阅读说明：本技术 一种载波网络故障的诊断方法、系统、设备和介质 (Method, system, equipment and medium for diagnosing carrier network fault ) 是由 施展 付佳佳 于 2021-07-19 设计创作，主要内容包括：本发明涉及公开了一种载波网络故障的诊断方法、系统、设备和介质,包括：获取网络特征数据集,并根据所述网络特征数据集,构建载波网络的故障诊断数据集；对所述故障诊断数据集进行预处理,得到预处理后的故障诊断数据集,其中,所述预处理包括：归一化处理；将所述预处理后的故障诊断数据集分为测试数据集和训练数据集；将训练数据集输入到预设的故障诊断分类模型进行训练,获得训练后的故障诊断分类模型；将所述测试数据集输入到训练后的故障诊断分类模型,获得载波网络的故障分类结果。本发明提出的基于网络特征的载波网络故障的诊断方法具有较好的应用效果和性能,解决了电力载波网络故障的诊断算法准确率低的问题。(The invention relates to and discloses a method, a system, equipment and a medium for diagnosing carrier network faults, wherein the method comprises the following steps: acquiring a network characteristic data set, and constructing a fault diagnosis data set of a carrier network according to the network characteristic data set; preprocessing the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, wherein the preprocessing comprises: normalization processing; dividing the preprocessed fault diagnosis data set into a test data set and a training data set; inputting the training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model; and inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network. The method for diagnosing the carrier network fault based on the network characteristics has good application effect and performance, and solves the problem of low accuracy of a diagnosis algorithm of the power carrier network fault.)

1. A method for diagnosing carrier network faults is characterized by comprising the following steps:

acquiring a network characteristic data set, and constructing a fault diagnosis data set of a carrier network according to the network characteristic data set;

preprocessing the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, wherein the preprocessing comprises: normalization processing;

dividing the preprocessed fault diagnosis data set into a test data set and a training data set;

inputting the training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model;

and inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network.

2. The method of diagnosing carrier network faults according to claim 1, wherein the fault diagnosis data set includes: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

3. The method according to claim 2, wherein the centrality of the bottom node is calculated as follows:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iA set of related symptoms, wherein the value of the symptoms in the set is 1, S represents the set of all symptoms collected by the network management system, and the symptoms in the setThe value is 1 and | x | represents the number of elements included in the computation set.

4. The method according to claim 1, wherein the predetermined fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating a fault condition.

5. A system for diagnosing carrier network faults, comprising: a data acquisition module, a preprocessing module, a splitting module, a training module and a fault classification module, wherein,

the data acquisition module is used for acquiring a network characteristic data set and constructing a fault diagnosis data set of the carrier network according to the network characteristic data set;

the preprocessing module is configured to preprocess the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, where the preprocessing includes: normalization processing;

the splitting module is used for dividing the preprocessed fault diagnosis data set into a test data set and a training data set;

the training module is used for inputting a training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model;

and the fault classification module is used for inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network.

6. The system of claim 5, wherein the set of fault diagnosis data comprises: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

7. The system of claim 6, wherein the centrality of the bottom node is calculated as follows:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iAnd the value of the symptom in the set is 1, S represents the set of all symptoms collected by the network management system, the value of the symptom in the set is 1, and | x | represents the number of elements contained in the calculation set.

8. The system for diagnosing carrier network faults according to claim 5, wherein the preset fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating a fault condition.

9. A computer terminal device, comprising:

one or more processors;

a memory coupled to the processor for storing one or more programs;

when executed by the one or more processors, cause the one or more processors to implement the method of diagnosing a carrier network fault as claimed in any one of claims 1 to 4.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a method of diagnosing a carrier network fault according to any one of claims 1 to 4.

Technical Field

The invention relates to the technical field of fault processing of power communication, in particular to a method, a system, equipment and a medium for diagnosing faults of a carrier network.

Background

The power line carrier network has become an important basic network resource for the application of the power internet of things. Under the power line carrier network environment, the service data of the power internet of things can be transmitted through the power line, and the construction cost and the construction expense of network resources are reduced. With the rapid development and application of network virtualization technology, a carrier network based on network slicing technology has become an important research field. In a carrier network environment based on a network slicing technique, conventional network resources are divided into underlying network resources and virtual network resources. The underlying network resources are responsible for providing network resources for the virtual network resources. The virtual network resources bear specific power Internet of things services, and isolation among the services can be achieved. Through analysis, the carrier network based on the network slicing technology improves the utilization rate of network resources and the reliability of power services. However, in a carrier network environment based on network slicing technology, the traffic state of the virtual network is transparent to the underlying network service provider, and the state of the underlying network resources is also transparent to the virtual network service provider. Under the background, the fault diagnosis of the virtual network service is greatly different from the processing method of the existing research. Therefore, how to accurately locate the fault and improve the reliability of the power line carrier network becomes a key problem.

At present, the existing research is analyzed, and the existing research adopts an intelligent algorithm to perform mathematical modeling according to data acquired by a network management system, so that a suspected fault set is inferred, and the inferred suspected fault set has low accuracy and long inference time.

Disclosure of Invention

The purpose of the invention is: the method, the system, the equipment and the medium for diagnosing the carrier network faults can solve the problem of low accuracy of a diagnosis algorithm of the power carrier network faults.

In order to achieve the above object, the present invention provides a method for diagnosing a carrier network fault, including:

acquiring a network characteristic data set, and constructing a fault diagnosis data set of a carrier network according to the network characteristic data set;

preprocessing the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, wherein the preprocessing comprises: normalization processing;

dividing the preprocessed fault diagnosis data set into a test data set and a training data set;

inputting the training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model;

and inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network.

Further, the fault diagnosis data set includes: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

Further, the centrality of the bottom-layer node is calculated by the following formula:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iAnd the value of the symptom in the set is 1, S represents the set of all symptoms collected by the network management system, the value of the symptom in the set is 1, and | x | represents the number of elements contained in the calculation set.

Further, the preset fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating a fault condition.

The invention also provides a system for diagnosing the carrier network fault, which comprises the following components: a data acquisition module, a preprocessing module, a splitting module, a training module and a fault classification module, wherein,

the data acquisition module is used for acquiring a network characteristic data set and constructing a fault diagnosis data set of the carrier network according to the network characteristic data set;

the preprocessing module is configured to preprocess the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, where the preprocessing includes: normalization processing;

the splitting module is used for dividing the preprocessed fault diagnosis data set into a test data set and a training data set;

the training module is used for inputting a training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model;

and the fault classification module is used for inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network.

Further, the fault diagnosis data set includes: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

Further, the centrality of the bottom-layer node is calculated by the following formula:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iAnd the value of the symptom in the set is 1, S represents the set of all symptoms collected by the network management system, the value of the symptom in the set is 1, and | x | represents the number of elements contained in the calculation set.

Further, the preset fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating the status of failureState.

The present invention also provides a computer terminal device, comprising: one or more processors; a memory coupled to the processor for storing one or more programs; when executed by the one or more processors, cause the one or more processors to implement the method for diagnosing a carrier network fault as in any one of the above.

The invention also provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out a method of diagnosing a carrier network fault as defined in any one of the above.

Compared with the prior art, the method, the system, the terminal device and the computer readable storage medium for diagnosing the carrier network fault have the advantages that:

through performance analysis of the fault diagnosis algorithm, the carrier network fault diagnosis method based on the network characteristics has good application effect and performance, and the problem of low accuracy of the power carrier network fault diagnosis algorithm is solved well.

Drawings

Fig. 1 is a schematic flowchart of a carrier network fault diagnosis method provided in the present invention;

FIG. 2 is a schematic diagram of a Bayesian fault propagation model provided by the present invention;

FIG. 3 is a graph showing the comparison of accuracy between the present invention and the prior art;

FIG. 4 is a graph illustrating the comparison of false alarm rates provided by the present invention and in the prior art;

fig. 5 is a schematic flow chart of a carrier network fault diagnosis system provided in the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It should be understood that the step numbers used herein are for convenience of description only and are not intended as limitations on the order in which the steps are performed.

It is to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

The terms "comprises" and "comprising" indicate the presence of the described features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

The term "and/or" refers to and includes any and all possible combinations of one or more of the associated listed items.

