CN104052612A - Method and system for fault identification and positioning of telecommunication service - Google Patents
Method and system for fault identification and positioning of telecommunication service Download PDFInfo
- Publication number
- CN104052612A CN104052612A CN201310080584.3A CN201310080584A CN104052612A CN 104052612 A CN104052612 A CN 104052612A CN 201310080584 A CN201310080584 A CN 201310080584A CN 104052612 A CN104052612 A CN 104052612A
- Authority
- CN
- China
- Prior art keywords
- feature value
- grader
- service feature
- fault
- failure
- 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.)
- Granted
Links
Landscapes
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention provides a method and system for fault identification and positioning of a telecommunication service. The method comprises: according to a correlation coefficient, representing a fault source, between a service feature value and a fault phenomenon, service feature values that are used as ones for back propagation (BP) neutral network training are selected form sample data; a weak classifier based on the BP neutral network is determined for each selected service feature value and the weak classifiers are processed linear combination to obtain corresponding strong classifiers; according to the value order of correlation coefficients between all service feature values and fault phenomena, all strong classifiers corresponding to all service feature values are connected in series and then are trained so as to obtain a layer classifier for multi-fault source identification; and fault identification and positioning are carried out on the service data and/or network data generated during the telecommunication service process by using the layer classifier. According to the invention, the training sample amount for the neutral network can be reduced; the fault detection real-time performance is improved; and the fault factor can be localized conveniently.
Description
Technical field
The present invention relates to core net, webmaster and business support field in communication network, be specifically related to a kind of Fault Identification of telecommunication service and the method and system of location.
Background technology
Along with the development of telecommunications industry and information services industry, telecommunication traffic is rapid growth also, the number of faults occurring also will promote thereupon, and this becomes a huge obstacle of the stickiness between communication service and the user who promotes user satisfaction, strengthens company of operator.
Current, an important step that solves fault is to carry out Fault Identification and location, determines fault.The common implementation method of this step is: manual identified fault, according to business datum and maintenance staff's experience, judge which aspect exists fault, and the flow process of concrete judgement is illustrated in fig. 1 shown below.
In Fig. 1, OCS is Online Charging System (Online Charging System); HLR is home location register system (Host Location Register); BOSS is Business Operating Support System.Above-mentioned three systems are referred to as operation system.The data that produce in operation system running are sampled, can obtain business datum.In the time that operation system breaks down, current processing mode is: the contact staff on foreground receives after the fault message that client submits to, give background maintenance personnel by troubleshooting task, the information that background maintenance personnel provide according to client and foreground contact staff is confirmed fault, and locate the reason of fault, then deal with problems.This is a kind of passive, pattern of postponing, must be after customer complaint or foreground could find fault when the relevant information that reports an error is provided.Obviously, the mode of this handling failure just can not be timely, thereby have influence on client's satisfaction, reduces the degree of recognition of user to operator's service.
Neural net is called again artificial neural net, neural calculating, connectionism artificial intelligence, parallel distributed processing etc.A neural net is a large-scale parallel distributed processor being made up of simple process unit, has storage Heuristics and makes it available characteristic.By adopting the neural net shown in Fig. 2, can detect in real time fault, to shorten the time of processing when fault occurs.
In Fig. 2, the data that system gathers from OCS, HLR, BOSS and network, be input in real time in trained neural net by neural net input layer, according to the kind of the fault of prediction, the input that the different input neuron of output layer produces shows the different system failures.
The Fault Identification mode of the employing manual identified shown in Fig. 1 be a kind of passive, postpone pattern, must be after customer complaint or foreground just go to find fault while having relevant reporting an error, handling failure just can not be timely so like this, thereby affect the approval of user to service.Fault Model based on neural net shown in Fig. 2 exists following problem:
(1) all faults all detect in a neural net, cause the sample data amount for training very big, and this has hindered the practicality of this class model.
