CN117309377A - Gear box variable rotation speed compound fault diagnosis method - Google Patents
Gear box variable rotation speed compound fault diagnosis method Download PDFInfo
- Publication number
- CN117309377A CN117309377A CN202311196564.2A CN202311196564A CN117309377A CN 117309377 A CN117309377 A CN 117309377A CN 202311196564 A CN202311196564 A CN 202311196564A CN 117309377 A CN117309377 A CN 117309377A
- Authority
- CN
- China
- Prior art keywords
- signal
- fault
- diagnosis
- composite
- gear
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 96
- 238000000034 method Methods 0.000 title claims abstract description 80
- 150000001875 compounds Chemical class 0.000 title claims abstract description 14
- 239000002131 composite material Substances 0.000 claims abstract description 53
- 230000004927 fusion Effects 0.000 claims abstract description 42
- 238000001125 extrusion Methods 0.000 claims abstract description 21
- 230000001360 synchronised effect Effects 0.000 claims abstract description 18
- 230000009466 transformation Effects 0.000 claims abstract description 15
- 238000004458 analytical method Methods 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000004088 simulation Methods 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 46
- 238000010586 diagram Methods 0.000 claims description 40
- 238000004422 calculation algorithm Methods 0.000 claims description 37
- 238000006243 chemical reaction Methods 0.000 claims description 28
- 244000309464 bull Species 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 13
- 238000012952 Resampling Methods 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 7
- 230000005484 gravity Effects 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 238000005315 distribution function Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 5
- 230000008707 rearrangement Effects 0.000 claims description 5
- 230000003595 spectral effect Effects 0.000 claims description 3
- 238000004590 computer program Methods 0.000 description 13
- 238000000354 decomposition reaction Methods 0.000 description 7
- 238000001228 spectrum Methods 0.000 description 7
- 238000002474 experimental method Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 239000000284 extract Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 238000009826 distribution Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000000280 densification Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 239000006247 magnetic powder Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/02—Gearings; Transmission mechanisms
- G01M13/021—Gearings
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/02—Gearings; Transmission mechanisms
- G01M13/028—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a gearbox variable rotation speed compound fault diagnosis method, which comprises the following steps: acquiring an original vibration signal of the fault simulation of the gear box; preprocessing the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transformation to obtain a reconstruction signal of a variable rotation speed signal; decomposing the reconstruction signal by adopting a composite fault decoupling method based on self-adaptive multipoint optimal minimum entropy deconvolution, and extracting a fault impact component from the reconstruction signal; extracting multi-angle features of the fault impact components, and selecting effective features in the multi-angle features; and carrying out multi-sensor fusion diagnosis by utilizing a multi-sensor fusion diagnosis model according to the local diagnosis results of the plurality of single sensors to obtain a composite fault diagnosis result. The characteristic decoupling, diagnosis and identification of the composite fault of the gear box under the unknown variable speed condition are realized, and the high accuracy is realized.
Description
Technical Field
The invention relates to a gearbox variable rotation speed compound fault diagnosis method, and belongs to the technical field of gearbox compound fault diagnosis.
Background
The reliability of the large-scale electromechanical equipment is related to national development, economic construction and personnel safety, the gear box is taken as the most important component of the large-scale electromechanical equipment, whether the gear box breaks down or not has great influence on the running condition of the whole mechanical equipment, and the internal gears and rolling bearings bear the action of high load and high friction and are the parts which are most easily damaged, so that the fault diagnosis of the gear box has important significance for preventing sudden faults, protecting personnel safety and reducing economic losses.
In practical engineering application, the gear box always works under the working condition of high instantaneous load and is expressed as a variable rotation speed working condition. The traditional fault diagnosis method is generally aimed at a uniform speed working condition, and the analyzed signals are all under a steady state condition, but the traditional fault diagnosis method loses the superiority due to the influence of a variable rotating speed. In order to convert the variable speed signal to a steady state signal, order tracking techniques are proposed. Synchronous extrusion wavelet transform (synchrosqueezing wavelet transform, SWT) is a special time-frequency rearrangement method that can directly extract the instantaneous frequency of the signal, and has been used for fault diagnosis of rotating machinery at varying rotational speeds. However, SWT is sensitive to instantaneous frequency conversion of a signal, but under the environments of strong noise and the like, SWT conversion is easy to be interfered, so that the extracted instantaneous frequency conversion is inaccurate, and the end point effect at two ends of the signal is easy to occur because the focus of an algorithm is at the center of gravity frequency of the signal during spectrum rearrangement.
