CN111950377A - Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition - Google Patents
Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition Download PDFInfo
- Publication number
- CN111950377A CN111950377A CN202010681363.1A CN202010681363A CN111950377A CN 111950377 A CN111950377 A CN 111950377A CN 202010681363 A CN202010681363 A CN 202010681363A CN 111950377 A CN111950377 A CN 111950377A
- Authority
- CN
- China
- Prior art keywords
- graph
- fault
- vibration
- adopting
- matrix
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Abstract
The invention relates to a rotary machine fault intelligent diagnosis method based on fuzzy soft morphology pattern recognition. The method comprises the following steps: arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data; generating a three-dimensional parameter graph by using the acquired time series vibration signals; performing difference value reconstruction preprocessing on the three-dimensional parameter graph; carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment; adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancement pretreatment; performing texture feature extraction on the reinforced rotating machinery vibration parameter graph by adopting a fuzzy soft morphology composite edge detection operator; aiming at the texture characteristics, extracting fault characteristics by adopting a gray level-primitive-gradient co-occurrence matrix; and according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method.
Description
Technical Field
The invention relates to a rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition, and belongs to the technical field of high-speed rotary machine fault diagnosis and signal processing.
Background
With the continuous development of modern industrial production and the gradual progress of science and technology, in order to improve the production efficiency and the product quality to the maximum extent, some large-scale rotating machines such as aero-engines, gas turbines, fans, steam turbines, compressors, generators and the like are continuously developed towards the directions of large scale, high speed, automation, intellectualization, continuous operation and complex structure, so that the damage caused by the faults of rotating machine equipment is more serious, once the faults occur, huge losses are caused to enterprises, and therefore, the monitoring and fault diagnosis of the operating state of the rotating machines are of great significance.
The conventional diagnosis method is to perform equipment diagnosis based on a comparison statistical threshold, a model or an artificial intelligence method by monitoring signals such as vibration and rotating speed of acquisition equipment, but in actual rotary machine fault diagnosis, a field expert usually obtains a large number of parameter graphic features reflecting the running state of the equipment through observing and analyzing vibration signals to make diagnosis. The parameter patterns containing a large amount of state information include two-dimensional amplitude-frequency characteristic curves, phase-frequency characteristic curves, axis locus diagrams, frequency spectrograms, trend charts, three-dimensional vibration three-dimensional spectrograms, three-dimensional order ratio charts, and recently developed holographic spectrograms containing a larger amount of information. In the actual fault diagnosis, because the information in the graph, especially the information in the multi-dimensional graph is difficult to automatically extract, a large amount of information in the graph is not fully utilized, the main reason is that the graph information is difficult to describe in a language form and is difficult to express by a knowledge rule, so that the invention provides a method for extracting the vibration parameter graph characteristics based on a fuzzy soft morphology method to further diagnose the fault, and the fault diagnosis accuracy of the rotary machine can be effectively improved.
Disclosure of Invention
The invention aims to: aiming at the problems of the conventional rotary machine fault diagnosis technology, the invention provides a rotary machine fault diagnosis method based on a fuzzy soft morphology graph recognition technology, which is characterized in that the fault characteristics of a vibration graph are extracted for diagnosis by adopting the fuzzy soft morphology method based on the graph recognition technology according to the peak-valley, gray level, texture characteristics and the like of the vibration parameter graph, and the fault diagnosis accuracy can be improved.
The technical scheme adopted by the invention is as follows:
the intelligent rotary machine fault diagnosis method based on fuzzy soft morphological pattern recognition comprises the following steps:
(1) arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data;
(2) generating a three-dimensional parameter graph by using the acquired time series vibration signals;
(3) performing difference value reconstruction preprocessing on the three-dimensional parameter graph; the specific interpolation reconstruction method is disclosed in reference documents, and a rotary mechanical parameter graph interpolation method based on rotor dynamics is adopted;
(4) carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment;
(5) adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancing pretreatment in the step (4);
(6) performing texture feature extraction on the rotating machinery vibration parameter graph subjected to the enhancement processing in the step (5) by adopting a fuzzy soft morphology composite edge detection operator;
(7) aiming at the texture features extracted in the step (6), extracting fault features by adopting a gray-element-gradient co-occurrence matrix;
(8) and according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method.
Further, the step (2) of generating a three-dimensional parameter graph from the acquired time series vibration signals specifically includes the following steps:
(2.1) arranging a vibration sensor on the rotary machine, and collecting vibration time sequence signals XnN is a positive integer, 1,2... No. N;
(2.2) to the vibration time series signal XnIntercepting to generate a matrix expression Y of the vibration time series signal:
and (2.3) carrying out spectral analysis on the matrix expression Y to obtain a three-dimensional parameter graph set f.
Further, the step (5) adopts a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancement pretreatment, and is specifically performed by the following formula:
wherein, gmThe image set after the enhancement processing is performed, wherein m is 1,2,3,4,5,6, and alpha is a gray scale weight of the fuzzy soft-on operation; f is a graph set, b is a structural element, and f is corroded and expanded by b and is respectively defined as follows:
further, the step (6) adopts a fuzzy soft morphology composite edge detection operator to extract the texture features of the reinforced rotating machinery vibration parameter graph, and is specifically realized by the following formula:
h=a·g1+b·g2+c·g3+d·g4+e·g5+k·g6
wherein h is a graph set, namely texture characteristics, after the fuzzy soft morphology composite edge detection operator is processed;
g1=f-fΘb;
g5=f·b-f;
the weighting coefficients are a, b, c, d, e, k.
Further, the specific calculation method for the weighting coefficients a, b, c, d, e, and k is as follows:
(5.1) calculating to obtain an enhanced graph set gmCharacteristic value M of1、M2、M3、M4、M5 M6;
(5.2) averaging: mn=(M1+M2+M3+M4+M5+M6)/6;
(5.3) calculating the weight Wi=1/(|Mi-Mn|+1),i=1,2,3,4,5,6;
(5.4) calculating the weighting coefficients as:
a=W1/(W1+W2+W3+W4+M5+M6)
b=W2/(W1+W2+W3+W4+M5+M6)
c=W3/(W1+W2+W3+W4+M5+M6)
d=W4/(W1+W2+W3+W4+M5+M6)
e=W5/(W1+W2+W3+W4+M5+M6)
k=W6/(W1+W2+W3+W4+M5+M6)。
further, the step (7) adopts a gray-primitive-gradient co-occurrence matrix to extract the fault feature for the extracted texture feature, specifically:
let B (m1, n1, p1) be the gray-cell-gradient three-dimensional co-occurrence matrix, and the three dimensions are respectively: f (u, w) is a graph gray matrix, TG (s, t) is a normalized graph element matrix, and QG (r, v) is a normalized graph gradient matrix;
converting the texture feature matrix h into an F (u, w) graph gray matrix, a TG (s, t) normalized graph primitive matrix and a QG (r, v) normalized graph gradient matrix;
counting the number of pixels, namely the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix, namely F (u, w) ═ m1, TG (s, t) ═ n1 and QG (r, v) ═ p1, namely counting the number of the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix
B(m1,n1,p1)#={(u,w),(s,t),(r,v)|F(u,w)=m1,TG(s,t)=n1,QG(r,v)=p1}
Wherein m1, n1, p1 are 1,2 … 64, and 64 are in B (m1, n1, p1) matrix3An element;
and extracting angular second moment, contrast, absolute value, contrast moment, autocorrelation coefficient and entropy from B (m1, n1, p1) to be used as a fault feature sample.
Further, the step (8) adopts an artificial neural network method to perform fault diagnosis according to the extracted fault characteristics, specifically: and carrying out sample off-line training on the fault characteristics according to the BP neural network, then carrying out on-line acquisition on vibration data, extracting the fault characteristics by the same method, and diagnosing by using the trained neural network.
Compared with the prior art, the invention has the advantages that:
(1) the diagnostic pattern information is effectively utilized, and in the conventional fault diagnosis, the pattern feature is difficult to extract, so that the diagnostic pattern information is difficult to be directly used for fault diagnosis of the rotary machine, and therefore, the diagnostic accuracy is low. The invention constructs a fuzzy soft morphological parameter graph characteristic extraction and pretreatment method, can effectively obtain vibration information and improves the fault diagnosis accuracy.
(2) The judgment which can only be carried out by field experts in the past is avoided, and the automatic diagnosis level is improved;
(3) with the development of electronic information technology, communication technology and artificial intelligence technology and the utilization of Internet of things and mass big data, the invention combines an advanced fault feature extraction method and an artificial intelligence method, thereby improving the diagnosis speed and accuracy.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
Detailed Description
The diagnostic pattern information is effectively utilized, and in the conventional fault diagnosis, the pattern feature is difficult to extract, so that the diagnostic pattern information is difficult to be directly used for fault diagnosis of the rotary machine, and therefore, the diagnostic accuracy is low. The invention constructs a fuzzy soft morphological parameter graph characteristic extraction and pretreatment method, can effectively obtain vibration information and improves the fault diagnosis accuracy.
The invention relates to the technical field of fault diagnosis and signal processing of high-speed rotating machinery, in particular to a rotary machinery fault diagnosis method based on a fuzzy soft morphological graph recognition technology in China, which can be applied to large rotating machinery such as an aircraft engine, a gas turbine, a fan, a steam turbine, a compressor, a generator and the like serving as main production tools in the industries such as petrochemical industry, metallurgy, electric power, aviation, aerospace and the like.
As shown in FIG. 1, the invention provides an intelligent diagnosis method for rotary machine faults based on fuzzy soft morphology pattern recognition, which comprises the following steps:
(1) arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data;
(2) generating a three-dimensional parameter graph by using the acquired time series vibration signals;
the method for generating the three-dimensional parameter graph by using the acquired time series vibration signals specifically comprises the following steps:
(2.1) arranging a vibration sensor on the rotary machine, and collecting vibration time sequence signals XnN is a positive integer, 1,2... No. N;
(2.2) to the vibration time series signal XnIntercepting to generate a matrix expression Y of the vibration time series signal:
and (2.3) carrying out spectral analysis on the matrix expression Y to obtain a three-dimensional parameter graph set f.
(3) According to the rotor dynamics theory, utilizing the characteristic that the rotating machinery has periodic exciting force to carry out difference value reconstruction pretreatment on the three-dimensional parameter graph;
specific methods for reconstructing the difference are as follows: a rotary mechanical parameter graph interpolation method based on rotor dynamics is disclosed, which is reported in Chinese Motor engineering, 2010,30(29) and 90-95.
(4) Carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment; the gray scale adaptive histogram equalization enhancement preprocessing belongs to the prior art, and therefore, is not described in detail herein;
(5) adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancing pretreatment in the step (4);
specifically, the method is carried out by the following formula:
wherein, gmThe method is a graph set after enhancement processing, wherein m is 1,2,3,4,5,6, and alpha is a gray level weight of fuzzy soft-on operation, and the weight alpha has a large influence on the final filtering effect; f is a graph set, b is a structural element, and f is corroded and expanded by b and is respectively defined as follows:
(6) performing texture feature extraction on the rotating machinery vibration parameter graph subjected to the enhancement processing in the step (5) by adopting a fuzzy soft morphology composite edge detection operator;
the method is realized by the following formula:
h=a·g1+b·g2+c·g3+d·g4+e·g5+k·g6
wherein h is a graph set, namely texture characteristics, after the fuzzy soft morphology composite edge detection operator is processed; the formula reflects texture features of a composite edge detection operator of the graph, and is the sum of weighted values of features such as an outer edge, an inner edge and an edge straddling an actual Euclidean boundary of the graph, which are respectively extracted.
g1=f-fΘb;
g5=f·b-f;
The weighting coefficients are a, b, c, d, e, k.
The specific calculation method for the weighting coefficients a, b, c, d, e and k is as follows:
(6.1) calculating to obtain an enhanced graph set gmCharacteristic value M of1、M2、M3、M4、M5 M6;
(6.2) averaging: mn=(M1+M2+M3+M4+M5+M6)/6;
(6.3) calculating the weight Wi=1/(|Mi-Mn|+1),i=1,2,3,4,5,6;
(6.4) calculating the weighting coefficients as:
a=W1/(W1+W2+W3+W4+M5+M6)
b=W2/(W1+W2+W3+W4+M5+M6)
c=W3/(W1+W2+W3+W4+M5+M6)
d=W4/(W1+W2+W3+W4+M5+M6)
e=W5/(W1+W2+W3+W4+M5+M6)
k=W6/(W1+W2+W3+W4+M5+M6)。
(7) aiming at the texture features extracted in the step (6), extracting fault features by adopting a gray-element-gradient co-occurrence matrix;
the method specifically comprises the following steps:
let B (m1, n1, p1) be the gray-cell-gradient three-dimensional co-occurrence matrix, and the three dimensions are respectively: f (u, w) is a graph gray matrix, TG (s, t) is a normalized graph element matrix, and QG (r, v) is a normalized graph gradient matrix;
converting the texture feature matrix h into an F (u, w) graph gray matrix, a TG (s, t) normalized graph primitive matrix and a QG (r, v) normalized graph gradient matrix;
the transformation of the three matrixes is the prior art, and can be specifically referred to;
1. research on a gray level-element co-occurrence matrix method for fault diagnosis of rotary machines, aeronautical dynamics, 2008,23(9), 1609-;
2. a gray gradient co-occurrence matrix method for rotary mechanical vibration fault diagnosis, which is reported in the aeronautics and dynamics, 2008,23(10): 1939-;
3. a rotary machine vibration time-frequency pattern recognition method based on a gray-gradient co-occurrence matrix is disclosed, wherein 2009,22(1):85-91 in the vibration engineering bulletin.
Counting the number of pixels, namely the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix, namely F (u, w) ═ m1, TG (s, t) ═ n1 and QG (r, v) ═ p1, namely counting the number of the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix
B(m1,n1,p1)#={(u,w),(s,t),(r,v)|F(u,w)=m1,TG(s,t)=n1,QG(r,v)=p1}
Wherein m1, n1, p1 are 1,2 … 64, and 64 are in B (m1, n1, p1) matrix3An element;
and extracting angular second moment, contrast, absolute value, contrast moment, autocorrelation coefficient and entropy from B (m1, n1, p1) to be used as a fault feature sample.
(8) And according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method. The method specifically comprises the following steps: and carrying out sample off-line training on the fault characteristics according to the BP neural network, then carrying out on-line acquisition on vibration data, extracting the fault characteristics by the same method, and diagnosing by using the trained neural network.
Furthermore, the invention also provides an intelligent fault diagnosis system for the rotating machinery, which comprises:
the signal acquisition module: arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data;
a difference value reconstruction preprocessing module: generating a three-dimensional parameter graph by using the acquired time series vibration signals; performing difference value reconstruction preprocessing on the three-dimensional parameter graph;
a gray level enhancement processing module: carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment;
the texture feature extraction module: adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancement pretreatment; performing texture feature extraction on the reinforced rotating machinery vibration parameter graph by adopting a fuzzy soft morphology composite edge detection operator;
the fault feature extraction module: aiming at the extracted textural features, extracting fault features by adopting a gray level-primitive-gradient co-occurrence matrix;
a fault diagnosis module: and according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method.
The method avoids the judgment which can only be carried out by field experts in the past, and improves the automatic diagnosis level; with the development of electronic information technology \ communication technology \ artificial intelligence technology and the utilization of internet of things \ mass big data, the invention adopts the combination of advanced fault feature extraction method and artificial intelligence method, thus improving the diagnosis speed and accuracy. The method is the first rotary machine fault diagnosis method based on the fuzzy soft morphology pattern recognition technology in China, can quickly and conveniently obtain different types of fault characteristics of equipment, can realize equipment state monitoring and fault diagnosis, and effectively improves the fault diagnosis accuracy. The rotary shaft can be applied to large rotary machines in the industries of petrochemical industry, metallurgy, electric power, aviation, aerospace and the like.
Claims (10)
1. The intelligent rotary machine fault diagnosis method based on fuzzy soft morphological pattern recognition is characterized by comprising the following steps of:
(1) arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data;
(2) generating a three-dimensional parameter graph by using the acquired time series vibration signals;
(3) performing difference value reconstruction preprocessing on the three-dimensional parameter graph;
(4) carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment;
(5) adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancing pretreatment in the step (4);
(6) performing texture feature extraction on the rotating machinery vibration parameter graph subjected to the enhancement processing in the step (5) by adopting a fuzzy soft morphology composite edge detection operator;
(7) aiming at the texture features extracted in the step (6), extracting fault features by adopting a gray-element-gradient co-occurrence matrix;
(8) and according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method.
2. The intelligent diagnosis method for the fault of the rotary machine based on the fuzzy soft morphology graph recognition is characterized in that: the step (2) of generating the three-dimensional parameter graph from the acquired time series vibration signals specifically comprises the following steps:
(2.1) arranging a vibration sensor on the rotary machine, and collecting vibration time sequence signals XnN is a positive integer, 1,2... No. N;
(2.2) to the vibration time series signal XnIntercepting to generate a matrix expression Y of the vibration time series signal:
and (2.3) carrying out spectral analysis on the matrix expression Y to obtain a three-dimensional parameter graph set f.
3. The intelligent diagnosis method for the fault of the rotary machine based on the fuzzy soft morphology graph recognition is characterized in that: and (5) enhancing the rotating machinery vibration parameter graph after the enhancement pretreatment by adopting a composite fuzzy soft form filter, wherein the enhancement treatment is specifically carried out by the following formula:
wherein, gmThe image set after the enhancement processing is performed, wherein m is 1,2,3,4,5,6, and alpha is a gray scale weight of the fuzzy soft-on operation; f is a graph set, b is a structural element, and f is corroded and expanded by b and is respectively defined as follows:
4. the intelligent diagnosis method for the fault of the rotary machine based on the fuzzy soft morphology graph recognition is characterized in that: the step (6) adopts a fuzzy soft morphology composite edge detection operator to extract the texture characteristics of the reinforced rotating machinery vibration parameter graph, and is specifically realized by the following formula:
h=a·g1+b·g2+c·g3+d·g4+e·g5+k·g6
wherein h is a graph set, namely texture characteristics, after the fuzzy soft morphology composite edge detection operator is processed;
g1=f-fΘb;
g5=f·b-f;
the weighting coefficients are a, b, c, d, e, k.
5. The intelligent diagnosis method for the fault of the rotating machinery based on the fuzzy soft morphology graph recognition is characterized in that: the specific calculation method for the weighting coefficients a, b, c, d, e and k is as follows:
(5.1) calculating to obtain an enhanced graph set gmCharacteristic value M of1、M2、M3、M4、M5 M6;
(5.2) averaging: mn=(M1+M2+M3+M4+M5+M6)/6;
(5.3) calculating the weight Wi=1/(|Mi-Mn|+1),i=1,2,3,4,5,6;
(5.4) calculating the weighting coefficients as:
a=W1/(W1+W2+W3+W4+M5+M6)
b=W2/(W1+W2+W3+W4+M5+M6)
c=W3/(W1+W2+W3+W4+M5+M6)
d=W4/(W1+W2+W3+W4+M5+M6)
e=W5/(W1+W2+W3+W4+M5+M6)
k=W6/(W1+W2+W3+W4+M5+M6)。
6. the intelligent diagnosis method for the fault of the rotary machine based on the fuzzy soft morphology graph recognition is characterized in that: the step (7) is to extract the fault features by adopting a gray-primitive-gradient co-occurrence matrix aiming at the extracted textural features, and specifically comprises the following steps:
let B (m1, n1, p1) be the gray-cell-gradient three-dimensional co-occurrence matrix, and the three dimensions are respectively: f (u, w) is a graph gray matrix, TG (s, t) is a normalized graph element matrix, and QG (r, v) is a normalized graph gradient matrix;
converting the texture feature matrix h into an F (u, w) graph gray matrix, a TG (s, t) normalized graph primitive matrix and a QG (r, v) normalized graph gradient matrix;
counting the number of pixels, namely the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix, namely F (u, w) ═ m1, TG (s, t) ═ n1 and QG (r, v) ═ p1, namely counting the number of the (m1, n1, p1) th element value of the gray-cell-gradient three-dimensional co-occurrence matrix
B(m1,n1,p1)#={(u,w),(s,t),(r,v)|F(u,w)=m1,TG(s,t)=n1,QG(r,v)=p1}
Wherein m1, n1, p1 are 1,2 … 64, and 64 are in B (m1, n1, p1) matrix3An element;
and extracting angular second moment, contrast, absolute value, contrast moment, autocorrelation coefficient and entropy from B (m1, n1, p1) to be used as a fault feature sample.
7. The intelligent diagnosis method for the fault of the rotating machinery based on the fuzzy soft morphology graph recognition is characterized in that: the step (8) adopts an artificial neural network method to diagnose the fault according to the extracted fault characteristics, and specifically comprises the following steps: and carrying out sample off-line training on the fault characteristics according to the BP neural network, then carrying out on-line acquisition on vibration data, extracting the fault characteristics by the same method, and diagnosing by using the trained neural network.
8. A rotary machine fault intelligent diagnosis system realized by the rotary machine fault intelligent diagnosis method based on the fuzzy soft morphology graph recognition according to claim 1, which is characterized by comprising:
the signal acquisition module: arranging a vibration sensor on a rotary machine to be diagnosed to acquire vibration signals, and forming time series vibration signals by the acquired data;
a difference value reconstruction preprocessing module: generating a three-dimensional parameter graph by using the acquired time series vibration signals; performing difference value reconstruction preprocessing on the three-dimensional parameter graph;
a gray level enhancement processing module: carrying out gray scale adaptive histogram equalization enhancement pretreatment on the graph subjected to the difference value reconstruction pretreatment;
the texture feature extraction module: adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancement pretreatment; performing texture feature extraction on the reinforced rotating machinery vibration parameter graph by adopting a fuzzy soft morphology composite edge detection operator;
the fault feature extraction module: aiming at the extracted textural features, extracting fault features by adopting a gray level-primitive-gradient co-occurrence matrix;
a fault diagnosis module: and according to the extracted fault characteristics, carrying out fault diagnosis by adopting an artificial neural network method.
9. The rotating machine fault intelligent diagnostic system of claim 8, wherein: the method for generating the three-dimensional parameter graph by using the acquired time series vibration signals specifically comprises the following steps:
(2.1) arranging a vibration sensor on the rotary machine, and collecting vibration time sequence signals XnN is a positive integer, 1,2... No. N;
(2.2) to the vibration time series signal XnIntercepting to generate a matrix expression Y of the vibration time series signal:
(2.3) carrying out spectral analysis on the matrix expression Y to obtain a three-dimensional parameter graph set f;
and (3) adopting a composite fuzzy soft form filter to enhance the rotating machinery vibration parameter graph after the enhancement pretreatment, and specifically carrying out the enhancement treatment by the following formula:
wherein, gmThe image set after the enhancement processing is performed, wherein m is 1,2,3,4,5,6, and alpha is a gray scale weight of the fuzzy soft-on operation; f is a graph set, b is a structural element, and f is corroded and expanded by b and is respectively defined as follows:
10. the rotating machine fault intelligent diagnostic system of claim 9, wherein: adopting a fuzzy soft morphology composite edge detection operator to extract the texture characteristics of the reinforced rotating machinery vibration parameter graph, and specifically realizing the method through the following formula:
h=a·g1+b·g2+c·g3+d·g4+e·g5+k·g6
wherein h is a graph set, namely texture characteristics, after the fuzzy soft morphology composite edge detection operator is processed;
g1=f-fΘb;
g5=f·b-f;
the weighting coefficients are a, b, c, d, e and k;
the specific calculation method for the weighting coefficients a, b, c, d, e and k is as follows:
(5.1) calculating to obtain an enhanced graph set gmCharacteristic value M of1、M2、M3、M4、M5 M6;
(5.2) averaging: mn=(M1+M2+M3+M4+M5+M6)/6;
(5.3) calculating the weight Wi=1/(|Mi-Mn|+1),i=1,2,3,4,5,6;
(5.4) calculating the weighting coefficients as:
a=W1/(W1+W2+W3+W4+M5+M6)
b=W2/(W1+W2+W3+W4+M5+M6)
c=W3/(W1+W2+W3+W4+M5+M6)
d=W4/(W1+W2+W3+W4+M5+M6)
e=W5/(W1+W2+W3+W4+M5+M6)
k=W6/(W1+W2+W3+W4+M5+M6)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010681363.1A CN111950377B (en) | 2020-07-15 | 2020-07-15 | Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010681363.1A CN111950377B (en) | 2020-07-15 | 2020-07-15 | Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111950377A true CN111950377A (en) | 2020-11-17 |
CN111950377B CN111950377B (en) | 2021-06-22 |
Family
ID=73341551
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010681363.1A Expired - Fee Related CN111950377B (en) | 2020-07-15 | 2020-07-15 | Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111950377B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113432709A (en) * | 2021-06-25 | 2021-09-24 | 湖南工业大学 | Visualization mechanical fault diagnosis method based on graphics |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200083902A1 (en) * | 2016-12-13 | 2020-03-12 | Idletechs As | Method for handling multidimensional data |
-
2020
- 2020-07-15 CN CN202010681363.1A patent/CN111950377B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200083902A1 (en) * | 2016-12-13 | 2020-03-12 | Idletechs As | Method for handling multidimensional data |
Non-Patent Citations (4)
Title |
---|
刘占生, 窦唯: "旋转机械振动参数图形边缘纹理提取的数学形态学方法", 《振动工程学报》 * |
刘占生,窦唯: "基于旋转机械振动参数图形融合灰度共生矩阵的故障诊断方法", 《中国电机工程学报》 * |
窦唯,刘占生,王政先,何鹏: "旋转机械状态参数图形识别的免疫-模糊形态学方法", 《航空动力学报》 * |
窦唯,刘占生: "旋转机械故障诊断的图形识别方法研究", 《振动与冲击》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113432709A (en) * | 2021-06-25 | 2021-09-24 | 湖南工业大学 | Visualization mechanical fault diagnosis method based on graphics |
CN113432709B (en) * | 2021-06-25 | 2023-08-08 | 湖南工业大学 | Visual mechanical fault diagnosis method based on graphics |
Also Published As
Publication number | Publication date |
---|---|
CN111950377B (en) | 2021-06-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Saufi et al. | Challenges and opportunities of deep learning models for machinery fault detection and diagnosis: A review | |
Zhao et al. | Intelligent diagnosis using continuous wavelet transform and gauss convolutional deep belief network | |
Liu et al. | Fault diagnosis of industrial wind turbine blade bearing using acoustic emission analysis | |
Liu et al. | Acoustic emission analysis for wind turbine blade bearing fault detection under time-varying low-speed and heavy blade load conditions | |
CN109297713B (en) | Steam turbine fault diagnosis method based on stable and non-stable vibration signal characteristic selection | |
CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
Liu et al. | Wind turbine blade bearing fault diagnosis under fluctuating speed operations via Bayesian augmented Lagrangian analysis | |
CN112461934B (en) | Aero-engine blade crack source positioning method based on acoustic emission | |
CN111397896A (en) | Fault diagnosis method and system for rotary machine and storage medium | |
CN110991295A (en) | Self-adaptive fault diagnosis method based on one-dimensional convolutional neural network | |
CN111626312A (en) | Wind turbine bearing fault diagnosis method based on fast multi-separation dictionary learning | |
Zhao et al. | Data augmentation via randomized wavelet expansion and its application in few-shot fault diagnosis of aviation hydraulic pumps | |
CN111950377B (en) | Rotary machine fault intelligent diagnosis method based on fuzzy soft morphological pattern recognition | |
CN113390631A (en) | Fault diagnosis method for gearbox of diesel engine | |
CN114548190A (en) | Wind turbine fault diagnosis method based on self-adaptive residual error neural network | |
CN116304751A (en) | Operation data processing method for overhauling motor train unit components | |
Xia et al. | Augmentation-based discriminative meta-learning for cross-machine few-shot fault diagnosis | |
CN111855208A (en) | Rotary machine weak fault signal extraction method based on order analysis and sparse coding | |
CN113240022A (en) | Wind power gear box fault detection method of multi-scale single-classification convolutional network | |
CN113537044A (en) | Aircraft engine fault diagnosis method based on STFT and improved DenseNet | |
CN113255458A (en) | Bearing fault diagnosis method based on multi-view associated feature learning | |
Liu et al. | Acoustic emission analysis for wind turbine blade bearing fault detection using sparse augmented Lagrangian algorithm | |
CN114964783B (en) | Gearbox fault detection model based on VMD-SSA-LSSVM | |
CN113639985B (en) | Mechanical fault diagnosis and state monitoring method based on optimized fault characteristic frequency spectrum | |
CN114136604A (en) | Rotary equipment fault diagnosis method and system based on improved sparse dictionary |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210622 |
|
CF01 | Termination of patent right due to non-payment of annual fee |