Input for detection of the neural net of fault has accessed all system service datas, and input produces the fault-signal of all representing fault phenomenons.In the time of training, the generation of any class phenomenon all needs the data of all inputs, from the dimension of input and output, the needed sample data amount of fault of determining any class is all very big, this makes to train such a detection neural net needs a large amount of sample datas, this is for a fault detection system in first stage of construction and the travelling mechanism of just paying close attention to fault detect, and difficulty is larger.
(2) for the neural net training, due to its complex structure, to calculate all kinds of faults simultaneously, cause amount of calculation to have greatly increase, this makes the time delay of detection in real time larger.
The object of fault detect is to shorten the response time of fault as far as possible, therefore, comprises that the real-time of fault detect of Fault Identification and two processes of fault location is extremely important.If neural network structure complexity, calculative complexity increases, and will extend computing time, and this will be unfavorable for shortening the time of Fault Identification and fault location.
(3) between phenomenon of the failure (output) and failure cause (input), do not set up clear and definite relation for the detection neural net of identifying and locate fault, cause being like this not easy to locate failure cause.
Summary of the invention
In view of this, the object of the embodiment of the present invention is to provide a kind of Fault Identification of telecommunication service and the method and system of location, can reduce the required training sample amount of neural net, improves fault detect real-time, and is convenient to locate failure cause.
For solving the problems of the technologies described above, the embodiment of the present invention provides scheme as follows:
The embodiment of the present invention provides a kind of Fault Identification of telecommunication service and the method for location, comprising:
According to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, from sample data, select coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer BP neural metwork training;
For each service feature value of having selected is determined a Weak Classifier based on BP neural net, and determined Weak Classifier is carried out to linear combination, obtain a strong classifier corresponding to this service feature value;
According to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source;
Utilize described layer grader to carry out Fault Identification and location to the business datum producing in telecommunication service process and/or network data.
Further, in such scheme, the described described layer grader that utilize carries out Fault Identification and location to the business datum producing in telecommunication service process and/or network data, comprise: by the business datum producing in telecommunication service process and/or the described layer of network data input grader, the output producing according to each strong classifier, location failure cause.
Further, in such scheme, described is that each service feature value of having selected is determined a Weak Classifier based on BP neural net, comprising:
For each service feature value of having selected, set up a Weak Classifier based on BP neural net;
Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Further, in such scheme, the sample data that described employing comprises multiple faults source data and phenomenon of the failure is trained the strong classifier after series connection respectively, obtains the layer grader for the identification of multiple faults source, comprising:
Employing comprises the sample data of multiple faults source data and phenomenon of the failure, strong classifier after series connection is trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Further, in such scheme, further select the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, strong classifier corresponding selected service feature value is carried out to described series connection.
The embodiment of the present invention also provides a kind of Fault Identification of telecommunication service and the system of location, comprising:
Eigenvalue unit for according to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, selects coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer BP neural metwork training from sample data;
Single fault source recognition unit, is used to each service feature value of having selected to determine a Weak Classifier based on BP neural net, and determined Weak Classifier is carried out to linear combination, obtains a strong classifier corresponding to this service feature value;
Multiple faults source recognition unit, be used for according to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source;
Service Processing Unit, carries out Fault Identification and location for business datum and/or the network data of utilizing described layer grader to produce telecommunication service process.
Further, in such scheme, described Service Processing Unit, further by the business datum producing in telecommunication service process and/or the described layer of network data input grader, the output producing according to each strong classifier, location failure cause.
Further, in such scheme, described single fault source recognition unit, is further used for each service feature value for having selected, and sets up a Weak Classifier based on BP neural net; Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Further, in such scheme, described multiple faults source recognition unit, be further used for adopting the sample data that comprises multiple faults source data and phenomenon of the failure, the strong classifier after series connection trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Further, in such scheme, described multiple faults source recognition unit, is further used for selecting the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, and strong classifier corresponding selected service feature value is carried out to described series connection.
Can find out from the above, the Fault Identification of the telecommunication service that the embodiment of the present invention provides and the method and system of location, can identify fault location accurately, can reduce the workload of artificial inspection business datum.The embodiment of the present invention also has that training sample amount is little, fault detect real-time is high and the advantage such as failure cause location is quick.
Brief description of the drawings
Fig. 1 is the schematic diagram of manual identified fault in prior art;
Fig. 2 is the schematic diagram that application neural net is carried out Fault Identification;
Fig. 3 is the schematic flow sheet of method described in the embodiment of the present invention;
Fig. 4 is the structural representation of system described in the embodiment of the present invention;
Fig. 5 is the schematic diagram of the stacked neural net integrated morphology in the embodiment of the present invention;
Fig. 6 for the embodiment of the present invention provide layer grader structural representation.
Embodiment
Evading the experience of (fault solution) from existing fault, the most easily reduce the link of user satisfaction, is also that the link of spended time is fault location, can not find rapidly the reason of fault, Failure elimination.Therefore,, if can recognize at short notice fault, location fault, this will shorten the time that solves fault greatly, improves user's satisfaction.Meanwhile, if establish clear and definite relation between phenomenon of the failure and failure cause, for shortening fault settling time, also have very important significance.
Based on above analysis, the embodiment of the present invention provides a kind of Fault Identification of telecommunication service and the method and system of location, can reduce required training sample amount, improves fault detect real-time, and is convenient to locate failure cause.The method of the embodiment of the present invention, can initiatively go to analyze existing business datum or network data, thereby can find fault, and determine failure cause fast, to help attendant to solve rapidly failure problems.Here, business datum is that network data is the data that produce in communication network (as Access Network, core net etc.) running from the data of operation system output.
For making the object, technical solutions and advantages of the present invention clearer, describe the present invention below in conjunction with the accompanying drawings and the specific embodiments.
Please refer to Fig. 3, the Fault Identification of telecommunication service and the method for location described in the embodiment of the present invention, comprise the following steps:
Step 31, according to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, from sample data, select coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer (BP, Back Propagation) neural metwork training.
In this step, by coefficient correlation, sample data is carried out to dimension-reduction treatment, choose suitable characteristic value and carry out subsequent calculations, to reduce amount of calculation, improve computational speed.
Step 32, according to AdaBoost algorithm, for each service feature value of having selected is determined a Weak Classifier based on BP neural net, and carries out linear combination by determined Weak Classifier, obtains a strong classifier corresponding to this service feature value.
In this step, for each service feature value of having selected, set up a Weak Classifier based on BP neural net; Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader; Then, the Weak Classifier after training is carried out to linear combination, obtain a strong classifier.
Step 33, according to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source.
In this step, further select the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, strong classifier corresponding selected service feature value is carried out to described series connection.Then, comprise the sample data of multiple faults source data and phenomenon of the failure by employing, strong classifier after series connection is trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Step 34, utilizes described layer grader to carry out Fault Identification and location to the business datum producing in telecommunication service process and/or network data.
In this step, by by the business datum producing in telecommunication service process and/or the described layer of network data input grader, then according to the output of each strong classifier generation, location failure cause.
Can find out from above method, the embodiment of the present invention is by correlation analysis, to training and the Data Dimensionality Reduction that detects, reduced the requirement to training sample quantity, makes the fault detection technique that proposes in the embodiment of the present invention pratical and feasible.And the embodiment of the present invention, by the design arrangement of the tracer of layering, resolves into simple calculating by complicated calculating, has improved the real-time of fault detect.Finally, the embodiment of the present invention provides a kind of layer grader of hierarchy, and the output of each strong classifier of this layer of grader is corresponding to a fault, like this, in the time that certain strong classifier produces output, can navigate to corresponding failure cause, improve the speed of failure cause location.
Based on above method, the embodiment of the present invention also provides a kind of Fault Identification of telecommunication service and the system of location accordingly, and as shown in Figure 4, this system comprises:
Eigenvalue unit for according to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, selects coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer BP neural metwork training from sample data;
Single fault source recognition unit, be used for according to AdaBoost algorithm, for each service feature value of having selected is determined a Weak Classifier based on BP neural net, and determined Weak Classifier is carried out to linear combination, obtain a strong classifier corresponding to this service feature value;
Multiple faults source recognition unit, be used for according to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source;
Service Processing Unit, carries out Fault Identification and location for business datum and/or the network data of utilizing described layer grader to produce telecommunication service process.
Preferably, described Service Processing Unit, further by the business datum producing in telecommunication service process and/or the described layer of network data input grader, the output producing according to each strong classifier, location failure cause.
Preferably, described single fault source recognition unit, is further used for each service feature value for having selected, and sets up a Weak Classifier based on BP neural net; Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Preferably, described multiple faults source recognition unit, be further used for adopting the sample data that comprises multiple faults source data and phenomenon of the failure, strong classifier after series connection is trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
Preferably, described multiple faults source recognition unit, is further used for selecting the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, and strong classifier corresponding selected service feature value is carried out to described series connection.
For understanding better above method, by by more detailed explanation, specific embodiments of the invention are further elaborated below.
Please refer to Fig. 5, Fig. 5 provides the embodiment of the present invention for the identification of fault and the stacked neural net integrated morphology of location.This neural net integrated morphology mainly realizes three functions: the correlation of (1) analysis of failure source and phenomenon of the failure data, and a point phenomenon of the failure is determined sample data; (2) neural network training (grader) is for the identification of single fault source (single characteristic value); (3) integrated neural network, forms stacked grader, for the identification of multiple faults source (multiple characteristic values).Below describe respectively.
(1) correlation analysis, determines the input feature vector value of sample data
The characteristic value (dimension) of sample data is numerous, and in order to reduce the amount of calculation of neural net, the embodiment of the present invention is carried out dimension-reduction treatment to the input of sample data, selects suitable characteristic value.
As a kind of preferred implementation, adopt the correlation analysis of following formula to select characteristic value.
In formula, X is characteristic value to be selected, and Y is the output valve relevant with fault.For example, the remaining space that X is server hard disc and the ratio of normed space, the number of times that Y is interior inquiry per second.Subscript i represents different time points, and N represents the quantity of sample data sampling,
represent X
imean value,
represent Y
imean value, r
xYrepresent the correlation between X and Y.
In this step, the present embodiment is selected r
xYbe 0.5 as threshold value, select to be more than or equal to 0.5 characteristic value, enter follow-up training and calculate.
(2) single fault source identifier
In the present embodiment, adopt BP neural net, setting up one has the network model of three layers: input layer, hidden layer, output layer.First, the business datum as sample data or the network data that obtain in advance in input layer input, the effect of the action function of process hidden layer, just can output valve at output, in the present embodiment, output valve only has two kinds: 1 and 0, wherein 1 represent that these data have fault, 0 representative does not have, then calculate error according to error computing function, compared with error expected, if do not reach the requirement of error expected, the information of output is just now fed back to input layer and hidden layer, thereby act on input layer and hidden layer, regulate weights and relevant function, reach the effect of optimizing output valve.
According to the integrated theory of neural net, by the multiple neural nets of independent training, and its result is synthesized, can improve significantly the generalization ability of neural net, in the present embodiment, this generalization ability is Fault Identification ability.Here the result synthetic method adopting is AdaBoost algorithm, is each characteristic value and determines a Weak Classifier, then these Weak Classifiers is carried out to linear combination, has just formed a kind of strong classifier.In this algorithm, each service feature value is endowed weights, if certain sample is not correctly classified, its weights will be enhanced, and can reduce on the contrary, and its arthmetic statement is as follows:
1) for selected characteristic value, obtain the sample of the positive and negative business datum of some, as the sample data of training classifier, supposition here has positive and negative data sample respectively to have some.
2) weight w of initialization training classifier combination, by positive and negative business datum sample according to default different proportion (for example, 10:1,10:2,10:3 ..., 10:10) carry out the combination of the multiple samples of sample combination producing.Conventionally, negative sample is few more a lot of than positive sample, and therefore, the positive sample of every class sample combination does not reduce, and the generation of negative sample can adopt the mode that repeats to generate to carry out sample size polishing;
3) for a grader (BP neural network algorithm) is set up in each sample combination, for each sample combination, carry out pre-determined number (as 100 training), then, the grader that executes training is carried out to simulation error estimation, and error is: training simulation output and the actual difference of exporting;
4) select all samples to combine the grader of the predetermined number of error minimum in corresponding grader, for example 3 of Select Error minimum graders;
5) using the simulation correct recognition rata of grader as combining weights, the grader of above-mentioned selection is carried out to linear combination, obtain strong classifier corresponding to this service feature value.
Just can obtain a strong classifier through process above, whether fault of these data that what this grader can be very fast tell, has improved the efficiency of Fault Identification, but can only be for the situation of fault with single features value.
(3) multiple faults source identifier
Obtain single strong classifier above, and in the time that the quantity of characteristic value exceedes one, cannot adopt single fault source identifier above.Therefore, the present embodiment provides the model of a kind of layer of grader as shown in Figure 6, and this layer of grader is by the mode of series connection, multiple neural nets (strong classifier) above to be combined.Arrange the position of each neural net in layer grader according to coefficient correlation.Adopt all source of trouble X and the phenomenon of the failure Y of multiple faults source data and normal data respectively each neural net to be trained.Concrete steps are:
1) step of the correlation analysis in employing (1), selection is greater than all relevant with a certain phenomenon of the failure Y characteristic value X of correlation coefficient threshold 0.5; A phenomenon of the failure may cause by several sources of trouble, and a source of trouble is corresponding to a characteristic value; The a certain phenomenon of the failure Y here can be optional from the phenomenon of the failure of needs location.
2) by the size order of the coefficient correlation between each characteristic value and selected phenomenon of the failure Y, the neural net corresponding source of trouble represented each characteristic value is entered in a layer grader (as shown in Figure 6); The size of coefficient correlation has reflected the size of the possibility of the source of trouble that causes phenomenon of the failure; The neural net that coefficient correlation is large, sequence is in the position that more approaches input (position of more keeping right in as Fig. 6);
3) by sample data, neural net is trained respectively the process of training and (2) single fault source identifier 1), 2) and 3) step is identical.
By such arrangement, every one deck fault grader just can be got rid of a source of trouble, the grader (sequence is the closer to the grader of input position) that level is higher, the possibility that its characteristic of correspondence value causes fault is larger, therefore, this class combination can detect and locate fault efficiently, ground floor grader has only been used one just can filter out the information that is not fault, then successively the remaining business datum not filtering out is gone to carry out through other grader, the output location failure cause (source of trouble) that so just can identify very efficiently fault and produce by neural net.
The conclusion drawing through experiment: use this hierarchy on average can improve several times than the efficiency of the identification fault of the single Neural structure without layering.
To sum up, the embodiment of the present invention is according to the demand of the fault location of telecommunication service and feature thereof, the cascade pattern of grader has been proposed, in the time of the choosing of service feature value, the embodiment of the present invention will be done association analysis between the phenomenon of the failure in business datum and the data of the source of trouble (failure cause) representative, determines the investigation scope (selecting coefficient correlation to be greater than the characteristic value of threshold value) of fault, thereby can reduce dimension, reduce the computation complexity of subsequent calculations, improve computational speed; And the embodiment of the present invention adopts the hierarchy schema of grader, the advantage of hierarchy is: (1) treatment effeciency is high.The treatment effeciency of hierarchy grader is higher than the grader being made up of single Neural, through experimental verification: input identical characteristic value, the efficiency of the sorter model of hierarchy schema can improve several times; (2) locate fast fault: the output by neural net in hierarchy can be located the source of trouble.
The above is only embodiments of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.
Claims (10)
1. the Fault Identification of telecommunication service and a method for location, is characterized in that, comprising:
According to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, from sample data, select coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer BP neural metwork training;
For each service feature value of having selected is determined a Weak Classifier based on BP neural net, and determined Weak Classifier is carried out to linear combination, obtain a strong classifier corresponding to this service feature value;
According to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source;
Utilize described layer grader to carry out Fault Identification and location to the business datum producing in telecommunication service process and/or network data.
2. the method for claim 1, it is characterized in that, the described described layer grader that utilize carries out Fault Identification and location to the business datum producing in telecommunication service process and/or network data, comprise: by the business datum producing in telecommunication service process and/or the described layer of network data input grader, the output producing according to each strong classifier, location failure cause.
3. the method for claim 1, is characterized in that, described is that each service feature value of having selected is determined a Weak Classifier based on BP neural net, comprising:
For each service feature value of having selected, set up a Weak Classifier based on BP neural net;
Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
4. the method for claim 1, is characterized in that, the sample data that described employing comprises multiple faults source data and phenomenon of the failure is trained the strong classifier after series connection respectively, obtains the layer grader for the identification of multiple faults source, comprising:
Employing comprises the sample data of multiple faults source data and phenomenon of the failure, strong classifier after series connection is trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
5. the method for claim 1, is characterized in that, further selects the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, and strong classifier corresponding selected service feature value is carried out to described series connection.
6. the Fault Identification of telecommunication service and a system for location, is characterized in that, comprising:
Eigenvalue unit for according to the coefficient correlation between the service feature value in representing fault source and phenomenon of the failure, selects coefficient correlation to be greater than the service feature value of predetermined threshold value, as the service feature value of back transfer BP neural metwork training from sample data;
Single fault source recognition unit, is used to each service feature value of having selected to determine a Weak Classifier based on BP neural net, and determined Weak Classifier is carried out to linear combination, obtains a strong classifier corresponding to this service feature value;
Multiple faults source recognition unit, be used for according to the size order of the coefficient correlation between each service feature value and phenomenon of the failure, by strong classifier series connection corresponding each service feature value, and adopt the sample data that comprises multiple faults source data and phenomenon of the failure to train the strong classifier after connecting, obtain the layer grader for the identification of multiple faults source;
Service Processing Unit, carries out Fault Identification and location for business datum and/or the network data of utilizing described layer grader to produce telecommunication service process.
7. system as claimed in claim 6, is characterized in that, described Service Processing Unit, further by the business datum producing in telecommunication service process and/or the described layer of network data input grader, the output producing according to each strong classifier, location failure cause.
8. system as claimed in claim 6, is characterized in that, described single fault source recognition unit is further used for each service feature value for having selected, and sets up a Weak Classifier based on BP neural net; Adopt the sample data that the comprises positive and negative business datum combination obtaining in advance, train the described Weak Classifier identification form source of trouble, until reach predetermined frequency of training, obtain the Weak Classifier after training, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
9. system as claimed in claim 6, it is characterized in that, described multiple faults source recognition unit, be further used for adopting the sample data that comprises multiple faults source data and phenomenon of the failure, the strong classifier after series connection trained, until reach predetermined frequency of training, obtain described layer grader, wherein, rear estimation training objective error is once trained in every execution, selects the grader of error minimum wherein and reduces the weights of this grader.
10. system as claimed in claim 6, it is characterized in that, described multiple faults source recognition unit, is further used for selecting the coefficient correlation between phenomenon of the failure to be greater than the service feature value of predetermined threshold value, and strong classifier corresponding selected service feature value is carried out to described series connection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310080584.3A CN104052612B (en) | 2013-03-13 | 2013-03-13 | A kind of Fault Identification of telecommunication service and the method and system of positioning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310080584.3A CN104052612B (en) | 2013-03-13 | 2013-03-13 | A kind of Fault Identification of telecommunication service and the method and system of positioning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104052612A true CN104052612A (en) | 2014-09-17 |
| CN104052612B CN104052612B (en) | 2017-08-25 |
Family
ID=51504997
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310080584.3A Active CN104052612B (en) | 2013-03-13 | 2013-03-13 | A kind of Fault Identification of telecommunication service and the method and system of positioning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104052612B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105554786A (en) * | 2015-12-11 | 2016-05-04 | 中国联合网络通信集团有限公司 | Method and device for positioning network problem |
| CN106713016A (en) * | 2016-12-07 | 2017-05-24 | 中国联合网络通信集团有限公司 | Fault reasoning method and apparatus of indoor distribution system |
| CN107067519A (en) * | 2017-05-12 | 2017-08-18 | 广东工业大学 | A kind of safety box based on the electric identification of brain |
| CN107169434A (en) * | 2017-05-10 | 2017-09-15 | 广东工业大学 | One kind possesses the electric personal identification method of exclusive brain |
| CN107370618A (en) * | 2017-06-05 | 2017-11-21 | 北京奇艺世纪科技有限公司 | Malfunction elimination method, apparatus and electronic equipment |
| CN108696379A (en) * | 2017-04-07 | 2018-10-23 | 南京航空航天大学 | Cellular network fault diagnosis system based on integrated study and SMOTE |
| EP3396558A4 (en) * | 2015-12-23 | 2018-10-31 | Tencent Technology Shenzhen Company Limited | Method for user identifier processing, terminal and nonvolatile computer readable storage medium thereof |
| CN110716127A (en) * | 2019-10-29 | 2020-01-21 | 恩平市高尔电子科技有限公司 | Microphone receiver circuit board detecting system |
| CN112129869A (en) * | 2020-09-23 | 2020-12-25 | 清华大学深圳国际研究生院 | Data-driven field mass spectrometer stable output control system and method |
| CN113490857A (en) * | 2019-02-06 | 2021-10-08 | 利萨·德雷克塞迈尔有限责任公司 | Method and test device |
| US11196638B2 (en) | 2015-11-05 | 2021-12-07 | Huawei Technologies Co., Ltd. | Network event prediction method and apparatus and method and apparatus for establishing network-event prediction model |
| CN113887414A (en) * | 2021-09-30 | 2022-01-04 | 北京百度网讯科技有限公司 | Target detection method, target detection device, electronic equipment and storage medium |
| CN116805926A (en) * | 2023-08-21 | 2023-09-26 | 上海飞旗网络技术股份有限公司 | Network service type identification model training method and network service type identification method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102957579A (en) * | 2012-09-29 | 2013-03-06 | 北京邮电大学 | Network anomaly traffic monitoring method and device |
-
2013
- 2013-03-13 CN CN201310080584.3A patent/CN104052612B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102957579A (en) * | 2012-09-29 | 2013-03-06 | 北京邮电大学 | Network anomaly traffic monitoring method and device |
Non-Patent Citations (1)
| Title |
|---|
| HAN-YONGZHONG等: "Research of Identification and Positioning System for Network Failure based on BP neural network", 《IEEE》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11196638B2 (en) | 2015-11-05 | 2021-12-07 | Huawei Technologies Co., Ltd. | Network event prediction method and apparatus and method and apparatus for establishing network-event prediction model |
| CN105554786B (en) * | 2015-12-11 | 2019-08-09 | 中国联合网络通信集团有限公司 | A kind of method and device of addressing network problems |
| CN105554786A (en) * | 2015-12-11 | 2016-05-04 | 中国联合网络通信集团有限公司 | Method and device for positioning network problem |
| EP3396558A4 (en) * | 2015-12-23 | 2018-10-31 | Tencent Technology Shenzhen Company Limited | Method for user identifier processing, terminal and nonvolatile computer readable storage medium thereof |
| CN106713016A (en) * | 2016-12-07 | 2017-05-24 | 中国联合网络通信集团有限公司 | Fault reasoning method and apparatus of indoor distribution system |
| CN108696379A (en) * | 2017-04-07 | 2018-10-23 | 南京航空航天大学 | Cellular network fault diagnosis system based on integrated study and SMOTE |
| CN107169434A (en) * | 2017-05-10 | 2017-09-15 | 广东工业大学 | One kind possesses the electric personal identification method of exclusive brain |
| CN107067519B (en) * | 2017-05-12 | 2023-02-03 | 广东工业大学 | Safe box based on brain electricity identification |
| CN107067519A (en) * | 2017-05-12 | 2017-08-18 | 广东工业大学 | A kind of safety box based on the electric identification of brain |
| CN107370618A (en) * | 2017-06-05 | 2017-11-21 | 北京奇艺世纪科技有限公司 | Malfunction elimination method, apparatus and electronic equipment |
| CN113490857B (en) * | 2019-02-06 | 2024-04-02 | 利萨·德雷克塞迈尔有限责任公司 | Method and test device |
| CN113490857A (en) * | 2019-02-06 | 2021-10-08 | 利萨·德雷克塞迈尔有限责任公司 | Method and test device |
| CN110716127A (en) * | 2019-10-29 | 2020-01-21 | 恩平市高尔电子科技有限公司 | Microphone receiver circuit board detecting system |
| CN112129869B (en) * | 2020-09-23 | 2022-11-18 | 清华大学深圳国际研究生院 | Data-driven field mass spectrometer stable output control system and method |
| CN112129869A (en) * | 2020-09-23 | 2020-12-25 | 清华大学深圳国际研究生院 | Data-driven field mass spectrometer stable output control system and method |
| CN113887414A (en) * | 2021-09-30 | 2022-01-04 | 北京百度网讯科技有限公司 | Target detection method, target detection device, electronic equipment and storage medium |
| CN116805926A (en) * | 2023-08-21 | 2023-09-26 | 上海飞旗网络技术股份有限公司 | Network service type identification model training method and network service type identification method |
| CN116805926B (en) * | 2023-08-21 | 2023-11-17 | 上海飞旗网络技术股份有限公司 | Network service type identification model training method and network service type identification method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104052612B (en) | 2017-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104052612A (en) | Method and system for fault identification and positioning of telecommunication service | |
| CN108520357B (en) | Method and device for judging line loss abnormality reason and server | |
| CN105930277B (en) | A kind of defect source code localization method based on defect report analysis | |
| CN106470122B (en) | A kind of network failure locating method and device | |
| CN104331635B (en) | The method of power optical fiber Communication ray power prediction | |
| CN110542819B (en) | Transformer fault type diagnosis method based on semi-supervised DBNC | |
| CN109117941A (en) | Alarm prediction method, system, storage medium and computer equipment | |
| CN103514566A (en) | Risk control system and method | |
| CN107967485A (en) | Electro-metering equipment fault analysis method and device | |
| CN103926526A (en) | Analog circuit fault diagnosis method based on improved RBF neural network | |
| CN112051480A (en) | Neural network power distribution network fault diagnosis method and system based on variational modal decomposition | |
| CN112433927A (en) | Cloud server aging prediction method based on time series clustering and LSTM | |
| CN106327357A (en) | Load identification method based on improved probabilistic neural network | |
| CN112308341A (en) | Power data processing method and device | |
| CN109829627A (en) | A kind of safe confidence appraisal procedure of Electrical Power System Dynamic based on integrated study scheme | |
| CN109242165A (en) | A kind of model training and prediction technique and device based on model training | |
| CN106251861A (en) | A kind of abnormal sound in public places detection method based on scene modeling | |
| CN115422788A (en) | Power distribution network line loss analysis management method, device, storage medium and system | |
| CN113393316B (en) | Loan overall process accurate wind control and management system based on massive big data and core algorithm | |
| CN115878992A (en) | Monitoring method and monitoring system for comprehensive pipe rack power supply system | |
| Gu et al. | Investigation on the quality assurance procedure and evaluation methodology of machine learning building energy model systems | |
| CN116611911A (en) | Credit risk prediction method and device based on support vector machine | |
| CN116225752A (en) | Fault root cause analysis method and system for micro-service system based on fault mode library | |
| CN116228312A (en) | Processing method and device for large-amount point exchange behavior | |
| CN109726824A (en) | The transfer learning method and terminal device of training pattern |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |