A front-haul network fault prediction method and device, electronic equipment and medium

Abstract

The present disclosure provides a kind of front transmission network fault prediction method, device, electronic equipment and medium, by obtaining historical network data, historical network data includes the historical basic data of front transmission network, alarm data and non-alarm data;Using training data, the initial network fault prediction model is trained, and the target network fault prediction model is obtained, and training data is obtained by extracting features to historical network data;Real-time network data of front transmission network is predicted using target network fault prediction model, and prediction result is obtained.By obtaining the related data of front transmission link, the data is modeled and supervised learning, to learn a lot of normal and non-normal network data to deduce training network model, based on the model trained, front transmission link is monitored, so as to predict fault, reduce the influence of front transmission link alarm on service.

Description

Technical Field

[0001] This disclosure relates to information and communication network management technology, and in particular to a method, apparatus, electronic device and medium for predicting fronthaul network faults. Background Technology

[0002] In terms of communication network alarm and fault handling, after the equipment generates an alarm, it reports to the network management system for analysis and then dispatches a work order to the on-site personnel for handling. This is mainly done through retroactive processing, lacking effective alarm and fault prediction and prevention plans, resulting in a large number of alarms and low work efficiency of on-site personnel in the later stages. Summary of the Invention

[0003] To address the problems existing in related technologies, this disclosure proposes a fronthaul network fault prediction method, comprising: acquiring historical network data; training an initial network fault prediction model using training data to obtain a target network fault prediction model, wherein the training data is obtained by feature extraction from historical network data; and using the target network fault prediction model to predict real-time network data of the fronthaul network to obtain prediction results.

[0004] The method disclosed herein acquires relevant data from the fronthaul link, performs modeling and supervised learning on the data, derives and trains a network model using a large amount of normal and abnormal network data, monitors the fronthaul link based on the trained model, thereby predicting faults and reducing the impact of fronthaul link alarms on services.

[0005] The first aspect of this disclosure proposes a network fault prediction method, which includes: acquiring historical network data, including historical basic data, alarm data and non-alarm data of the fronthaul network; training an initial network fault prediction model using training data to obtain a target network fault prediction model, wherein the training data is obtained by feature extraction from the historical network data; and using the target network fault prediction model to predict real-time network data of the fronthaul network to obtain a prediction result.

[0006] In some embodiments of this disclosure, feature extraction of historical network data includes: constructing a data chain based on historical network data, the data chain including multidimensional data; and using an autoencoder neural network to extract features from the data chain to obtain a feature space, the feature space being used to train an initial network fault prediction model.

[0007] In some embodiments of this disclosure, feature extraction of a data chain using an autoencoder neural network to obtain a feature space includes: performing a first encoding on the multidimensional data to obtain a first feature vector; performing a second encoding on the first feature vector to obtain a second feature vector, until the difference between the decoded data of the second feature vector and the multidimensional data in the same dimension is less than a first threshold, thereby obtaining a feature space, which includes multidimensional data, the first feature vector, and the second feature vector.

[0008] In some embodiments of this disclosure, the real-time network data of the fronthaul network is predicted using a target network fault prediction model to obtain the prediction results. This includes: acquiring real-time network data of the fronthaul network, where the real-time network data includes at least one of optical module information, the association between the remote radio unit (RRU) and the baseband processing unit (BBU), traffic statistics information, and log information; and inputting the real-time network data into the target network fault prediction model to obtain the prediction results of the fronthaul network.

[0009] In some embodiments of this disclosure, training an initial network fault prediction model using training data to obtain a target network fault prediction model includes: classifying the training data according to a leaf growth strategy using the LightGBM algorithm; and training the initial network fault prediction model using the classified training data to obtain the target network fault prediction model.

[0010] In some embodiments of this disclosure, historical basic data includes at least one of the following: optical module information, the association between the radio remote unit (RRU) and the baseband processing unit (BBU) of the fronthaul network, traffic statistics information, and log information.