As shown in fig. 1, the present invention provides a method for diagnosing a carrier network fault, which at least includes the following steps:

s1, acquiring a network characteristic data set, and constructing a fault diagnosis data set of the carrier network according to the network characteristic data set;

specifically, a data set is constructed from a monitoring data set Ω of the integrated network management system according to the attributes of the nodesEach piece of data for constructing the data set is the data characteristics of 7 bottom-layer nodes and 1 bottom-layer node state. Data feature set x of 7 bottom nodes_iThe method comprises the following steps: degree of bottom layer nodeCentrality of underlying nodesImportance of underlying nodesHistorical failure times of underlying nodesFault dependencies of underlying nodesFault independence of underlying nodesResource utilization of underlying nodes

It should be noted that the power carrier network is composed of four parts, namely an access terminal, an access node, a control node and an intelligent gateway. Along with the rapid development and application of the power internet of things technology, the types of the access terminals are more and more, namely, the access terminals comprise traditional data acquisition equipment and novel business equipment such as intelligent electric meters and intelligent household appliances. In a network slicing environment, infrastructures such as access nodes, control nodes, intelligent gateways and the like are divided into an underlying network and a virtual network.

The underlying network is denoted by G ═ (N, E), and the virtual network by G^V＝(N^V,E^V) And (4) showing. Wherein N and N^VRespectively representing the underlying nodes and virtual nodes. Each bottom node n_iThe e N mainly comprises computing resources and each virtual nodeNeed to go to the bottom node n_iAnd e.g. N applies for computing resources so as to bear the power service. E and E^VRespectively representing the underlying link and the virtual link. Each underlying link e_iThe E is mainly composed of bandwidth resources and each virtual linkNeed to go to the underlying link e_iE applies for bandwidth resources, thereby carrying power service.

To describe the relationship of the underlying network and the virtual network, G is used^V↓ G denotes a virtual network G^VAnd applying for and obtaining bottom link and bottom node resources from the bottom network G. Bottom node n_iBelongs to N as a virtual nodeAllocating resources, usingAnd (4) showing. Underlying Link e_iE is a virtual linkAllocating resources, usingAnd (4) showing.Representing virtual linksThe two virtual nodes of the network node are mapped to form an end-to-end bottom path.

In order to improve the service quality of power services, most power companies have built integrated network management systems. The comprehensive network management system can acquire the running state of the underlying network and the running state of the power service in real time. Considering that the service modes from a server to a client in the power internet of things are more, the invention mainly researches end-to-end power services. Use ofRepresenting an end-to-end power service, the two end points of the power service being virtual nodes respectivelyAnd virtual node

In general, the types of data information that can be obtained by the integrated network management system are various. The data index thresholds which can be obtained by the comprehensive network management system in different scenes are different. If the fault diagnosis is directly carried out according to the alarm information reported by the integrated network management system, the problem of low relevance between a fault diagnosis algorithm and the actual network environment is easily caused. In order to improve the relevance of the fault diagnosis algorithm data, the invention obtains the basic network resource information and the service alarm information from the comprehensive network management system according to the fault operation and maintenance experience, and obtains the data characteristics with higher relevance to the fault diagnosis based on the information, thereby improving the performance of the fault diagnosis algorithm.

With the gradual improvement of network management technology and monitoring capability, more and more information can be monitored by the underlying network nodes. The information of the underlying network nodes is very beneficial to improving the accuracy rate of fault diagnosis. Meanwhile, the bottom layer network node can monitor the available state of the bottom layer link connected with the bottom layer network node. Therefore, the invention takes the underlying network nodes as a research object, and realizes the improvement of the performance of the fault diagnosis algorithm by analyzing the data characteristics of the underlying network nodes and establishing the fault diagnosis algorithm model.

The data characteristics with high relevance with fault diagnosis obtained by the invention comprise 7 data characteristics, namely: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

1. Degree of bottom layer node

The degree of the bottom layer network node has larger relevance with the fault probability of the power service borne by the bottom layer network node. When the degree of the underlying network node is large, if a part of links of the node have faults, the power service can still adopt a dynamic routing strategyAnd slightly, reselecting a new link to carry the power service. Therefore, the greater the degree of the underlying network nodes, the higher the reliability of the power service carried thereon. Degree usage of underlying network nodesAnd (4) showing. Underlying network node n_iThe value of the degree of the E N is the number of all the direct connection edges of the node.

2. Centrality of underlying nodes

The centrality of the underlying network node refers to the position of the current underlying network node in the network. The more centralized the position of the underlying network node in the network, the more traffic passing through the underlying network node, the more electric power traffic carried thereon. When a node in the network center fails, the number of failed power services is large. Centrality usage of underlying nodesExpressed, calculated using equation (1). Wherein the content of the first and second substances,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count.

3. Importance of underlying nodes

The greater the amount of power traffic carried on the underlying node, the more important the underlying node is in terms of reliable operation of the power traffic. Therefore, the importance of the underlying node is measured by the amount of power traffic carried thereon. Importance usage of underlying nodesExpressed, the calculation was performed using equation (2). Wherein, delta_jRepresenting the number of power services of type j. z represents the class of power traffic carried on the underlying nodeThe number of types.

4. Historical failure times of underlying nodes

The more times of failure of the bottom node, the lower the reliability of the current bottom node is, and the higher the probability of failure again is. In order to evaluate the reliability of the bottom node, the number of failures of the bottom node in the last half year is closely related to the reliability of the bottom node according to the operation experience. Use ofRepresenting the underlying node n_iThe number of failures occurring in the last half year. The larger the value, the lower the reliability of the current underlying node, and the higher the probability that the power traffic carried thereon may fail.

5. Fault dependencies of underlying nodes

When the power service has a fault, the fault can be positioned according to the alarm information reported by the network management system. In order to evaluate the relevance of each bottom layer node and the current alarm, the invention provides a Bayesian fault propagation model. The Bayesian fault propagation model comprises an upper layer node, a lower layer node and a connecting line between the two layers of nodes. Upper level nodes represent symptom nodes, using S_o＝{s₁,s₂,...,s_mDenotes a set of m symptom nodes. s_m0 indicates that the service corresponding to the symptom node is available. s_mThe symbol 1 indicates that the service corresponding to the symptom node is in an unavailable state. The lower node represents a faulty node, and X ═ X is used₁,x₂,...,x_nDenotes the set of n failed nodes. x is the number of_nAnd 0 represents that the underlying network component corresponding to the fault node is in the available state. x is the number of_nAnd 1 represents that the under layer network component corresponding to the fault node has an unavailable state. The connecting line between two layers of nodes represents the probability value of the upper layer symptom node abnormal after the lower layer fault node is in fault, and P(s) is used_j|f_i) And (4) showing. Because the network management system alarms according to the information collected by the network management protocol, and the information collection is easily influenced by the instability of the network environment, the connection value between the upper node and the lower node is generally less than 1.

To determine fault dependencies of underlying nodesThe invention provides a formula (3) for calculating the fault correlation of the bottom layer nodes. Wherein the content of the first and second substances,representation and underlying node n_iA set of related symptoms, the value of a symptom in the set being 1. And S represents a set of all symptoms collected by the network management system, and the value of the symptoms in the set is 1. | x | represents the number of elements included in the computation set.

6. Fault independence of underlying nodes

The fault correlation of the bottom layer node can judge the correlation between the current alarm and the current bottom layer node, but the independence between the current bottom layer node and the alarm cannot be explained. If the fault independence of the bottom layer node is known, the degree of irrelevance of the alarm and the current bottom layer node can be judged.

Bottom node n_iFault independence use ofIt is shown that, using the formula (4) for calculation,representing the underlying node n_iAnd corresponding fault nodes in the fault propagation model.To representBottom node n_iCorresponding symptom nodes in the fault propagation model.Representing the underlying node n_iAnd corresponding symptom nodes in the fault propagation model are reported by the network management system as alarms, so that a set of symptoms with the value of 1 can be observed. Therefore, the larger the value of the formula (4), the lower-level node n is shown_iThe smaller the association with the currently observed alarm, the greater the independence. Otherwise, the bottom node n is indicated_iThe greater the association with the currently observed alarm, the less independent.

7. Resource utilization of underlying nodes

According to the operation experience, the reliability of the bottom node is related to the resource utilization rate of the bottom node. When the resource utilization rate of the bottom node is too high, the aging speed of the bottom node is easy to accelerate, and therefore the probability of the failure of the bottom node is increased. Resource utilization usage by underlying nodesExpressed, the value is the amount of used resources divided by the total amount of resources.

S2, preprocessing the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, wherein the preprocessing comprises: normalization processing;

specifically, the fault diagnosis data set is preprocessed, and the influence of different data dimensions on the performance of the algorithm is reduced by adopting a normalization algorithm. Because the values of the 7 input attributes do not belong to the same order of magnitude, the difference of the input variables is large. In order to improve the performance of the algorithm, a maximum and minimum normalization algorithm is adopted to assemble the dataIs subjected to normalizationThen, obtain the data set { (x)_i,y_i),i＝1,2,...,l}。

S3, dividing the preprocessed fault diagnosis data set into a test data set and a training data set;

specifically, the data set is divided into a training set and a test set. From a data set { (x)_i,y_i) 80% of the data are taken out as a training set { (x } { (1, 2.) }._i,y′_i) 1, 2.. multidot.z }, with the remaining 20% of the data as the test set { (x ″ ")_i,y″_i),i＝1,2,...,t}。

S4, inputting the training data set into a preset fault diagnosis classification model for training to obtain a trained fault diagnosis classification model;

specifically, the model key parameters are first calculated, and the training model is second obtained. Using training set { (x'_i,y′_i) And (e), searching the optimal penalty factor and kernel function variance parameter value by adopting a cross-validation method according to the data in the i-1, 2. And calculating an objective function formula (14) according to the parameter values to obtain a classification model f (x) of fault diagnosis.

In the fault diagnosis data characteristic analysis part, the invention describes how to calculate 7 fault diagnosis data characteristics according to network management data collected by a comprehensive network management system. In order to perform fault diagnosis based on these fault diagnosis data characteristics, a data set for fault diagnosis needs to be constructed. A Support Vector Machine (SVM) is a supervised learning algorithm, and can classify underlying network resources according to the characteristics of fault diagnosis data. According to the invention, a fault diagnosis model based on SVM is constructed according to the characteristics of fault diagnosis data.

Using { (x)_i,y_i) I 1, 2.. multidot.l represents a failure diagnosis set composed of l pieces of failure diagnosis-related data. Wherein x is_iRepresenting the ith fault diagnosis data characteristic, the current bottom node n_iThe 7 data characteristics of (1). y is_iRepresenting the current underlying node n_iThe available state of (c). When y is_iWhen 1, indicate the current bottom node n_iIs in the state of failureState. When y is_iWhen-1, indicates the current bottom level node n_iIs in a normal state.

The goal of fault diagnosis is according to x_iState of (2), determining y_iThe state of (1). The goal of fault diagnosis can be achieved assuming that the hyperplane wx + b is 0. The hyperplane can be employed according to x_iState of (2), determining y_iThe state of (1). When wx_iWhen + b is greater than or equal to 1, y_i1. When wx_iWhen b is less than or equal to-1, y_iIs-1. At this time, the solution can be performed by solving the objective function of equation (5).

s.t.y_i(wx_i+b)≥1 (6)

To facilitate solving the objective function, the saddle point of the Lagrangian function can be calculated by equation (7), where a_iRepresenting lagrangian function coefficients.

When the optimal solution is obtainedTime, optimal w in the objective function^*、b^*The calculation can be performed using the formulas (8), (9). x is the number of_rAnd x_sA pair of support vectors in either of two categories.

At this time, the optimal classification objective function is calculated using equation (10). Wherein sgn [ η ] is a step function, and when η is larger than or equal to 0, sgn [ η ] is 1. When η <0, sgn [ η ] ═ 1.

The problem cannot be solved in a linear environment due to uncertainty of fault diagnosis data in the carrier network. To solve this problem, phi R is mapped by non-linearity^d→ H implements an optimized classification in the high-dimensional space H. In order to realize the mapping from the low-dimensional space to the high-dimensional space, the invention adopts a radial basis kernel function K (x, x)_i) The implementation and the definition are shown as formula (11).

