CN102243133B - High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis - Google Patents
High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis Download PDFInfo
- Publication number
- CN102243133B CN102243133B CN 201110086995 CN201110086995A CN102243133B CN 102243133 B CN102243133 B CN 102243133B CN 201110086995 CN201110086995 CN 201110086995 CN 201110086995 A CN201110086995 A CN 201110086995A CN 102243133 B CN102243133 B CN 102243133B
- Authority
- CN
- China
- Prior art keywords
- impact
- signal
- automat
- frequency
- analysis
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a high-speed automaton fault diagnosis method based on movement patterns and impact signal analysis, which aims at automatic fault diagnosis, warning and forecasting. The method comprises the following steps of: selecting three testing points including the point in front of a case, the point in the case and the point behind the case, and testing the impact vibration generated by the high-speed actions and a hitting process of a cannon automaton through a high temperature resistant acceleration transducer; using a computer DASP (dynamic adaptive speculative preprocessor) system to acquire and record three groups of impact acceleration response signals at positions of the three testing points; preprocessing the signals with zero mean, filtering, re-sampling and outlier deletion; performing time domain analysis on the impact signals, and forming an expert rule according to the movement patterns, wherein the expert rule is used for identifying states of main components of the automaton; and performing frequency domain short-time Fourier transform analysis and entropy-spectrum analysis on the impact response signals after widening, building a neural network model according to inherent characteristics and movement patterns of each component of the automaton to identify network training classes, and diagnosing faults of the automaton through a three-layer BP neural network trained by a particle swarm optimization proportion gradient momentum conjugation algorithm.
Description
Technical field
The present invention relates to the analysis of High-Speed Automatic machine impact signal and fault diagnosis field.Specifically a kind of high-speed impact action for each mechanism in High-Speed Automatic machine work arranges that in the appropriate location of cabinet outside measuring point measures impulse response signal, carries out the technology of data analysis processing and fault diagnosis according to the automat motion morphology.
Background technology
The scientific and technological content of current military equipment is more and more higher, and the maintenance support means are relatively backward, the classic method employing listens, touches, sees and frequent tearing open greatly greatly unloaded, the disintegration mode of unpacking checks, maintenance cost is high, the cycle is long, be subjected to the interference of subjective factor easily to cause Misdiagnosis, can not adapt to the reality needs far away.In experiment, the automat of high rate of fire antiaircraft gun due to manufacturing and rigging error and abominable condition of work (as excessive in load, high speed impact, insufficient lubrication), makes it be easy to break down.Therefore be necessary automat action fault is carried out necessary research, seeks a kind of can disturb more, in the complex vibration signal of low signal-to-noise ratio, accurate, rapid extraction fault signature is also identified the method for fault.
Summary of the invention
The present invention seeks in order to overcome the deficiency of above-mentioned prior art, a kind of realize automatic trouble diagnosis, warning and prediction are provided, are difficult for being subjected to the interference of subjective factor to cause the based on motion form of Misdiagnosis and the High-Speed Automatic machine real-time fault diagnosis method that impact signal is analyzed.
The present invention is directed to the problem in the High-Speed Automatic machine development and application maintenance of small caliber piece, principle of work according to High-Speed Automatic machine, to the form time series analysis of taking exercises of each mechanism action of automat, motion cycle chart such as accompanying drawing 1, comprise each parts action situation in the continuous fire process, the anglec of rotation, each motion flow of energy transmission and power transmission, and the bonding space modeling analysis of automat action collision.By the shock Response Analysis to High-Speed Automatic machine enclosure outside surface appropriate location, each member action mechanism of research cabinet inside, the signal processing technologies such as combining information entropy are carried out the real-time diagnosis of automat fault.