[0011] A second aspect of this disclosure provides a fronthaul network fault prediction device, comprising: an acquisition module for acquiring historical network data, including historical basic data, alarm data, and non-alarm data of the fronthaul network; a processing module for training an initial network fault prediction model using training data to obtain a target network fault prediction model, wherein the training data is obtained by feature extraction from the historical network data; and a prediction module for using the target network fault prediction model to predict real-time network data of the fronthaul network to obtain a prediction result.

[0012] A third aspect of this disclosure provides an electronic device comprising: one or more processors; a storage device communicatively connected to the one or more processors and storing one or more programs thereon; wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the methods described in the first aspect of this disclosure.

[0013] A fourth aspect of this disclosure provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform the methods described in the first aspect of this disclosure.

[0014] A fifth aspect of this disclosure provides a computer program product including a computer program that, when executed, implements the methods described in the first aspect of this disclosure.

[0015] A sixth aspect of this disclosure provides a chip including one or more interface circuits and one or more processors; the interface circuits are configured to receive signals from the memory of an electronic device and send signals to the processors, the signals including computer instructions stored in the memory, which, when executed by the processors, cause the electronic device to perform the methods described in the first aspect of this disclosure.

[0016] In summary, the fronthaul network fault prediction method proposed in this disclosure involves acquiring historical network data; training an initial network fault prediction model using training data (obtained by feature extraction from historical network data); and using the target network fault prediction model to predict real-time network data of the fronthaul network to obtain prediction results. By collecting and analyzing relevant network information of the fronthaul link, extracting relevant feature information to train the model, and using the trained model to predict potential problems in the fronthaul link, network faults can be identified in advance, reducing the impact of fronthaul link alarms on services.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0019] Figure 1 This is a flowchart of a fronthaul network fault prediction method proposed in an embodiment of this disclosure;

[0020] Figure 2 This is a schematic diagram of the process for feature extraction from historical network data according to an embodiment of this disclosure;

[0021] Figure 3 This is a schematic diagram of the feature extraction process proposed in the embodiments of this disclosure;

[0022] Figure 4 This is a schematic diagram of the process for predicting real-time network data of the fronthaul network according to an embodiment of this disclosure.

[0023] Figure 5 This is a schematic diagram of the process for training an initial network fault prediction model according to an embodiment of this disclosure;

[0024] Figure 6A This is a schematic diagram illustrating the construction of a data chain according to an embodiment of this disclosure;

[0025] Figure 6B This is a schematic diagram of the autoencoder neural network proposed in the embodiments of this disclosure;

[0026] Figure 6C This is a schematic diagram illustrating the model training proposed in an embodiment of this disclosure;

[0027] Figure 7 This is a schematic diagram of the structure of a fronthaul network fault prediction device provided in an embodiment of the present disclosure;

[0028] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0029] The embodiments of this disclosure are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0030] The fronthaul link refers to the link between the baseband processing unit (BBU) and the remote radio unit (RRU), typically connected via optical modules and optical fibers. In remote scenarios, fiber optic cables may also be used between these links. During long-term network operation, including exposure to environmental corrosion, optical modules and optical fibers may exhibit unhealthy conditions. When these unhealthy conditions accumulate over time, equipment malfunctions can occur, impacting services. To proactively identify potential problems and reduce the impact of subsequent fronthaul link alarms on services, this disclosure proposes a fronthaul network fault prediction method. This method constructs a feature model of fronthaul link alarms before they occur, identifies potential problems, and addresses them in advance, reducing the probability of fronthaul alarms and improving network availability.

[0031] First, let's introduce the technical terms used in this publication:

[0032] Autoencoder neural networks (ANNs) are neural networks consisting of an autoencoder and a decoder. An autoencoder is an unsupervised learning neural network model that achieves data compression and decompression by encoding input data into a latent space representation and then reconstructing it back into the original input using a decoder. The autoencoder learns data representations to encode and construct a feature space, minimizing the error between the output and the input when the decoder reconstructs the feature space.

[0033] LightGBM algorithm: A distributed gradient boosting framework that can run on different systems. It uses gradient boosting to build a decision tree on the model to solve classification and regression problems.

[0034] The fronthaul network fault prediction method provided in this application will be described in detail below with reference to the accompanying drawings.

[0035] Figure 1 This is a flowchart illustrating a fronthaul network fault prediction method proposed in an embodiment of this disclosure. Figure 1 As shown, the network fault prediction method includes steps 101-104.

[0036] Step 101: Obtain historical network data.

[0037] In embodiments of this disclosure, historical network data includes historical basic data of the fronthaul network, alarm data, and non-alarm data.

[0038] In the embodiments disclosed herein, historical basic data includes optical module information, RRU and BBU association, call statistics information, and log information.

[0039] In the embodiments of this disclosure, the optical module information may include information such as the manufacturer, wavelength, received optical power, and temperature threshold of the optical module; alarm information and non-alarm information may include information such as device, link, power, and temperature; call statistics information may include information such as call drop rate, congestion, call completion rate, and availability; and log information may include information such as temperature changes and performance degradation.

[0040] In the embodiments of this disclosure, alarm data and non-alarm data are obtained from the network management system, the association between RRU and BBU is obtained from the resource management system, and traffic statistics information is obtained from the network management system.

[0041] For example, information about optical modules, device relationships, alarms and non-alarms, call statistics and logs in the network is obtained. There is no correlation between all the raw data and alarm data and non-alarm data.

[0042] Step 102: Use the training data to train the initial network fault prediction model to obtain the target network fault prediction model.

[0043] In the embodiments of this disclosure, the training data is obtained by feature extraction from historical network data.

[0044] In the embodiments of this disclosure, feature extraction of historical network data can construct the correlation between historical basic data and alarm data or non-alarm data, thereby obtaining parameter information under alarm state and parameter information under non-alarm state.

[0045] In the embodiments of this disclosure, training an initial network fault prediction model with training data enables the initial network fault prediction model to learn the relationship between specific network data and alarm data or non-alarm data, thereby enabling the obtained target network fault prediction model to determine under what parameters an alarm will occur.

[0046] For example, historical network data is compressed and dimensionality reduced to construct a feature space. Deep learning is then used to learn and analyze the feature space to build a prediction model.

[0047] Step 103: Use the target network fault prediction model to predict the real-time network data of the fronthaul network and obtain the prediction results.

[0048] In the embodiments of this disclosure, a trained network fault prediction model can be used to analyze real-time data of the fronthaul network and predict in advance whether an alarm fault will occur.

[0049] For example, the prediction model is applied to live network data. Based on the constructed prediction model, the network data is analyzed periodically to identify data behaviors that meet the prediction and generate prediction results.

[0050] In summary, the fronthaul network fault prediction method proposed in this disclosure involves acquiring historical network data; training an initial network fault prediction model using training data (obtained by feature extraction from historical network data); and using the target network fault prediction model to predict real-time network data of the fronthaul network to obtain prediction results. By collecting and analyzing relevant network information of the fronthaul link, extracting relevant feature information to train the model, and using the trained model to predict potential problems in the fronthaul link, network faults can be identified in advance, reducing the impact of fronthaul link alarms on services.

[0051] based on Figure 1 The embodiment shown, Figure 2 The flowchart illustrating the feature extraction process for historical network data presented in this disclosure is further illustrated. Figure 2 based on Figure 1 The illustrated embodiment further describes step 102, as follows: Figure 2 As shown, it includes the following steps:

[0052] Step 201: Construct a data chain based on historical network data.

[0053] In embodiments of this disclosure, the data chain includes multidimensional data. This multidimensional data includes historical basic data of the fronthaul network, alarm data, and non-alarm data.

[0054] In the embodiments of this disclosure, constructing a data chain can be achieved by forming a data chain from optical module information, RRU and BBU association, alarm information, non-alarm information, call statistics information, and log information, which will be used for subsequent feature extraction.

