scenarios, thus improving the flexibility of the method.

[0086] Meanwhile, the baseline prediction model or network element health assessment model in this embodiment can be implemented using an integrated model that is more suitable for the scenario. The selectable models include, but are not limited to, XGBoost, LightGBM, GBDT and Random Forest, among which Random Forest is more interpretable.

[0087] Based on the above embodiments, the step of obtaining a network element instance sample dataset based on the historical operation data of network elements includes: obtaining an initial sample dataset based on the historical operation data, wherein the initial sample dataset includes normal network element samples and abnormal network element samples, and the number of abnormal network element samples is less than the number of normal network element samples; for each abnormal network element sample, generating at least one synthetic network element sample based on multiple adjacent samples of the abnormal network element sample, wherein the similarity between the synthetic network element sample and the abnormal network element sample is greater than a first preset threshold; and using the normal network element samples, abnormal network element samples, and synthetic network element samples as the network element instance sample dataset, wherein the difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold.

[0088] Specifically, for each of the abnormal network element samples, generating at least one synthetic network element sample based on the neighboring samples of the abnormal network element sample includes: calculating the sample distance between the abnormal network element sample and every other abnormal network element sample; determining a preset number of samples with the shortest sample distance as the neighboring samples; for each neighboring sample, calculating the feature distance between the neighboring sample and each feature of the abnormal network element sample; multiplying each feature distance by a preset random factor to obtain multiple synthetic features; and constructing the synthetic network element sample from the multiple synthetic features.

[0089] Typically, in historical operational data, network elements are in a normal state most of the time. Therefore, the number of abnormal network element samples in the corresponding initial sample dataset is far less than the number of normal network element samples, which may lead to insufficient model prediction accuracy due to the insufficient number of abnormal network element samples. Therefore, this embodiment of the disclosure utilizes existing abnormal network element samples to perform SMOTE (Synthetic Minority Oversampling Technique) algorithm data augmentation to make the number of abnormal network element samples close to the number of normal network element samples.

[0090] Specifically, for each abnormal network element sample, the Euclidean distance between the features of the abnormal network element sample and the features of every other abnormal network element sample is calculated, and finally, a preset number of samples with the shortest Euclidean distance are determined as adjacent samples.

[0091] Furthermore, each neighboring sample includes multiple sample features, such as the system average load and the number of authentication requests. The feature distance between each feature of the neighboring sample and the abnormal network element sample is calculated, and then each feature distance is multiplied by a randomly generated factor between 0 and 1 to obtain multiple synthetic features. The sample composed of these synthetic features becomes the synthetic network element sample. This is equivalent to randomly selecting a point on the line connecting the abnormal network element sample and its neighboring samples as the synthetic network element sample. Thus, the similarity between the synthetic network element sample and the corresponding abnormal network element sample is greater than a first preset threshold.

[0092] The above process of constructing synthetic network element samples is repeated until the difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold. Typically, the number of synthetic network element samples is:

[0093] Y = N / 100 × T

[0094] Where N is the sampling ratio, which is usually a whole number of hundreds, such as 100, 200, etc., and is determined according to the number of training network element samples T.

[0095] The expanded abnormal network element samples include the original abnormal network element samples and the synthesized network element samples, and the number of expanded abnormal network element samples is about 1:1 with the number of normal network element samples.

[0096] This embodiment of the disclosure expands the abnormal network element samples so that the number of expanded abnormal network element samples is basically consistent with the number of normal network element samples. This solves the problem of insufficient model prediction accuracy caused by imbalanced sample labels, improves the accuracy of the baseline prediction model and the accuracy of contribution calculation, and thus improves the accuracy of network element health assessment.

[0097] Figure 2 A flowchart of a network element health assessment method provided in another embodiment of this disclosure is shown below. Figure 2 As shown, the method includes the following steps:

[0098] S201. Obtain a sample dataset of network element instances based on the historical operation data of network elements.

[0099] Specifically, an initial sample dataset is obtained based on historical operational data. The initial sample dataset includes normal network element samples and abnormal network element samples, with the number of abnormal network element samples being less than the number of normal network element samples. For each abnormal network element sample, at least one synthetic network element sample is generated based on multiple adjacent samples of the abnormal network element sample. The similarity between the synthetic network element sample and the abnormal network element sample is greater than a first preset threshold. The normal network element samples, abnormal network element samples, and synthetic network element samples are used as the network element instance sample dataset, and the difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold.

[0100] S202. For each network element instance sample in the network element instance sample dataset, the network element instance sample is input into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features.

[0101] S203. For each of the sample features, calculate the marginal contribution of the sample feature in each of the single predictions.

[0102] S204. Calculate the cumulative or average value of the marginal contribution in each of the single predictions to obtain the contribution of the sample feature in the prediction process.

[0103] SHAP (SHapley Additive exPlanation) is an additive explanatory measure derived from game theory. It is the only feature interpretability measure among ensemble model interpretability tools that possesses both consistency and local accuracy. Its formula is:

[0104]

[0105] Where, φ i Let M represent the contribution of feature i, and M be the set of all available data features; S be the set of features other than sample i that appear in a single prediction process, i.e. f is the number of elements in the feature set S; x (S) is the predicted value of the baseline prediction model for the sample when only the feature set S is used, f x (S∪{i}) is the model's prediction of a sample when feature i is added to the feature set S.

[0106] Weighting Items It is calculated based on the Shapley value assignment principle in combinatorics.

[0107] Among them, the definition

[0108] f x (S)=f(h x (z′))=E(f(x)|x s )

[0109] Where z′ represents the index of the feature set that did not appear in a single prediction, z′ i =0 indicates that feature i did not appear, z′ i =1 indicates that feature i appears in a single prediction calculation. f x (S∪{i})-f x (S) represents the marginal contribution of feature i, which indicates the change in the predicted value when feature i is added to a subset of features.

[0110] Therefore, the marginal contribution of a feature in a single prediction process can be represented, while the contribution on the overall dataset can be represented by the sum of the single prediction results or the mean statistic.

[0111] S205. Determine multiple target features based on the contribution of each of the sample features.

[0112] S206. Extract the target features based on the historical operation data to obtain the network element feature sample dataset.

[0113] S207. Update and train the benchmark prediction model based on the network element feature sample dataset to obtain the trained network element health assessment model.

[0114] After the target features have been determined, they can be used to extract features from historical operating data, retaining only the target features to obtain a network element feature sample dataset. The baseline prediction model can then be updated and trained again to obtain a trained network element health assessment model.

[0115] Optionally, feature importance can be ranked using data feature SHAP values, and target features can be selected by combining expert experience. Model parameter tuning can be achieved by combining hyperparameter search cross-validation, including but not limited to grid search cross-validation and random search cross-validation.

[0116] Grid search cross-validation is an exhaustive search method that searches for the optimal model parameters by traversing every possible combination of parameters in a given parameter grid. Random search cross-validation is an improvement on grid search; instead of traversing all possible parameter combinations, it randomly selects a subset of parameter combinations for training.

[0117] S208. Extract the target features based on the actual operating data of the network elements to obtain the network element target feature dataset.

[0118] S209. Input the target feature dataset into the trained network element health assessment model so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

[0119] Specifically, the network element health assessment model predicts the health probability of the network element in real time, outputting the failure probability p of the network element. It then maps the normal probability (1-p) of the network element to a scoring interval of [0, 100] to obtain the health assessment result of the network element. The calculation process is as follows:

[0120] Health assessment result = (1-p)×100

[0121] This embodiment of the disclosure solves the problem of insufficient model prediction accuracy caused by imbalanced sample labels by combining the SMOTE minority sample synthesis technology. It uses the normal probability of the network element object to map its health score, realizes real-time quantitative assessment of network element health, and improves its sensitivity to quickly capture changes in network element health.

[0122] Meanwhile, the embodiments of this disclosure employ SHAP feature interpretability technology, which solves the problem of inconsistent feature importance measurement in classification tasks, adaptively optimizes feature selection, improves model prediction accuracy, and solves the problem of scene feature differentiation.

[0123] Based on the above embodiments, the method further includes: calculating the contribution of each target feature in the real-time prediction process during the real-time prediction of the health probability of the network element by the network element health assessment model; updating the target features based on the contribution of each target feature; and predicting the health probability of the network element according to the updated target features.

[0124] During the network element health assessment process, if there is a feature that contributes more to the target features than the existing target features, it can be added to the target features; or, if the contribution of the existing target features is too low, it can be removed from the target features.

[0125] Correspondingly, when the target characteristics change, the network element health assessment model is also updated accordingly to make it more suitable for the current network element assessment application environment.

[0126] In addition, the data used by the model can be optimized, such as increasing the amount of training data by accumulating historical running data, cleaning the data, and adding abnormal network element samples and personalized data standards for specific scenarios through expert annotation.

[0127] Based on the above approach, the data and models can be further optimized to improve the accuracy of network element health assessment methods.

[0128] Figure 3 This is a schematic diagram of the structure of the network element health assessment device provided in this embodiment. The network element health assessment device can be a terminal device with data processing function as described in the above embodiment, or the network element health assessment device can be a component or assembly within the terminal device. The network element health assessment device provided in this embodiment can execute the processing flow provided in the network element health assessment method embodiment, such as... Figure 3As shown, the network element health assessment device 30 includes: a first acquisition module 31, a first prediction module 32, a calculation module 33, a determination module 34, a second acquisition module 35, and a second prediction module 36. The first acquisition module 31 acquires a network element instance sample dataset based on the historical operating data of the network element. The first prediction module 32 inputs each network element instance sample in the dataset into a benchmark prediction model to predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features. The calculation module 33 calculates the contribution of each sample feature in the prediction process. The determination module 34 determines multiple target features based on the contribution of each sample feature. The second acquisition module 35 extracts the target features based on the actual operating data of the network element to obtain a network element target feature dataset. The second prediction module 36 inputs the target feature dataset into a trained network element health assessment model to predict the health probability of the network element in real time, obtaining the health assessment result of the network element.

[0129] Optionally, the first acquisition module 31 includes a first acquisition unit 311 and a sample generation unit 312. The first acquisition unit 311 is used to obtain an initial sample dataset based on historical operating data. The initial sample dataset includes normal network element samples and abnormal network element samples, wherein the number of abnormal network element samples is less than the number of normal network element samples. The sample generation unit 312 is used to generate at least one synthetic network element sample for each abnormal network element sample based on multiple adjacent samples of the abnormal network element sample. The similarity between the synthetic network element sample and the abnormal network element sample is greater than a first preset threshold. The normal network element samples, abnormal network element samples, and synthetic network element samples are used as the network element instance sample dataset. The difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold.

[0130] Optionally, the sample generation unit 312 is specifically used to calculate the sample distance between the abnormal network element sample and each other abnormal network element sample; determine the preset number of samples with the shortest sample distance as the neighboring samples; for each neighboring sample, calculate the feature distance between the neighboring sample and each feature of the abnormal network element sample; multiply each feature distance by a preset random factor to obtain multiple synthetic features; and construct the synthetic network element sample from the multiple synthetic features.

[0131] Optionally, the network element health assessment device 30 further includes a first training module 37, used to train a model based on the network element instance sample dataset to obtain a benchmark prediction model.

[0132] Optionally, the prediction process includes a single prediction for each network element instance sample. The calculation module 33 is specifically used to calculate the marginal contribution of each sample feature in each single prediction for each sample feature; and to calculate the cumulative value or average value of the marginal contribution in each single prediction to obtain the contribution of the sample feature in the prediction process.

[0133] Optionally, the network element health assessment device 30 further includes a second training module 38, used to extract the target features based on the historical operating data to obtain the network element feature sample dataset; and to update and train the benchmark prediction model according to the network element feature sample dataset to obtain a trained network element health assessment model.

[0134] Optionally, the second prediction module 36 is further configured to calculate the contribution of each target feature in the real-time prediction process during the real-time prediction of the health probability of the network element by the network element health assessment model; update the target features based on the contribution of each target feature; and predict the health probability of the network element based on the updated target features.

[0135] Figure 3 The network element health assessment device shown in the embodiment can be used to execute the technical solution of the above method embodiment. Its implementation principle and technical effect are similar, and will not be described again here.

[0136] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. The electronic device can be a terminal device with data processing capabilities as described in the above embodiments. The electronic device provided in this disclosure can execute the processing flow provided in the network element health assessment method embodiment, such as… Figure 4 As shown, the electronic device 40 includes: a memory 41, a processor 42, a computer program, and a communication interface 43; wherein the computer program is stored in the memory 41 and is configured to be executed by the processor 42 as described above in the network element health assessment method.

[0137] In addition, this disclosure also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the network element health assessment method described in the above embodiments.

[0138] Furthermore, this disclosure also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the network element health assessment method as described above.

[0139] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0140] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for assessing the health of network elements, characterized in that, The method includes: Obtain a sample dataset of network element instances based on the historical operational data of network elements; For each network element instance sample in the network element instance sample dataset, the network element instance sample is input into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features. Calculate the contribution of each of the sample features in the prediction process; Based on the contribution of each of the sample features, multiple target features are determined; The target features are extracted based on the actual operating data of the network elements to obtain the network element target feature dataset; The target feature dataset is input into the trained network element health assessment model so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

2. The method according to claim 1, characterized in that, The method for obtaining a network element instance sample dataset based on historical operational data of network elements includes: An initial sample dataset is obtained based on historical operational data. The initial sample dataset includes normal network element samples and abnormal network element samples, with the number of abnormal network element samples being less than the number of normal network element samples. For each of the abnormal network element samples, at least one synthetic network element sample is generated based on multiple neighboring samples of the abnormal network element sample, and the similarity between the synthetic network element sample and the abnormal network element sample is greater than a first preset threshold. The normal network element samples, abnormal network element samples, and synthetic network element samples are used as the network element instance sample dataset, and the difference between the number of abnormal network element samples and the number of synthetic network element samples and the number of normal network element samples is less than a second preset threshold.

3. The method according to claim 2, characterized in that, For each of the abnormal network element samples, generating at least one synthetic network element sample based on the adjacent samples of the abnormal network element sample includes: Calculate the sample distance between the abnormal network element sample and every other abnormal network element sample; The preset number of samples with the shortest sample distance are determined as the adjacent samples; For each of the neighboring samples, calculate the feature distance between the neighboring sample and each feature of the abnormal network element sample; Each of the feature distances is multiplied by a preset random factor to obtain multiple synthetic features; The synthetic network element sample is composed of the multiple synthetic features.

4. The method according to claim 1, characterized in that, After obtaining the network element instance sample dataset based on the historical operational data of the network element, the method further includes: The baseline prediction model is obtained by training the model based on the network element instance sample dataset.

5. The method according to claim 1, characterized in that, The prediction process includes a single prediction for each network element instance sample, and calculating the contribution of each sample feature in the prediction process includes: For each of the sample features, calculate the marginal contribution of the sample feature in each of the single predictions; Calculate the cumulative or average value of the marginal contribution in each of the single predictions to obtain the contribution of the sample feature in the prediction process.

6. The method according to claim 1, characterized in that, After determining multiple target features based on the contribution of each of the sample features, the method further includes: Based on the historical operational data, the target features are extracted to obtain a network element feature sample dataset; The baseline prediction model is updated and trained based on the network element feature sample dataset to obtain a trained network element health assessment model.

7. The method according to claim 1, characterized in that, The method further includes: During the real-time prediction of the health probability of the network element by the network element health assessment model, the contribution of each target feature in the real-time prediction process is calculated. The target features are updated based on the contribution of each of the target features; The health probability of the network element is predicted based on the updated target characteristics.

8. A network element health assessment device, characterized in that, include: The first acquisition module is used to acquire a sample dataset of network element instances based on the historical operational data of the network elements. The first prediction module is used to input each network element instance sample in the network element instance sample dataset into the benchmark prediction model so that the benchmark prediction model can predict the health probability of the network element instance sample. The network element instance sample includes multiple sample features. A calculation module is used to calculate the contribution of each of the sample features in the prediction process; A determination module is used to determine multiple target features based on the contribution of each of the sample features; The second acquisition module is used to extract the target features based on the actual operating data of the network element to obtain the network element target feature dataset; The second prediction module is used to input the target feature dataset into the trained network element health assessment model, so that the network element health assessment model can predict the health probability of the network element in real time and obtain the health assessment result of the network element.

9. An electronic device, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

Images