CN110487917A - A kind of measure on stress pulse neural network based and analysis system - Google Patents
A kind of measure on stress pulse neural network based and analysis system Download PDFInfo
- Publication number
- CN110487917A CN110487917A CN201910814133.5A CN201910814133A CN110487917A CN 110487917 A CN110487917 A CN 110487917A CN 201910814133 A CN201910814133 A CN 201910814133A CN 110487917 A CN110487917 A CN 110487917A
- Authority
- CN
- China
- Prior art keywords
- stress
- stress pulse
- neural network
- frequency
- module
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/14—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object using acoustic emission techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4409—Processing the detected response signal, e.g. electronic circuits specially adapted therefor by comparison
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/4481—Neural networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N29/00—Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
- G01N29/44—Processing the detected response signal, e.g. electronic circuits specially adapted therefor
- G01N29/46—Processing the detected response signal, e.g. electronic circuits specially adapted therefor by spectral analysis, e.g. Fourier analysis or wavelet analysis
Abstract
The system that the present invention proposes a kind of measure on stress pulse based on machine learning and analysis, the operation workflow of whole system is as follows: mixing vibration signal and the stress wave signal of other low-frequency noises to what part of appliance generated, shockwave sensor detects and amplifies and be filtered, and then demodulation obtains simulation stress wave signal;Analog signal is received using digital processing element later, carry out analog-to-digital conversion and data processing and is sent to computer;Computer is handled and is analyzed to the raw data file of certain format, diagnosis report is made, wherein the method that analysis and diagnosis step uses neural network, when using PNN network, as network inputs, output layer is the decision for compareing corresponding event and making for time domain and frequency domain character.
Description
Technical field
The present invention relates to a kind of testing and analysis system, in particular to a kind of measure on stress pulse neural network based and analysis
System.
Background technique
The trouble shooting of equipment and on-call maintenance are the safe and reliable operations being not only to equipment itself, and to product matter
The guarantee of amount, production efficiency.Traditional vibration analysis method needs to ring from very big range of signal by the frequency of relatively flat
Fault-signal should be looked for detect the frequecy characteristic variation of wide scope, it is unwise to the slight change of machine friction in failure early stage
Sense.Only after failure deterioration, level of vibration is significantly increased, and vibrating sensor can just detect exception.
It is high that therefore, it is necessary to a kind of sensitivitys, and the maintenance cost equipment low with replacement equipment cost carries out failure to equipment
Detection.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology, provide a kind of measure on stress pulse neural network based with
Analysis system, shockwave sensor have very narrow frequency range and high frequency response, therefore non-to the defect of mechanical surface
It is often sensitive, convenient for the fault detection to equipment;Stress wave monitoring is a kind of predictive maintenance means, supports actively to repair, Ke Yigeng
Timely and effectively equipment fault is found and handled, reduces the cost of maintenance and more exchange device;In the system, neural network
The equipment health status reflected to stress wave signal, fault type can be helped to make more automation, intelligentized judgement, increased
The applicability of feeder apparatus.
In order to solve the above technical problems, the technical solution adopted by the present invention is that: a kind of stress wave inspection neural network based
Survey and analysis system, including shockwave sensor, digital processing element and analysis module;Wherein, the shockwave sensor
For the stress wave signal for the generation that rubs between part of appliance to be converted to stress pulse sequence;The digital processing element is used for
The stress pulse sequence from multiple shockwave sensors is received, and stress wave pulse train is converted and located
Reason;The digital stress pulse wave number that the analysis module sends the digital processing element from time domain and frequency domain is according to carrying out letter
Breath extracts, and is handled using information of the neural network to extraction, obtains corresponding Trouble Report.
Preferably, the shockwave sensor includes inverting element, the analog signal being connected with inverting element modulation
Network and demodulator, wherein the inverting element is used to the stress wave signal for the generation that rubs between part of appliance being converted to electricity
Signal is sent to the analog signal modulating network, and the high freguency bandpass filter in the analog signal modulating network is by telecommunications
Number amplification and filtration treatment, the demodulator for will amplification and filtration treatment after electric signal is converted to stress wave impulse sequence
Column.
Preferably, the digital processing element include power module, communication module, high speed AD sampling module, CPU and
CPLD;Wherein, the power module is for powering;The communication module with host computer for communicating, and by treated, signal is sent out
It send to host computer;The high speed AD sampling module, for receiving the stress wave impulse sequence from multiple shockwave sensors
It arranges and completes analog-to-digital conversion;The CPLD is used to carry out data processing to stress wave pulse train;The CPU be used for CPLD it
Between data transmission and synchronously control.
Preferably, the analysis module includes information extraction modules and neural computing module;Wherein, the information mentions
Modulus block is by stress wave energy diagram and/or histogram and/or spectrum analysis, to digital stress pulse wave from time domain and frequency domain
Data carry out information extraction;The neural computing module, the information for extracting the information extraction modules pass through mind
It is handled through network method, generates Trouble Report.
Preferably, the shockwave sensor meets following standard: having under the shockwave sensor resonant frequency
Resonance gain;The total energy value of resonant energy integral is provided in +/- the 10% of the standard value margin of tolerance, and can be used
Standard test fixture measurement obtains the normal data for calibration;Resonance exports the amplitude that half is decayed in some cycles,
And the 20% of initial communication is brought down below in the periodicity of corner frequency for corresponding to demodulator low pass.
Preferably, the digital stress pulse wave number that the analysis module sends the digital processing element is according to carrying out time domain
When feature extraction: by digital stress pulse wave number according to certain compression is carried out, forming multiple waveforms feature;Concrete mode are as follows: with
Certain time length calculates the maximum of data in each window, minimum, mean value, variance as a window.
Preferably, the digital stress pulse wave number that the analysis module sends the digital processing element is according to carrying out frequency domain
When feature extraction: first to digital stress pulse wave number according to the processing for carrying out FFT module, wherein spectrum density is by root-mean-square value to one
The data for determining number are averagely obtained, and there are certain data overlap rate, resolution ratio, the frequency ranges of setpoint frequency;It
Each stress spectral density is converted to a table afterwards, table content is preceding certain amount frequency line and corresponding each letter
Number amplitude;Frequency line is divided into a certain number of sections with certain frequency interval, calculates each section of amplitude peak, most slightly
Degree, average amplitude, amplitude variance, using all sections of 4 parameters as frequency domain character.
Preferably, the neural computing module generates the obtained temporal signatures and the frequency domain character
Nonlinear Mapping output carries out fault diagnosis;Wherein, the data set that the temporal signatures and the frequency domain character are constituted, a part
Neural network is constructed as training data, a part is assessed as test set, and decision-making value is arranged, and compares to form failure and examine
Disconnected report.
Preferably, the neural computing module includes probabilistic neural network, counterpropagation network, learning vector quantizations
One of network, radial primary function network, adaptive resonance theory prototype network, adaptive organising map network are a variety of.
Compared with prior art, the beneficial effects of the present invention are:
1, shockwave sensor has very narrow frequency range and high frequency response, therefore non-to the defect of mechanical surface
It is often sensitive, convenient for the fault detection to equipment;
2, stress wave monitoring is a kind of predictive maintenance means, supports actively to repair, can be more timely and effectively to equipment event
Barrier is found and is handled, and the cost of maintenance and more exchange device is reduced;
3, in the system, neural network can help the equipment health status reflected to stress wave signal, fault type
Make more automation, intelligentized judgement.
It should be appreciated that aforementioned description substantially and subsequent detailed description are exemplary illustration and explanation, it should not
As the limitation to the claimed content of the present invention.
Detailed description of the invention
With reference to the attached drawing of accompanying, the more purposes of the present invention, function and advantage are by the as follows of embodiment through the invention
Description is illustrated, in which:
Fig. 1 diagrammatically illustrates overall structure diagram of the invention;
Fig. 2 diagrammatically illustrates inside idiographic flow schematic diagram of the invention.
In figure:
Specific embodiment
By reference to exemplary embodiment, the purpose of the present invention and function and the side for realizing these purposes and function
Method will be illustrated.However, the present invention is not limited to exemplary embodiment as disclosed below;Can by different form come
It is realized.The essence of specification is only to aid in those skilled in the relevant arts' Integrated Understanding detail of the invention.
Hereinafter, the embodiment of the present invention will be described with reference to the drawings.In the accompanying drawings, identical appended drawing reference represents identical
Or similar component or same or like step.
Stress wave is the form of the high-frequency structure sound as caused by the friction between moving component.The analysis of stress wave is related to
It include the noise of equipment normal operation and the elimination of vibration and the detection of high-frequency sound and amplification to interference signal.Stress wave
Signal frequency range (35-40KHz) is much higher than structural vibration frequency (usually in 0-20KHz).Because shockwave sensor has very
Narrow frequency range and high frequency response, thus it is very sensitive to the defect of mechanical surface, and this is based on stress wave principle
Equipment monitoring compared to conventional method advantage.And stress wave monitoring is a kind of predictive maintenance means, supports actively to repair,
Can more timely and effectively equipment fault be found and be handled, reduce the cost of maintenance and more exchange device.
The analysis of stress wave signal is related to the relevant knowledge of signal processing, is mainly analyzed and is sentenced from time domain, frequency domain
It is disconnected.Three big master tools of analysis are: stress wave energy diagram, histogram, spectrum analysis.Neural network can then be helped to stress
Equipment health status that wave signal is reflected, fault type make more automation, intelligentized judgement.
The system that the present invention proposes a kind of measure on stress pulse based on machine learning and analysis passes through stress wave principle, letter
Number processing the relevant technologies, neural network method realize the monitoring of the health status of shop equipment and the analysis of fault type, main
It constitutes including 3 three shockwave sensor 1, digital processing element 2, analysis module parts (as shown in Figure 1).
One basic shockwave sensor has certain frequency response and damping characteristic, is designated specifically to stress wave
Analysis, preferentially meets following standard:
(1) have certain resonance gain (representative value: 30dB) with the suitable of proof stress wave under its dominant resonant frequency
When selective amplification;
(2) in the specified margin of tolerance (+/- 10%) of standard value provides the total energy value of resonant energy integral, and
Standard test fixture measurement can be used and obtain the normal data for calibration;
(3) resonance output (representative value: 5) decays to the amplitude of half, and low corresponding to demodulator in some cycles
The 20% of initial communication is brought down below in the periodicity of logical corner frequency.
One basic shockwave sensor 1 includes at least: generating the inverting element 101 of dominant resonant frequency;With transducing
The connected analog signal modulating network 102 of element 101.The wherein stress that inverting element 101 generates the friction between part of appliance
Wave signal is converted to electric signal and is sent to analog signal modulation 102, using piezoelectric transducer;In analog signal modulating network 102
High freguency bandpass filter electric signal is amplified and is filtered, get rid of equipment work normally generate low-frequency noise and vibration
Dynamic signal obtains stress wave signal, then obtains stress pulse sequence by demodulator 103.
Digital processing element 2 mainly includes power module 203, communication module 205, high speed AD sampling module 201, CPU
204, CPLD 202 etc..Wherein power module 203 is responsible for power supply;Communication module 205 be responsible for the unit and other units or on
Position machine communication;High speed AD sampling module 201 be responsible for receive the signal from multiple shockwave sensors and complete analog-to-digital conversion and
Processing;CPU 204 is mainly responsible for the transmission of the data between CPLD 202 and synchronously control etc.;CPLD 202 is responsible for stress wave
Energy datum is converted and is handled.
Analysis module 3 is mainly the data for handling and analyzing digital processing element and send, and is formed to equipment health shape
The assessment report of state.It is made of two parts, information extraction modules and neural computing module;Wherein, the information extraction
Module is by stress wave energy diagram and/or histogram and/or spectrum analysis, to digital stress pulse wave number from time domain and frequency domain
According to progress information extraction;The neural computing module, the information for extracting the information extraction modules pass through nerve
Network method is handled, and Trouble Report is generated, i.e., extracts first with the relevant knowledge of signal processing from time domain and frequency domain 302
Information, handling the information of extraction using neural network, realizes the intelligence and automation of analysis, also can be used as later
The reference of three big master tools.
When using PNN network in neural computing module 303, process is described as follows:
(1) temporal signatures extraction is carried out according to 301 to digital stress pulse wave number.
The existence form of digital stress wave train of pulse is with the binary system lattice of the sampling of certain sampling rate, certain time
Formula document, each document have a certain size.
Temporal signatures, which extract, to be considered to carry out each document into certain compression, forms multiple waveforms feature.Concrete mode are as follows:
Using certain time length as a window, the maximum of data in each window, minimum, mean value, variance are calculated.One text
This 4 parameters of all windows are used as compressed temporal signatures in shelves.
Temporal signatures generally reflect whether macroscopical operating condition of equipment is normal.
(2) frequency domain character extraction is carried out according to 301 to digital stress pulse wave number.
Concrete mode are as follows: first the data of original document are carried out with the processing of FFT module, spectrum density is by root-mean-square value to certain
The document data of number is averagely obtained, and there are certain data overlap rate, resolution ratio, the frequency range of frequency can be set
It is fixed.
Each stress spectral density be converted to a table later, table content be certain amount frequency line and
The amplitude of corresponding each signal.
Frequency line is divided into a certain number of sections with certain frequency interval, calculates each section of amplitude peak, most slightly
Degree, average amplitude, amplitude variance, using all sections of 4 parameters as frequency domain character.
Frequency domain character can reflect the concrete type of failure and positioning.
(3) temporal signatures obtained by the above method and frequency domain character are generated using neural computing module 303 non-thread
Property mapping output, carry out the diagnosis of failure.Data set a part constructs neural network, a part of conduct as 304 data of training
305 collection of test is assessed, and certain decision-making value is arranged, compares to form fault diagnosis report.Settable different decision threshold
Value 306 is generated to the correct alarm rate of corresponding event, correctly the numerical value of judgement rate, false alarm rate, rate of failing to report etc. compares, to network
It optimizes, obtains Trouble Report 307.
There are many selections for neural computing module 303, remove probabilistic neural network (PNN), there are also counterpropagation networks
(BPN), LVQ Networks (LVQ), radial primary function network (RBF), adaptive resonance theory prototype network (ART),
One of adaptive organising map network (SOM) is a variety of.
The beneficial effects of the present invention are: shockwave sensor has very narrow frequency range and high frequency response, therefore
It is very sensitive to the defect of mechanical surface, convenient for the fault detection to equipment;Stress wave monitoring is a kind of predictive maintenance means, branch
It holds and actively repairs, can more timely and effectively equipment fault be found and be handled, reduce the cost of maintenance and more exchange device;
In the system, it is more automatic that neural network can help the equipment health status reflected to stress wave signal, fault type to make
Change, intelligentized judgement.
In conjunction with the explanation and practice of the invention disclosed here, the other embodiment of the present invention is for those skilled in the art
It all will be readily apparent and understand.Illustrate and embodiment is regarded only as being exemplary, true scope of the invention and purport are equal
It is defined in the claims.
Claims (9)
1. a kind of measure on stress pulse neural network based and analysis system characterized by comprising shockwave sensor, number
Word processing unit and analysis module;
Wherein, the shockwave sensor is used to the stress wave signal for the generation that rubs between part of appliance being converted to stress wave impulse
Sequence;
The digital processing element is used to receive the stress pulse sequence from multiple shockwave sensors, and to stress
Pulse sequence is converted and is handled;
The digital stress pulse wave number that the analysis module sends the digital processing element from time domain and frequency domain is according to progress
Information extraction is handled using information of the neural network to extraction, obtains corresponding Trouble Report.
2. measure on stress pulse neural network based according to claim 1 and analysis system, which is characterized in that described to answer
Wave sensor includes inverting element, the analog signal modulating network and demodulator that are connected with the inverting element,
Wherein, the inverting element is used to the stress wave signal for the generation that rubs between part of appliance being converted to electric signal, is sent to
Electric signal is amplified and is filtered by the analog signal modulating network, the high freguency bandpass filter in the analog signal modulating network
Processing, the demodulator are used to the electric signal after amplification and filtration treatment being converted to stress pulse sequence.
3. measure on stress pulse neural network based according to claim 1 and analysis system, which is characterized in that the number
Word processing unit includes power module, communication module, high speed AD sampling module, CPU and CPLD;
Wherein, the power module is for powering;The communication module with host computer for communicating, and by treated, signal is sent
To host computer;The high speed AD sampling module, for receiving the stress pulse sequence from multiple shockwave sensors
And complete analog-to-digital conversion;The CPLD is used to carry out data processing to stress wave pulse train;The CPU is used between CPLD
Data transmission and synchronously control.
4. measure on stress pulse neural network based according to claim 1 and analysis system, which is characterized in that described point
Analysing module includes information extraction modules and neural computing module;
Wherein, the information extraction modules are by stress wave energy diagram and/or histogram and/or spectrum analysis, from time domain and frequency
To digital stress pulse wave number according to progress information extraction on domain;
The neural computing module, the information for extracting the information extraction modules are carried out by neural network method
Processing generates Trouble Report.
5. measure on stress pulse neural network based according to claim 1 or 2 and analysis system, which is characterized in that institute
It states shockwave sensor and meets following standard:
There is resonance gain under the shockwave sensor resonant frequency;
The total energy value of resonant energy integral is provided in +/- the 10% of the standard value margin of tolerance, and standard can be used
Test device measurement obtains the normal data for calibration;
Resonance exports the amplitude that half is decayed in some cycles, and in the period for the corner frequency for corresponding to demodulator low pass
The 20% of initial communication is brought down below in number.
6. measure on stress pulse neural network based according to claim 1 or 4 and analysis system, which is characterized in that institute
When stating digital stress pulse wave number that analysis module sends the digital processing element according to temporal signatures extraction is carried out:
By digital stress pulse wave number according to certain compression is carried out, multiple waveforms feature is formed;Concrete mode are as follows: with certain time
Length calculates the maximum of data in each window, minimum, mean value, variance as a window.
7. measure on stress pulse neural network based according to claim 1 and analysis system, which is characterized in that described point
When the digital stress pulse wave number that analysis module sends the digital processing element is according to frequency domain character extraction is carried out:
First to digital stress pulse wave number according to the processing for carrying out FFT module, wherein spectrum density is by root-mean-square value to a certain number of
Data are averagely obtained, and there are certain data overlap rate, resolution ratio, the frequency ranges of setpoint frequency;
Each stress spectral density is converted to a table later, table content is preceding certain amount frequency line and corresponding
The amplitude of each signal;
Frequency line is divided into a certain number of sections with certain frequency interval, each section of amplitude peak is calculated, minimum radius, puts down
Equal amplitude, the variance of amplitude, using all sections of 4 parameters as frequency domain character.
8. measure on stress pulse neural network based according to claim 4 and analysis system, which is characterized in that the mind
Through network query function module, failure is carried out to the Nonlinear Mapping output that the obtained temporal signatures and the frequency domain character generate
Diagnosis;
Wherein, the data set that the temporal signatures and the frequency domain character are constituted, a part construct nerve net as training data
Network, a part are assessed as test set, and decision-making value is arranged, compares to form fault diagnosis report.
9. measure on stress pulse neural network based according to claim 8 and analysis system, which is characterized in that the mind
Through network query function module include probabilistic neural network, counterpropagation network, LVQ Networks, radial primary function network,
One of adaptive resonance theory prototype network, adaptive organising map network are a variety of.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910814133.5A CN110487917A (en) | 2019-08-30 | 2019-08-30 | A kind of measure on stress pulse neural network based and analysis system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910814133.5A CN110487917A (en) | 2019-08-30 | 2019-08-30 | A kind of measure on stress pulse neural network based and analysis system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110487917A true CN110487917A (en) | 2019-11-22 |
Family
ID=68555738
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910814133.5A Pending CN110487917A (en) | 2019-08-30 | 2019-08-30 | A kind of measure on stress pulse neural network based and analysis system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110487917A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111948286A (en) * | 2020-08-10 | 2020-11-17 | 湖南大学 | Stress detection method, device and equipment based on ultrasonic waves and deep learning |
CN112067310A (en) * | 2020-09-11 | 2020-12-11 | 阴悦 | Health state monitoring system applying non-invasive high-frequency ultrasonic stress wave gas turbine |
CN113465635A (en) * | 2021-06-10 | 2021-10-01 | 电子科技大学 | Device for converting signals of sensor and neural network computing unit |
CN113944600A (en) * | 2021-10-15 | 2022-01-18 | 希音科技(杭州)有限公司 | Method and system for detecting fan main bearing fault by utilizing stress wave technology |
CN114970647A (en) * | 2022-07-29 | 2022-08-30 | 中南大学 | Electromagnetic data identification method and system based on probabilistic neural network |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6351713B1 (en) * | 1999-12-15 | 2002-02-26 | Swantech, L.L.C. | Distributed stress wave analysis system |
CN1776390A (en) * | 2005-11-30 | 2006-05-24 | 沈阳工业大学 | Low-speed heavy-load rotary machinery fault diagnosis method |
CN101871733A (en) * | 2010-06-11 | 2010-10-27 | 昆明理工大学 | Safety detecting method for flue gas waste heat recovery power system of industrial furnace |
CN102243133A (en) * | 2011-04-02 | 2011-11-16 | 中北大学 | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis |
CN104374576A (en) * | 2014-11-25 | 2015-02-25 | 沈阳化工大学 | Method for extracting fault stress waves of low-speed bearing |
CN108318249A (en) * | 2018-01-24 | 2018-07-24 | 广东石油化工学院 | A kind of method for diagnosing faults of bearing in rotating machinery |
CN108942409A (en) * | 2018-08-26 | 2018-12-07 | 西北工业大学 | The modeling and monitoring method of tool abrasion based on residual error convolutional neural networks |
CN110032752A (en) * | 2018-12-20 | 2019-07-19 | 湖南大学 | A kind of power electronic devices and module status detection monitoring system and method |
-
2019
- 2019-08-30 CN CN201910814133.5A patent/CN110487917A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6351713B1 (en) * | 1999-12-15 | 2002-02-26 | Swantech, L.L.C. | Distributed stress wave analysis system |
CN1776390A (en) * | 2005-11-30 | 2006-05-24 | 沈阳工业大学 | Low-speed heavy-load rotary machinery fault diagnosis method |
CN101871733A (en) * | 2010-06-11 | 2010-10-27 | 昆明理工大学 | Safety detecting method for flue gas waste heat recovery power system of industrial furnace |
CN102243133A (en) * | 2011-04-02 | 2011-11-16 | 中北大学 | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis |
CN104374576A (en) * | 2014-11-25 | 2015-02-25 | 沈阳化工大学 | Method for extracting fault stress waves of low-speed bearing |
CN108318249A (en) * | 2018-01-24 | 2018-07-24 | 广东石油化工学院 | A kind of method for diagnosing faults of bearing in rotating machinery |
CN108942409A (en) * | 2018-08-26 | 2018-12-07 | 西北工业大学 | The modeling and monitoring method of tool abrasion based on residual error convolutional neural networks |
CN110032752A (en) * | 2018-12-20 | 2019-07-19 | 湖南大学 | A kind of power electronic devices and module status detection monitoring system and method |
Non-Patent Citations (1)
Title |
---|
吴天舒等: "基于应力波分析的状态监控与故障预测研究", 《仪器仪表学报》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111948286A (en) * | 2020-08-10 | 2020-11-17 | 湖南大学 | Stress detection method, device and equipment based on ultrasonic waves and deep learning |
CN112067310A (en) * | 2020-09-11 | 2020-12-11 | 阴悦 | Health state monitoring system applying non-invasive high-frequency ultrasonic stress wave gas turbine |
CN113465635A (en) * | 2021-06-10 | 2021-10-01 | 电子科技大学 | Device for converting signals of sensor and neural network computing unit |
CN113944600A (en) * | 2021-10-15 | 2022-01-18 | 希音科技(杭州)有限公司 | Method and system for detecting fan main bearing fault by utilizing stress wave technology |
CN113944600B (en) * | 2021-10-15 | 2024-01-12 | 希音科技(杭州)有限公司 | Method and system for detecting fan main bearing faults by using stress wave technology |
CN114970647A (en) * | 2022-07-29 | 2022-08-30 | 中南大学 | Electromagnetic data identification method and system based on probabilistic neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110487917A (en) | A kind of measure on stress pulse neural network based and analysis system | |
CN105956526B (en) | Low signal-to-noise ratio microseismic event identification method based on multi-scale permutation entropy | |
CN108956787B (en) | Steel rail damage detection method based on neural network | |
CN109974782B (en) | Equipment fault early warning method and system based on big data sensitive characteristic optimization selection | |
CN106841403A (en) | A kind of acoustics glass defect detection method based on neutral net | |
CN110398647B (en) | Transformer state monitoring method | |
CN104412087A (en) | Ultrasonic gas leak detector with false alarm discrimination | |
CN103953490A (en) | Implementation method for monitoring status of hydraulic turbine set based on HLSNE | |
CN103760243A (en) | Microcrack nondestructive detecting device and method | |
CN106323452A (en) | Equipment abnormal sound detection method and detecting device | |
CN102122133A (en) | Self-adaption wavelet neural network abnormity detection and fault diagnosis classification system and method | |
CN111695465B (en) | Pipe network fault diagnosis and positioning method and system based on pressure wave mode identification | |
CN103776903B (en) | A kind of wind electricity blade delamination detection method and detection system | |
WO2023279382A1 (en) | Motor bearing running state fault detection method and system | |
CN115754022A (en) | Steel truss welding node damage degree identification method based on acoustic emission | |
CN110632484A (en) | ELM-based GIS partial discharge defect diagnosis and classification system and method | |
CN113283310A (en) | System and method for detecting health state of power equipment based on voiceprint features | |
CN104316167A (en) | Structure response abnormity detection method under the condition of low signal-to-noise ratio | |
CN102879475B (en) | Identification system for fatigue damage state of 16 manganese steel bearing member based on PCA and TDF | |
CN109934136A (en) | Fault Diagnosis of Roller Bearings based on Duffing oscillator and intrinsic mode component | |
CN102279222B (en) | 16Mn steel load carrying piece fatigue damage state identification system | |
CN107121493A (en) | A kind of equipment damage detection method and system | |
CN104535176A (en) | Vibration and noise detection device and method | |
Salah et al. | Automated valve fault detection based on acoustic emission parameters and artificial neural network | |
KR102392951B1 (en) | Equipment failure prediction system having multi-channel sensors for receiving acoustic signals in the ultrasonic band |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191122 |