CN109547248A - Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device - Google Patents
Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device Download PDFInfo
- Publication number
- CN109547248A CN109547248A CN201811387631.8A CN201811387631A CN109547248A CN 109547248 A CN109547248 A CN 109547248A CN 201811387631 A CN201811387631 A CN 201811387631A CN 109547248 A CN109547248 A CN 109547248A
- Authority
- CN
- China
- Prior art keywords
- key performance
- information vector
- performance index
- network fault
- diagnosis result
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 66
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 17
- 239000013598 vector Substances 0.000 claims abstract description 75
- 238000010606 normalization Methods 0.000 claims abstract description 27
- 238000003745 diagnosis Methods 0.000 claims description 59
- 238000007781 pre-processing Methods 0.000 claims description 18
- 238000013507 mapping Methods 0.000 claims description 4
- 238000012423 maintenance Methods 0.000 claims description 3
- 238000002405 diagnostic procedure Methods 0.000 abstract description 2
- 210000002569 neuron Anatomy 0.000 description 27
- 230000006870 function Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 208000024891 symptom Diseases 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000001276 Kolmogorov–Smirnov test Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004092 self-diagnosis Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/318—Received signal strength
- H04B17/327—Received signal code power [RSCP]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/336—Signal-to-interference ratio [SIR] or carrier-to-interference ratio [CIR]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Theoretical Computer Science (AREA)
- Electromagnetism (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Environmental & Geological Engineering (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention discloses based on artificial intelligence, in orbit aerocraft ad hoc network method for diagnosing faults and device, diagnostic method includes step S1: acquiring the raw information vector of Key Performance Indicator；S2: normalization pretreatment；S3: match decision is carried out to the Key Performance Indicator record normalized in pretreated information vector and Key Performance Indicator database, for the information vector for characterizing unknown Key Performance Indicator, Unsupervised clustering and label setting are successively carried out, new Key Performance Indicator is generated and records and be stored in Key Performance Indicator database；S4: it is recorded according to critical index corresponding to the information vector and determines network failure condition diagnosing result.The present invention can solve aircraft networks failure in the prior art and be difficult to the technical issues of independently diagnosing.
Description
Technical Field
The invention relates to the field of space networks, in particular to an on-orbit aircraft ad hoc network fault diagnosis method and device based on artificial intelligence.
Background
The fault diagnosis is a key function in the self-healing network, and the automatic fault identification of the system is accurate and reliable. Due to the characteristics of the operation orbit, a space network formed by the on-orbit aircraft often needs to make corresponding decision judgment without human intervention, and particularly, self-diagnosis and self-repair are carried out on network faults. In general, these symptoms are closely related to Key Performance Indicators (KPIs) of the system. Therefore, it is necessary to diagnose the occurrence of a failure according to the degradation of a specific symptom and the degree of its deterioration. Automation of the diagnostic process means that the diagnostic system must know the behavior of the fault. One possible approach is to extract information from the stored cases that contain the resolved fault. This data set will allow an automatic system to be obtained by supervised learning. However, the complexity of the space environment tends to make available historian features scarce due to the operation of the on-orbit aircraft, and the historical data obtained from the ground wireless network is insufficient to build a space network diagnostic system with supervised technology. Therefore, there is a need for autonomous diagnosis of in-orbit aircraft networks in conjunction with Artificial Intelligence (AI) technology.
Although existing research gives solutions for system automatic diagnostics, it is mainly for terrestrial wireless systems and is based on expert knowledge and historical data of fault cases. On one hand, the elimination of network faults of the on-orbit aircraft is difficult to construct a complex data model through special knowledge; on the other hand, even though the historical data information of KPIs can be obtained through on-track operation for a period of time, how to reveal the internal relation between the network fault state and the influencing factors is a key technical problem to be solved.
Disclosure of Invention
The invention aims to provide an on-orbit aircraft ad hoc network fault diagnosis method and device based on artificial intelligence, so as to solve the technical problem that the network fault of an aircraft is difficult to diagnose autonomously in the prior art.
In order to solve the problems, the invention provides an on-orbit aircraft ad hoc network fault diagnosis method based on artificial intelligence, which comprises the following steps of:
s1: collecting original information vectors of key performance indexes;
s2: carrying out normalization preprocessing on the original information vector;
s3: matching decision is carried out on the information vector after normalization preprocessing and key performance index records in a key performance index database so as to judge whether the information vector represents known key performance indexes or unknown key performance indexes, if the information vector represents the unknown key performance indexes, unsupervised clustering and label setting are carried out on the information vector in sequence, and new key performance index records are generated and stored in the key performance index database;
s4: and determining a network fault state diagnosis result according to the key index record corresponding to the information vector.
Preferably, in step S3, the specific step of determining whether the information vector represents a known key performance indicator or an unknown key performance indicator is as follows:
if only one key performance index record with the distance from the information vector smaller than the threshold value exists in the key performance index database, the information vector represents the known key performance index; otherwise, the information vector represents an unknown key performance index.
Preferably, the method for determining the network fault state diagnosis result specifically includes: if only one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result directly according to the key performance index record;
and if more than one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result by combining a nearest neighbor method and a quantile method.
Preferably, the method for determining the network fault state diagnosis result by combining the nearest neighbor method and the quantile method specifically comprises the following steps: if the first diagnostic result determined by the nearest neighbor method is not equal to the second diagnostic result determined by the quantile method, and the average contour coefficient of the first diagnostic result is smaller than the average contour coefficient of the second diagnostic result, the final diagnostic result is the second diagnostic result; otherwise, the final diagnosis result is the first diagnosis result.
Preferably, the unsupervised clustering adopts a self-organizing mapping algorithm to perform rough classification, and performs successive merging on the classes with the minimum inter-class distance based on the Ward hierarchy method, thereby realizing fine clustering.
Preferably, the normalization preprocessing adopts an interval normalization method or a standard deviation method.
Preferably, the key performance indicators include link maintenance, handover success rate, received reference signal power, received reference signal quality, signal to interference and noise ratio, user average throughput and distance.
The invention also provides an on-orbit aircraft ad hoc network fault diagnosis device based on artificial intelligence, which is characterized by comprising
The normalization preprocessing unit is used for carrying out normalization preprocessing on the original information vector;
the learning unit is used for carrying out matching decision on the information vector after normalization preprocessing and key performance index records in a key performance index database so as to judge whether the information vector represents a known key performance index or an unknown key performance index, and if the information vector represents an unknown key performance index, carrying out unsupervised clustering and label setting on the information vector in sequence to generate a new key performance index record and storing the new key performance index record in the key performance index database;
and the diagnosis unit is used for determining the network fault state diagnosis result according to the key index record corresponding to the information vector.
Compared with the prior art, the invention has the following technical effects:
1. the method is based on the artificial intelligence technology to carry out (non-) supervised learning on the data acquired by the aircraft, realizes the fault autonomous diagnosis of the on-orbit aircraft in the scene without human intervention, and improves the reliability and robustness of the network.
2. The method carries out unsupervised clustering on the unknown key performance index information, enriches the records of the database and lays a foundation for further mining and analyzing historical data; the existing key performance indexes are subjected to deep training clustering, and the precision of the diagnosis result is improved.
3. The invention integrates a plurality of neuron information through the contour controller, thereby avoiding the influence of the clustered classified boundary neurons on the diagnosis result.
Of course, it is not necessary for any product in which the invention is practiced to achieve all of the above-described advantages at the same time.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts. In the drawings:
FIG. 1 is a flowchart of an in-orbit aircraft ad hoc network fault diagnosis method based on artificial intelligence according to an embodiment of the present invention;
FIG. 2 is a flow chart of matching decision, unsupervised clustering and label setting of an on-orbit aircraft ad hoc network fault diagnosis method based on artificial intelligence in the embodiment of the invention;
fig. 3 is a flowchart of a network fault state diagnosis result determination method based on an artificial intelligence ad hoc network fault diagnosis method in an embodiment of the present invention.
Detailed Description
The method and the device for diagnosing the ad hoc network fault of the on-orbit aircraft based on artificial intelligence provided by the invention will be described in detail with reference to the accompanying drawings, the embodiment is implemented on the premise of the technical scheme of the invention, and a detailed implementation manner and a specific operation process are given, but the protection scope of the invention is not limited to the following embodiment, and a person skilled in the art can modify and revise the method and the device within the scope of not changing the spirit and content of the invention.
Example one
Referring to fig. 1, the method for diagnosing the ad hoc network fault of the on-orbit aircraft based on artificial intelligence mainly includes three stages: the method comprises an input stage, a processing stage and an output stage, and specifically comprises the following steps:
s1: collecting original information vector of Key Performance Indicator (KPI), and recording as x ═ x1,…,xM]；
S2: carrying out normalization pretreatment on the original information vector, wherein the data vector after normalization pretreatment is recorded as
S3: information vector and key performance after normalization preprocessingCarrying out matching decision on key performance index records in an index database to judge whether the information vector represents known key performance indexes or unknown key performance indexes, if the information vector represents unknown key performance indexes, carrying out unsupervised clustering and label setting on the information vector in sequence to generate new key performance index records and storing the new key performance index records into a key performance index database, wherein the generated new key performance index records are recorded asn represents the number of new key performance indicator records generated;
s4: determining a network fault state diagnosis result according to the key index record corresponding to the information vector, wherein the output vector of the diagnosis result is recorded asF is a mark bit, wherein F is 1 to indicate that the system is normal, otherwise, the system is abnormal; l represents the number of key performance indicators,represents a key performance index ylCorresponding network failure status.
When the original information vector of the key performance index is acquired in the input stage, the original information vector x consists of the key performance indexes related to the aircraft, and the key performance indexes of different time aggregation levels (such as hours, days, weeks, months and the like) can be acquired according to the actual diagnosis requirement.
In view of the technical requirements of the system, and in order to facilitate subsequent data processing, the raw information vector needs to be subjected to normalization preprocessing, so that the sensitivity of the algorithm to the data scale is reduced.
As an example, the normalization preprocessing may employ an interval normalization method or a standard deviation method. The specific operation of these two methods is as follows:
(1) interval normalization, which operates to vary all input variables x within a desired interval range
Wherein x isupAnd xlowAre respectively after normalizationUpper and lower bounds of the numerical interval. In particular, when xlow＝-1，xupWhen 1, then the variables are normalized
(2) A standard deviation method, in order to ensure the individual difference of key performance indexes in the input variable x, the input variable is corrected into a vector with zero mean and standard deviation, and the linear operation is that
Where mean (x) represents the mean value of x, std (x) represents the standard deviation of x.
To improve the real-time and accuracy of fault diagnosis, the preprocessed vectors are normalizedAnd the method is further used for learning in matching decision, unsupervised clustering and label setting stages.
As an embodiment, the specific steps of performing the matching decision in step S3 to determine whether the information vector represents a known key performance indicator or an unknown key performance indicator are as follows:
if only one key performance index record with the distance from the information vector smaller than the threshold value exists in the key performance index database, the information vector represents the known key performance index; otherwise, the information vector represents an unknown key performance index.
In this embodiment, the euclidean distance is used when the matching decision calculates the distance between the key performance index record and the information vector.
As an embodiment, the unsupervised clustering adopts a self-organizing mapping algorithm to carry out rough classification, and the classes with the minimum distance between the classes are successively merged based on a Ward hierarchy method, so that the fine clustering is realized.
Referring to fig. 2, in the unsupervised clustering stage, the data is roughly classified by using a self-organizing mapping algorithm, and the unsupervised clustering stage is mainly divided into three stages of competition, cooperation, and adaptive adjustment of neurons:
1) competition: the input vector of the training module isIn order to find the neuron which is matched with the neuron most, a discriminant function based on Euclidean distance is adopted
Wherein,and N is the weight vector of the jth neuron in the t iteration, and is the total number of the neurons. When the weight vector of a neuron is closest to the input vector, the neuron is a winning neuron.
2) And (3) cooperation: when determining the winning neuronThereafter, to update neurons in proximity thereto, assignments are madeThe weights of the neighboring neurons are
Wherein the radius sigmatFor time-dependent decay functions, usually in exponential form σt＝σ0exp(-t/τσ). In particular, it is possible to use, for example,① has symmetry and peaks at the winning neuron, ② monotonically decreases with increasing distance and approaches 0, and ③ is a shift inequality regardless of the position of the winning neuron.
Thus, the distance of the winning neuron from its neighbors will affect the weights and thus the degree of update.
3) Self-adaptive adjustment: updating the parameters of the nodes according to a gradient descent method
Wherein the learning rate ηt＝η0exp(-t/τη) In relation to time t, for training patternsAnd (6) updating.
Winning neurons by weight updatingAnd its neighboring neuron directional input vectorClose, thereby causing the training data to exhibit topological ordering.
And repeating the competition, the cooperation and the self-adaptive adjustment until the algorithm is converged, and obtaining the ordered neurons after the rough classification is finished. In this embodiment, whenAnd when the threshold value is smaller than the preset threshold value, the algorithm is considered to be converged.
After the rough classification is finished, the classes with the minimum inter-class distance are further combined successively based on the Ward hierarchy method, and therefore the fine clustering is achieved.
Specifically, continuing with fig. 2, a fine clustering analysis is performed on the ordered neurons obtained from the coarse classification. By dividing the neurons into different clusters, neurons within a cluster have similar characteristics and are distinct from the characteristics of neurons within other clusters. Assume that the set of classes is G ═ G1,…,gLWhere L is the total number of classes. The clustering process is mainly based on Ward hierarchy, the classes with the minimum inter-class distance are combined successively through the hierarchy, meanwhile, the inter-class distance is updated through a Lance-Williams recursion formula, and the clustering effect is judged through a Davies-Bouldin Index (DBI). DBI is generally defined as
Wherein R ispqIs class gpAnd class gqIs defined as
Wherein S ispAnd DpqRespectively the degree of dispersion and the center distance of the two classes. By calculating the ratio of the sum of the intra-class distances to the extra-class distanceThe clustering selection algorithm based on the minimum DBI can avoid the defect that the k-mean algorithm can only obtain a local optimal solution. And further, whether the classes with similar statistical characteristics exist is judged through K-S test, so that the same key performance index characteristics do not exist in any two classes.
After the coarse classification and the fine clustering are completed, label setting is further carried out on the neurons. Considering the relationship between the generated clusters and the network fault states, one of the clusters will correspond to the normal state F being 1, the other clusters will correspond to the possible network fault states, and there is a one-to-one correspondence relationship between the clusters and the network fault states. Due to the lack of additional information, it is often difficult to assess the effectiveness of the clustering. However, with the increase of the running time, after a certain data amount is obtained, the learning performance is improved through supervised learning.
And finally, updating the learned key performance index record to a key performance index database (KPI database). If a plurality of winning neurons exist simultaneously after learning, a plurality of new key performance index records are generated correspondingly.
As an embodiment, the method for determining the network fault state diagnosis result specifically includes: if only one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result directly according to the key performance index record;
and if more than one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result by combining a nearest neighbor method and a quantile method.
In this embodiment, the method for determining the network fault state diagnosis result by combining the nearest neighbor method and the quantile method specifically includes: if the first diagnostic result determined by the nearest neighbor method is not equal to the second diagnostic result determined by the quantile method, and the average contour coefficient of the first diagnostic result is smaller than the average contour coefficient of the second diagnostic result, the final diagnostic result is the second diagnostic result; otherwise, the final diagnosis result is the first diagnosis result.
If more than one key performance index record corresponding to the information vector is recorded, a fuzzy network fault state diagnosis result may be generated. In order to further improve the accuracy and success rate of diagnosis, a quantile method and an average contour coefficient are adopted to further determine a diagnosis result. And determining the neurons with smaller quantiles away from the possible diagnostic network fault state results, such as a tertile, a quartile, a quintile and the like. Compared with a single neuron based on the nearest neighbor method, the quantile method can integrate more neuron information in the class, thereby improving the fault tolerance of the diagnosis result.
Specifically, referring to fig. 3, assuming that the first diagnosis result based on the nearest neighbor method is B and the second diagnosis result based on the quantile is P, the final diagnosis result is further determined according to the mean contour coefficient. If B ═ P, the final diagnosis result can be directly obtained as B; if B ≠ P, then a comparison discrimination is performed. The mean profiles of B and P are calculated, respectively, as
And
wherein,as a function of the profile, is
Wherein, aiIs the average of the distances of component i from other components in the class, biThe minimum average of the distances of component i to all components within the neighborhood class.
If it is notThe final diagnosis result is P, otherwise, the final diagnosis result is B.
In the output stage, according to the diagnosis result, the fault unit with the best matching key performance index can be correspondingly determined based on the corresponding relation between the key performance index and the fault unit.
As an embodiment, for a high dynamic mobile network composed of on-orbit aircrafts, the key performance indexes include link maintenance, handover success rate, received reference signal power, received reference signal quality, signal-to-interference-and-noise ratio, user average throughput and distance, and the like.
When the network is degraded for a number of reasons, the diagnostics provided by the present invention depend on the impact of the number of reasons on the key performance indicators. Network fault conditions due to a variety of causes may present major symptoms of a fault, and therefore only the most major fault is identified, and once the current most major fault is cleared, the next major fault will be identified.
Example two
The invention also discloses an on-orbit aircraft ad hoc network fault diagnosis device based on artificial intelligence, which comprises
The normalization preprocessing unit is used for carrying out normalization preprocessing on the original information vector; for the specific execution process of the normalization preprocessing unit, please refer to step S2 in the first embodiment, which is not described herein again.
The learning unit is used for carrying out matching decision on the information vector after normalization preprocessing and key performance index records in a key performance index database so as to judge whether the information vector represents a known key performance index or an unknown key performance index, and if the information vector represents an unknown key performance index, carrying out unsupervised clustering and label setting on the information vector in sequence to generate a new key performance index record and storing the new key performance index record in the key performance index database; for the specific implementation process of the learning unit, please refer to step S3 in the first embodiment, which is not described herein.
The diagnosis unit determines a network fault state diagnosis result according to the key index record corresponding to the information vector; for the specific execution process of the diagnosis unit, refer to step S4 in the first embodiment, which is not described herein again.
The disclosure above is only one specific embodiment of the present application, but the present application is not limited thereto, and any variations that can be made by those skilled in the art are intended to fall within the scope of the present application.
Claims (8)
1. An on-orbit aircraft ad hoc network fault diagnosis method based on artificial intelligence is characterized by comprising the following steps:
s1: collecting original information vectors of key performance indexes;
s2: carrying out normalization preprocessing on the original information vector;
s3: matching decision is carried out on the information vector after normalization preprocessing and key performance index records in a key performance index database so as to judge whether the information vector represents known key performance indexes or unknown key performance indexes, if the information vector represents the unknown key performance indexes, unsupervised clustering and label setting are carried out on the information vector in sequence, and new key performance index records are generated and stored in the key performance index database;
s4: and determining a network fault state diagnosis result according to the key index record corresponding to the information vector.
2. The method according to claim 1, wherein in step S3, the specific step of determining whether the information vector represents a known key performance indicator or an unknown key performance indicator is as follows:
if only one key performance index record with the distance from the information vector smaller than the threshold value exists in the key performance index database, the information vector represents the known key performance index; otherwise, the information vector represents an unknown key performance index.
3. The method according to claim 1, wherein the method of determining the network fault status diagnosis result specifically comprises: if only one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result directly according to the key performance index record;
and if more than one key performance index record corresponding to the information vector exists, determining a network fault state diagnosis result by combining a nearest neighbor method and a quantile method.
4. The method according to claim 3, wherein the method for determining the network fault state diagnosis result by combining the nearest neighbor method and the quantile method specifically comprises: if the first diagnostic result determined by the nearest neighbor method is not equal to the second diagnostic result determined by the quantile method, and the average contour coefficient of the first diagnostic result is smaller than the average contour coefficient of the second diagnostic result, the final diagnostic result is the second diagnostic result; otherwise, the final diagnosis result is the first diagnosis result.
5. The method according to claim 2, wherein the unsupervised clustering adopts a self-organizing mapping algorithm for rough classification, and based on Ward hierarchy method, the classes with the minimum inter-class distance are successively merged, so as to realize fine clustering.
6. The method of claim 1, wherein the normalization preprocessing is performed by an interval normalization method or a standard deviation method.
7. The method of claim 1, wherein the key performance indicators comprise link maintenance, handover success rate, received reference signal power, received reference signal quality, signal to interference and noise ratio, user average throughput, and distance.
8. An on-orbit aircraft ad hoc network fault diagnosis device based on artificial intelligence is characterized by comprising
The normalization preprocessing unit is used for carrying out normalization preprocessing on the original information vector;
the learning unit is used for carrying out matching decision on the information vector after normalization preprocessing and key performance index records in a key performance index database so as to judge whether the information vector represents a known key performance index or an unknown key performance index, and if the information vector represents an unknown key performance index, carrying out unsupervised clustering and label setting on the information vector in sequence to generate a new key performance index record and storing the new key performance index record in the key performance index database;
and the diagnosis unit is used for determining the network fault state diagnosis result according to the key index record corresponding to the information vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811387631.8A CN109547248A (en) | 2018-11-20 | 2018-11-20 | Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811387631.8A CN109547248A (en) | 2018-11-20 | 2018-11-20 | Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109547248A true CN109547248A (en) | 2019-03-29 |
Family
ID=65848946
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811387631.8A Pending CN109547248A (en) | 2018-11-20 | 2018-11-20 | Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109547248A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580287A (en) * | 2019-08-20 | 2019-12-17 | 北京亚鸿世纪科技发展有限公司 | Emotion classification method based ON transfer learning and ON-LSTM |
| CN111145541A (en) * | 2019-12-18 | 2020-05-12 | 深圳先进技术研究院 | Traffic flow data prediction method, storage medium, and computer device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120029977A1 (en) * | 2010-07-30 | 2012-02-02 | International Business Machines Corporation | Self-Extending Monitoring Models that Learn Based on Arrival of New Data |
| CN105873105A (en) * | 2016-04-22 | 2016-08-17 | 中国科学技术大学 | Method for anomaly detection and positioning of mobile communication network based on network experience quality |
| CN108681747A (en) * | 2018-05-11 | 2018-10-19 | 武汉理工大学 | Rotary machinery fault diagnosis based on deep learning and condition monitoring system and method |
-
2018
- 2018-11-20 CN CN201811387631.8A patent/CN109547248A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120029977A1 (en) * | 2010-07-30 | 2012-02-02 | International Business Machines Corporation | Self-Extending Monitoring Models that Learn Based on Arrival of New Data |
| CN105873105A (en) * | 2016-04-22 | 2016-08-17 | 中国科学技术大学 | Method for anomaly detection and positioning of mobile communication network based on network experience quality |
| CN108681747A (en) * | 2018-05-11 | 2018-10-19 | 武汉理工大学 | Rotary machinery fault diagnosis based on deep learning and condition monitoring system and method |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580287A (en) * | 2019-08-20 | 2019-12-17 | 北京亚鸿世纪科技发展有限公司 | Emotion classification method based ON transfer learning and ON-LSTM |
| CN111145541A (en) * | 2019-12-18 | 2020-05-12 | 深圳先进技术研究院 | Traffic flow data prediction method, storage medium, and computer device |
| CN111145541B (en) * | 2019-12-18 | 2021-10-22 | 深圳先进技术研究院 | Traffic flow data prediction method, storage medium, and computer device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7240691B1 (en) | Data drive active power distribution network abnormal state detection method and system | |
| CN110929918B (en) | 10kV feeder fault prediction method based on CNN and LightGBM | |
| CN116757534B (en) | Intelligent refrigerator reliability analysis method based on neural training network | |
| US20200393329A1 (en) | Diagnosing method of engine condition and diagnostic modeling method thereof | |
| CN111597760B (en) | Method for obtaining gas path parameter deviation value under small sample condition | |
| CN115270965A (en) | Power distribution network line fault prediction method and device | |
| CN112232370A (en) | Fault analysis and prediction method for engine | |
| CN111882114B (en) | Short-time traffic flow prediction model construction method and prediction method | |
| CN117076955A (en) | Fault detection method and system for high-voltage frequency converter | |
| CN114553671A (en) | Diagnosis method for power communication network fault alarm | |
| CN115096627A (en) | Method and system for fault diagnosis and operation and maintenance in manufacturing process of hydraulic forming intelligent equipment | |
| CN112596016A (en) | Transformer fault diagnosis method based on integration of multiple one-dimensional convolutional neural networks | |
| CN109547248A (en) | Based on artificial intelligence in orbit aerocraft ad hoc network method for diagnosing faults and device | |
| CN115859777A (en) | Method for predicting service life of product system in multiple fault modes | |
| CN117633688A (en) | Large-scale power data anomaly detection method based on ridge regression-k-means clustering-LOF-LSTM fusion algorithm | |
| CN112613227B (en) | Model for predicting remaining service life of aero-engine based on hybrid machine learning | |
| CN114529067A (en) | Method for performing predictive maintenance on electric vehicle battery based on big data machine learning | |
| CN117216713A (en) | Fault delimiting method, device, electronic equipment and storage medium | |
| CN115130617B (en) | Detection method for continuous increase of self-adaptive satellite data mode | |
| CN115033893A (en) | Information vulnerability data analysis method of improved clustering algorithm | |
| CN115392710A (en) | Wind turbine generator operation decision method and system based on data filtering | |
| CN115081514A (en) | Industrial equipment fault identification method under data imbalance condition | |
| CN114692729A (en) | New energy station bad data identification and correction method based on deep learning | |
| CN113239626A (en) | Sensor layout optimization method based on system reliability | |
| CN112561153A (en) | Scenic spot crowd gathering prediction method based on model integration |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190329 |
|
| RJ01 | Rejection of invention patent application after publication |