CN110905735A - Blade fault diagnosis method based on sound periodicity - Google Patents
Blade fault diagnosis method based on sound periodicity Download PDFInfo
- Publication number
- CN110905735A CN110905735A CN201911302163.4A CN201911302163A CN110905735A CN 110905735 A CN110905735 A CN 110905735A CN 201911302163 A CN201911302163 A CN 201911302163A CN 110905735 A CN110905735 A CN 110905735A
- Authority
- CN
- China
- Prior art keywords
- frame
- category
- sound
- blade
- periodic
- 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
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
- F03D—WIND MOTORS
- F03D17/00—Monitoring or testing of wind motors, e.g. diagnostics
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
- F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
- F05B2260/00—Function
- F05B2260/80—Diagnostics
Abstract
The invention provides a blade fault diagnosis method based on sound periodicity, which comprises the following steps: firstly, extracting a feature vector matrix MFCC from an acquired sound signal of the leaf by using a Mel Frequency Cepstrum Coefficient (MFCC) method, and defining a periodic frame type according to the periodicity of the sound signal of the leaf; then, clustering all frame signals into two types by adopting a K _ means algorithm, and establishing a periodic frame type probability matrix according to a clustering result; and finally, determining the category of each final frame signal according to the probability matrix, and realizing the fault diagnosis of the fan blade through a bar graph on a category number time domain. The invention can realize the non-contact on-line monitoring and diagnosis of the blade, and the adopted sound signal is easy to obtain, thereby having potential engineering application value.
Description
Technical Field
The invention relates to a blade fault diagnosis method based on sound periodicity, which can improve the accuracy of fault diagnosis by considering the characteristic of sound periodicity of a wind driven generator blade when the blade fault diagnosis is carried out by using an unsupervised learning K _ means algorithm, and belongs to the technical field of fault diagnosis.
Background
The wind power plant blades often have the faults of cracking, surface shedding, abrasion and the like when operating in a severe open environment for a long time, and the faults of the blades not only influence the efficiency of a fan for capturing wind energy, but also reduce the service life of the blades and increase the operation and maintenance cost.
At present, fault diagnosis of blades based on sound signals has potential application value, and the fault diagnosis steps aiming at the blades in domestic and foreign researches are generally divided into two parts, wherein the first step is to extract characteristic information from the sound signals of the fan blades by using a theoretical algorithm of signal processing, and the second step is to classify the sound signals by using machine learning intelligent algorithms such as a neural network, an SVM, a cluster and the like, so that the fault diagnosis of the blades is realized. In practice, the blade rotates for a circle periodically, and the sound collecting position of the blade is right below the windward side of the impeller, so that the sound signal of the blade also has periodicity.
Disclosure of Invention
The invention aims to research a blade fault diagnosis method based on sound periodicity, and introduces the invention content from the following 5 steps:
step 1: extracting characteristic information from a section of sound signal by using an existing signal processing mode MFCC algorithm;
step 2: according to the actual situation, the time of one rotation of the blade is T, k represents the kth frame signal, the value range of k is [1, n ], and n represents the total frame number of the divided sound signals; setting the sampling frequency of the sound sensor to fs, then the number of divided frames in one period of the blade sound is N, and the calculation formula of N is as follows:
N＝(T*fs-wlen+inc)/inc (1)
wherein wlen is the frame length initially set, inc is the frame shift, q represents the q frame signal in a period, the value range of q is [1, N ], the same signal frame in each period can be found according to the time sequence, and the expression is as follows:
k＝q+(λ-1)*N (2)
wherein λ represents the λ -th period of the sound signal and is in a rangeThe circumference is lambda is more than or equal to 1, and the maximum value is lambdamaxN/N, the N frame signals can be divided into N types of frame signals according to the above formula, each type of frame signal contains λ identical frames, and these types are defined as periodic frame types.
And step 3: based on a K _ means algorithm, all frame signals are clustered into two types, and each frame signal is labeled with a type according to a clustering result;
and 4, step 4: after the clustering in the step 3, defining the number of the frames clustered into the category 1 divided by the value of λ as the probability of the periodic frame category divided into the category 1, and similarly, defining the number of the frames clustered into the category 2 divided by the value of λ as the probability of the periodic frame category divided into the category 2, wherein the size of the matrix is N × 2, and establishing a probability matrix of the periodic frame category.
And 5: and finally classifying the periodic frame types according to the frame type corresponding to the maximum probability, and diagnosing the fault of the fan blade by depicting a time domain graph of the frame type label.
From the practical application angle, the method is different from the current intelligent diagnosis algorithm based on a large number of data samples, does not need to acquire a large number of data information, does not need to establish a diagnosis and identification model, can diagnose whether the fan is in fault only by reference and comparison among three blades of a single fan, and is simple in operation process and high in diagnosis precision; from the algorithm perspective, the method optimizes the clustering algorithm with periodicity of the samples, so that the clustering effect is better.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a method for diagnosing blade faults based on sound periodicity according to the present disclosure;
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.
As shown in fig. 1, a flow chart of a blade fault diagnosis method based on sound periodicity according to the present invention includes the steps of:
step 1, acquiring a section of sound signal s (t) under an impeller of a fan needing fault diagnosis, wherein the sound signal at least needs three periods including a blade rotation period, and extracting a feature vector matrix MFCC (s, n) by utilizing a Mel Frequency Cepstrum Coefficient (MFCC) method of the existing sound feature extraction technology, wherein a row of the MFCC matrix represents the number of frames into which the section of sound signal is divided, namely n represents the feature dimension of each frame of signal, and is generally 12-16;
step 2, dividing the same type of frames according to the periodicity of the blade sound, wherein according to the actual situation, the time of one rotation of the blade is T, k represents a k frame signal, and the value range of k is [1, n ]; setting the sampling frequency of the sound sensor to fs, then the number of divided frames in one period of the blade sound is N, and the calculation formula of N is as follows:
N＝(T*fs-wlen+inc)/inc (1)
wherein wlen is the frame length initially set, inc is the frame shift, and q is defined to represent the q frame signal in a period, the value range of q is [1, N ], and the same signal frame in each period can be found according to the time sequence, and the expression is as follows:
k＝q+(λ-1)*N (2)
in the formula, λ represents the λ -th period of the sound signal, the value range is λ ≧ 1, and the maximum value is λmaxN/N, the N frame signals can be divided into N types of frame signals according to the above formula, each type of frame signal contains λ identical frames, and these types are defined as periodic framesA category;
step 3, utilizing K _ means to gather the n frame signals into two categories, and dividing the n frame signals into two categories, namely category 1 and category 2, wherein each frame signal marks a corresponding category label according to a clustering result, and the frame signals in the periodic frame category at the moment can belong to one category or different categories;
step 4, for the lambda frame signals in each periodic frame category, the numerical value of dividing the number of frames clustered into the category 1 by the lambda is defined as the probability of dividing the periodic frame category into the category 1, and similarly, the numerical value of dividing the number of frames clustered into the category 2 by the lambda is defined as the probability of dividing the periodic frame category into the category 2, and a probability matrix of the periodic frame category is established according to the principle, wherein the matrix is N2 in large size;
and 5, carrying out final classification on each periodic frame type according to the frame type corresponding to the maximum probability, namely uniformly dividing each frame in the periodic frame types into a type 1 or a type 2, re-marking the number of each frame at the moment, depicting a time sequence diagram of each frame type label on a time domain according to the time of each frame signal, and further diagnosing whether the blade of the fan is in a fault state or not by analyzing the periodicity of the time sequence diagram.
The principle and the implementation mode of the invention are explained by applying specific embodiments in the invention, and the description of the embodiments is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (4)
1. A blade fault diagnosis method based on sound periodicity is characterized by comprising the following steps:
step 1, extracting sound characteristic information from a signal to be detected by using an MFCC algorithm;
step 2, dividing periodic frame categories according to the periodicity of the blades, wherein each category comprises frame signals with periodicity under different periods;
step 3, based on the K _ means algorithm, clustering all frame signals into two types, and marking the type of each frame signal according to the clustering result;
step 4, establishing a periodic probability matrix of the frame type according to the clustering result of the K _ means algorithm;
and 5, re-marking the category of each frame signal according to the probability matrix, and diagnosing the fault of the fan blade by depicting a time domain graph of a frame category label.
2. The method for diagnosing blade fault based on sound periodicity as claimed in claim 1, wherein the frame classification characteristics in step 2 are: according to the actual situation, the time of one rotation of the blade is T, k represents the kth frame signal, the value range of k is [1, n ], and n represents the total frame number of the divided sound signals; setting the sampling frequency of the sound sensor to fs, then the number of divided frames in one period of the blade sound is N, and the calculation formula of N is as follows:
N＝(T*fs-wlen+inc)/inc (1)
wherein wlen is the frame length initially set, inc is the frame shift, q represents the q frame signal in a period, the value range of q is [1, N ], the same signal frame in each period can be found according to the time sequence, and the expression is as follows:
k＝q+(λ-1)*N (2)
in the formula, λ represents the λ -th period of the sound signal, the value range is λ ≧ 1, and the maximum value is λmaxN/N, the N frame signals can be divided into N types of frame signals according to the above formula, each type of frame signal contains λ identical frames, and these types are defined as periodic frame types.
3. The method of claim 1, wherein the probability matrix in step 4 is characterized by: for λ frame signals in each periodic frame category, after the clustering in step 3, the frame signals in the periodic frame categories may belong to one category or may belong to different categories, so the number of frames clustered into category 1 divided by the value of λ is defined as the probability of the periodic frame category being divided into category 1, and similarly, the number of frames clustered into category 2 divided by the value of λ is defined as the probability of the periodic frame category being divided into category 2, where the matrix size is N × 2.
4. The method for diagnosing blade fault based on sound periodicity as claimed in claim 1, wherein the category of the frame signal in the step 5 is characterized by: and finally classifying the periodic frame types according to the frame type corresponding to the maximum probability, namely uniformly dividing the frames in the periodic frame types into a type 1 or a type 2, and re-marking the type numbers of the frames at the moment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911302163.4A CN110905735A (en) | 2019-12-17 | 2019-12-17 | Blade fault diagnosis method based on sound periodicity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911302163.4A CN110905735A (en) | 2019-12-17 | 2019-12-17 | Blade fault diagnosis method based on sound periodicity |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110905735A true CN110905735A (en) | 2020-03-24 |
Family
ID=69826114
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911302163.4A Pending CN110905735A (en) | 2019-12-17 | 2019-12-17 | Blade fault diagnosis method based on sound periodicity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110905735A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112302976A (en) * | 2020-09-18 | 2021-02-02 | 河北建投新能源有限公司 | Fan blade fault early warning method based on entropy weight method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2016125947A (en) * | 2015-01-07 | 2016-07-11 | 三菱重工業株式会社 | Diagnosis system of wind power generating installation and diagnosis method |
| CN106762452A (en) * | 2016-12-08 | 2017-05-31 | 东北大学 | Fan master control system fault diagnosis and on-line monitoring method based on data-driven |
| KR101789923B1 (en) * | 2016-06-02 | 2017-10-25 | 두산중공업 주식회사 | Wind farm supervision monitoring method |
| CN110259648A (en) * | 2019-07-05 | 2019-09-20 | 河北工业大学 | A kind of fan blade method for diagnosing faults based on optimization K-means cluster |
-
2019
- 2019-12-17 CN CN201911302163.4A patent/CN110905735A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2016125947A (en) * | 2015-01-07 | 2016-07-11 | 三菱重工業株式会社 | Diagnosis system of wind power generating installation and diagnosis method |
| KR101789923B1 (en) * | 2016-06-02 | 2017-10-25 | 두산중공업 주식회사 | Wind farm supervision monitoring method |
| CN106762452A (en) * | 2016-12-08 | 2017-05-31 | 东北大学 | Fan master control system fault diagnosis and on-line monitoring method based on data-driven |
| CN110259648A (en) * | 2019-07-05 | 2019-09-20 | 河北工业大学 | A kind of fan blade method for diagnosing faults based on optimization K-means cluster |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112302976A (en) * | 2020-09-18 | 2021-02-02 | 河北建投新能源有限公司 | Fan blade fault early warning method based on entropy weight method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107220469B (en) | Method and system for estimating state of fan | |
| CN111914883B (en) | Spindle bearing state evaluation method and device based on deep fusion network | |
| CN110569925B (en) | LSTM-based time sequence abnormality detection method applied to power equipment operation detection | |
| CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
| CN111352977B (en) | Time sequence data monitoring method based on self-attention bidirectional long-short term memory network | |
| CN110259648B (en) | Fan blade fault diagnosis method based on optimized K-means clustering | |
| CN112660745B (en) | Intelligent diagnosis method and system for carrier roller fault and readable storage medium | |
| CN111060337A (en) | Running equipment real-time fault diagnosis method based on expert system | |
| CN113076834B (en) | Rotating machine fault information processing method, processing system, processing terminal, and medium | |
| CN112836720B (en) | Building operation and maintenance equipment abnormality diagnosis method, system and computer readable storage medium | |
| CN109100648A (en) | Ocean current generator impeller based on CNN-ARMA-Softmax winds failure fusion diagnosis method | |
| CN108335019A (en) | The fault diagnosis recognition methods of fan yaw system | |
| CN113011763A (en) | Bridge damage identification method based on space-time diagram convolution attention | |
| Joshuva et al. | A study of various blade fault conditions on a wind turbine using vibration signals through histogram features | |
| CN110905735A (en) | Blade fault diagnosis method based on sound periodicity | |
| CN110874584B (en) | Blade fault diagnosis method based on improved prototype clustering | |
| CN108106717A (en) | A kind of method based on voice signal identification set state | |
| CN114330141A (en) | Fan main shaft bearing service life prediction method based on GRU (generalized regression Unit) super-parameter optimization | |
| CN114764570A (en) | Blade fault diagnosis method, device and system and storage medium | |
| CN113887586A (en) | FFCNN-SVM transfer learning fault diagnosis method based on feature fusion under small sample | |
| Saha et al. | Investigating Rotor Conditions on Wind Turbines Using Integrating Tree Classifiers | |
| CN111397884A (en) | Blade fault diagnosis method for improving Mel cepstrum coefficient algorithm | |
| CN116819318B (en) | Motor fault detection method and system | |
| CN116070102A (en) | Fan blade icing detection method and system based on Beta variogram self-encoder | |
| CN117332342B (en) | Machine pump equipment operation fault classification method and device based on semi-supervised learning |
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 |