The Performances of Small-Caliber Automatic Mechanism structure is in the shell percussion with for producing strong shock and vibration and noise in defeated bullet process, and the automat cabinet can be regarded as take machine case body and the vibrational system of each mechanism member as quality.The powder gases that the automat firing action produces, a series of member high-speed motions of the back such as promotion machine frame are completed the defeated bullet process of automatic confession and continuous fire, and this is the main exciting source of automat.Because shooting fire occasion case rotation axis is done cyclical movement with members such as turning the thorax slide plate, support stiffness changes repeatedly, each structure function moment of torsion also changes continuously, and they all cause the exciting force effect, all will produce the impact shock of certain frequency and direction.The motion of member three directions in cabinet and impact can cause the radial and axial vibration of cabinet, and then the shock response that produces whole automat cabinet.Because each member in the enclosed cabinet transmits force and motion, the impact shock of each member can pass to the automat box body by member and the rotating shaft of contact, makes each sidewall of casing produce the impact shock response.The up-set conditions such as crackle appear in running status generation clamping stagnation and the member of automat, to directly be reflected on mechanism's transmission characteristic, as loosening, eccentric, the local fatigue crackle of rotation axis or fracture etc. will have influence on the shock response characteristic than the low frequency composition, when turning thorax slide plate generation excessive wear, gummed, spot corrosion equivalent damage, the upper frequency composition of response characteristic strengthens, all these strengthens dynamic loading, and impact shock is corresponding aggravation also.
Based on top impact shock analysis and automat mechanism operation circular chart, add the Real-time Collection of the actual shock response data that measure, the time-domain and frequency-domain analysis and processing method that research is applicable, as information entropy, WAVELET PACKET DECOMPOSITION etc., extract various time domains and frequency domain character, in conjunction with the diagnostic method with Expert Rules, can separate and failure prediction automat mechanism action state.Also can utilize energy and the entropy information feature of a plurality of frequency ranges of WAVELET PACKET DECOMPOSITION proposition, use the ratio gradient conjugate momentum algorithm with Multi-layer BP Neural Network to come training network, do information fusion and intelligent diagnostics that based on motion form and impact signal are analyzed.As long as input continuously the sample data that needs, according to the network of training, the data of impact response test are carried out classified calculating, just can carry out effectively, quickly and easily localization of fault.
Concrete steps of the present invention are:
(1) arrange test point on the complicated outside surface of the cabinet of the High-Speed Automatic machine of small caliber piece, this selecting test point is to clash into basic exercise form than sensitive part and Crack Damage position according to inner main member high-speed motion, in conjunction with casing structure natural vibration characteristic and transitive relation, determine after employing particle group optimizing (PSO) technical Analysis; Calculate through signal analysis and transmission characteristic, set up signal transmission and relational model between different measuring points, carry out optimizing the locations of the measuring points, finally (extractor) three measuring points after (turning the thorax body sidewall) and cabinet in (initiatively slide plate antetheca), cabinet before selected cabinet.Adopt the high temperature resistant acceleration transducer of processing through mechanical filter, the impact shock of gun automata high speed motion and knockout process generation is tested.
(2) use the portable DASP of computing machine (Data Acquisition ﹠amp; Signal Processing-data acquisition and signal are processed) system acquisition record the cabinet outside surface before (initiatively slide plate antetheca), in the impact acceleration response signal of (turning the thorax body sidewall), the three groups of measuring points in rear (extractor) position.
(3) impulse response signal of institute's acquisition and recording is done the pre-service such as zero-mean, filtering, resampling, the rejecting of wild point, adopt low pass, bandreject filtering method according to the natural frequency characteristic of automat structure, the low pass threshold frequency is made as 2000Hz, mallat algorithm with core in the recovery technique of average in short-term of baseline wander and wavelet analysis is processed signal, get three layers of wavelet decomposition low frequency reconstruction signal, eliminate the low frequency part distortion of ground unrest and signal.
(4) first impact signal is done time-domain analysis, extracts each percussive action cycle of time domain, each peak value and frequency.Utilize variation and the distribution of time-domain signal peak value and the moment, each impact energy and the entropy of great many of experiments extraction, respectively change temporal signatures by what automat moved circular chart again, the common Expert Rules that consists of the state recognition of automat main member, this rule is the result of the numerical value such as mechanism element high-speed motion Impact energy and entropy and fault degree statistical study, has the directive significance of identification positioning and quantitative.
(5) the impact response signal is done frequency domain Short Time Fourier Analysis and the entropy spectral analysis after broadening, provide each peak value and frequency in frequency domain, do each band energy and entropy analysis based on WAVELET PACKET DECOMPOSITION, after the signal of medium filtering scheduling algorithm is processed, in conjunction with concentration of energy frequency range given in power spectrum chart, use four layers of WAVELET PACKET DECOMPOSITION, signal is divided into 16 frequency ranges, getting the 3-11 frequency range is characteristic spectra, provide the status flag vector for the fault diagnosis layer, it is inputted as neural network, 5 kinds of situations for automat, every kind of situation provides 15 learning samples, be output as 1, 1, 1, 1, 1 is corresponding normal respectively, turn the wearing and tearing of thorax slide plate, push away and play the slide wearing and tearing, connecting cylinder is loosening, five kinds of situations of active slide plate face fracture, again in conjunction with inherent characteristic and the motion morphology of each member of automat, set up neural network model and carry out the network training Classification and Identification, utilize three layers of BP neural network of particle group optimizing ratio gradient momentum conjugation Algorithm for Training to carry out fault diagnosis to automat.
The present invention has been merged signal testing and processing, feature extraction and fault diagnosis in one, can realize automatic trouble diagnosis, reports to the police and prediction.For the different fault type of automat, developed the fault diagnosis algorithm based on particle group optimizing ratio gradient momentum conjugation algorithm of neural network model, can realize conveniently fault diagnosis and prediction of automat.Performances of Small-Caliber Automatic Mechanism maintenance support means have been solved relatively backward, the deficiency that the classic method employing is listened, touched, sees and frequent tearing open greatly greatly unloaded, the disintegration mode of unpacking checks, maintenance cost is low, the cycle is short, be difficult for being subjected to the interference of subjective factor to cause Misdiagnosis, intelligent degree is high, feature richness is practical, can adapt to real equipment preparation and maintenance needs.
Description of drawings
Fig. 1 is automat displacement cycle figure;
Fig. 2 is vibration time-domain curve before the automat cabinet;
Fig. 3 is the test macro block diagram;
Fig. 4 is cabinet three first layers wavelet decomposition restructuring graph;
Fig. 5 is automat Y-direction (barrel direction) each section of motion analysis enlarged drawing;
Fig. 6 is operating power spectral density curve before cabinet;
Fig. 7 is the energy spectrogram of WAVELET PACKET DECOMPOSITION;
Fig. 8 is overall technological scheme block diagram of the present invention.
Embodiment
Adopt the high temperature resistant acceleration transducer of processing through mechanical filter, the impact shock of gun automata high speed motion and knockout process generation is tested test gained impact shock acceleration responsive curve such as accompanying drawing 2.
Use the portable DASP of computing machine (Data Acquisition ﹠amp; Signal Processing-data acquisition and signal are processed) system acquisition record the cabinet outside surface before (initiatively slide plate antetheca), in impact shock acceleration responsive signal, the composition frame chart of test macro such as the accompanying drawing 3 of (turning the thorax body sidewall), the three groups of measuring points in rear (extractor) position.
The impulse response signal of institute's acquisition and recording is done the pre-service such as zero-mean, filtering, resampling, the rejecting of wild point, adopt low pass, bandreject filtering method according to the natural frequency characteristic of automat structure, the low pass threshold frequency is made as 2000Hz, mallat algorithm with core in the recovery technique of average in short-term of baseline wander and wavelet analysis is processed signal, get three layers of wavelet decomposition low frequency reconstruction signal, eliminate the low frequency part distortion of ground unrest and signal, low frequency reconstruct data curve is seen accompanying drawing 4.
First impact signal is done time-domain analysis, extracts each percussive action cycle of time domain, each peak value and frequency, and each section of impact signal amplifies analysis chart and see accompanying drawing 5.Utilize variation and the distribution of time-domain signal peak value and the moment, each impact energy and the entropy of great many of experiments extraction, respectively change temporal signatures by what automat moved circular chart again, the common Expert Rules that consists of the state recognition of automat main member, this rule is the result of the numerical value such as mechanism element high-speed motion Impact energy and entropy and fault degree statistical study, has the directive significance of identification positioning and quantitative.Use such Expert Rules to carry out separation and the prediction of the malfunctions such as member crackle, as shown in table 1.
Table 1 High-Speed Automatic motor-driven do eigenwert and Expert Rules diagnosis
the impact response signal is done frequency domain Short Time Fourier Analysis and the entropy spectral analysis after broadening, provide each peak value and frequency in frequency domain, power spectrum chart is seen accompanying drawing 6, do each band energy and entropy analysis based on WAVELET PACKET DECOMPOSITION, each band energy figure sees accompanying drawing 7, after processing through the signal of medium filtering scheduling algorithm, in conjunction with concentration of energy frequency range given in power spectrum chart, use four layers of WAVELET PACKET DECOMPOSITION, signal is divided into 16 frequency ranges, getting the 3-11 frequency range is characteristic spectra, provide the status flag vector for the fault diagnosis layer, it is inputted as neural network, 5 kinds of situations for automat, every kind of situation provides 15 learning samples, be output as 1, 1, 1, 1, 1 is corresponding normal respectively, turn the wearing and tearing of thorax slide plate, push away and play the slide wearing and tearing, connecting cylinder is loosening, five kinds of situations of active slide plate face fracture, as shown in table 2.
The input of table 2 train samples data and target output
In conjunction with inherent characteristic and the motion morphology of each member of automat, set up neural network model and carry out the network training Classification and Identification again, this model adopts 3 layers of BP neural network, uses ratio gradient conjugate momentum algorithm to come BP network.Input and output layer neuron is respectively 9 and 5, and through network training repeatedly, finding to select hidden neuron is that 6 network can be obtained classifying quality and speed of convergence preferably.The neuronic transport function of input and output is all elected linear transfer function purelin as, and the transport function of hidden neuron is elected tanh S type transport function tansig as, and target error is le-3.Neural network adopts the improved BP neural network, and learning function is Gradient Descent momentum learning function learngdm, and factor of momentum is 0.9.Recognition training calculates as table 3, carries out location and the diagnosis of automat member initial failure, and the Neural Network Diagnosis that obtains output is as shown in table 4.
The input of table 3 neural network check (turning the thorax slide plate) sample data
This network model can effectively carry out Classification and Identification to automat action state, and carry out localization of fault, adopted a large amount of test sample book data that this network model is tested in test, and diagnose calculating, fault diagnosis rate of accuracy reached to 90%, so this three layers of BP neural network utilizing particle group optimizing ratio gradient momentum conjugation Algorithm for Training that automat is carried out fault diagnosis is convenient, feasible, effectively.
Claims (2)
1. the High-Speed Automatic machine method for diagnosing faults analyzed of a based on motion form and impact signal is characterized in that:
(1) arrange test point on the complicated outside surface of the cabinet of the High-Speed Automatic machine of small caliber piece, three measuring points of extractor rear wall after active slide plate antetheca, cabinet transfer thorax body sidewall and cabinet before the selection cabinet; Adopt the high temperature resistant acceleration transducer of processing through mechanical filter, the impact shock of gun automata high speed motion and knockout process generation is tested;
(2) use three groups of impulse response signals of three point positions of the portable DASP system acquisition record of computing machine;
(3) impulse response signal of institute's acquisition and recording is done zero-mean, filtering, resampling, wild some rejecting pre-service, structure operating frequency characteristic according to the association of automat firing rate, the threshold frequency of low pass is made as 2000Hz, employing is processed signal based on the mallat algorithm of core in the restorative procedure of average in short-term of medium filtering and wavelet analysis, get three layers of wavelet decomposition low frequency reconstruction signal, eliminate the low frequency part distortion of ground unrest and signal;
(4) first impact response signal is done time-domain analysis, extracts cycle, peak value and the frequency of each time percussive action in time-domain signal; Utilize variation and the distribution of time-domain signal peak value and the moment, each impact energy and the entropy of experiment acquisition, respectively change temporal signatures by what automat moved circular chart again, the common Expert Rules that consists of the state recognition of automat main member, this rule is the numerical value of mechanism element high-speed motion Impact energy and entropy and the result of fault degree statistical study;
(5) the impact response signal is done frequency domain Short Time Fourier Analysis and the entropy spectral analysis after broadening, provides each peak value and frequency in frequency domain, does each band energy and entropy analysis based on WAVELET PACKET DECOMPOSITION, after processing through the signal of median filtering algorithm, in conjunction with concentration of energy frequency range given in power spectrum chart, use four layers of WAVELET PACKET DECOMPOSITION, signal is divided into 16 frequency ranges, getting the 3-11 frequency range is characteristic spectra, provide the status flag vector for the fault diagnosis layer, it is inputted as neural network, 5 kinds of situations for automat, every kind of situation provides 15 learning samples, be output as 1, 1, 1, 1, 1 is corresponding normal respectively, turn the wearing and tearing of thorax slide plate, push away and play the slide wearing and tearing, connecting cylinder is loosening, five kinds of situations of active slide plate face fracture, again in conjunction with inherent characteristic and the motion morphology of each member of automat, set up neural network model and carry out network training and Classification and Identification, utilization is through three layers of BP neural network of the ratio gradient momentum conjugation Algorithm for Training of particle group optimizing, automat is carried out fault diagnosis.
2. the High-Speed Automatic machine method for diagnosing faults analyzed of based on motion form as claimed in claim 1 and impact signal, when it is characterized in that the impulse response signal of institute's acquisition and recording is done zero-mean, filtering, resampling, wild some rejecting pre-service, the automat firing rate is 1000/minutes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110086995 CN102243133B (en) | 2011-04-02 | 2011-04-02 | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110086995 CN102243133B (en) | 2011-04-02 | 2011-04-02 | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102243133A CN102243133A (en) | 2011-11-16 |
| CN102243133B true CN102243133B (en) | 2013-05-08 |
Family
ID=44961301
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110086995 Expired - Fee Related CN102243133B (en) | 2011-04-02 | 2011-04-02 | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102243133B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102840798B (en) * | 2012-08-20 | 2014-08-20 | 中国兵器工业第二〇二研究所 | Method of testing vibration response time of artillery structure under shooting condition |
| CN103543026B (en) * | 2013-10-23 | 2015-07-01 | 青岛理工大学 | Structural damage identification method based on vibration transfer rate function and support vector machine |
| CN103544526A (en) * | 2013-11-05 | 2014-01-29 | 辽宁大学 | Improved particle swarm algorithm and application thereof |
| CN103795373B (en) * | 2013-11-29 | 2016-07-06 | 电子科技大学中山学院 | Particle filter generating method for incomplete system fault diagnosis |
| CN103969067B (en) * | 2014-04-01 | 2016-04-06 | 中北大学 | A kind of automatic load system method for diagnosing faults based on Generalized Morphological |
| CN104778360B (en) * | 2015-04-10 | 2018-06-22 | 清华大学 | Firer impacts data prediction method and system |
| CN105415191B (en) * | 2015-11-26 | 2018-06-12 | 西北工业大学 | A kind of grinding status investigating method and device based on sound emission |
| CN105910803B (en) * | 2016-04-15 | 2019-05-28 | 中北大学 | A kind of feeding bullet diagnosis method for system fault based on sound pressure information processing |
| CN106017837B (en) * | 2016-06-30 | 2018-12-21 | 北京空间飞行器总体设计部 | A kind of analogy method of equivalent sound simulation source |
| CN106247848B (en) * | 2016-07-26 | 2017-10-10 | 中北大学 | A kind of complicated automatic Incipient Fault Diagnosis method for supplying bullet system |
| CN106123677B (en) * | 2016-08-09 | 2017-12-08 | 中国人民解放军军械工程学院 | Antiaircraft gun automaton fault diagnosis experiment platform and simulated shooting method |
| CN107153154B (en) * | 2017-06-14 | 2019-11-01 | 中国矿业大学 | A kind of two sides earth cable insulation deterioration diagnosis method |
| CN110488121A (en) * | 2019-08-22 | 2019-11-22 | 广东工业大学 | A kind of fault detection method of MMC, system, device and readable storage medium storing program for executing |
| CN110487917A (en) * | 2019-08-30 | 2019-11-22 | 重庆高开清芯科技产业发展有限公司 | A kind of measure on stress pulse neural network based and analysis system |
| CN110907081B (en) * | 2019-12-09 | 2021-08-06 | 湖北文华***工程有限公司 | Pantograph contact pressure monitoring method based on Internet of things |
| CN111259921B (en) * | 2019-12-19 | 2023-09-05 | 杭州安脉盛智能技术有限公司 | Transformer sound abnormality detection method based on improved wavelet packet and deep learning |
| CN111506043B (en) * | 2020-04-09 | 2021-04-09 | 中国船舶工业综合技术经济研究院 | Fault prediction system for key components of naval gun |
| CN113432886A (en) * | 2021-06-09 | 2021-09-24 | 中北大学 | Vehicle full-life cycle vibration impact testing method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SU1226115A1 (en) * | 1984-10-17 | 1986-04-23 | Специальное Опытное Проектно-Конструкторско-Технологическое Бюро "Васхнил" | Arrangement for revealing deffects in machines and mechanisms |
| DK0975947T3 (en) * | 1998-01-06 | 2006-03-06 | Gfm Ges Fuer Maschinendiagnose | Apparatus and method for monitoring the collection of quantities of mobile bodies |
| CN101271035A (en) * | 2007-03-22 | 2008-09-24 | 唐德尧 | Non-rotation movement mechanical failure diagnosis system and method |
| CN101561298A (en) * | 2009-05-06 | 2009-10-21 | 中北大学 | Micro intelligent electronic tester |
| CN101620024A (en) * | 2009-03-04 | 2010-01-06 | 唐德尧 | Resonance demodulation detection method of mechanical failure impact |
-
2011
- 2011-04-02 CN CN 201110086995 patent/CN102243133B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SU1226115A1 (en) * | 1984-10-17 | 1986-04-23 | Специальное Опытное Проектно-Конструкторско-Технологическое Бюро "Васхнил" | Arrangement for revealing deffects in machines and mechanisms |
| DK0975947T3 (en) * | 1998-01-06 | 2006-03-06 | Gfm Ges Fuer Maschinendiagnose | Apparatus and method for monitoring the collection of quantities of mobile bodies |
| CN101271035A (en) * | 2007-03-22 | 2008-09-24 | 唐德尧 | Non-rotation movement mechanical failure diagnosis system and method |
| CN101620024A (en) * | 2009-03-04 | 2010-01-06 | 唐德尧 | Resonance demodulation detection method of mechanical failure impact |
| CN101561298A (en) * | 2009-05-06 | 2009-10-21 | 中北大学 | Micro intelligent electronic tester |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102243133A (en) | 2011-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102243133B (en) | High-speed automaton fault diagnosis method based on movement patterns and impact signal analysis | |
| Yang et al. | Fault diagnosis of gearbox based on RBF-PF and particle swarm optimization wavelet neural network | |
| CN110779746B (en) | Diagnosis method for improving composite fault of deep sparse self-encoder network rotating machinery | |
| CN102507230B (en) | Method for diagnosing fault of automatic ammunition supply and transportation device | |
| CN104712542B (en) | A kind of reciprocating compressor sensitive features based on Internet of Things are extracted and method for diagnosing faults | |
| Li et al. | A new intelligent fusion method of multi-dimensional sensors and its application to tribo-system fault diagnosis of marine diesel engines | |
| Shi et al. | Fault diagnosis in a hydraulic directional valve using a two-stage multi-sensor information fusion | |
| CN103969067B (en) | A kind of automatic load system method for diagnosing faults based on Generalized Morphological | |
| CN109100143A (en) | Fault Diagnosis of Roller Bearings and equipment based on CEEMDAN and CFSFDP | |
| Huang et al. | Tensor singular spectrum decomposition: multisensor denoising algorithm and application | |
| CN103115789A (en) | Second generation small-wave support vector machine assessment method for damage and remaining life of metal structure | |
| Zhang et al. | A multi-module generative adversarial network augmented with adaptive decoupling strategy for intelligent fault diagnosis of machines with small sample | |
| CN109000921A (en) | A kind of diagnostic method of wind generator set main shaft failure | |
| CN105241665A (en) | Rolling bearing fault diagnosis method based on IRBFNN-AdaBoost classifier | |
| CN115762533A (en) | Bird song classification and identification method and device | |
| Chen et al. | Fault identification method for planetary gear based on DT-CWT threshold denoising and LE | |
| Liao et al. | Research on a rolling bearing fault detection method with wavelet convolution deep transfer learning | |
| Chen et al. | A visualized classification method via t-distributed stochastic neighbor embedding and various diagnostic parameters for planetary gearbox fault identification from raw mechanical data | |
| Liang et al. | Multi-sensor signals multi-scale fusion method for fault detection of high-speed and high-power diesel engine under variable operating conditions | |
| Luan et al. | Bearing fault damage degree identification method based on SSA-VMD and Shannon entropy–exponential entropy decision | |
| Zhang et al. | Fault diagnosis method of belt conveyor idler based on sound signal | |
| Song et al. | Research on rolling bearing fault diagnosis method based on improved LMD and CMWPE | |
| CN117628005A (en) | Signal-fused hydraulic motor fault diagnosis method and system | |
| Chang et al. | Similarity metric-based metalearning network combining prior metatraining strategy for intelligent fault detection under small samples prerequisite | |
| An et al. | Deep learning-based composite fault diagnosis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130508 Termination date: 20140402 |