CN111898705A - Fault feature parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering - Google Patents
Fault feature parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering Download PDFInfo
- Publication number
- CN111898705A CN111898705A CN202010833932.XA CN202010833932A CN111898705A CN 111898705 A CN111898705 A CN 111898705A CN 202010833932 A CN202010833932 A CN 202010833932A CN 111898705 A CN111898705 A CN 111898705A
- Authority
- CN
- China
- Prior art keywords
- features
- hierarchical clustering
- clustering
- fuzzy
- sensitivity
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Algebra (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Computing Systems (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Devices For Executing Special Programs (AREA)
Abstract
The invention discloses a fault characteristic parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering. The invention provides self-adaptive hierarchical clustering based on fuzzy relation based on logsig function, and is applied to fault diagnosis of equipment; sensitive features are calculated and selected based on the fuzzy relation without prior knowledge, so that the intelligence of the method is improved; the use of the optimized features simplifies the feature set, avoids dimension disasters, reduces the calculation burden and improves the fault diagnosis efficiency; the adaptive hierarchical clustering preferred in combination with features has higher diagnostic accuracy.
Description
Technical Field
The invention relates to a fault characteristic parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering, and belongs to the technical field of big data processing.
Background
With the development of science and technology, large-scale equipment is more and more complicated, the cooperation between the part is inseparabler, and the trouble of part all can bring the loss of shutting down, causes great economic loss, can endanger personal safety in the serious case. In addition, if the fault cannot be accurately positioned, blind repair and disassembly can cause precision errors, reliability reduction and the like. Therefore, the fault diagnosis technology is a precondition for ensuring the safe and stable operation of the equipment, and is also important for the maintenance of the equipment.
Due to the fact that the number of measuring points is large, the number of monitoring parameters (force, temperature, vibration, sound, energy, hydraulic pressure and the like) is large, diverse and complex state monitoring big data are formed, and fault diagnosis of equipment enters a big data era. The high-dimensional features can provide richer feature parameters for fault diagnosis, but the feature dimension is too high, and when the scale of the training sample is not large, the influences of overfitting and the like are brought to fault diagnosis and identification, so that the accuracy of fault diagnosis is influenced.
In neural networks, it is common to useThe function characterizes the fuzzy relationship between samples, and thus the ordered structure between samples. In different kinds of faults, the larger the difference between the same characteristics is, the more sensitive the characteristics are to the classification of the categories is, and the larger sensitivity coefficient is taken.
The hierarchical clustering algorithm belongs to an unsupervised classification algorithm, is suitable for clustering of data sets with any shapes, does not need to determine parameters such as a clustering center, the number of clusters and the like in advance, but has no uniform standard of end conditions, still needs to set corresponding thresholds, and has larger calculated amount.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a fault characteristic parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering.
The invention provides the following technical scheme:
1) fuzzy preference relationship calculation
1.1) given System S ═<X,Q,U>Wherein X ═ { X ═ X1,x2,…,xNDenotes a sample set, Q ═ Q1,q2,…,qJIs the set of features, U ═ U1,u2,…,uCIs the failure set;
wherein q isi1,qj1E is Q; i is not equal to j; k is the number of clusters;
1.2) to dijFurther simplification is realized, as shown in a formula (2);
wherein Δ q ═ qi1-qj1;
2) Coefficient of sensitivity calculation
Assume a set of raw features q containing class C failuresm,j,m=1,2,…,N;j=1,2,…,J}CWhere N is the number of samples per fault, J is the number of features, qn,jRepresents the jth characteristic value of the nth sample;
the total number M of samples of the system S is nxc, and the total number L of data is nxc × J;
calculating the sensitivity coefficient of each characteristic according to the formula 2) to form a fuzzy relation matrix
The coefficient of sensitivity for each feature is:
3) sensitive feature selection
Sensitivity coefficient (SP) of all features1,SP2,…,SPJ) The front v sensitivity coefficients are selected as sensitivity characteristics Q ' ═ Q ' in sequence from small to large '1,q′2,…,q′vV is the preset number of sensitive features;
the problem of feature redundancy is not considered in the sensitive feature selection, and redundant features may still be included; in order to further improve the efficiency and reduce the feature dimension, the invention uses the self-adaptive hierarchical clustering algorithm to remove redundant features.
4) Removing redundant features based on adaptive hierarchical clustering;
for a certain degree of clustering of the data set, the contour coefficient SkThe definition is as follows:
wherein S isIThe contour coefficient of the sample individual is shown, T is the number of samples in the data set, and k is the clustering number;
wherein a (I) represents a sample xIAnd the average distance between all other samples belonging to class C, b (I) denotes sample xIAnd the minimum of the average distances of all samples in each class other than class C;
normalizing the selected sensitive features Q' to obtain a normalized feature set, clustering according to a self-adaptive hierarchical clustering method, and clustering to obtain class numbersc is the preferred number of features, the center of the c class is taken as the preferred feature, and a preferred feature set is formed
The invention has the beneficial effects that:
1. the method removes redundant features by using a self-adaptive hierarchical clustering algorithm, adopts a clustering contour coefficient as an index for evaluating the clustering effectiveness, does not need to preset the clustering number, self-adaptively determines the clustering number, and obtains a certain clustering result, so that the inter-class distance is as large as possible, the intra-class distance is as small as possible, and good separability is realized among classes;
2. the invention provides self-adaptive hierarchical clustering based on fuzzy relation based on logsig function, and is applied to fault diagnosis of equipment; sensitive features are calculated and selected based on the fuzzy relation without prior knowledge, so that the intelligence of the method is improved; the use of the optimized features simplifies the feature set, avoids dimension disasters, reduces the calculation burden and improves the fault diagnosis efficiency; the adaptive hierarchical clustering preferred in combination with features has higher diagnostic accuracy.
3. The monitoring data often has the characteristics of ambiguity, uncertainty and the like, the fuzzy preference relationship has inherent advantages, the preference of a decision maker can be better reflected, and the system is more comprehensively described; aiming at the problem of fault diagnosis in a big data form, the method has inherent advantages by combining the fuzzy preference relationship, reduces the feature dimension, removes redundant features, selects the feature combination with the largest fault diagnosis information amount from high-dimensional features, and improves the efficiency of fault diagnosis.
Drawings
FIG. 1 is a graph of the logsig function over the interval [ -1, 1 ];
FIG. 2 is a fuzzy preference relationship based on the features of equation (1);
FIG. 3 is a flow chart of the adaptive hierarchical clustering algorithm of the present invention;
fig. 4 is a schematic diagram of the sensitive feature selection described in example 2.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
As shown in fig. 1 and 3.
A fault characteristic parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering comprises the following steps:
1) fuzzy preference relationship calculation
1.1) given System S ═<X,Q,U>Wherein X ═ { X ═ X1,x2,…,xNDenotes a sample set, Q ═ Q1,q2,…,qJIs the set of features, U ═ U1,u2,…,uCIs the failure set;
wherein q isi1,qj1E is Q; i is not equal to j; k is the number of clusters;
as can be seen from FIG. 2, dij(q)=dji(q) when i ═ j, dij(q) 0.5, with increasing | Δ q |, dij(q) increases from 0.5 when q is increasedi,l>>qj,lWhen d is greater thanij(q) → 1. Therefore, in feature selection, it is only necessary to characterize the difference between two features, and it is not necessary to describe q in detaili,lWhether greater or less than qj,l。
1.2) to dijFurther simplified as shown in(2) Shown;
wherein Δ q ═ qi1-qj1;
As can be seen from FIG. 2, the parameter k takes different values, dijThe change is large, and the preference degree of the fuzzy relation of the features also changes.
2) Coefficient of sensitivity calculation
Assume a set of raw features q containing class C failuresm,j,m=1,2,…,N;j=1,2,…,J}CWhere N is the number of samples per fault, J is the number of features, qn,jRepresents the jth characteristic value of the nth sample;
the total number M of samples of the system S is nxc, and the total number L of data is nxc × J;
calculating the sensitivity coefficient of each characteristic according to the formula 2) to form a fuzzy relation matrix
The coefficient of sensitivity for each feature is:
3) sensitive feature selection
Sensitivity coefficient (SP) of all features1,SP2,…,SPJ) The front v sensitivity coefficients are selected as sensitivity characteristics Q ' ═ Q ' in sequence from small to large '1,q′2,…,q′vV is the preset number of sensitive features;
the problem of feature redundancy is not considered in the sensitive feature selection, and redundant features may still be included; in order to further improve the efficiency and reduce the feature dimension, the invention uses the self-adaptive hierarchical clustering algorithm to remove redundant features.
4) Removing redundant features based on adaptive hierarchical clustering;
for a certain degree of clustering of the data set, the contour coefficient SkThe definition is as follows:
wherein S isIThe contour coefficient of the sample individual is shown, T is the number of samples in the data set, and k is the clustering number;
wherein a (I) represents a sample xIAnd the average distance between all other samples belonging to class C, b (I) denotes sample xIAnd the minimum of the average distances of all samples in each class other than class C;
normalizing the selected sensitive features Q' to obtain a normalized feature set, clustering according to an adaptive hierarchical clustering method, wherein the number c of the clusters is the preferred number of the features, and the center of the c classes is taken as the preferred feature to form the preferred feature set
Example 2
As shown in fig. 4.
The method of embodiment 1 is used for fault diagnosis and fault type determination of the bearing-integrated simulation system, and comprises the following steps:
the vibration sensor was used to acquire 4 states of the bearing simulation system: normal state, outer ring fault, inner ring fault, rolling element fault;
serial number | Operating | Status flag | |
1 | Is normal | 0 | |
2 | |
1 | |
3 | Inner ring failure | 2 | |
4 | Failure of rolling body | 3 |
A1) Feature extraction
Extracting time domain characteristics, frequency domain characteristics, EEMD decomposed IMF component characteristics and wavelet packet decomposed energy of the original vibration signal, and forming a characteristic set;
a1.1) time-domain features
Kurtosis:skewness:wherein x (N) is a time domain sequence of the signal, and N is the number of vibration sample points;
1.2) frequency domain characteristics
standard deviation of frequency:(wherein fkIs the frequency value of the K-th line, s (K) is the frequency spectrum of signal x (n), K is the number of lines;
a1.3) energy index
A2) Sensitive feature selection
The large number of features not only can reduce the calculation efficiency, but also can cause dimension disaster, the sensitive coefficients of 134 features are calculated according to the fuzzy preference relationship-based method, and 41 sensitive features are selected;
A3) preferred feature selection
And (4) performing normalization processing on the selected 41 sensitive features, and clustering by adopting an adaptive hierarchical clustering algorithm. In this example, the finally determined c is 12.
A4) Fault diagnosis
And introducing the 12 optimal characteristics into an adaptive hierarchical clustering algorithm, and identifying the vibration signals which are actually acquired according to the trained fault model to obtain a clustering category, thereby realizing fault diagnosis and determining the fault type. In this embodiment, the classification accuracy reaches 99.4%.
Finally, it should be noted that the above-mentioned contents are only used for illustrating the technical solutions of the present invention, and not for limiting the protection scope of the present invention, and that the simple modifications or equivalent substitutions of the technical solutions of the present invention by those of ordinary skill in the art can be made without departing from the spirit and scope of the technical solutions of the present invention.
Claims (1)
1. A fault characteristic parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering is characterized by comprising the following steps:
1) fuzzy preference relationship calculation
1.1) given System S ═<X,Q,U>Wherein X ═ { X ═ X1,x2,...,xNDenotes a sample set, Q ═ Q1,q2,...,qJIs the set of features, U ═ U1,u2,...,uCIs the failure set;
wherein q isil,qjlE is Q; i is not equal to j; k is the number of clusters;
1.2) to dijFurther simplification is realized, as shown in a formula (2);
wherein Δ q ═ qil-qjl;
2) Coefficient of sensitivity calculation
Assume a set of raw features q containing class C failuresm,j,m=1,2,...,N;j=1,2,...,J}CWhere N is the number of samples per fault, J is the number of features, qn,jRepresents the jth characteristic value of the nth sample;
the total number M of samples of the system S is nxc, and the total number L of data is nxc × J;
calculating the sensitivity coefficient of each characteristic according to the formula 2) to form a fuzzy relation matrix
The coefficient of sensitivity for each feature is:
3) sensitive feature selection
Sensitivity coefficient (SP) of all features1,SP2,...,SPI) The front v sensitivity coefficients are selected as sensitivity characteristics Q ' ═ Q ' in sequence from small to large '1,q'2,...,q'vV is the preset number of sensitive features;
4) removing redundant features based on adaptive hierarchical clustering;
for a certain degree of clustering of the data set, the contour coefficient SkThe definition is as follows:
wherein S isIThe contour coefficient of the sample individual is shown, T is the number of samples in the data set, and k is the clustering number;
wherein a (I) represents a sample xIAnd the average distance between all other samples belonging to class C, b (I) denotes sample xIAnd the minimum of the average distances of all samples in each class other than class C;
normalizing the selected sensitive features Q' to obtain a normalized feature set, clustering according to an adaptive hierarchical clustering method, wherein the number c of the clusters is the preferred number of the features, and the center of the c classes is taken as the preferred feature to form the preferred feature set
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010833932.XA CN111898705B (en) | 2020-08-18 | 2020-08-18 | Fault feature parameter selection method based on fuzzy preference relation and self-adaptive hierarchical clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010833932.XA CN111898705B (en) | 2020-08-18 | 2020-08-18 | Fault feature parameter selection method based on fuzzy preference relation and self-adaptive hierarchical clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111898705A true CN111898705A (en) | 2020-11-06 |
CN111898705B CN111898705B (en) | 2023-04-25 |
Family
ID=73229929
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010833932.XA Active CN111898705B (en) | 2020-08-18 | 2020-08-18 | Fault feature parameter selection method based on fuzzy preference relation and self-adaptive hierarchical clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111898705B (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105975995A (en) * | 2016-05-26 | 2016-09-28 | 山东省计算中心(国家超级计算济南中心) | Fuzzy-preference-relation-based multi-vibration-signal fusion method |
US20170061072A1 (en) * | 2015-09-02 | 2017-03-02 | Guardant Health, Inc. | Machine Learning for Somatic Single Nucleotide Variant Detection in Cell-free Tumor Nucleic acid Sequencing Applications |
CN106769052A (en) * | 2017-03-21 | 2017-05-31 | 桂林电子科技大学 | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on cluster analysis |
CN108170823A (en) * | 2018-01-04 | 2018-06-15 | 江西师范大学 | A kind of Freehandhand-drawing interactive three-dimensional model retrieval method understood based on high-level semantic attribute |
CN109143867A (en) * | 2018-09-26 | 2019-01-04 | 上海海事大学 | A kind of energy management method of the hybrid power ship based on ANN Control |
CN110991478A (en) * | 2019-10-29 | 2020-04-10 | 西安建筑科技大学 | Method for establishing thermal comfort model and method and system for setting user preference temperature |
-
2020
- 2020-08-18 CN CN202010833932.XA patent/CN111898705B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170061072A1 (en) * | 2015-09-02 | 2017-03-02 | Guardant Health, Inc. | Machine Learning for Somatic Single Nucleotide Variant Detection in Cell-free Tumor Nucleic acid Sequencing Applications |
CN105975995A (en) * | 2016-05-26 | 2016-09-28 | 山东省计算中心(国家超级计算济南中心) | Fuzzy-preference-relation-based multi-vibration-signal fusion method |
CN106769052A (en) * | 2017-03-21 | 2017-05-31 | 桂林电子科技大学 | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on cluster analysis |
CN108170823A (en) * | 2018-01-04 | 2018-06-15 | 江西师范大学 | A kind of Freehandhand-drawing interactive three-dimensional model retrieval method understood based on high-level semantic attribute |
CN109143867A (en) * | 2018-09-26 | 2019-01-04 | 上海海事大学 | A kind of energy management method of the hybrid power ship based on ANN Control |
CN110991478A (en) * | 2019-10-29 | 2020-04-10 | 西安建筑科技大学 | Method for establishing thermal comfort model and method and system for setting user preference temperature |
Non-Patent Citations (2)
Title |
---|
文政颖等: "一种基于模糊层次聚类分析的大数据挖掘算法" * |
董炜;刘明明;王良顺;赵辉;辜勋;: "基于群决策的道岔控制电路故障诊断方法" * |
Also Published As
Publication number | Publication date |
---|---|
CN111898705B (en) | 2023-04-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110132598B (en) | Fault noise diagnosis algorithm for rolling bearing of rotating equipment | |
CN111666982B (en) | Electromechanical equipment fault diagnosis method based on deep neural network | |
CN106769052A (en) | A kind of mechanical system rolling bearing intelligent failure diagnosis method based on cluster analysis | |
CN109974782B (en) | Equipment fault early warning method and system based on big data sensitive characteristic optimization selection | |
WO2023044978A1 (en) | Adversarial-flow-model-based unsupervised fault diagnosis method for mechanical device | |
CN113255848A (en) | Water turbine cavitation sound signal identification method based on big data learning | |
CN110297469B (en) | Production line fault judgment method based on resampling integrated feature selection algorithm | |
CN110991471B (en) | Fault diagnosis method for high-speed train traction system | |
CN111504635A (en) | Planetary gear fault diagnosis method based on differential evolution probability neural network | |
CN109298633A (en) | Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization | |
CN114297918A (en) | Aero-engine residual life prediction method based on full-attention depth network and dynamic ensemble learning | |
CN109255201A (en) | A kind of ball screw assembly, health evaluating method based on SOM-MQE | |
CN114871850B (en) | Tool wear state assessment method based on vibration signals and BP neural network | |
Jiao et al. | Partly interpretable transformer through binary arborescent filter for intelligent bearing fault diagnosis | |
CN115290326A (en) | Rolling bearing fault intelligent diagnosis method | |
CN115375026A (en) | Method for predicting service life of aircraft engine in multiple fault modes | |
CN115392393A (en) | Temperature measuring instrument state detection method | |
CN114861349A (en) | Rolling bearing RUL prediction method based on model migration and wiener process | |
CN114443338A (en) | Sparse negative sample-oriented anomaly detection method, model construction method and device | |
CN114742115A (en) | Rolling bearing fault diagnosis model and method based on temperature and vibration characteristic fusion | |
CN114970309A (en) | Thermal power equipment state early warning evaluation method | |
CN111983365B (en) | Transformer winding deformation detection method based on oscillatory wave multistage decomposition | |
Cheng et al. | Control chart pattern recognition using wavelet analysis and neural networks | |
CN111898705A (en) | Fault feature parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering | |
Zhou et al. | Degradation State Recognition of Rolling Bearing Based on K‐Means and CNN Algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |