CN113709779B - Cellular network fault diagnosis method - Google Patents

Cellular network fault diagnosis method Download PDF

Info

Publication number
CN113709779B
CN113709779B CN202110862707.3A CN202110862707A CN113709779B CN 113709779 B CN113709779 B CN 113709779B CN 202110862707 A CN202110862707 A CN 202110862707A CN 113709779 B CN113709779 B CN 113709779B
Authority
CN
China
Prior art keywords
matrix
network
data set
characteristic parameters
network fault
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110862707.3A
Other languages
Chinese (zh)
Other versions
CN113709779A (en
Inventor
朱晓荣
吴铭骁
何明坤
肖芳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Posts and Telecommunications
Original Assignee
Nanjing University of Posts and Telecommunications
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing University of Posts and Telecommunications filed Critical Nanjing University of Posts and Telecommunications
Priority to CN202110862707.3A priority Critical patent/CN113709779B/en
Publication of CN113709779A publication Critical patent/CN113709779A/en
Application granted granted Critical
Publication of CN113709779B publication Critical patent/CN113709779B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/04—Arrangements for maintaining operational condition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Test And Diagnosis Of Digital Computers (AREA)

Abstract

The invention discloses a cellular network fault diagnosis method, which comprises the following steps: step 1, determining a network fault data set; step 2, obtaining a network fault data set after dimension reduction; step 3, representing the network fault data set subjected to the dimension reduction in the step 2 in a feature matrix form; the label information of the network fault data set after the dimension reduction in the step 2 is expressed in a label matrix form; converting the introduced weight matrix into an adjacent matrix with matrix elements of 0 and 1 only; and 4, fault diagnosis based on the graph convolution neural network. The novel cellular network fault diagnosis method is used for further researching intelligent fault diagnosis of heterogeneous wireless networks, analyzing similar characteristics among samples by combining a big data processing method, converting the existing network fault parameter data set into graph structure data, extracting features from the graph structure data by using a graph convolution neural network, and accordingly completing classification tasks of sample nodes and predicting fault types of cells.

Description

Cellular network fault diagnosis method
Technical Field
The invention relates to the technical field of communication, in particular to a cellular network fault diagnosis method.
Background
In recent years, heterogeneous cellular networks have become one of the important methods for increasing system capacity in response to the proliferation of mobile data traffic and the different demands of various services. However, as the size and complexity of cellular networks increases, the operation and maintenance tasks for cellular networks become complex and cumbersome. Although the end-to-end user experience is significantly improved in terms of throughput and delay, cellular networks also become more prone to failure. How to predict and locate faults when they occur or are about to occur has become a significant challenge.
In the past, the traditional network fault detection and diagnosis are mainly completed through manual operation, so that the mapping relation between the network symptoms and the fault categories is difficult to obtain accurately, and a large amount of manpower and material resources are consumed.
With the rapid development of artificial intelligence, most of the current intelligent fault diagnosis methods are based on machine learning, and intelligent fault diagnosis has been changed from the traditional model-based fault diagnosis method to a data-driven fault diagnosis method. The network fault diagnosis technology based on machine learning learns the mapping relation between network events by mining information in a large amount of training data, then establishes a fault diagnosis model according to the mapping relation, and finally applies the trained model to newly observed network symptoms so as to realize the prediction of network states. However, existing fault diagnosis solutions mostly use supervised learning methods in machine learning, which mostly rely on massive data sets and require that samples in the data sets have sufficient marker information. However, in practical situations, it is very difficult to obtain the marking information, the marking sample is very time-consuming, the cost of performing manual category marking is too high, the historical data set is too few, and the obtaining cost is also very expensive. Most of the fault diagnosis methods are based on fine adjustment and classification of supervised learning, a large amount of unlabeled sample information cannot be fully utilized, and unlabeled sample data is wasted.
In summary, the problem of how to perform accurate network fault diagnosis under the condition of very little effective marking information in the fault diagnosis process of the 4G/5G heterogeneous wireless network is a technical problem to be solved by those skilled in the art.
Disclosure of Invention
The invention aims to solve the technical problem of overcoming the defects of the prior art and providing a cellular network fault diagnosis method.
The invention adopts the following technical scheme for solving the technical problems:
the invention provides a fault diagnosis method of a cellular network, which comprises the following steps:
step 1, determining a network fault data set; the method comprises the following steps:
step 101, data in a network fault data set comprises original characteristic parameters and label categories, and a non-network fault condition and 5 network faults are defined as the label categories, wherein the 5 network faults are uplink interference, downlink interference, coverage holes, air interface faults and base station faults respectively;
step 102, taking all key performance indexes used in the fault cell identification as original characteristic parameters, wherein the original characteristic parameters comprise 16 characteristic parameters, and the 16 characteristic parameters comprise reference signal receiving power, reference signal receiving quality, uplink packet loss rate, downlink packet loss rate, uplink signal to noise ratio, downlink signal to noise ratio, radio resource control connection establishment success rate, evolution radio access bearer establishment success rate, call drop rate, handover success rate, uplink average throughput, downlink average throughput, node outgoing average throughput, node incoming average throughput, handover delay and link error rate;
step 2, adopting an XGBoost algorithm, and selecting the first n characteristic parameters with highest diagnosis accuracy of the XGBoost algorithm from the original characteristic parameters as the characteristic parameters of data in the network fault data set, so as to obtain a network fault data set with reduced dimension, wherein n is more than 0 and less than 16;
step 3, graph data conversion, specifically comprising the following steps:
step 301, representing the network fault data set subjected to the dimension reduction in step 2 in a form of a feature matrix, wherein each row vector in the feature matrix corresponds to a piece of feature parameter vector except for category information in the network fault data set subjected to the dimension reduction;
step 302, representing the label information of the network fault data set subjected to the dimension reduction in the step 2 in a label matrix form; in the tag matrix, the tag line vector of the marked data is in a form of single-hot coding, and the tag line vector of the unmarked data is a zero vector; the marked data is data with a label, and the unmarked data is data without a label;
step 303, mapping the network fault data set after the dimension reduction into an undirected graph g= (V, E), wherein the undirected graph is composed of two types of elements, namely a node set V and an edge set E; introducing a weight matrix to represent the similarity between nodes in the undirected graph, wherein elements in the weight matrix are the similarity between node pairs, and the similarity between node pairs is obtained by calculating Euclidean distances between every two nodes and normalizing; comparing all elements in the weight matrix with the parameter threshold through the set parameter threshold, setting the element as 1 if the current element is larger than the parameter threshold, otherwise setting the element as 0; thereby converting the weight matrix into an adjacent matrix with matrix elements of only 0 and 1, wherein the adjacent matrix represents the adjacent relation between nodes;
and 4, fault diagnosis based on a graph convolution neural network, which comprises the following specific steps:
step 401, adjusting parameters of a graph convolution neural network and selecting a hierarchical structure to construct a graph convolution neural network model;
step 402, feature matrix of step 301 and label matrix of step 302 are inputs of a graph rolling neural network, and are used for training parameters in the graph rolling neural network model, and obtaining high-order aggregation feature attributes of each node according to the trained graph rolling neural network model, the adjacent matrix of step 303 and a propagation formula between layers; inputting the high-order aggregation characteristic attribute of each node into a Softmax layer in the graph convolution neural network model to obtain a final fault classification diagnosis result.
As a further optimization scheme of the cellular network fault diagnosis method, the step 2 is specifically as follows:
step 201, obtaining importance scores of each of the original characteristic parameters by using an XGBoost algorithm;
step 202, sorting the characteristic parameters in a descending order according to the importance scores of the characteristic parameters;
step 203, selecting the first n characteristic parameters with highest diagnosis accuracy of the XGBoost algorithm from the original characteristic parameters according to the ordered sequence in the step 202 as the characteristic parameters of the data in the network fault data set, thereby obtaining a network fault data set with dimension reduced, wherein n is more than 0 and less than 16;
as a further optimization scheme of the cellular network fault diagnosis method, parameters in the graph roll neural network model comprise the number of graph roll layers, the probability of dropout layers and the size of a filter matrix.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
(1) The invention is based on a semi-supervised fault diagnosis algorithm, so that the number of the marking data for training is reduced;
(2) According to the invention, the XGBoost algorithm is used for solving the optimal characteristic parameter combination, so that the problems of reduced model training speed, dimension disaster generation and the like are effectively avoided;
(3) The complex nonlinear relation between the network key performance index parameter and the fault type is fully utilized, the similarity relation between samples is well utilized, and the performance is further improved;
(4) And the graph convolutional neural network level is reasonably constructed, the convolutional neural network parameters are reasonably set, and the effectiveness and reliability of model fault diagnosis are improved.
Drawings
Fig. 1 is a 4G/5G heterogeneous wireless network scenario diagram of the present invention.
Fig. 2 is a block diagram of data in the present invention.
Fig. 3 is a hierarchical structure diagram of a graph convolutional neural network in the present invention.
Detailed Description
The technical scheme of the invention is further described in detail below with reference to the accompanying drawings:
the invention provides a novel intelligent fault diagnosis algorithm model based on a graph convolution neural network, which fully utilizes the marking information of marking data and the characteristic parameter information of non-marking data, and can achieve good diagnosis accuracy under the condition that only a small amount of marked marking data is used for training. The model can quickly detect network faults and further identify possible network fault types, so that the recovery speed of a fault cell is increased.
A4G/5G heterogeneous wireless network scene diagram is shown in fig. 1, and the content mainly comprises optimal feature combination selection. The invention considers the heterogeneous wireless network scene of the cross overlapping coverage of the macro cell, the micro cell and the femto cell, and under the scene, the system is more complex and the network management is more difficult due to the diversity of the network. The invention considers the network fault diagnosis and prediction under the scene, firstly, the key performance index for measuring the network performance and the common network fault are needed to be analyzed, and the early work for constructing the network fault diagnosis model is realized.
1) The feature selection comprises the following steps:
step 101, according to the real data set obtained by drive test and combining with the actual situation, selecting the non-network fault situation and other 5 common network faults as the label categories of the network fault data set, wherein the 5 network faults are uplink interference, downlink interference, coverage holes, air interface faults and base station faults respectively;
step 102, the key performance index parameters used in identifying the faulty cell are taken as original characteristic parameters of the sample data, and the key performance index parameters include 16 key performance index parameters including reference signal receiving power, reference signal receiving quality, uplink packet loss rate, downlink packet loss rate, uplink signal to noise ratio, downlink signal to noise ratio, radio resource control connection establishment success rate, evolved radio access bearer establishment success rate, dropped call rate, handover success rate, uplink average throughput, downlink average throughput, node outgoing average throughput, node incoming average throughput, handover delay and link error rate.
2) And selecting the optimal feature combination, wherein if the data sample in the data set contains too many feature parameters, the construction of the adjacency matrix of the next graph is not facilitated, the generated adjacency matrix is unreasonable, and in the fault diagnosis stage, the model training speed is reduced, and the problems of dimension disasters and the like are caused. Therefore, a small number of important network parameters need to be selected to identify a failed cell based on their importance. The method comprises the following steps:
in step 201, the XGBoost algorithm is used to select the optimal feature combination, where XGBoost creates multiple lifting trees, and in each iteration of training XGBoost creates a new decision tree model in the gradient descent direction of the previous model loss function, so that the residual error between the predicted value and the true value of the previous model is gradually reduced. The XGBoost algorithm can obtain an importance score for each characteristic parameter, and the more the attribute is used for constructing a decision tree in the model, the more important the attribute is;
step 202, utilizing the feature importance ranking function of the XGBoost framework to rank the 16 feature parameters in a descending order according to the importance scores of the feature parameters;
step 203, selecting the first n characteristic parameters with highest diagnosis accuracy of the XGBoost algorithm as the characteristic parameters of the network fault data set after dimension reduction according to the ordered original characteristic parameters in step 202, wherein n is more than 0 and less than 16;
3) The graph data conversion needs to convert the data in the data set into non-European data, namely graph data, so as to accord with the input format of the graph convolution neural network. As shown in fig. 2, the method proposed by the present invention first needs to map the original network failure data set into an undirected graph g= (V, E), which is composed of two types of elements, namely, a node set V and an edge set E. In the invention, each sample data in the original data set corresponds to one node in the graph, each node has a respective attribute characteristic, and the similarity measure between the nodes can be measured by the weight of the edge. The method comprises the following steps:
step 301, representing the characteristic parameters of the sample data in the network fault data set after the dimension reduction in a form of a characteristic matrix, wherein each row vector in the characteristic matrix corresponds to a characteristic parameter vector formed by removing the category information from one piece of sample data in the network fault data set after the dimension reduction;
in step 302, the label information of the samples in the original dataset is represented in the form of a label matrix, and the label information of the label samples in the matrix is represented by row vectors similar to single-hot codes, and the row vectors respectively correspond to different types of faults suffered by the cell currently. The corresponding row vector of the unlabeled data is a zero vector, which indicates that the labeling information is unknown;
step 303, firstly constructing a weight matrix to record the similarity between nodes in the graph, and the element w in the weight matrix ij I.e. the similarity between node pairs is obtained by calculating Euclidean distance in pairs and normalizing, i.e
Figure BDA0003186325420000051
Wherein w is ij Representing node x i And x j Similarity measure between x i And x j Representing the i and j two different samples respectively, delta is called Gaussian bandwidth parameter and can be autonomously defined according to actual conditions. Then, a reasonable parameter threshold value alpha is set, all elements in the weight matrix are compared with the threshold value alpha, and the element w with the value larger than the threshold value is compared with the threshold value alpha ij Reset to 1, otherwise set to 0 all. Then, the weight matrix is converted into an adjacency matrix a having elements of only 0 and 1, wherein 0 indicates that there is no relation between the two samples, and 1 indicates that the degree of similarity between the two samples is high.
4) In order to improve the accuracy of network fault diagnosis based on fault diagnosis of a graph convolution neural network, the optimal neural network structure and parameters need to be determined before the GCN is trained, and in order to illustrate the structure and the workflow of the GCN, a GCN model defined by the invention is shown as shown in fig. 3, and the method comprises the following steps:
and step 401, selecting reasonable GCN parameters and a hierarchical structure, and constructing a graph convolutional neural network. The structure and parameter selection of the GCN mainly comprises the number of layers of the picture convolution layer, the size of a filter matrix of each picture convolution layer and the probability size of a dropout layer. According to the actual situation, when the number of layers of the graph roll lamination is 2, the characteristic parameter information of the neighbor nodes of each center node can be reasonably aggregated, and the problem of overfitting caused by excessive training parameters due to the fact that the number of the convolution layers is excessive can be avoided. The probability of the dropout layer is generally selected to be 0.25 or 0.5 properly, the size of the filter matrix is determined according to the size of the convolution layer of the previous layer, and meanwhile, factors such as attribute dimension reduction in operation and the like are considered;
step 402, after GCN construction is completed, the steps areThe feature matrix in 301 and the tag matrix in step 302 serve as inputs to the GCN, which, according to the GCN defined layer-to-layer propagation formula,
Figure BDA0003186325420000052
and dividing a training set, training a graph convolution neural network model, and utilizing the trained GCN model to aggregate characteristic parameters of each node and neighbor nodes to obtain high-order aggregation characteristic attributes of each node. Wherein H is (l+1) Matrix formed by high-order aggregation characteristic attributes of all nodes of current graph convolution layer output, H (l) Is the output of the previous layer, +.>
Figure BDA0003186325420000053
Indicating that the adjacency matrix in step 303 is added with self-loop,/->
Figure BDA0003186325420000054
Is about->
Figure BDA0003186325420000055
Is a degree matrix, sigma represents an activation function, W (l) A trainable weight matrix representing a layer i of the graph roll layer; />
And step 403, inputting the obtained high-order aggregation characteristic attribute of each node into a Softmax layer to obtain a final fault classification diagnosis result. The Softmax activation function needs to be applied on each row of the feature matrix output by the convolution layer of the last layer of the graph. The diagnosis result is represented by a characteristic matrix Z at the node level, and network faults corresponding to row vectors similar to the one-hot codes in each row in the matrix Z represent different kinds of network fault categories suffered by the current cell respectively.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the scope of the present invention should be included in the scope of the present invention.