[0055] Step 202: Use an autoencoder neural network to extract features from the data chain to obtain the feature space.

[0056] In embodiments of this disclosure, the feature space is used to train an initial network fault prediction model.

[0057] In the embodiments of this disclosure, an autoencoder neural network can be used to encode and decode data chains, minimizing the error between the decoded output data and the input data chain, thereby achieving data compression and dimensionality reduction.

[0058] In the above embodiments, by forming a data chain from historical network data and extracting features through an autoencoder neural network, a feature space based on historical network data can be constructed, thereby achieving data compression and dimensionality reduction. Using the feature space for subsequent model training enables the model to learn multiple layers of features in the feature space during the training process, resulting in better model training performance.

[0059] based on Figure 1 , Figure 2 The embodiment shown, Figure 3 A schematic diagram of the feature extraction process proposed in this disclosure is further shown. Figure 3 based on Figure 2 The illustrated embodiment further defines step 202. For example... Figure 3 As shown, the method includes the following steps:

[0060] Step 301: Perform a first encoding on the multidimensional data to obtain a first feature vector.

[0061] In the embodiments of this disclosure, the first encoding of multidimensional data may be the first encoding of a data chain constructed based on historical network data to obtain an encoded feature vector.

[0062] For example, the input multidimensional data is The first feature vector output after encoding is Where a (2) =θ (1) X, θ (1) This represents the weights when mapping from layer 1 to layer 2.

[0063] Step 302: Perform a second encoding on the first feature vector to obtain a second feature vector, until the difference between the decoded data of the second feature vector and the multidimensional data in the same dimension is less than a first threshold, thus obtaining the feature space.

[0064] In embodiments of this disclosure, the feature space includes multidimensional data, a first feature vector, and a second feature vector.

[0065] In the embodiments of this disclosure, the second encoding can be performed multiple times. That is, the first feature vector can be encoded once, twice, or multiple times. Each encoding yields a second feature vector, until the difference between the decoded data of the final second feature vector and the multidimensional data in the same dimension is less than a first threshold. At this point, feature extraction ends, and the feature space is obtained.

[0066] In the embodiments of this disclosure, the first threshold is a preset critical value used to determine whether the autoencoder construction is complete, that is, to determine whether the error between the feature vector of the encoded output and the original input multidimensional data is small enough.

[0067] In the embodiments of this disclosure, the second encoding of the first feature vector can be performed by encoding the feature vector output after the first encoding of the multidimensional data again to obtain a second feature vector. If the difference between the output data after decoding the second feature vector and the input multidimensional data is less than a first threshold, the autoencoder is completed, feature extraction ends, and a feature space is obtained. If the difference is greater than or equal to the first threshold, the second feature vector is encoded again to obtain a third feature vector. The above decoding process is repeated for the third feature vector. If the difference between the output data and the input multidimensional data in the same dimension is less than the first threshold, feature extraction ends, and a feature space is obtained. If the difference is greater than or equal to the first threshold, encoding and decoding continue until the difference is less than the first threshold, feature extraction ends, and a feature space is obtained. The feature space includes the input multidimensional data, the first feature vector, and the feature vectors obtained by continuously encoding the first feature vector.

[0068] For example, the first feature vector is encoded a second time, that is, the first feature vector is encoded a second time. The weights θ when mapping from the second layer to the third layer are added. (2) The second eigenvector can be obtained. Where a (3) =θ (2) θ (1) X, θ (2) This represents the weights when mapping from layer 2 to layer 3, and so on. Multiple encodings are performed to obtain multiple feature vectors until the difference between the decoded feature vector and the input multidimensional data in the same dimension is less than a first threshold. This results in the final feature vector output after multiple encodings. That is, k+2 encodings were performed, and each feature vector obtained by encoding together with the input multidimensional data constitutes the feature space.

[0069] In the above embodiments, by using an autoencoder neural network to encode and decode the constructed data chain, the error between the decoded output data and the input data is less than a preset threshold. The autoencoder is then constructed, and a feature space can be obtained. The feature space realizes data compression and dimensionality reduction, while retaining the important features of the input data, which is convenient for subsequent classification and model training.

[0070] based on Figures 1-3 The embodiment shown, Figure 4 The flowchart of training the initial network fault prediction model proposed in this disclosure is further illustrated. Figure 4 based on Figures 1-3 The illustrated embodiment is for Figure 1 Step 102 in the text will be described in further detail, such as Figure 4 As shown, it includes the following steps:

[0071] Step 401: Use the LightGBM algorithm to classify the training data according to the leaf growth strategy.

[0072] In the embodiments of this disclosure, classifying training data using the LightGBM algorithm with a leaf growth strategy involves selecting the leaf that yields the greatest gain at each split. With the same number of splits, leaf growth can reduce the loss function more significantly. In other words, for each leaf node, all possible feature split points are calculated, and the gain of each split point is evaluated. The feature split point with the largest gain is selected for splitting, and so on, to construct multiple decision trees. When all decision trees are constructed, multiple categories can be obtained.

[0073] In the embodiments of this disclosure, classifying training data can be achieved by learning a tree structure in the feature space according to a leaf growth strategy, thereby obtaining the association between each node in the multidimensional data and each node in the feature vector, and thus constructing the association between historical network data and alarm data or non-alarm data.

[0074] In the embodiments of this disclosure, by using the LightGBM algorithm to classify the training data according to the leaf growth strategy, different parameter categories can be obtained for those that produce alarms and those that do not.

[0075] Step 402: Use the classified training data to train the initial network fault prediction model to obtain the target network fault prediction model.

[0076] In the embodiments of this disclosure, training the initial network fault prediction model with the classified training data can be achieved by using the LightGBM algorithm to analyze the feature space in step 401 to obtain multiple types of data, thereby constructing a prediction model between network data and alarms or non-alarms.

[0077] In the above embodiments, by constructing a prediction model through deep learning feature space, the model can learn the relationship between historical network data and fault alarms or non-alarms, thereby enabling early prediction of faults in the fronthaul network and making the trained model prediction more accurate and efficient.

[0078] based on Figures 1-4 The embodiment shown, Figure 5 A schematic diagram of the process for predicting real-time network data of the fronthaul network as proposed in this disclosure is further shown. Figure 5 based on Figure 1 The illustrated embodiment further defines step 103. For example... Figure 5 As shown, it includes the following steps:

[0079] Step 501: Obtain real-time network data from the fronthaul network.

[0080] In embodiments of this disclosure, the real-time network data of the fronthaul network includes at least one of the following: optical module information, the association between the remote radio unit (RRU) and the baseband processing unit (BBU), traffic statistics information, and log information.

[0081] In the embodiments of this disclosure, the real-time network data of the fronthaul network can be acquired periodically, at a custom acquisition interval, or only during a period when the failure times of the historical fronthaul network are relatively concentrated.

[0082] Step 502: Input the real-time network data into the target network fault prediction model to obtain the prediction results of the fronthaul network.

[0083] In the embodiments of this disclosure, real-time network data is input into the target network fault prediction model. This can be achieved by using the target network fault prediction model to analyze the real-time network data. When the parameters in the real-time network data are identified as meeting the parameters that trigger an alarm, it is determined that a fronthaul network fault may occur. If the parameters do not trigger an alarm, it is determined that a fronthaul network fault will not occur, and the model outputs the prediction result.

[0084] For example, the prediction model is applied to the live network data. Based on the constructed prediction model, the network data is analyzed periodically to identify data behaviors that meet the prediction and generate prediction results. The prediction results can be binary classification one-hot encoding with or without alarms.

[0085] In the above embodiments, by using a trained network fault prediction model and applying it to real network scenarios, fronthaul link faults can be predicted in advance, changing the traditional post-event discovery and handling method, reducing the impact on services, realizing the early prediction of potential problems in the fronthaul network, reducing the impact on services after alarm faults occur, and improving network availability by identifying potential problems and handling them in advance.

[0086] In summary, the method proposed in this disclosure can detect potential problems in the fronthaul network in advance, identify and address corresponding problems and risks ahead of time, reduce the probability of fronthaul alarms and the impact of subsequent fronthaul link alarms on services, and improve network availability.

[0087] The following describes the specific implementation of the fronthaul network fault prediction method provided in this disclosure:

[0088] Step 1: Obtain network information and build a data chain.

[0089] Obtain information from the network regarding optical modules, device relationships, alarms and non-alarms, call statistics, and logs, and construct a system such as Figure 6A The data chain shown.

[0090] The optical module information includes the manufacturer, wavelength, received optical power, temperature threshold, etc.; alarm information and non-alarm information include equipment, link, power, temperature, etc.; call statistics information includes call drop rate, congestion, call completion rate, availability, etc.; and log information includes temperature changes, performance degradation, etc.

[0091] Step 2: Compress and reduce the dimensionality of the data to construct the feature space.

[0092] The data in Step 1 is characterized by multiple time granularities, multiple dimensions, and large data volume. The autoencoder neural network algorithm is used to classify and learn the data chain in Step 1, identify and generate a feature space.

[0093] like Figure 6B The diagram shown is of an autoencoder neural network, X1, X2, X3...X n This represents the multidimensional data in the data chain of step 1; a1, a2, a3...a n This represents the hidden layer encoded features of the encoder output. This represents the data reconstructed by the decoder. When X and When the error between them is minimized, the autoencoder is completed, that is, the data chain in step 1 is reduced in dimension and compressed through encoding and then used as the input for step 3.

[0094] X1, X2, X3...X n Indicates input data, θ(j) a1 represents the weight when the j-th layer is mapped to the (j+1)-th layer. (j) Represents the i-th activation unit in the j-th layer. The autoencoder is constructed by encoding the output of the first encoding, followed by multiple encodings until the error between the decoded output vector and the input X is less than a preset value. Then, the output is calculated using a vectorization method.

[0095] Step 3: Build a prediction model using deep learning.

[0096] Deep learning is used to learn and analyze the feature space in step 2, and a detection model between the input data (feature space) and the alarm faults in the fronthaul network is constructed using a neural network. The LightGBM algorithm is employed, such as... Figure 6C The diagram shown illustrates the model training process, where the tree structure is learned according to the leaf growth strategy, data analysis is completed, and a predictive model is constructed.

[0097] The specific rule is: at each split, select the leaf that brings the greatest gain for splitting. Under the same number of splits, growing by leaf can reduce the loss function more.

[0098] Step 4: Generate prediction results.

[0099] The aforementioned feature model is applied to the network, and the network data is analyzed periodically based on the constructed prediction model to identify data behaviors that meet the prediction requirements and generate prediction results. That is, the input is network data and dimensionality-reduced feature data, and the output is a binary one-hot encoding indicating whether there are alarms.

[0100] based on Figures 1-5 The illustrated embodiments, such as Figure 7 As shown, this disclosure provides a schematic diagram of the structure of a fronthaul network fault prediction device 700. The device 700 includes an acquisition module 710, a processing module 720, and a prediction module 730.

[0101] The acquisition module 710 is used to acquire historical network data, which includes historical basic data, alarm data and non-alarm data of the fronthaul network.

[0102] The processing module 720 is used to train the initial network fault prediction model using training data to obtain the target network fault prediction model. The training data is obtained by feature extraction from historical network data.

[0103] The prediction module 730 is used to predict the real-time network data of the fronthaul network using the target network fault prediction model, and obtain the prediction results.

[0104] The processing module is also used to construct data chains based on historical network data, which include multi-dimensional data; and to extract features from the data chains using an autoencoder neural network to obtain a feature space, which is used to train an initial network fault prediction model.

[0105] The processing module is also used to perform a first encoding on the multidimensional data to obtain a first feature vector; perform a second encoding on the first feature vector to obtain a second feature vector, until the difference between the decoded data of the second feature vector and the multidimensional data in the same dimension is less than a first threshold, to obtain a feature space, which includes the multidimensional data, the first feature vector, and the second feature vector.

[0106] The prediction module is also used to acquire real-time network data of the fronthaul network. The real-time network data includes at least one of the following: optical module information, the relationship between the remote radio unit (RRU) and the baseband processing unit (BBU), traffic statistics information, and log information. The real-time network data is input into the target network fault prediction model to obtain the prediction results of the fronthaul network.

[0107] The processing module is also used to classify the training data according to the leaf growth strategy using the LightGBM algorithm; and to train the initial network fault prediction model using the classified training data to obtain the target network fault prediction model.

[0108] In embodiments of this disclosure, historical basic data includes at least one of the following: optical module information, the association between the remote radio unit (RRU) and the baseband processing unit (BBU) of the fronthaul network, traffic statistics information, and log information.

[0109] In summary, the fronthaul network fault prediction device employs an acquisition module to obtain historical network data, including basic historical data, alarm data, and non-alarm data of the fronthaul network. A processing module uses training data to train an initial network fault prediction model, obtaining a target network fault prediction model. The training data is obtained by feature extraction from historical network data. A prediction module uses the target network fault prediction model to predict real-time network data of the fronthaul network, yielding prediction results. By collecting and analyzing relevant network information of the fronthaul link, extracting relevant feature information to train the model, and using the trained model to predict potential problems in the fronthaul link, network faults can be identified in advance, reducing the impact of fronthaul link alarms on services.

[0110] The methods and systems provided in the embodiments of this application have been described above. To implement the functions of the methods provided in the embodiments of this application, the electronic device may include hardware structures and software modules, and may implement the above functions in the form of hardware structures, software modules, or a combination of hardware structures and software modules. One of the above functions may be executed in the form of hardware structures, software modules, or a combination of hardware structures and software modules.

[0111] Figure 8 This is a block diagram illustrating an electronic device 800 for implementing the above-described fronthaul network fault prediction method according to an exemplary embodiment.

[0112] Reference Figure 8 The electronic device 800 may include a communication interface 801, capable of interacting with other devices; a processor 802, connected to the communication interface 801 to enable interaction with other devices, used to execute the methods provided by one or more of the above-described technical solutions when running a computer program; and a memory 803, on which the computer program is stored. Specifically, the specific processing procedure of the processor 802 can refer to the fronthaul network fault prediction method described in the above embodiments of this disclosure.

[0113] Of course, in practical applications, the various components in electronic device 800 are coupled together through bus system 804. It can be understood that bus system 804 is used to realize the connection and communication between these components. In addition to a data bus, bus system 804 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, in... Figure 8 The general labeled all buses as Bus System 804.

[0114] The memory 803 in this embodiment is used to store various types of data to support the operation of the electronic device 800. Examples of such data include any computer program used to operate on the electronic device 800.

[0115] The methods disclosed in the embodiments of this application can be applied to processor 802, or implemented by processor 802. Processor 802 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 802 or by instructions in the form of software. The processor 802 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 802 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in memory 803. Processor 802 reads the information in memory 803 and combines its hardware to complete the steps of the aforementioned method.

[0116] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0117] Embodiments of this disclosure also propose a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the fronthaul network fault prediction method described in the above embodiments of this disclosure.

[0118] Embodiments of this disclosure also propose a computer program product, including a computer program that is executed by a processor using the fronthaul network fault prediction method described in the above embodiments of this disclosure.

[0119] Embodiments of this disclosure also propose a chip including one or more interface circuits and one or more processors; the interface circuits are used to receive signals from the memory of an electronic device and send signals to the processors, the signals including computer instructions stored in the memory, which, when executed by the processor, cause the electronic device to perform the fronthaul network fault prediction method described in the above embodiments of this disclosure.

[0120] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this disclosure 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 disclosure described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0121] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0122] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0123] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processing module, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (control method), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic device, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0124] It should be understood that various parts of the embodiments of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0125] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0126] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc.

[0127] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting faults in a fronthaul network, characterized in that, The method includes: Acquire historical network data, which includes historical basic data, alarm data and non-alarm data of the fronthaul network; The initial network fault prediction model is trained using training data to obtain the target network fault prediction model. The training data is obtained by feature extraction from the historical network data. The target network fault prediction model is used to predict the real-time network data of the fronthaul network to obtain the prediction results. The feature extraction of the historical network data includes: Based on the historical network data, a data chain is constructed, which includes multi-dimensional data; The data chain is used to extract features using an autoencoder neural network to obtain a feature space, which is used to train an initial network fault prediction model. The step of training the initial network fault prediction model using training data to obtain the target network fault prediction model includes: The LightGBM algorithm is used to classify the training data according to the leaf growth strategy. The classification of the training data is to learn the tree structure of the feature space according to the leaf growth strategy, obtain the relationship between each node in the multidimensional data and each node in the feature vector, and construct the relationship between historical network data and alarm data or non-alarm data. The initial network fault prediction model is trained using the classified training data to obtain the target network fault prediction model.

2. The method according to claim 1, characterized in that, The feature space obtained by using an autoencoder neural network to extract features from the data chain includes: The multidimensional data is first encoded to obtain a first feature vector; The first feature vector is encoded in a second way to obtain a second feature vector. The process continues until the difference between the decoded data of the second feature vector and the multidimensional data in the same dimension is less than a first threshold, thereby obtaining the feature space. The feature space includes the multidimensional data, the first feature vector, and the second feature vector.

3. The method according to claim 1, characterized in that, The step of using the target network fault prediction model to predict the real-time network data of the fronthaul network, and obtaining the prediction results, includes: The real-time network data of the fronthaul network is obtained, and the real-time network data includes at least one of the following: optical module information, the association between the remote radio unit (RRU) and the baseband processing unit (BBU), traffic statistics information, and log information. The real-time network data is input into the target network fault prediction model to obtain the prediction result of the fronthaul network.

4. The method according to claim 1, characterized in that, The historical basic data includes at least one of the following: optical module information, the association between the remote radio unit (RRU) and the baseband processing unit (BBU) of the fronthaul network, traffic statistics information, and log information.

5. A fronthaul network fault prediction device, characterized in that, The device includes an acquisition module, a processing module, and a prediction module. The acquisition module is used to acquire historical network data, which includes historical basic data, alarm data and non-alarm data of the fronthaul network. The processing module is used to train the initial network fault prediction model using training data to obtain the target network fault prediction model. The training data is obtained by feature extraction from the historical network data. The prediction module is used to predict the real-time network data of the fronthaul network using the target network fault prediction model to obtain the prediction result. The processing module is also used for: Based on the historical network data, a data chain is constructed, which includes multi-dimensional data; The data chain is used to extract features using an autoencoder neural network to obtain a feature space, which is used to train an initial network fault prediction model. The LightGBM algorithm is used to classify the training data according to the leaf growth strategy. The classification of the training data is to learn the tree structure of the feature space according to the leaf growth strategy, obtain the relationship between each node in the multidimensional data and each node in the feature vector, and construct the relationship between historical network data and alarm data or non-alarm data. The initial network fault prediction model is trained using the classified training data to obtain the target network fault prediction model.

6. An electronic device, characterized in that, include: One or more processors; A storage device communicatively connected to the one or more processors, wherein one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any one of claims 1-4.

7. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method as described in any one of claims 1-4.

8. A computer program product, characterized in that, Includes a computer program that, when executed, implements the method as described in any one of claims 1-4.

Images