CN115064182A - Fan fault feature identification method of self-adaptive Mel filter in strong noise environment - Google Patents
Fan fault feature identification method of self-adaptive Mel filter in strong noise environment Download PDFInfo
- Publication number
- CN115064182A CN115064182A CN202210380360.3A CN202210380360A CN115064182A CN 115064182 A CN115064182 A CN 115064182A CN 202210380360 A CN202210380360 A CN 202210380360A CN 115064182 A CN115064182 A CN 115064182A
- Authority
- CN
- China
- Prior art keywords
- mel filter
- signal
- self
- adaptive
- filter
- 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 36
- 230000005236 sound signal Effects 0.000 claims abstract description 40
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 230000003044 adaptive effect Effects 0.000 claims description 15
- 230000000694 effects Effects 0.000 claims description 7
- 238000005311 autocorrelation function Methods 0.000 claims description 6
- 238000001228 spectrum Methods 0.000 claims description 5
- 230000003111 delayed effect Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000009432 framing Methods 0.000 claims description 2
- 238000004088 simulation Methods 0.000 claims 1
- 238000003745 diagnosis Methods 0.000 abstract description 8
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/24—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being the cepstrum
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/72—Wind turbines with rotation axis in wind direction
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention provides a fan fault feature identification method of a self-adaptive Mel filter in a strong noise environment, which comprises the following steps: s1: collecting sound signals generated when workshop fan equipment runs; s2: preprocessing the sound signal collected in the step S1 to enhance the target signal; s3: extracting the fundamental frequency of the preprocessed sound signal, and solving the fundamental frequency of each frame of signal by adopting a short-time autocorrelation method; s4: designing a self-adaptive Mel filter, taking each frame of signal as a research object, adding a filter taking the frame fundamental frequency as a central frequency, reconstructing a Mel filter bank, and filtering a sound signal by the self-adaptive Mel filter to obtain characteristic parameters; s5: and adding the noise with different intensities into the sound signal acquired in the step S1, extracting the characteristics in the step S4, sending the characteristics to a trained SVM model, and observing the recognition result of the acoustic characteristics in the environment with different noise intensities. The invention designs the self-adaptive Mel filter, and improves the efficiency and reliability of fan fault diagnosis in a strong noise environment.
Description
Technical Field
The invention relates to the field of a method for diagnosing faults of a fan device in a workshop in a strong noise environment, in particular to a method for identifying fault characteristics of a fan in a strong noise environment by using a self-adaptive Mel filter.
Background
The fan is one of the most widely used mechanical devices in a factory, and once a fault occurs, property loss and even a serious safety accident may be caused, so that fault diagnosis becomes a necessary means for preventing the accident. At present, a common traditional fault diagnosis signal is a vibration signal, but in many cases, the acquisition of the vibration signal is difficult, and the analysis is only limited to the situation that the local diagnosis and the fault development are serious, so that certain limitations exist. The sound transmission is non-contact, the collection is convenient, and the state information of the equipment operation is better contained, so the acoustic characteristic is a better analysis object for equipment fault diagnosis.
The noise of a workshop where the fan is located is generally high, and the identification accuracy of the traditional Mel cepstrum coefficient method is not ideal when the characteristics of the sound signal with low signal-to-noise ratio are identified. Therefore, in order to improve the fault feature recognition effect of mechanical equipment such as a fan and the like in a strong noise environment, the invention improves the traditional Mel filter and provides a fault feature recognition method of the fan equipment in a workshop in the strong noise environment based on the self-adaptive Mel filter, and the fan equipment has higher recognition accuracy rate and higher fault diagnosis reliability in the noise environment.
Disclosure of Invention
In order to solve the problems, the invention discloses a fan fault feature identification method of a self-adaptive Mel filter in a strong noise environment, aiming at the strong noise environment, a self-adaptive Mel filter group is constructed, so that the extracted feature parameters can reflect the features of signals per se better, and the equipment fault diagnosis effect is improved.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a fan fault feature identification method of a self-adaptive Mel filter under a strong noise environment comprises the following steps:
s1: collecting sound signals generated when workshop fan equipment runs;
s2: preprocessing the sound signal collected in the step S1 to enhance the target signal;
s3: extracting the fundamental frequency of the preprocessed sound signal, and solving the fundamental frequency of each frame of signal by adopting a short-time autocorrelation method;
s4: designing a self-adaptive Mel filter, taking each frame of signal as a research object, adding a filter taking a frame fundamental frequency as a central frequency, reconstructing a Mel filter bank, and filtering a sound signal by the self-adaptive Mel filter to obtain characteristic parameters;
s5: the sound signals collected in the step S1 are added with noises with different intensities, the characteristics are extracted in the step S4 and then sent to a trained SVM model, the recognition results of the acoustic characteristics under the noise conditions with different intensities are observed, and meanwhile, the recognition results are compared with the recognition results of the traditional Mel cepstrum coefficient method, so that the self-adaptive Mel filter has a better recognition effect under the condition of low signal to noise ratio.
The invention is further improved in that: in step S2, the preprocessing process uses a delay-sum beam forming algorithm to combine the sound signals collected by the multiple microphones, so as to suppress the interference signals in the non-target direction and enhance the sound signals in the target direction.
The fan fault feature identification method of the self-adaptive Mel filter under the strong noise environment is characterized in that: step S3 is to extract the fundamental frequency of the signal by using a short-time autocorrelation method, and determine the fundamental frequency of the signal by comparing the similarity between the original signal and the delayed signal, where the short-time autocorrelation function formula is as follows:
in the formula: s (N) is the sound signal, w (m) is a window function, τ is the delay of time, and N is the frame length. The distance between two maxima of the short-time autocorrelation function is found, and the ratio of the sampling frequency to the distance between the two maxima is the fundamental frequency.
The invention further improves that: in step S4, the method of adaptive mel-frequency filter is used to extract the sound features, and the specific method is as follows:
and S4a, designing an adaptive Mel filter. The fundamental frequency f extracted in the step S3 b With the central frequency vector (f) of the original Mel filter 1 ,f 2 ,...,f N ) Combining and reconstructing the central frequency vector (f) of the Mel filter 1 ,f 2 ,...f b ...,f N+1 ) Taking the constructed new vector as the center frequency of the adaptive Mel filter, and substituting the center frequency into the following formula:
in the formula: h m (k) Representing filter parameters, f m Represents the center frequency of the triangular filter; an adaptive mel filter bank is thus obtained.
And S4b, pre-emphasizing the pre-processed signal, framing and windowing to obtain a plurality of sound segments, sending the sound segments to a self-adaptive Mel filter for filtering, calculating the logarithmic spectrum of the sound segments, and finally performing discrete cosine transform to obtain characteristic parameters.
The invention further improves that: in the step S5, the sound signals collected in the step S1 are added into different signal-to-noise ratios and noises to simulate different workshop environments, the characteristics are extracted in the step S4 and then sent to a trained SVM model, recognition results under different noise intensity backgrounds are obtained through SVM classification, and meanwhile, the recognition results are compared with recognition results of a traditional Mel cepstrum coefficient method, and the result shows that the recognition effect of the self-adaptive Mel filter under the strong noise environment is better.
The invention has the beneficial effects that:
1. aiming at a strong noise environment of a workshop, a self-adaptive Mel filter bank is constructed, so that the extracted characteristic parameters can highlight the characteristics of the sound signals of the equipment;
2. the fault diagnosis recognition effect is improved in a strong noise environment, the reliability of the recognition result is improved, and the method has great significance in practical engineering application.
Drawings
Fig. 1 is a flow chart of a fan equipment fault feature identification method of a self-adaptive mel filter provided by the invention.
FIG. 2 is a flow chart of the feature extraction of the adaptive mel-frequency cepstrum coefficient in the invention.
FIG. 3 shows the fan sound feature recognition rate of two Mel filters in the invention under different signal-to-noise ratios.
Detailed Description
The present invention will be further illustrated with reference to the accompanying drawings and specific embodiments, which are to be understood as merely illustrative of the invention and not as limiting the scope of the invention. It should be noted that as used in the following description, the terms "front," "back," "left," "right," "upper" and "lower" refer to directions in the drawings, and the terms "inner" and "outer" refer to directions toward and away from, respectively, the geometric center of a particular component.
As shown in fig. 1, the method for identifying the fan fault characteristics of the adaptive mel filter in the strong noise environment includes the following steps:
s1: collecting sound signals generated when fan equipment in a workshop runs;
s2: preprocessing the sound signal collected in the step S1 to enhance the target signal;
and S2a, in the preprocessing process, a delay-sum beam forming algorithm is adopted to combine the sound signals collected by the multiple microphones, so that interference signals in a non-target direction are suppressed, and the sound signals in a target direction are enhanced.
S3: extracting the fundamental frequency of the preprocessed sound signal, and solving the fundamental frequency of each frame of signal by adopting a short-time autocorrelation method;
s3a, extracting signal fundamental frequency by adopting a short-time autocorrelation method, and determining the signal fundamental frequency by comparing the similarity between the original signal and the delayed signal thereof, wherein the short-time autocorrelation function formula is as follows:
in the formula: s (N) is the sound signal, w (m) is a window function, τ is the delay of time, and N is the frame length. The distance between two maxima of the short-time autocorrelation function is found, and the ratio of the sampling frequency to the distance between the two maxima is the fundamental frequency.
As shown in fig. 2, S4: designing a self-adaptive Mel filter, taking each frame of signal as a research object, adding a filter taking the frame fundamental frequency as a central frequency, reconstructing a Mel filter bank, and filtering a sound signal by the self-adaptive Mel filter to obtain characteristic parameters;
and S4a, designing a self-adaptive Mel filter. The fundamental frequency f extracted in the step S3 b With the central frequency vector (f) of the original Mel filter 1 ,f 2 ,...,f N ) Combining and reconstructing the central frequency vector (f) of the Mel filter 1 ,f 2 ,...f b ...,f N+1 ) Taking the constructed new vector as the center frequency of the adaptive Mel filter, and substituting the center frequency into the following formula:
in the formula: h m (k) Representing filter parameters, f m Represents the center frequency of the triangular filter; an adaptive mel filter bank is thus obtained.
S4b, pre-emphasizing the pre-processed signal to enhance the high-frequency component:
H(z)＝1-az -1
where a is the pre-emphasis factor, 0.9< a < 1.0.
Dividing the signal into frames, dividing the signals into short-time segments, and adding a Hanning window to each frame of signal to prevent frequency spectrum leakage, wherein the Hanning window formula is as follows:
the processed sound signal is sent to a self-adaptive Mel filter for filtering after fast Fourier transform, and then the logarithmic spectrum of the sound signal is calculated to enable the sound signal to have stronger robustness, wherein the logarithmic spectrum is as follows:
in the formula: h m (k) Is a filter, s (n) is a sound signal;
finally, discrete cosine transform is carried out to obtain characteristic parameters, and the formula is as follows:
in the formula: c (n) is the MFCC feature vector for each frame of signal.
S5: the sound signals collected in the step S1 are added with different signal-to-noise ratios and noises to simulate different workshop environments, and the sound signals are sent to a trained SVM model after the characteristics are extracted in the step S4, wherein the model can be expressed as
W*X+b＝0
In the formula: x is the feature vector of the data set, W is the weight, and b is the bias vector.
Constructing the objective function by the lagrange multiplier can be expressed as:
constraint conditions are as follows:
in the formula: n is the number of samples, λ i 、λ j Is the ith and j samples X i 、X j Corresponding Lagrange multiplier, y i 、y j The type value of the fault corresponding to the ith and jth samples.
The recognition results under different intensity noise backgrounds are obtained through SVM classification, and compared with the recognition results of the traditional Mel cepstrum coefficient method, the result shows that the feature recognition effect of the self-adaptive Mel filter under the strong noise environment is better.
The technical means disclosed in the invention scheme are not limited to the technical means disclosed in the above embodiments, but also include the technical scheme formed by any combination of the above technical features.
Claims (5)
1. The fan fault feature identification method of the self-adaptive Mel filter under the strong noise environment is characterized by comprising the following steps of: the method comprises the following steps:
s1: collecting sound signals generated when workshop fan equipment runs;
s2: preprocessing the sound signal collected in the step S1 to enhance the target signal;
s3: extracting the fundamental frequency of the preprocessed sound signal, and solving the fundamental frequency of each frame of signal by adopting a short-time autocorrelation method;
s4: designing a self-adaptive Mel filter, taking each frame of signal as a research object, adding a filter taking the frame fundamental frequency as a central frequency, reconstructing a Mel filter bank, and filtering a sound signal by the self-adaptive Mel filter to obtain characteristic parameters;
s5: and adding the noise with different intensities into the sound signal acquired in the step S1, extracting the characteristics in the step S4, sending the characteristics to a trained SVM model, observing the recognition result of the sound signal characteristics under the background of the noise with different intensities, and comparing the recognition result with the recognition result of the traditional Mel cepstrum coefficient method to obtain the self-adaptive Mel filter with better recognition effect under the strong noise environment.
2. The method for identifying the fault characteristics of the fan in the strong noise environment by the adaptive Mel filter as claimed in claim 1, wherein: in step S2, the preprocessing process uses a delay-sum beam forming algorithm to combine the sound signals collected by the multiple microphones, so as to suppress the interference signals in the non-target direction and enhance the sound signals in the target direction.
3. The method for identifying the fault characteristics of the fan in the strong noise environment by the adaptive Mel filter as claimed in claim 1, wherein: step S3 is to extract the fundamental frequency of the signal by using a short-time autocorrelation method, and determine the fundamental frequency of the signal by comparing the similarity between the original signal and the delayed signal, where the short-time autocorrelation function formula is as follows:
in the formula: s (N) is a sound signal, w (m) is a window function, tau is a time delay amount, and N is a frame length; the distance between two maxima of the short-time autocorrelation function is found, and the ratio of the sampling frequency to the distance between the two maxima is the fundamental frequency.
4. The method for identifying the fault characteristics of the fan in the strong noise environment by the adaptive Mel filter as claimed in claim 1, wherein: in step S4, a fan sound feature is extracted by using a self-adaptive mel filter, and the specific method is as follows:
s4a: designing an adaptive Mel filter; the fundamental frequency f extracted in the step S3 b With the central frequency vector (f) of the original Mel filter 1 ,f 2 ,...,f N ) Combining and reconstructing the central frequency vector (f) of the Mel filter 1 ,f 2 ,...f b ...,f N+1 ) Taking the constructed new vector as the center frequency of the adaptive Mel filter, and substituting the center frequency into the following formula:
in the formula: h m (k) Representing filter parameters, f m Represents the center frequency of the triangular filter; thus, an adaptive Mel filter bank can be obtained;
s4b: pre-emphasis, framing and windowing are carried out on the preprocessed signals to obtain a plurality of sound segments, the sound segments are sent to a self-adaptive Mel filter for filtering, the logarithmic spectrum of the sound segments is calculated, and finally discrete cosine transform is carried out to obtain characteristic parameters.
5. The method for identifying the fault characteristics of the fan in the strong noise environment by the adaptive Mel filter as claimed in claim 1, wherein: in step S5, the sound signal collected in step S1 is added into a noise simulation workshop environment with different intensities, and the sound signal is sent to a trained SVM model after the characteristics are extracted in step S4, wherein the model can be expressed as
W*X+b＝0
In the formula: x is the feature vector of the data set, W is the weight, b is the offset vector;
constructing the objective function by the lagrange multiplier can be expressed as:
constraint conditions are as follows:
in the formula: n is the number of samples, λ i 、λ j Is the ith and j samples X i 、X j Corresponding Lagrange multiplier, y i 、y j The type value of the fault corresponding to the ith and jth samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210380360.3A CN115064182A (en) | 2022-04-12 | 2022-04-12 | Fan fault feature identification method of self-adaptive Mel filter in strong noise environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210380360.3A CN115064182A (en) | 2022-04-12 | 2022-04-12 | Fan fault feature identification method of self-adaptive Mel filter in strong noise environment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115064182A true CN115064182A (en) | 2022-09-16 |
Family
ID=83197044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210380360.3A Pending CN115064182A (en) | 2022-04-12 | 2022-04-12 | Fan fault feature identification method of self-adaptive Mel filter in strong noise environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115064182A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114136888A (en) * | 2021-12-09 | 2022-03-04 | 四川启睿克科技有限公司 | Spectral data calibration method of multi-light-source portable near-infrared spectrometer |
-
2022
- 2022-04-12 CN CN202210380360.3A patent/CN115064182A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114136888A (en) * | 2021-12-09 | 2022-03-04 | 四川启睿克科技有限公司 | Spectral data calibration method of multi-light-source portable near-infrared spectrometer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109357749B (en) | DNN algorithm-based power equipment audio signal analysis method | |
| CN112257521B (en) | CNN underwater acoustic signal target identification method based on data enhancement and time-frequency separation | |
| CN108701469B (en) | Cough sound recognition method, device, and storage medium | |
| CN111261189B (en) | Vehicle sound signal feature extraction method | |
| CN102968990B (en) | Speaker identifying method and system | |
| CN104706321A (en) | MFCC heart sound type recognition method based on improvement | |
| CN111044814B (en) | Method and system for identifying transformer direct-current magnetic bias abnormality | |
| CN110987434A (en) | Rolling bearing early fault diagnosis method based on denoising technology | |
| CN109949823A (en) | A kind of interior abnormal sound recognition methods based on DWPT-MFCC and GMM | |
| CN112329914B (en) | Fault diagnosis method and device for buried transformer substation and electronic equipment | |
| CN108670291A (en) | The heart sound kind identification method of improved MFCC is combined based on EMD | |
| CN112052712B (en) | Power equipment state monitoring and fault identification method and system | |
| CN115064182A (en) | Fan fault feature identification method of self-adaptive Mel filter in strong noise environment | |
| CN112394324A (en) | Microphone array-based remote sound source positioning method and system | |
| CN108903914A (en) | A kind of heart sound kind identification method of the MFCC decomposed based on EMD | |
| CN114263621B (en) | Test method and system for centrifugal pump cavitation fault diagnosis simulation | |
| CN114023353A (en) | Transformer fault classification method and system based on cluster analysis and similarity calculation | |
| CN111785262B (en) | Speaker age and gender classification method based on residual error network and fusion characteristics | |
| CN110865375B (en) | Underwater target detection method | |
| CN113345443A (en) | Marine mammal vocalization detection and identification method based on mel-frequency cepstrum coefficient | |
| CN112786072A (en) | Ship classification and identification method based on propeller radiation noise | |
| CN112033656A (en) | Mechanical system fault detection method based on broadband spectrum processing | |
| CN115310477A (en) | Pump equipment fault sound detection method and system based on fractal features and predator algorithm | |
| CN110555457A (en) | Engine lubricating oil wear signal characteristic processing system and method | |
| CN112086105B (en) | Target identification method based on Gamma atom sub-band continuous spectrum characteristics |
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 |