Claims (3)

1. A method for diagnosing a fault in a cellular network, comprising the steps of:
step 1, determining a network fault data set; the method comprises the following steps:
step 101, data in a network fault data set comprises original characteristic parameters and label categories, and a non-network fault condition and 5 network faults are defined as the label categories, wherein the 5 network faults are uplink interference, downlink interference, coverage holes, air interface faults and base station faults respectively;
step 102, taking all key performance indexes used in the fault cell identification as original characteristic parameters, wherein the original characteristic parameters comprise 16 characteristic parameters, and the 16 characteristic parameters comprise reference signal receiving power, reference signal receiving quality, uplink packet loss rate, downlink packet loss rate, uplink signal to noise ratio, downlink signal to noise ratio, radio resource control connection establishment success rate, evolution radio access bearer establishment success rate, call drop rate, handover success rate, uplink average throughput, downlink average throughput, node outgoing average throughput, node incoming average throughput, handover delay and link error rate;
step 2, adopting an XGBoost algorithm, and selecting the first n characteristic parameters with highest diagnosis accuracy of the XGBoost algorithm from the original characteristic parameters as the characteristic parameters of data in the network fault data set, so as to obtain a network fault data set with reduced dimension, wherein n is more than 0 and less than 16;
step 3, graph data conversion, specifically comprising the following steps:
step 301, representing the network fault data set subjected to the dimension reduction in step 2 in a form of a feature matrix, wherein each row vector in the feature matrix corresponds to a piece of feature parameter vector except for category information in the network fault data set subjected to the dimension reduction;
step 302, representing the label information of the network fault data set subjected to the dimension reduction in the step 2 in a label matrix form; in the tag matrix, the tag line vector of the marked data is in a form of single-hot coding, and the tag line vector of the unmarked data is a zero vector; the marked data is data with a label, and the unmarked data is data without a label;
step 303, mapping the network fault data set after the dimension reduction into an undirected graph g= (V, E), wherein the undirected graph is composed of two types of elements, namely a node set V and an edge set E; introducing a weight matrix to represent the similarity between nodes in the undirected graph, wherein elements in the weight matrix are the similarity between node pairs, and the similarity between node pairs is obtained by calculating Euclidean distances between every two nodes and normalizing; comparing all elements in the weight matrix with the parameter threshold through the set parameter threshold, setting the element as 1 if the current element is larger than the parameter threshold, otherwise setting the element as 0; thereby converting the weight matrix into an adjacent matrix with matrix elements of only 0 and 1, wherein the adjacent matrix represents the adjacent relation between nodes;
and 4, fault diagnosis based on a graph convolution neural network, which comprises the following specific steps:
step 401, adjusting parameters of a graph convolution neural network and selecting a hierarchical structure to construct a graph convolution neural network model;
step 402, feature matrix of step 301 and label matrix of step 302 are inputs of a graph rolling neural network, and are used for training parameters in the graph rolling neural network model, and obtaining high-order aggregation feature attributes of each node according to the trained graph rolling neural network model, the adjacent matrix of step 303 and a propagation formula between layers; inputting the high-order aggregation characteristic attribute of each node into a Softmax layer in the graph convolution neural network model to obtain a final fault classification diagnosis result.
2. The method for diagnosing a fault in a cellular network according to claim 1, wherein step 2 is specifically as follows:
step 201, obtaining importance scores of each of the original characteristic parameters by using an XGBoost algorithm;
step 202, sorting the characteristic parameters in a descending order according to the importance scores of the characteristic parameters;
step 203, selecting the first n characteristic parameters with highest diagnosis accuracy of the XGBoost algorithm from the original characteristic parameters according to the order in the step 202 as the characteristic parameters of the data in the network fault data set, thereby obtaining the network fault data set with dimension reduced, wherein n is more than 0 and less than 16.
3. The method of claim 1, wherein the parameters in the convolutional neural network model include a number of layers of the convolutional neural network, a dropout layer probability size, and a filter matrix size.
CN202110862707.3A 2021-07-29 2021-07-29 Cellular network fault diagnosis method Active CN113709779B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110862707.3A CN113709779B (en) 2021-07-29 2021-07-29 Cellular network fault diagnosis method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110862707.3A CN113709779B (en) 2021-07-29 2021-07-29 Cellular network fault diagnosis method

Publications (2)

Publication Number Publication Date
CN113709779A CN113709779A (en) 2021-11-26
CN113709779B true CN113709779B (en) 2023-06-02

Family

ID=78650855

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110862707.3A Active CN113709779B (en) 2021-07-29 2021-07-29 Cellular network fault diagnosis method

Country Status (1)

Country Link
CN (1) CN113709779B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114513367B (en) * 2021-12-10 2023-02-10 西安电子科技大学 Cellular network anomaly detection method based on graph neural network
CN114077811B (en) * 2022-01-19 2022-04-12 华东交通大学 Electric power Internet of things equipment abnormality detection method based on graph neural network
CN115426671B (en) * 2022-09-01 2023-10-31 中国电信股份有限公司 Method, system and equipment for training graphic neural network and predicting wireless cell faults
CN115524002B (en) * 2022-09-19 2023-08-22 国家电投集团河南电力有限公司 Operation state early warning method, system and storage medium of power plant rotating equipment
CN117540175A (en) * 2024-01-09 2024-02-09 海纳云物联科技有限公司 Model training method, prediction method, device, equipment and medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110839256A (en) * 2019-09-26 2020-02-25 西安交通大学 Cellular network data anomaly detection method
CN111242225A (en) * 2020-01-16 2020-06-05 南京邮电大学 Fault detection and diagnosis method based on convolutional neural network
EP3836599A1 (en) * 2019-12-12 2021-06-16 Telefonica Digital España, S.L.U. Method for detecting permanent failures in mobile telecommunication networks

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110839256A (en) * 2019-09-26 2020-02-25 西安交通大学 Cellular network data anomaly detection method
EP3836599A1 (en) * 2019-12-12 2021-06-16 Telefonica Digital España, S.L.U. Method for detecting permanent failures in mobile telecommunication networks
CN111242225A (en) * 2020-01-16 2020-06-05 南京邮电大学 Fault detection and diagnosis method based on convolutional neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于卷积神经网络的局域网故障诊断技术研究;周云风;电子科技大学;全文 *

Also Published As

Publication number Publication date
CN113709779A (en) 2021-11-26

Similar Documents

Publication Publication Date Title
CN113709779B (en) Cellular network fault diagnosis method
CN110943857B (en) Power communication network fault analysis and positioning method based on convolutional neural network
CN108540330B (en) Network fault diagnosis method based on deep learning in heterogeneous network environment
CN111405585B (en) Neighbor relation prediction method based on convolutional neural network
CN111444939B (en) Small-scale equipment component detection method based on weak supervision cooperative learning in open scene of power field
CN112039687A (en) Small sample feature-oriented fault diagnosis method based on improved generation countermeasure network
WO2022057321A1 (en) Method and apparatus for detecting anomalous link, and storage medium
CN113779247B (en) Network fault diagnosis method and system based on intention driving
CN110062410B (en) Cell interruption detection positioning method based on self-adaptive resonance theory
CN113297429B (en) Social network link prediction method based on neural network architecture search
Liao et al. Network state awareness and proactive anomaly detection in self-organizing networks
CN110139315B (en) Wireless network fault detection method based on self-learning
CN113676357B (en) Decision method for edge data processing in power internet of things and application thereof
CN117236520B (en) Distributed multi-unmanned aerial vehicle cluster cooperative scheduling system and method thereof
CN111405605B (en) Wireless network interruption detection method based on self-organizing mapping
CN115734274A (en) Cellular network fault diagnosis method based on deep learning and knowledge graph
CN115423041A (en) Edge cloud fault prediction method and system based on deep learning
CN114698016A (en) Network fault detection method and device for poor-quality cell
Ruan et al. An Intelligent Cell Outage Detection Method in Cellular Networks
Wan et al. Capturing Spatial-Temporal Correlations with Attention Based Graph Convolutional Networks for Network Traffic Prediction
CN113762485B (en) Attention mechanism-based multi-dataset joint prediction method
CN115865627B (en) Cellular network fault diagnosis method for carrying out characterization learning based on pattern extraction
CN117692026B (en) Link sensing method and device for power line communication
Le et al. Enhanced handover clustering and forecasting models based on machine learning and big data
CN114501525B (en) Wireless network interruption detection method based on condition generation countermeasure network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant