Network element data acquisition method and device

By fusing discrete and continuous parameter features into a predictive model to train a network data acquisition strategy, the problem of low matching degree of network data acquisition strategies in existing technologies is solved, and a higher acquisition success rate and data availability are achieved.

CN121691085APending Publication Date: 2026-03-17CHINA TELECOM CORP LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511950494.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-22
Publication Date
2026-03-17

AI Technical Summary

Technical Problem

Existing network data collection strategies cannot be applied to all network elements, leading to collection failures or data unavailability. Furthermore, they require extensive manual adjustments, resulting in low collection success rates and low data availability.

Method used

A predictive model is used to analyze network element information. By fusing discrete parameter features and continuous parameter features, a data acquisition strategy is obtained, including a vendor strategy inheritance relationship graph and temporal dependency relationship. Feature encoding and fusion are performed by combining graph neural networks and long short-term memory models to optimize the acquisition strategy.

Benefits of technology

It improved the prediction accuracy and network element matching of the data acquisition strategy, reduced the need for manual adjustments, and increased the success rate of the first data acquisition and data availability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121691085A_ABST
    Figure CN121691085A_ABST
Patent Text Reader

Abstract

The invention discloses a network element data acquisition method and device. The method comprises the following steps: receiving network element information of target network element equipment; a prediction model is adopted to analyze network element information of the target network element equipment to obtain a data acquisition strategy of the target network element equipment, the prediction model is obtained based on fusion feature training, fusion features are determined according to discrete parameter features and continuous parameter features, and the discrete parameter features comprise a manufacturer strategy inheritance relation graph, a manufacturer strategy inheritance relation graph and a manufacturer strategy inheritance relation graph; the manufacturer strategy inheritance relation graph is used for representing an acquisition strategy inheritance relation between network element devices of different manufacturers, and the continuous parameter characteristics are used for representing a time sequence dependency relation between parameters; and collecting data of the target network element equipment according to the data collection strategy of the target network element equipment. The method provided by the invention at least solves the technical problem of low acquisition success rate caused by low matching degree between the network element data acquisition strategy and the network element in the related technology.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a method and apparatus for collecting network data. Background Technology

[0002] Currently, in data acquisition and control platforms of related technologies, the configuration method for acquisition strategies during data acquisition of network elements usually reuses existing acquisition strategies within the platform. However, actual acquisition strategies are often influenced by multiple parameters, such as network element type, network type, region, manufacturer, and network element device type. Each parameter contains numerous subcategories. Network types include 3G wireless networks, 4G core networks, and 5G core networks, while network element types include CSCF (Call Session Control Function), DCGW (Diameter Connection Gateway), FIREWALL, and HNRF (Home NodeB Radio Resource Function), and so on. The same applies to other parameters. At present, the acquisition strategies of the acquisition and control platform cannot be applied to all network elements, sometimes leading to acquisition failures or unusable acquired data (e.g., empty acquisition files, garbled data, empty values ​​for required indicators). Technicians still need to manually adjust the acquisition strategies based on network element information and the results of the acquired data to adapt to various network element devices.

[0003] Currently, operators have a large number of network elements, and the data collection tasks are constantly increasing. Existing methods for matching collection strategies have low success rates and low data availability for network data collection, and they also consume a lot of manpower. Summary of the Invention

[0004] This application provides a network data acquisition method and apparatus to at least solve the technical problem of low acquisition success rate caused by the low matching degree between network data acquisition strategies and network elements in related technologies.

[0005] According to one aspect of the embodiments of this application, a method for collecting network element data is provided, comprising: receiving network element information of a target network element device; analyzing the network element information of the target network element device using a prediction model to obtain a data collection strategy for the target network element device, wherein the prediction model is trained based on fusion features, the fusion features are determined according to discrete parameter features and continuous parameter features, the discrete parameter features include: a vendor strategy inheritance relationship graph, the vendor strategy inheritance relationship graph is used to represent the inheritance relationship of collection strategies between network element devices of different vendors, the continuous parameter features are used to represent the temporal dependency relationship between parameters; and collecting data from the target network element device according to the data collection strategy of the target network element device.

[0006] Optionally, the method further includes: encoding discrete parameters in the training dataset using a discrete parameter encoding model in the training service model to obtain discrete parameter features, wherein the discrete parameters include at least one of the following: network element type, network type, network element manufacturer, network element location, and network element model, and the discrete parameter encoding model includes a graph neural network; encoding continuous parameters in the training dataset using a continuous parameter encoding model in the training service model to obtain continuous parameter features, wherein the continuous parameters include at least one of the following: acquisition period and acquisition timeout, and the continuous parameter encoding model includes a long short-term memory model; fusing the discrete parameter features and the continuous parameter features using a fusion layer in the training service model to obtain the fused features; and obtaining the output result of the training service model based on the fused features, wherein the output result of the training service model includes at least one of the following: the prediction model, the data validity model, and the adversarial verification model.

[0007] Optionally, a predictive model is used to analyze the network element information of the target network element device to obtain a data acquisition strategy for the target network element device. This includes: selecting multiple initial acquisition strategies with the highest weights from multiple candidate acquisition strategies in the analysis results of the predictive model; sequentially using the multiple initial acquisition strategies to acquire data from the target network element device to obtain multiple acquisition results; using the data validity model to perform validity analysis on the multiple acquisition results to obtain the data validity probability of each acquisition result; and determining the initial acquisition strategy corresponding to the acquisition result with the highest data validity probability and valid data as the data acquisition strategy for the target network element device.

[0008] Optionally, the method further includes: determining that the data of the collection result is valid when the data validity probability is higher than a preset probability; and performing incremental learning on the training service model and selecting the data collection strategy of the target network element device from the alternative strategies when the data of multiple collection results are invalid.

[0009] Optionally, the method further includes: when the target network element device is a newly joined network element device, using the target network element device's data acquisition strategy to collect data from the target network element device, and obtaining the collected data; using the adversarial verification model to verify the distribution of the collected data, and if the distribution of the collected data conforms to a preset distribution, determining that the data acquisition strategy of the target network element device has passed the verification.

[0010] Optionally, the method further includes: if the target network element is not a newly joined network element, outputting the data acquisition strategy of the target network element.

[0011] Optionally, the prediction model is determined by: obtaining a training dataset, wherein the training dataset includes: network element information, acquisition strategy information, and historical acquisition data; the network element information includes at least: network element type, network type, network element manufacturer, network element location, and network element model; the acquisition strategy information includes at least: acquisition service information, acquisition protocol, acquisition instructions, acquisition cycle, acquisition source classification strategy, distribution strategy, blocking handling strategy, acquisition timeout time, and number of failure retries; and the historical acquisition data includes at least: valid data, invalid data, and acquisition strategy change records; and training a pre-built training service model using the training dataset to obtain the prediction model.

[0012] According to another aspect of the embodiments of this application, a network data acquisition device is also provided, comprising: a receiving module for receiving network element information of a target network element device; an analysis module for analyzing the network element information of the target network element device using a prediction model to obtain a data acquisition strategy for the target network element device, wherein the prediction model is trained based on fusion features, the fusion features are determined according to discrete parameter features and continuous parameter features, the discrete parameter features include: a vendor strategy inheritance relationship graph, the vendor strategy inheritance relationship graph is used to represent the inheritance relationship of acquisition strategies between network element devices of different vendors, and the continuous parameter features are used to represent the temporal dependency relationship between parameters; and an acquisition module for acquiring data of the target network element device according to the data acquisition strategy of the target network element device.

[0013] According to another aspect of the embodiments of this application, a computer device is also provided, including: a memory and a processor, wherein the memory is used to store program instructions; and the processor, connected to the memory, is used to execute the above-described network data acquisition method.

[0014] According to another aspect of the embodiments of this application, a computer program product is also provided, including computer instructions that, when executed by a processor, implement the above-described network data acquisition method.

[0015] In this embodiment, network element information of a target network element device is received; a prediction model is used to analyze the network element information of the target network element device to obtain the data acquisition strategy of the target network element device. The prediction model is trained based on fused features, which are determined according to discrete parameter features and continuous parameter features. The discrete parameter features include a vendor strategy inheritance graph, which represents the inheritance relationship of acquisition strategies between network element devices from different vendors. The continuous parameter features represent the temporal dependency relationship between parameters. Data from the target network element device is collected according to its data acquisition strategy. The prediction model is trained using the fused features obtained by fusing discrete and continuous parameter features, thereby improving the prediction accuracy of the acquisition strategy and enhancing the matching degree between the data acquisition strategy and the network element. This solves the technical problem of low acquisition success rate due to low matching degree between network element data acquisition strategies and network elements in related technologies. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0017] Figure 1 This is a hardware structure block diagram of a computer terminal for implementing a network data acquisition method according to an embodiment of this application;

[0018] Figure 2 This is a flowchart of a network data collection method according to an embodiment of this application;

[0019] Figure 3 This is a schematic diagram of the architecture of a training service model according to an embodiment of this application;

[0020] Figure 4 This is a data preparation flowchart according to an embodiment of this application;

[0021] Figure 5 This is a flowchart illustrating a network data collection strategy generation method according to an embodiment of this application;

[0022] Figure 6 This is a schematic diagram of a network data acquisition system architecture according to an embodiment of this application;

[0023] Figure 7 This is a structural diagram of a network data acquisition device according to an embodiment of this application. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] The information collected in this application embodiment is information and data authorized by the user or fully authorized by all parties. The collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data all comply with the relevant laws, regulations and standards of the relevant regions, and necessary confidentiality measures have been taken. It does not violate public order and good morals, and provides corresponding operation entry points for users to choose to authorize or reject the automated decision results. If the user chooses to reject, the process will proceed to the expert decision-making process.

[0027] To better understand the embodiments of this application, the technical terms used in the embodiments of this application are explained as follows:

[0028] Data Acquisition and Control Platform: The full name is "Data Acquisition and Control Platform". It is a unified management and control platform for data acquisition and control of various network elements and devices. It is the only channel that provides real-time data acquisition and operation interaction of cloud network in a service-oriented manner.

[0029] Network element: refers to the basic unit in a 5G wireless access network, mainly responsible for receiving and transmitting wireless signals. Network elements are interconnected with other network elements in the 5G wireless access network, forming the core part of the 5G network.

[0030] To address the problems existing in related technologies, this application provides a method for collecting network data, which can be run on... Figure 1 The computer terminal shown is explained below.

[0031] The network data collection method embodiments provided in this application can be executed on mobile terminals, computer terminals, or similar computing devices. Figure 1 A hardware block diagram of a computer terminal for implementing a network data acquisition method is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission module 106 for communication functions connected via wired and / or wireless networks. In addition, it may also include: a display, a keyboard, a cursor control device, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, and a BUS bus. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0032] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0033] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the network data acquisition method in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned network data acquisition method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0034] The transmission module 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission module 106 may be a radio frequency (RF) module, used for wireless communication with the Internet.

[0035] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0036] It should be noted here that, in some optional embodiments, the above... Figure 1 The computer terminal shown may include hardware elements (including circuitry), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware and software elements. It should be noted that... Figure 1 This is only one instance of a specific particular instance, and is intended to illustrate the types of components that may exist in the aforementioned computer terminal.

[0037] In the above operating environment, this application provides an embodiment of a network data acquisition method. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] Figure 2 This is a flowchart of a network data collection method according to an embodiment of this application, such as... Figure 2As shown, the method includes the following steps:

[0039] Step S202: Receive network element information of the target network element device;

[0040] In step S202, the network element information includes, but is not limited to: basic parameters: network element type, network type, network element equipment manufacturer, location, model, etc.

[0041] Step S204: Analyze the network element information of the target network element device using a prediction model to obtain the data acquisition strategy of the target network element device. The prediction model is trained based on fusion features, which are determined according to discrete parameter features and continuous parameter features. The discrete parameter features include: a vendor strategy inheritance relationship graph, which represents the inheritance relationship of acquisition strategies between network element devices from different vendors. The continuous parameter features represent the temporal dependency relationship between parameters.

[0042] Step S206: Collect data from the target network element device according to the data acquisition strategy of the target network element device.

[0043] Through steps S202 to S206 above, network element information of the target network element device is received; a prediction model is used to analyze the network element information of the target network element device to obtain the data acquisition strategy of the target network element device. The prediction model is trained based on fused features, which are determined according to discrete parameter features and continuous parameter features. The discrete parameter features include a vendor strategy inheritance relationship graph, which represents the inheritance relationship of acquisition strategies between network element devices from different vendors; the continuous parameter features represent the temporal dependency relationship between parameters. Data from the target network element device is collected according to its data acquisition strategy. The prediction model is trained using the fused features obtained by fusing discrete and continuous parameter features, thereby improving the prediction accuracy of the acquisition strategy and enhancing the matching degree between the data acquisition strategy and the network element. This solves the technical problem of low acquisition success rate due to low matching degree between network element data acquisition strategies and network elements in related technologies. A detailed explanation follows.

[0044] In some embodiments of this application, the specific steps for training a pre-built training service model using the training dataset to obtain the prediction model are as follows: Figure 3As shown, the process includes: encoding discrete parameters in the training dataset using a discrete parameter encoding model in the training service model to obtain discrete parameter features, wherein the discrete parameters include at least one of the following: network element type, network type, network element manufacturer, network element location, and network element model; the discrete parameter encoding model includes a graph neural network; encoding continuous parameters in the training dataset using a continuous parameter encoding model in the training service model to obtain continuous parameter features, wherein the continuous parameters include at least one of the following: acquisition period and acquisition timeout; the continuous parameter encoding model includes a long short-term memory model; fusing the discrete parameter features and the continuous parameter features using a fusion layer in the training service model to obtain the fused features; and obtaining the output result of the training service model based on the fused features, wherein the output result of the training service model includes at least one of the following: the prediction model, the data validity model, and the adversarial verification model.

[0045] Specifically, the training service model employs a multimodal deep learning framework. The model architecture includes: Discrete parameter encoding: Graph Neural Networks (GNNs) are used to process discrete parameters such as element type, vendor, and location to construct a vendor strategy inheritance graph. Continuous parameter encoding: LSTM (Long Short-Term Memory) with an attention mechanism is used to process continuous parameters such as acquisition period and timeout to capture temporal dependencies. Fusion layer: The GNN output and LSTM output are fused through a cross-attention mechanism to generate the final prediction.

[0046] The specific fusion process is as follows: A shared attention layer (such as a multi-head self-attention mechanism) is used to calculate the correlation between the GNN output and the LSTM output, obtaining the attention weights. Based on the attention weights, the GNN output and LSTM output vectors are weighted and summed to obtain the fused vector. The fused vector is then input into a prediction layer, such as a fully connected layer, to generate the final prediction result.

[0047] Understandably, the training service model is used to train the prediction model, data validity model, and adversarial verification model, etc. Specifically, the output of the training service model is as follows: Prediction collection strategy model: Input network element information, output Top-N strategies and their weights (e.g., vendor A's 5G core network equipment priority recommendation strategy P1, weight 0.92). Data validity model: Input collected data, output validity probability (0~1). Adversarial verification module: For cold-start network elements, generate synthetic data and judge its rationality through pre-trained GAN.

[0048] The prediction model is determined by obtaining a training dataset, wherein, as shown in the example... Figure 4As shown, the training dataset includes: network element information, policy association relationships, location policy characteristics, collection policy information, and historical collection data. The network element information includes at least: network element type, network type, network element manufacturer, network element location, and network element model. The collection policy information includes at least: collection service information, collection protocol, collection instructions, collection cycle, collection source classification policy, distribution policy, blocking handling policy, collection timeout time, and number of failure retries. The historical collection data includes at least: valid data, invalid data, and collection policy change records. The pre-built training service model is trained using the training dataset to obtain the prediction model.

[0049] It should be noted that policy association represents the policy inheritance pattern between network elements from different vendors (e.g., vendor B's devices typically inherit the collection policies of similar devices from vendor A). Geographical policy characteristics represent the specific policy requirements of different regional network environments. Valid data: Data corresponding to collection results that meet preset quality standards; Invalid data: Data with empty collection files, garbled data, or missing required indicators; Policy change records must include at least: Policy failure events caused by vendor model upgrades.

[0050] In some embodiments of this application, the specific steps for analyzing the network element information of the target network element device using a predictive model to obtain the data acquisition strategy of the target network element device are as follows: Selecting multiple initial acquisition strategies with the highest weights from multiple candidate acquisition strategies in the analysis results of the predictive model; sequentially using the multiple initial acquisition strategies to acquire data from the target network element device, obtaining multiple acquisition results; performing validity analysis on the multiple acquisition results using the data validity model to obtain the data validity probability of each acquisition result; determining the initial acquisition strategy corresponding to the acquisition result with the highest data validity probability and valid data as the data acquisition strategy of the target network element device. If the data validity probability is higher than a preset probability, the data of the acquisition result is determined to be valid; if the data of multiple acquisition results is invalid, incremental learning is performed on the multimodal model, and the data acquisition strategy of the target network element device is selected from the candidate strategies.

[0051] It is understandable that the analysis results output by the prediction model include: multiple candidate acquisition strategies and the weight of each candidate acquisition strategy. The weight of each candidate acquisition strategy is used to represent the probability that the prediction model selects that candidate acquisition strategy as the final acquisition strategy.

[0052] like Figure 5 As shown, this includes: Policy prediction: Input target network element information (network type: 5G; vendor: XX; location: A), model outputs Top-3 policies:

[0053] Strategy 1 (weight 0.89);

[0054] Strategy 2 (weight 0.76);

[0055] Strategy 3 (weight 0.68);

[0056] Strategy Validation: Data was collected using three strategies in sequence, yielding results D1, D2, and D3. D1 to D3 were then input into the data validity model, outputting probabilities: D1 = 0.98 (valid), D2 = 0.45 (invalid), D3 = 0.87 (valid).

[0057] The dynamic optimization process is as follows: Select strategy 1 (highest weight and most effective). If all strategies are ineffective, trigger incremental learning: fine-tune the corresponding vendor branch of the model based on failure data and recommend historically best alternative strategies.

[0058] When the target network element is a newly added network element, its data acquisition strategy is used to collect data. The adversarial verification model is then used to verify the distribution of the collected data. If the data distribution conforms to a preset distribution, the data acquisition strategy of the target network element is deemed to have passed verification. When the target network element is not a newly added network element, its data acquisition strategy is directly output without requiring verification using an adversarial verification model.

[0059] It is understandable that newly added network elements have no historical data.

[0060] To better understand the network data acquisition method proposed in this application, this application also provides a network data acquisition system, such as... Figure 6 As shown, it includes: a data layer, a feature engineering module, a model training module, and a strategy selection module. The data layer includes: a historical data storage module, used to provide network element information, acquisition strategies, and acquisition results; a feature engineering module, used to build a vendor / location relationship model, extract strategy features, and generate validity labels; a model training module, used to perform multimodal fusion (concatenation + attention fusion) on the features generated by feature engineering, and use a prediction model to analyze the fused features to obtain analysis results; and a strategy selection module, used to screen multiple candidate strategies in the analysis results and determine the target acquisition strategy based on the screening results.

[0061] In related technologies, the network data collection process relies on manually copying existing collection strategies and then manually adjusting them based on network element information and collection results to adapt to different network element devices. This results in low collection efficiency, a low initial collection success rate, and high costs associated with manual intervention.

[0062] The network data acquisition method provided in this application uses GNN (Graph Neural Network) to encode discrete parameters of network elements (manufacturer, location, etc.) and mines implicit correlations between strategies (such as the inheritance relationship of network element equipment from different manufacturers). It combines LSTM + attention mechanism to process continuous parameters (acquisition period, timeout) and capture long-term dependencies. This significantly improves strategy prediction accuracy and reduces the need for manual adjustments. For new network elements without historical data, the rationality of the strategy is verified by generating synthetic data and using a GAN discriminator. The initial acquisition success rate of cold-start network elements is significantly improved, avoiding manual trial and error. Real-time monitoring of acquisition failure events triggers local model fine-tuning (updating only the strategy branch of the affected network element type). This significantly improves the strategy update time, reducing it to 10 minutes, adapting to network changes. Through weight ranking and validity verification, the optimal strategy is output. This improves the usability of the acquired data.

[0063] Table 1 shows the comparison results between the network data collection method provided in the embodiments of this application and the traditional collection method.

[0064] Table 1

[0065]

[0066] In practical application scenarios, the network data acquisition method provided in this application includes: extracting relevant information about network elements and acquisition strategies from historical data of the acquisition and control platform, and performing preprocessing. GNN is used to encode discrete parameters, and LSTM is used to encode continuous parameters; the two are then fused using a cross-attention mechanism. The model predicts the acquisition strategy based on the input network element information, and verifies the feasibility of the strategy through a data validity model. The optimal strategy is implemented for data acquisition on network elements, and the acquisition results are fed back to the system for GAN data synthesis and strategy rationality verification in cold-start scenarios, as well as incremental model learning in general scenarios. The system continuously optimizes the model based on the feedback results to improve acquisition efficiency and data quality.

[0067] Table 2 shows the differences between the network data collection method proposed in this application embodiment and the traditional collection method.

[0068] Table 2

[0069]

[0070] Figure 7 A network data acquisition device is shown, which includes:

[0071] Receiver module 70 is used to receive network element information of the target network element device;

[0072] Analysis module 72 is used to analyze the network element information of the target network element device using a prediction model to obtain the data acquisition strategy of the target network element device. The prediction model is trained based on fusion features, which are determined according to discrete parameter features and continuous parameter features. The discrete parameter features include a vendor strategy inheritance relationship graph, which is used to represent the inheritance relationship of acquisition strategies between network element devices from different vendors. The continuous parameter features are used to represent the temporal dependency relationship between parameters.

[0073] The acquisition module 74 is used to acquire data from the target network element device according to the data acquisition strategy of the target network element device.

[0074] The aforementioned network data acquisition device receives network element information from a target network element device; it then analyzes this information using a prediction model to obtain the data acquisition strategy for that target network element device. The prediction model is trained based on fused features, which are determined by discrete and continuous parameter features. The discrete parameter features include a vendor strategy inheritance graph, representing the inheritance relationship of acquisition strategies between network element devices from different vendors, and continuous parameter features, representing the temporal dependency relationship between parameters. The device collects data from the target network element device according to its data acquisition strategy. The prediction model is trained using the fused features obtained by fusing discrete and continuous parameter features, thereby improving the accuracy of the acquisition strategy prediction and enhancing the matching degree between the data acquisition strategy and the network element. This solves the technical problem of low acquisition success rate due to low matching degree between network data acquisition strategies and network elements in related technologies.

[0075] The analysis module 72 includes a training submodule, used to train a pre-built training service model using the training dataset to obtain the prediction model. This includes: encoding discrete parameters in the training dataset using a discrete parameter encoding model within the training service model to obtain discrete parameter features, wherein the discrete parameters include at least one of the following: network element type, network type, network element manufacturer, network element location, and network element model; the discrete parameter encoding model includes a graph neural network; encoding continuous parameters in the training dataset using a continuous parameter encoding model within the training service model to obtain continuous parameter features, wherein the continuous parameters include at least one of the following: acquisition period and acquisition timeout; the continuous parameter encoding model includes a long short-term memory model; fusing the discrete parameter features and the continuous parameter features using a fusion layer within the training service model to obtain the fused features; and obtaining the output result of the training service model based on the fused features, wherein the output result of the training service model includes at least one of the following: the prediction model, a data validity model, and an adversarial verification model.

[0076] The analysis module 72 further includes an analysis submodule, used to analyze the network element information of the target network element device using a prediction model to obtain the data acquisition strategy of the target network element device, including: selecting multiple initial acquisition strategies with the highest weights from multiple candidate acquisition strategies in the analysis results of the prediction model; sequentially using the multiple initial acquisition strategies to acquire data from the target network element device to obtain multiple acquisition results; using the data validity model to perform validity analysis on the multiple acquisition results to obtain the data validity probability of each acquisition result; and determining the initial acquisition strategy corresponding to the acquisition result with the highest data validity probability and valid data as the data acquisition strategy of the target network element device.

[0077] The analysis submodule includes a filtering unit and a verification unit. The filtering unit is used to determine that the data of the collection result is valid when the data validity probability is higher than a preset probability; and to perform incremental learning on the multimodal model and select the data collection strategy of the target network element device from the alternative strategies when the data of multiple collection results are invalid.

[0078] The verification unit is configured to, when the target network element is a newly added network element, collect data from the target network element using its data acquisition strategy, and obtain the collected data; verify the distribution of the collected data using the adversarial verification model; and determine that the data acquisition strategy of the target network element passes verification if the distribution of the collected data conforms to a preset distribution. If the target network element is not a newly added network element, the unit outputs the data acquisition strategy of the target network element.

[0079] The aforementioned network data acquisition device further includes: a determination submodule, used to acquire a training dataset, wherein the training dataset includes: network element information, acquisition strategy information, and historical acquisition data. The network element information includes at least: network element type, network type, network element manufacturer, network element location, and network element model. The acquisition strategy information includes at least: acquisition service information, acquisition protocol, acquisition command, acquisition cycle, acquisition source classification strategy, distribution strategy, blocking handling strategy, acquisition timeout time, and number of failure retries. The historical acquisition data includes at least: valid data, invalid data, and acquisition strategy change records. The pre-built training service model is trained using the training dataset to obtain the prediction model.

[0080] It should be noted that, Figure 7 The network data acquisition device shown is used to perform... Figure 2 The network data acquisition method shown above is also applicable to this network data acquisition device, and will not be repeated here.

[0081] This application also provides a computer device, including: a memory and a processor, wherein the memory is used to store program instructions; and the processor, connected to the memory, is used to execute the above-described network data acquisition method.

[0082] This application also provides a method for collecting network data executed by a computer device. This method involves receiving network element information from a target network element device; analyzing the network element information using a prediction model to obtain the data collection strategy of the target network element device; wherein the prediction model is trained based on fused features, which are determined according to discrete parameter features and continuous parameter features. The discrete parameter features include a vendor strategy inheritance graph, which represents the inheritance relationship of collection strategies between network element devices from different vendors; and the continuous parameter features represent the temporal dependency relationship between parameters. Data from the target network element device is collected according to its data collection strategy. The prediction model is trained using the fused features obtained by fusing discrete and continuous parameter features, thereby improving the accuracy of the collection strategy prediction and enhancing the matching degree between the data collection strategy and the network element. This solves the technical problem of low collection success rate due to low matching degree between network data collection strategies and network elements in related technologies.

[0083] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps of the network data acquisition method in this application.

[0084] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0085] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0086] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0089] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0090] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method of collecting network element data, the method comprising: The method comprises the following steps: receiving network element information of a target network element device; analyzing the network element information of the target network element device by using a prediction model to obtain a data collection strategy of the target network element device, wherein the prediction model is trained based on fusion features, the fusion features are determined according to discrete parameter features and continuous parameter features, the discrete parameter features comprise a vendor strategy inheritance relationship graph, and the vendor strategy inheritance relationship graph is used to represent a collection strategy inheritance relationship between network element devices of different vendors, and the continuous parameter features are used to represent a time sequence dependency relationship between parameters; collecting data of the target network element device according to the data collection strategy of the target network element device.

2. The method of claim 1, wherein, The method further comprises the following steps: encoding discrete parameters in a training data set by using a discrete parameter encoding model in a training service model to obtain discrete parameter features, wherein the discrete parameters comprise at least one of the following: a network element type, a network type, a network element manufacturer, a network element location, and a network element model, and the discrete parameter encoding model comprises a graph neural network; encoding continuous parameters in the training data set by using a continuous parameter encoding model in the training service model to obtain continuous parameter features, wherein the continuous parameters comprise at least one of the following: a collection period and a collection timeout time, and the continuous parameter encoding model comprises a long short-term memory model; fusing the discrete parameter features and the continuous parameter features by using a fusion layer in the training service model to obtain the fusion features; obtaining an output result of the training service model according to the fusion features, wherein the output result of the training service model comprises at least one of the following: the prediction model, a data validity model, and an adversarial verification model.

3. The method of claim 2, wherein, analyzing the network element information of the target network element device by using a prediction model to obtain a data collection strategy of the target network element device, comprising the following steps: selecting a plurality of initial collection strategies with the highest weights from a plurality of candidate collection strategies in an analysis result of the prediction model; collecting data of the target network element device by using the plurality of initial collection strategies in sequence to obtain a plurality of collection results; performing validity analysis on the plurality of collection results by using a data validity model to obtain a data validity probability of each collection result in the plurality of collection results; determining an initial collection strategy corresponding to a collection result with the highest data validity probability and valid data as the data collection strategy of the target network element device.

4. The method of claim 3, wherein, The method further comprises the following steps: determining that the data of the collection result is valid in a case where the data validity probability is higher than a preset probability; in a case where the data of the plurality of collection results is all invalid, performing incremental learning on the training service model and selecting a data collection strategy of the target network element device from alternative strategies.

5. The method of claim 2, wherein, The method further comprises the following steps: in a case where the target network element device is a newly connected network element device, collecting data of the target network element device by using the data collection strategy of the target network element device to obtain collected data; The acquired data distribution is verified by using the confrontation verification model, and in a case where the acquired data distribution conforms to the preset distribution, it is determined that the data collection strategy of the target network element device passes the verification.

6. The method of claim 5, wherein, The method further comprises: In a case where the target network element device is not a newly-accessed network element device, outputting the data collection strategy of the target network element device.

7. The method of claim 1, wherein, The prediction model is determined by the following way: Obtaining a training data set, wherein the training data set comprises network element information, collection strategy information and historical collection data, the network element information at least comprises network element type, network type, network element manufacturer, network element location, network element model, the collection strategy information at least comprises collection service information, collection protocol, collection instruction, collection period, collection source classification strategy, distribution strategy, blocking processing strategy, collection timeout time and failure retry number, and the historical collection data at least comprises valid data, invalid data and collection strategy change record; Training a pre-constructed training service model by using the training data set to obtain the prediction model.

8. A network element data collection apparatus, characterized by Comprise: A receiving module configured to receive network element information of a target network element device; An analysis module configured to analyze the network element information of the target network element device by using a prediction model to obtain a data collection strategy of the target network element device, wherein the prediction model is trained based on fusion features, the fusion features are determined according to discrete parameter features and continuous parameter features, the discrete parameter features comprise a manufacturer strategy inheritance relationship graph, and the manufacturer strategy inheritance relationship graph is used to represent the collection strategy inheritance relationship between network element devices of different manufacturers, and the continuous parameter features are used to represent the time sequence dependency relationship between parameters; A collection module configured to collect data of the target network element device according to the data collection strategy of the target network element device.

9. A computer device, comprising: Comprise: A memory and a processor, wherein the memory is configured to store program instructions; The processor is connected with the memory and is configured to execute the network element data collection method in any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that, The computer instructions are executed by the processor to implement the network element data collection method in any one of claims 1 to 7.