At this time, the objective function of equation (5) becomes the objective function in equation (12), where a_iA is greater than or equal to 0_iC is less than or equal to C, and C represents a penalty factor and is used for optimizing the solution of the objective function.

When in useWhen the temperature of the water is higher than the set temperature,at this time, the objective function of the failure diagnosis is updated to equation (14).

And S5, inputting the test data set into the trained fault diagnosis classification model to obtain a fault classification result of the carrier network.

Specifically, using SVM diagnostic model f (x) to test set { (x ″)_i,y″_i) And i is 1,2, a.

Specifically, the method of the present invention is shown in Table 1,

TABLE 1 network characteristic-based carrier network fault diagnosis algorithm

In one embodiment of the present invention, the fault diagnosis data set includes: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

In one embodiment of the present invention, the centrality of the bottom node is calculated by the following formula:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hopsCounting;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iAnd the value of the symptom in the set is 1, S represents the set of all symptoms collected by the network management system, the value of the symptom in the set is 1, and | x | represents the number of elements contained in the calculation set.

In an embodiment of the present invention, the preset fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating a fault condition.

In order to verify the performance of the algorithm, a GT-ITM (E.W.Zegura, K.L.Calvert, S.Bhattacharjee.how to model an internet [ C ]// Proceedings of IEEE INFOCOM, 1996) tool is used in the experiment to generate a network topology environment. The network topology environment includes an underlying network environment and a virtual network environment. The number of bottom layer nodes contained in the bottom layer network environment is increased from 100 to 600, and the bottom layer nodes are used for simulating network environments with different scales, so that the influence of different bottom layer network environments on algorithm performance is verified. The number of virtual network nodes contained in the virtual network environment is subject to uniform distribution of (5, 10). To simulate network failures, the a priori failure probability of each underlying network node is set to follow a uniform distribution of [0.005, 0.01] [ Rish, m.brodie, s.ma, n.odinstova, a.beygel zimer, g.grabainik, k.hernandez.adaptive diagnostics in Distributed Systems [ J ]. IEEE trans.neural Networks, 16(5), 2005 ].

In terms of algorithm comparison, the algorithm CNFDAoNC of the present invention is compared with a fault diagnosis algorithm based on Bayesian theory (FDAoBT). The algorithm FDAoBT adopts Bayesian theory to model the alarm information of the network management system, and the algorithm CNFDAoNC of the invention carries out modeling after preprocessing the information collected by the network management system according to the network characteristics. Considering that the main processes of the two algorithms are different, the fault diagnosis duration is not comparable, so the method only compares the fault diagnosis accuracy with the false alarm rate in two dimensions.

The algorithm accuracy comparison results are shown in fig. 3. In the figure, the X-axis represents the increase in the number of bottom nodes from 100 to 600. The Y-axis represents the accuracy of the fault diagnosis results of the algorithm. As can be seen from the figure, as the number of the nodes of the underlying network increases, the accuracy of the two algorithms is relatively stable, which shows that the two algorithms can obtain better diagnosis effect for different network environments. In the aspect of comparison of the two algorithms, the fault diagnosis accuracy of the algorithm CNFDAoNC is maintained to be about 82%, and the fault diagnosis accuracy of the comparison algorithm FDAoBT is maintained to be about 78%. The algorithm of the invention obtains better results in the aspect of fault diagnosis accuracy.

The algorithm false alarm rate comparison results are shown in fig. 4. In the figure, the X-axis represents the increase in the number of bottom nodes from 100 to 600. The Y-axis represents the false alarm rate of the fault diagnosis result of the algorithm. As can be seen from the figure, as the number of the nodes of the underlying network increases, the false alarm rates of the two algorithms are relatively stable, which shows that the two algorithms can obtain better diagnosis effects for different network environments. In the aspect of comparison of the two algorithms, the fault diagnosis false alarm rate of the algorithm CNFDAoNC is maintained to be about 16%, and the fault diagnosis false alarm rate of the comparison algorithm FDAoBT is maintained to be about 23%. The algorithm of the invention obtains better results in the aspect of fault diagnosis false alarm rate.

By comparing the accuracy and the false alarm rate of the two algorithms in fault diagnosis, the algorithm of the invention preprocesses the fault diagnosis data, which is the main reason that the algorithm can obtain better results. Therefore, as the correlation of the fault handling data to the fault increases, the performance of the fault diagnosis algorithm may increase significantly.

Compared with the prior art, the method for diagnosing the carrier network fault has the beneficial effects that:

through performance analysis of the fault diagnosis algorithm, the carrier network fault diagnosis method based on the network characteristics has good application effect and performance, and the problem of low accuracy of the power carrier network fault diagnosis algorithm is solved well.

As shown in fig. 5, the present invention further provides a system 200 for diagnosing a carrier network fault, including: a data acquisition module 201, a pre-processing module 202, a splitting module 203, a training module 204, and a fault classification module 205, wherein,

the data acquisition module 201 is configured to acquire a network characteristic data set, and construct a fault diagnosis data set of a carrier network according to the network characteristic data set;

the preprocessing module 202 is configured to preprocess the fault diagnosis data set to obtain a preprocessed fault diagnosis data set, where the preprocessing includes: normalization processing;

the splitting module 203 is configured to divide the preprocessed fault diagnosis data set into a test data set and a training data set;

the training module 204 is configured to input a training data set to a preset fault diagnosis classification model for training, so as to obtain a trained fault diagnosis classification model;

the fault classification module 205 is configured to input the test data set to the trained fault diagnosis classification model, so as to obtain a fault classification result of the carrier network.

In one embodiment of the present invention, the fault diagnosis data set includes: the degree of the bottom node, the centrality of the bottom node, the importance of the bottom node, the historical failure frequency of the bottom node, the failure correlation of the bottom node, the failure independence of the bottom node and the resource utilization rate of the bottom node.

In one embodiment of the present invention, the centrality of the bottom node is calculated by the following formula:

wherein the content of the first and second substances,representing the centrality of the underlying node, N representing the underlying node, N_iThe elements that represent N are represented by,representing the underlying node n_iTo the bottom node n_jEnd-to-end hop count;

the importance of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,represents the importance of the underlying node, δ_jRepresenting the number of power services with the type j, and z representing the number of power service types borne on the bottom-layer node;

and the fault correlation of the bottom layer node adopts the following calculation formula:

wherein the content of the first and second substances,representation and underlying node n_iAnd the value of the symptom in the set is 1, S represents the set of all symptoms collected by the network management system, the value of the symptom in the set is 1, and | x | represents the number of elements contained in the calculation set.

In an embodiment of the present invention, the preset fault diagnosis classification model adopts the following calculation formula:

wherein f (x) represents a fault diagnosis classification model, l represents the number of data, sgn [ eta ] n]Is a step function, K (x)_iAnd x) represents a radial basis kernel function,b^*represents the optimal solution, y_iIndicating a fault condition.

The present invention also provides a computer terminal device, comprising: one or more processors; a memory coupled to the processor for storing one or more programs; when executed by the one or more processors, cause the one or more processors to implement the method for diagnosing a carrier network fault as in any one of the above.

It should be noted that the processor may be a Central Processing Unit (CPU), other general-purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an application-specific programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc., the general-purpose processor may be a microprocessor, or the processor may be any conventional processor, the processor is a control center of the terminal device, and various interfaces and lines are used to connect various parts of the terminal device.

The memory mainly includes a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function, and the like, and the data storage area may store related data and the like. In addition, the memory may be a high speed random access memory, may also be a non-volatile memory, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card (FlashCard), and the like, or may also be other volatile solid state memory devices.

It should be noted that the terminal device may include, but is not limited to, a processor and a memory, and those skilled in the art will understand that the terminal device is only an example and does not constitute a limitation of the terminal device, and may include more or less components, or combine some components, or different components.

The present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method of diagnosing a carrier network fault as in any one of the above.

It should be noted that the computer program may be divided into one or more modules/units (e.g., computer program), and the one or more modules/units are stored in the memory and executed by the processor to implement the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used for describing the execution process of the computer program in the terminal device. The above-mentioned embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, and it should be understood that the above-mentioned embodiments are only examples of the present invention and are not intended to limit the scope of the present invention. It should be understood that any modifications, equivalents, improvements and the like, which come within the spirit and principle of the invention, may occur to those skilled in the art and are intended to be included within the scope of the invention.