In practical engineering application, the transmission system in the gearbox is very complex, fault signals generated by different parts are coupled together, and the traditional fault diagnosis method is difficult to separate, so that the difficulty of fault diagnosis is increased. The multipoint optimal minimum entropy deconvolution (Multipoint Optimal Minimum Entropy Deconvolution Adjusted, MOMEDA) method can extract fault impacts from vibration signals with the multipoint kurtosis as an indicator. However, the MOMEDA method needs to set reasonable parameters to effectively extract the fault pulse signal, and the noise reduction accuracy of the MOMEDA is affected by the filtering period and the filter length.
The ratio (SPSE) of kurtosis to power spectrum entropy is adopted as an objective function, the uniformity degree of periodic pulses can be expressed by depending on the sensitivity of SPSE to the impact of signals, and the parameters of MOMEDA are selected in a self-adaptive manner by combining a grid optimization algorithm, so that the problem of insufficient signal extraction caused by improper MOMEDA parameter selection can be solved, and the composite fault decoupling of self-adaptive parameter selection is realized.
The essence of fault feature extraction is to extract parameters reflecting signal differences, and multiple angles feature information which can describe signals more completely compared with single signal features, however, irrelevant redundant features can not only provide relevant information of gear box faults, but also increase the calculation amount of fault diagnosis, so that algorithm time is prolonged, and even fault diagnosis accuracy is reduced.
Disclosure of Invention
The purpose is as follows: in view of at least one of the above technical problems, the invention provides a gearbox variable rotation speed compound fault diagnosis method, which is based on improved keyless phase order analysis of synchronous extrusion wavelet transform (SWT), compound fault decoupling of self-adaptive Multipoint Optimal Minimum Entropy Deconvolution (MOMEDA), multi-sensor and multi-angle feature fusion, realizes feature decoupling and diagnosis identification of gearbox compound faults under unknown variable rotation speed conditions, and is an effective gearbox variable rotation speed compound fault diagnosis method.
The invention adopts the technical scheme that:
in a first aspect, the present invention provides a gearbox variable rotation speed compound fault diagnosis method, including:
acquiring an original vibration signal of the fault simulation of the gear box;
preprocessing the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transformation to obtain a reconstruction signal of a variable rotation speed signal;
decomposing the reconstruction signal by adopting a composite fault decoupling method based on self-adaptive multipoint optimal minimum entropy deconvolution, and extracting a fault impact component from the reconstruction signal;
extracting multi-angle features of the fault impact components, and selecting effective features in the multi-angle features; and carrying out multi-sensor fusion diagnosis by utilizing a multi-sensor fusion diagnosis model according to the local diagnosis results of the plurality of single sensors to obtain a composite fault diagnosis result.
In some embodiments, the original vibration signal includes a normal signal, a bearing outer ring fault signal, a bearing inner ring fault signal, a bearing roller fault signal, a bull gear fault signal, a bearing outer ring and bull gear composite fault signal, a bearing inner ring and bull gear composite fault signal, a bearing roller and bull gear composite fault signal in a gearbox constant speed operating state, a normal signal, a bearing outer ring fault signal, a bearing outer ring and bull gear composite fault signal, a bearing inner ring and bull gear composite fault signal, a bearing roller and bull gear composite fault signal in a gearbox variable speed operating state.
In some embodiments, the preprocessing is performed on the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transformation to obtain a reconstruction signal of a variable rotation speed signal, and the method comprises the following steps:
(21) The method comprises the steps of performing enhancement and noise reduction treatment on the original vibration signal frequency domain part by using an improved synchronous extrusion wavelet transformation SAM-SWT algorithm based on SAM, and extracting high-precision instantaneous frequency conversion data, wherein the method specifically comprises the following steps:
the method comprises the steps that firstly, fourier transformation is conducted on an original vibration signal to obtain a signal amplitude;
performing power operation on the signal amplitude, and correcting the phase of the original vibration signal and the signal amplitude subjected to power operation through Fourier inversion conversion to obtain a corrected signal;
Thirdly, carrying out continuous wavelet transformation on the correction signal to obtain a series of wavelet coefficients;
step four, performing bias guide on the wavelet coefficient to obtain the instantaneous frequency of the signal, performing extrusion rearrangement in a time-frequency domain to obtain a SAM-SWT time-frequency diagram, and obtaining the instantaneous frequency conversion of the vibration signal;
(22) Processing the instantaneous frequency conversion data by using an interpolation algorithm, and then fitting to obtain an instantaneous frequency conversion curve;
(23) And performing equal-angle resampling on the instantaneous frequency conversion curve by using an equal-angle resampling algorithm to obtain a reconstructed signal of the variable-speed signal.
In some embodiments, the instantaneous frequency conversion data is processed using a four-segment Hermite interpolation algorithm, specifically including:
the instantaneous frequency-converted data was divided into four equal parts by time, and each part of instantaneous frequency-converted data was individually subjected to a piecewise three-dimensional Hermite interpolation method.
In some embodiments, the method for resolving the reconstructed signal by adopting a composite fault decoupling method based on adaptive multipoint optimal minimum entropy deconvolution, and extracting a fault impact component from the reconstructed signal comprises the following steps:
(31) Calculating pulse period Tg of the gear and the bearing according to the inherent characteristic frequency of the gear and the bearing;
(32) Setting grid parameters of a large step length by using a grid optimization algorithm, wherein the grid parameters comprise a filter length L, a period T and a search step length, and constructing a two-dimensional grid on a coordinate system by taking the filter length L and the period T as an abscissa;
(33) Decomposing the reconstructed signal by a self-adaptive multipoint optimal minimum entropy deconvolution MOMEDA method, and comparing the inverse of the SPSE value to obtain the large-step optimal parameter, wherein the filter order is Ls and the period is Ts; wherein the SPSE value is the ratio of the kurtosis to the power spectral entropy;
(34) Judging whether the obtained value meets the optimal solution, if so, outputting the step (35), and if not, returning to the step (32); wherein the optimal solution is determined by comparing a maximum or minimum of an objective function between grid points;
(35) Setting grid parameters of a small step length, updating and optimizing a network, and updating a two-dimensional grid by taking the updated filter length L and the updated filter period T as the abscissa and the ordinate;
(36) Judging whether the obtained value meets the optimal solution, if so, decomposing out a signal containing an obvious fault pulse period to realize composite fault decoupling, and if not, continuing the step (35).
In some embodiments, extracting multi-angle features of the fault impact component, selecting valid ones of the multi-angle features; according to the local diagnosis results of a plurality of single sensors, performing multi-sensor fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result, comprising:
(41) Extracting multi-angle features of the fault impact component by using a multi-angle feature extraction model;
(42) Selecting, for each sensor, a valid feature from the multi-angle features using a modified ReliefF method; performing single-sensor local diagnosis according to the effective characteristics to obtain a single-sensor local diagnosis result;
(43) Correcting a basic probability assignment function of the D-S evidence theory, and constructing a multi-sensor fusion diagnosis model; and according to the local diagnosis results of the plurality of single sensors, performing fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result.
In some embodiments, wherein the multi-angle features include absolute mean, root mean square, standard deviation, permutation entropy, and center of gravity frequency; selecting valid features from the multi-angle features using a modified ReliefF method, comprising: improving the ReliefF based on cosine similarity to obtain a CRelieff algorithm, wherein the CRelieff algorithm constructs an evaluation function of a feature set through feature weights and redundancy metrics, searches and generates subsets by using sequence forward search, and performs overall evaluation on the feature subsets, wherein a scoring function L (F) is defined as follows:
Wherein Σw is the sum of feature weights and Y (F) is the redundancy of feature set F.
In some embodiments, the single sensor local diagnosis uses a wavelet neural network with a learning rate of 0.001, a weight of 0.01, and a number of iterations of 100.
In some embodiments, a basic probability assignment function of the D-S evidence theory is modified, and a multi-sensor fusion diagnosis model is constructed; according to the local diagnosis results of a plurality of single sensors, performing fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result comprises the following steps:
basic probability assignment function based on pearson correlation coefficient correction D-S evidence theory, pearson correlation coefficient r XY The expression is as follows:
wherein X, Y is a variable, the number of elements is n, x i 、y i The ith variable representing X and Y,and->Respectively representing sample average values of the two;
first defining a similarity r between two evidences using pearson correlation coefficients ij :
Wherein R is the number of focal elements, m 1 ,m 2 ,…,m N Assigning functions to the basic probabilities corresponding to the N evidence bodies, wherein the basic probability assigning function of the r focus element corresponding to the N evidence bodies is expressed as m n (A r ),E i ={m i (A 1 ),m i (A 2 ),…,m i (A R )}、E j ={m j (A 1 ),m j (A 2 ),…,m j (A R ) Assigning a function to the base probabilities corresponding to any two evidence volumes,assigning functions to the basic probabilities corresponding to the evidence bodies, respectively The average value; then define all evidence pairs evidence body E i Support degree s (E) i ) And normalizing the obtained product:
wherein,r in for E i And E is n Pearson correlation coefficient of (b);
finally, correcting the original basic probability distribution function to obtain corrected probability
Wherein m is i (A) Is an original basic probability distribution function based on WNN error distance, and
the multiple sensor arrangement positions of the multiple sensor fusion diagnostic model are at least selected from the position above the large gear, the position of the large gear bearing and the position above the small gear.
In a second aspect, the invention provides a gearbox variable rotation speed compound fault diagnosis device, which comprises a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is configured to operate in accordance with the instructions to perform the method according to the first aspect.
In a third aspect, the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of the first aspect.
In a fourth aspect, the present invention provides an apparatus comprising,
a memory;
a processor;
and
A computer program;
wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method of the first aspect described above.
The keyless phase order analysis can be performed by improving the synchronous extrusion wavelet transformation, the signal instantaneous rotation rate can be accurately extracted and fitted, and the extracted instantaneous rotation frequency curve is subjected to angle domain resampling by adopting an equal-angle resampling algorithm, so that the reconstruction of the variable rotation speed signal is completed.
An improved ReliefF method is constructed, characteristics of the multi-angle characteristics are selected, parameter correction is conducted on a basic probability assignment function of the D-S evidence theory, a multi-sensor fusion diagnosis model is constructed, reliability of data is comprehensively judged, and diagnosis precision is improved.
The beneficial effects are that: the invention provides a gearbox variable-rotation-speed composite fault diagnosis method, which is based on the composite fault decoupling of improved synchronous extrusion wavelet transform (SWT) and adaptive multi-point optimal minimum entropy deconvolution (MOMEDA), obtains the inherent fault frequency of an internal bearing and a gear through a gearbox fault mechanism, preprocesses an original vibration signal by adopting the improved synchronous extrusion wavelet transform-based non-key phase order analysis method to obtain a reconstruction signal of a variable-rotation-speed signal, decomposes the reconstruction signal by adopting the adaptive multi-point optimal minimum entropy deconvolution-based composite fault decoupling method, extracts fault impact components from the reconstruction signal, extracts multi-angle features of the fault impact components by adopting the multi-angle feature fault diagnosis method based on multi-sensor fusion, selects effective features of the fault impact components, constructs a multi-sensor fusion diagnosis model, and performs multi-sensor fusion diagnosis, so that the feature decoupling, diagnosis and diagnosis recognition of the gearbox composite fault under the condition of unknown variable rotation speed are realized. Has the following advantages:
(1) The characteristic decoupling, diagnosis and identification of the composite fault of the gear box under the unknown variable rotation speed condition are realized, and the method is an effective gear box variable rotation speed composite fault diagnosis method;
(2) The invention has higher fault recognition accuracy, and the overall recognition rate reaches 92.53% when WNN fault recognition is carried out on the features selected by CReliefF; and correcting the basic probability assignment function of the D-S evidence theory by using the Pearson correlation coefficient, wherein the overall recognition rate of the corrected multi-sensor fusion result reaches 96.58%.
Drawings
Fig. 1 is a flow chart of a composite fault diagnosis method for variable rotational speed of a gear box according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of an arrangement of a vibration laboratory bench in an embodiment of the present invention.
FIG. 3 is a schematic diagram of the type of gearbox bearings and gear failure in an embodiment of the present invention.
Fig. 4 is a schematic diagram of a SAM-based modified synchronous extrusion wavelet transform process in an embodiment of the present invention.
Fig. 5 is a schematic diagram of the principle of equal angle resampling in an embodiment of the invention.
Fig. 6 is a schematic time-frequency diagram of the SWT transformed variable speed signal according to an embodiment of the present invention.
Fig. 7 (a) is a schematic diagram of the original vibration signal in time domain and frequency domain according to an embodiment of the present invention.
Fig. 7 (b) is a schematic diagram of the time domain and the frequency domain of the SAM transformed data in an embodiment of the present invention.
Fig. 8 is a schematic diagram of amplitude variance of a SAM-SWT transformed signal in an embodiment of the present invention.
Fig. 9 is a time-frequency diagram of SAM-SWT transform at optimal power in an embodiment of the present invention.
Fig. 10 is a schematic diagram of the scatter points extracted from the SAM-SWT time-frequency plot in an embodiment of the present invention.
FIG. 11 is a schematic of the scatter points after four-segment Hermite interpolation in an embodiment of the present invention.
FIG. 12 is a schematic diagram of a fitted curve in an embodiment of the present invention.
Fig. 13 is a time-frequency diagram of a signal after equal-angle resampling in an embodiment of the invention.
FIG. 14 is a schematic diagram of SPSE results in an embodiment of the invention.
Fig. 15 is a schematic diagram of a decomposition result of a fault-optimized MOMEDA after grid updating in an embodiment of the present invention.
Fig. 16 (a) is a schematic diagram of a spectrum of a bearing failure MOMEDA decomposition signal in an embodiment of the invention.
Fig. 16 (b) is a schematic diagram of a spectrum of a gear failure MOMEDA decomposition signal in an embodiment of the invention.
FIG. 17 is a schematic diagram of multi-angle characteristic parameters of different fault types in an embodiment of the present invention.
FIG. 18 is a diagram of the results of Relieff feature selection in accordance with an embodiment of the present invention.
FIG. 19 is a schematic diagram showing the result of selecting CReliefF features in an embodiment of the present invention.
Fig. 20 is a schematic diagram of a wavelet neural network in an embodiment of the invention.
FIG. 21 is a schematic diagram of a multisensor fusion diagnostic model, in an embodiment of the present invention.
FIG. 22 is a diagram of raw base probability distributions for three sensors in an embodiment of the present invention;
FIG. 23 is a schematic diagram of basic probability distribution corresponding to three corrected sensors in an embodiment of the present invention;
FIG. 24 is a comparative schematic diagram of diagnostic results in an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and examples. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
In the description of the present invention, the meaning of a number is one or more, the meaning of a number is two or more, and greater than, less than, exceeding, etc. are understood to exclude the present number, and the meaning of a number is understood to include the present number. The description of the first and second is for the purpose of distinguishing between technical features only and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present invention, the descriptions of the terms "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Example 1
In a first aspect, the present embodiment provides a gearbox variable rotation speed compound fault diagnosis method, including:
acquiring an original vibration signal of the fault simulation of the gear box;
preprocessing the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transformation to obtain a reconstruction signal of a variable rotation speed signal;
decomposing the reconstruction signal by adopting a composite fault decoupling method based on self-adaptive multipoint optimal minimum entropy deconvolution, and extracting a fault impact component from the reconstruction signal;
Extracting multi-angle features of the fault impact components, and selecting effective features in the multi-angle features; and carrying out multi-sensor fusion diagnosis by utilizing a multi-sensor fusion diagnosis model according to the local diagnosis results of the plurality of single sensors to obtain a composite fault diagnosis result.
In some embodiments, as shown in fig. 1, a gearbox composite fault diagnosis method based on keyless phase order analysis of improved synchronous extrusion wavelet transform (SWT), composite fault decoupling of adaptive Multipoint Optimal Minimum Entropy Deconvolution (MOMEDA), multi-sensor and multi-angle feature fusion, comprises the steps of:
(1) Analyzing the inherent fault frequency of the internal bearings and gears of the gearbox according to the structural characteristics and vibration mechanism, selecting a sensor, a data acquisition card and a vibration experiment table, building a fault diagnosis experiment table of the gearbox, and carrying out fault simulation and acquisition of an original vibration signal;
the original vibration signals comprise a normal signal, a bearing outer ring fault signal, a bearing inner ring fault signal, a bearing roller fault signal, a bull gear fault signal, a bearing outer ring and bull gear composite fault signal, a bearing inner ring and bull gear composite fault signal, a bearing roller and bull gear composite fault signal under a gear box constant speed working state, and 13 fault signal types in total.
(2) Preprocessing the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transform (SWT), and obtaining a reconstructed signal of the original vibration signal (variable-speed composite fault signal), wherein the method comprises the following steps:
(21) The method comprises the steps of adopting an improved synchronous extrusion wavelet transform (SAM-SWT) algorithm based on SAM to carry out enhancement and noise reduction treatment on the frequency domain part of the original vibration signal, extracting high-precision instantaneous frequency conversion data, wherein the detailed flow of the SAM-SWT algorithm is as follows:
the method comprises the steps that firstly, fourier transformation is conducted on an original signal to obtain a signal amplitude; performing power operation on the signal amplitude, and correcting the amplitude subjected to the power operation and the phase of the original vibration signal through Fourier inversion to obtain a corrected signal; thirdly, carrying out continuous wavelet transformation on the correction signal to obtain a series of wavelet coefficients; and fourthly, performing bias guide operation on the wavelet coefficient to obtain the instantaneous frequency of the signal, and performing extrusion rearrangement in a time-frequency domain to obtain a SAM-SWT time-frequency diagram so as to obtain the instantaneous frequency conversion of the vibration signal.
The effectiveness of the SAM-SWT algorithm on the instantaneous frequency of the rotating speed depends on the value of the amplitude power, and in general, when the value of the power is in the range of-0.5 to 1.5, the enhancement effect on the rotating speed frequency is best, and the noise reduction effect is best.
(22) After the SAM-SWT time-frequency diagram is obtained, in order to increase the accuracy of the extracted rotation speed ridge line, data correction processing is carried out on the SAM-SWT time-frequency diagram, the maximum energy point extraction is carried out firstly, and the highlight part of the obtained SAM-SWT time-frequency diagram can be regarded as the instantaneous frequency of the signal; processing the instantaneous frequency conversion data extracted by the SAM-SWT algorithm by using a four-section Hermite interpolation algorithm, wherein the four-section Hermite interpolation algorithm divides the instantaneous frequency conversion data into four equal parts according to time, each part is independently used for segmenting the Hermite interpolation method for three times, the four-section Hermite interpolation algorithm improves the 'Dragon phenomenon' generated by directly using the Hermite interpolation, and solves the problem of poor effect when the data change is overlarge; and finally, a SAM-SWT algorithm is fitted to identify the obtained instantaneous frequency conversion curve.
(23) And performing equal-angle resampling on the extracted instantaneous frequency conversion curve by adopting an equal-angle resampling algorithm to obtain a reconstructed signal of the variable-speed signal. The problem that serious errors are generated when the traditional time-frequency analysis method analyzes the non-stationary signals can be solved through equal-angle resampling.
(3) Decomposing the reconstructed signal by adopting a composite fault decoupling method based on adaptive Multipoint Optimal Minimum Entropy Deconvolution (MOMEDA), and extracting a fault impact component from the reconstructed signal, wherein the method comprises the following steps of:
(31) The pulse period Tg of the gear and the bearing is calculated according to the natural characteristic frequency of the gear and the bearing.
(32) Setting grid parameters of a large step size by using a grid optimization algorithm, wherein the grid parameters comprise a filter length L, a period T and a search step size, setting the search range of the filter length L and the period T as [2,2500], [0.75Tg ] and [ 1.25Tg ], respectively, setting the search step size as 10 and 1, and constructing a two-dimensional grid on a coordinate system by taking the filter length L and the period T as an abscissa and an ordinate.
(33) Decomposing the reconstruction signal by the MOMEDA method, and comparing the reconstruction signal with the reciprocal of the SPSE value as an objective function to obtain the large-step optimal parameter, wherein the filter order is Ls and the period is Ts;
wherein the S isThe PSE value is the kurt (x) of kurt and the power spectrum entropy H PSE (x) Ratio of (2), namely:
the kurtosis kurt (x) is a multipoint kurtosis MK, and the formula is:
wherein, N sampling points are counted, L is the length y of the filter n To output the result, t n Constant for determining weight and impact position;
the power spectrum entropy formula is as follows:
for any time domain signal X= { X containing N data points 1 ,x 2 ,...,x N Fourier transforming to obtain F (ω), then its power spectrum is
The total power energy of the vibration signal is
The power spectral entropy is defined as
(34) Judging whether the obtained value meets the optimal solution, if so, outputting, and if not, continuing the step (2). Wherein the optimal solution is determined by comparing the maximum or minimum of the objective function between grid points.
(35) Setting grid parameters of small step length, updating and optimizing a network, setting the search ranges of the filter length L and the period T to be [0.9ls,1.1ls ], [0.9Ts and 1.1Ts ], and setting the search step length to be 1 and 0.11, wherein the updated filter length L and the period T are used as the abscissa and the ordinate to update the two-dimensional grid.
(36) Judging whether the obtained value meets the optimal solution, if so, decomposing out a signal containing an obvious fault pulse period to realize composite fault decoupling, and if not, continuing the step (35).
(4) The multi-angle feature fault diagnosis method based on multi-sensor fusion is adopted, the multi-angle feature of the fault impact component is extracted, the effective feature is selected, and a multi-sensor fusion diagnosis model is constructed for multi-sensor fusion diagnosis, and the method comprises the following steps:
(41) And constructing a multi-angle feature extraction model, and extracting multi-angle features of the decomposed signals containing obvious fault pulse periods, wherein different fault types of the multi-angle feature extraction comprise normal signals and bearing-gear composite fault signals in a gear box speed change working state, the multi-angle features comprise an absolute mean value, a root mean square value, a standard deviation, an arrangement entropy and a gravity center frequency, and the multi-sensor arrangement positions of the multi-sensor fusion diagnosis model are selected from the positions above a large gear, the positions of a large gear bearing and the positions above a small gear.
(42) An improved ReliefF method is constructed, feature selection is carried out on the extracted multi-angle features, and single-sensor local diagnosis is realized. The single sensor local diagnosis adopts a Wavelet Neural Network (WNN), the learning rate of the wavelet neural network is 0.001, the weight is 0.01, and the iteration number is 100; the improved ReliefF method is to improve the ReliefF based on cosine similarity, namely a CRelief algorithm, wherein the CRelief algorithm constructs an evaluation function of a feature set through feature weights and redundancy metrics, searches and generates subsets by using sequence forward search, and performs overall evaluation on the feature subsets, wherein the evaluation function can be defined as follows:
wherein, sigma W is the sum of the feature weights, Y (F) is the redundancy of the feature set F;
in some embodiments, vector similarity is measured by cosine similarity s (A, B),
obtaining a characteristic F by using a cosine similarity formula i Redundancy with feature set F is:
the redundancy of feature set F is:
(43) Correcting a basic probability assignment function of the D-S evidence theory, constructing a multi-sensor fusion diagnosis model, and realizing multi-sensor fusion diagnosis.
The basic probability assignment function of the D-S evidence theory is based on WNN, and the formula is as follows:
Wherein m is 1 ,m 2 ,…,m n Is a corresponding basic probability assignment function, and the focus element of the corresponding identification framework is A 1 ,A 2 ,...,A n ;
The method adopted by the basic probability assignment function for correcting the D-S evidence theory is based on the pearson correlation coefficient, wherein the pearson correlation coefficient is expressed as r:
wherein X, Y is a variable, the number of elements is n, x i 、y i The ith variable representing X and Y,and->Respectively representing sample average values of the two;
the step of correcting the basic probability assignment function of the D-S evidence theory using pearson correlation coefficients is:
first, the pearson correlation coefficient is used to define the similarity between two pieces of evidence:
wherein R is the number of focal elements, m 1 ,m 2 ,…,m N Assigning functions to the basic probabilities corresponding to the N evidence bodies, wherein the basic probability assigning function of the r focus element corresponding to the N evidence bodies is expressed as m n (A r ),E i ={m i (A 1 ),m i (A 2 ),…,m i (A R )}、E j ={m j (A 1 ),m j (A 2 ),…,m j (A R ) Assigning a function to the base probabilities corresponding to any two evidence volumes,assigning a mean value of the function to the basic probabilities corresponding to the evidence body respectively; then define all evidence pairs evidence body E i Support degree s (E) i ) And normalizing the obtained product:
wherein,rin is E i And E is n Pearson correlation coefficient of (b);
and finally, correcting the original basic probability distribution function to obtain corrected probability:
Wherein m is i (A) Is an original basic probability distribution function based on WNN error distance, and
application examples
The method is combined with experiments to comprehensively use for carrying out the composite fault diagnosis of the variable rotation speed of the gear box. In the experiment, the vibration sensor selects the GBC80 mining intrinsic safety type vibration sensor, the data acquisition system is a multisource sensing information acquisition platform manufactured by the data acquisition system, the vibration experiment table mainly comprises a basic vibration isolation platform, a driving motor, a planetary gear box fault simulation suite, a magnetic powder brake load device and the like, as shown in fig. 2, the sensor fault simulation suite comprises a secondary gear box, a set of normal straight teeth, a plurality of bearings and gears with prefabricated faults, and the arrangement positions are selected above a large gear, above a large gear bearing and above a small gear, and the fault type is shown in fig. 3. In the experiment, the sampling frequency of the signal is set to be 5000Hz, the sample length is 20000, the time interval is 4 seconds, the motor rotating speed is 1800r/min at the uniform rotating speed, the motor rotating speed is sinusoidal rotated from 0 to 1800r/min at the variable rotating speed, and the pulse period is 20s.
And (3) carrying out enhancement and noise reduction treatment on the frequency domain part of the original vibration signal by adopting an improved synchronous extrusion wavelet transform (SAM-SWT) algorithm based on SAM, and extracting high-precision instantaneous frequency conversion data, wherein the detailed flow of the SAM-SWT algorithm is shown in a figure 4. By means of fig. 6: the time-frequency diagram obtained by SWT conversion can clearly see the rotation speed change of the variable rotation speed signal. The original signal is subjected to SAM transformation to obtain a corrected signal, when the power of the amplitude is 1.2, the time domain and the frequency domain of the variable rotation speed signal processed by the SAM algorithm are shown in figure 7, and it can be seen that the amplitude of the original signal is increased after the SAM processing, so that fault information can be extracted more sensitively, and the accuracy of data is ensured.
In order to select the optimal power of the SAM-SWT algorithm, a window length n=256 and an overlap length m=128 are selected based on the value of the amplitude variance of the signal, and at this time, the amplitude variance of the signal after SAM-SWT conversion is shown in fig. 8 in the interval of-0.5 to 1.5, the power mo=0.9 is selected, and the amplitude variance of the correction signal is the smallest, i.e. the energy of the correction signal is the most concentrated, and at this time, the time-frequency diagram after SAM-SWT conversion is shown in fig. 9.
The energy threshold sigma is selected according to the curve information of the time-frequency diagram, other energy in the energy matrix is completely set to zero, the approximate curve of the rotating speed after the miscellaneous points are removed is shown in fig. 10, then four-segment Hermite interpolation is carried out, namely, data are equally divided into four areas, and segmentation three-time Hermite interpolation is respectively carried out, so that the accuracy of curve fitting is improved, and a scatter diagram after the four-segment Hermite interpolation is shown in fig. 11. Next, curve fitting is performed, as shown in fig. 12, with a fitting function of:
f(x)=122sin(6.087x+1.042)
the parameter calculated as the fitting function is p 2 The parameter of the original speed profile was 19.10, the error was 3.4% less than 5%, and the expected value was satisfied =19.74. The principle of the equal-angle resampling of the extracted instantaneous frequency conversion curve is shown in fig. 5, and the obtained reconstruction signal of the variable rotation speed signal is shown in fig. 13.
And according to the grid optimization algorithm self-adaptive parameter selection flow, taking the SPSE value as an objective function, and carrying out self-adaptive extraction on the filter length and the filter period. Firstly, determining a filtering period range according to natural characteristic frequencies of a gear and a bearing; then dividing grids by a principle of thinning before densification, and primarily screening an optimal parameter range; and finally, reducing the step length, and updating the grid to obtain the optimal parameters. The bearing filter period range [12 ] is set, the step length is 1, the filter length is [2 2500], the step length is 10, the SPSE result is shown in fig. 14, the corresponding optimal period is 33, and the optimal filter length is 1180. Next, the grid parameters are optimized, the filter period range is updated [29 ], the step length is 0.1, the filter length is [1062 1298], the step length is 1, the momeda decomposition result is shown in fig. 15, the corresponding optimal period is 33, and the optimal filter length is 1222. The decomposed signal is subjected to Hilbert transform, and the transform result is shown in fig. 16. The method can be used for effectively extracting weak faults in the gearbox vibration signals, the SPSE is used as a target, the periodic pulse impact of the self-adaptive extraction filtering parameters after MOMEDA decomposition is clearer, the fault pulse period distinction of bearing faults and gear faults is higher, and the decoupling effect is better.
The essence of fault feature extraction is to extract feature parameters reflecting signal differences as much as possible, and based on the time domain, frequency domain and information entropy principle, the multi-angle feature extraction is selected to more accurately describe and quantify fault features in vibration signals, so that the accuracy of fault identification is improved. After the original vibration signal of the gear box is subjected to variable rotation speed reconstruction of SAM-SWT and self-adaptive MOMEDA composite fault decoupling, a series of time domain signals containing fault pulses can be obtained, a multi-angle feature extraction method is constructed, and for components obtained by decomposing the composite fault vibration signal of the gear box under the variable rotation speed condition, the time domain, the frequency domain and the entropy features of each component are respectively extracted, so that multi-angle feature extraction is realized, and an original fault feature set is formed. The types of characteristic parameters extracted from one sample of different fault types are shown in fig. 17. Vc-a-1-b represents bearing failure components of bearing outer ring failure-gear failure vibration signals after SAM-SAT reconstruction and self-adaptive MOMEDA decomposition under variable rotation speed, vc-a-1-g represents gear failure components of bearing outer ring failure-gear failure vibration signals after SAM-SAT reconstruction and self-adaptive MOMEDA decomposition under variable rotation speed, and Vc-b-1-b, vc-b-1-g, vc-c-1-b and Vc-c-1-g represent bearing and gear failure pulse components extracted under different composite failure types. The 23-dimensional features are input as a ReliefF algorithm, and the weights of the features are shown in fig. 18. It can be seen from the graph that only p15 and p23, namely the characteristic weights of the gravity center frequency and the time domain energy entropy exceed 0.25, and other characteristic redundancies are serious. Therefore, the redundancy measurement of the ReliefF algorithm is optimized by using cosine similarity, the optimized feature selection result is shown in fig. 19, and it can be seen from the graph that after creieff selection, the feature weight shows a clearer classification result, and for the weights of 5 features, namely an absolute mean value, a root mean square value, a standard deviation, an arrangement entropy and a gravity center frequency, are all greater than 0.25, the feature selection is performed on the extracted five multi-angle features, and the single-sensor local diagnosis is realized. The single sensor local diagnosis mode uses wavelet neural network, as shown in fig. 20, wherein hidden layer nodes are composed of wavelet functions, the number of neurons of an output layer is 5, and the number corresponds to 5 output results of normal state, bearing outer ring fault, bearing inner ring fault, bearing roller fault and gear fault. The learning rate of the wavelet neural network is adjusted to be 0.001, the weight is 0.01, the iteration number is 100, and the absolute mean value, the root mean square value, the standard deviation, the permutation entropy and the gravity center frequency are used as inputs.
The basic probability assignment function of the D-S evidence theory is modified to construct a multi-sensor fusion diagnostic model, as shown in fig. 21. The method for assigning functions based on the original basic probabilities of the WNN error distance obtains the basic probability assignment of three sensors to the recognition frame theta as shown in fig. 22. When correction is performed by using pearson correlation coefficients, and the correction coefficients of the three sensors are 0.79, 0.86, and 0.77, respectively, the basic probability distributions of the three sensors after correction are as shown in fig. 23. Comparing the three sensor recognition results with the three sensor fusion results, as shown in fig. 24, it can be seen that the recognition accuracy after the three sensors are fused is greatly improved compared with that of a single sensor.
Example 2
In a second aspect, based on embodiment 1, the present embodiment provides a gearbox variable rotation speed compound fault diagnosis device, including a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is configured to operate in accordance with the instructions to perform the method according to embodiment 1.
Example 3
In a third aspect, based on embodiment 1, the present embodiment provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method described in embodiment 1.
Example 4
In a fourth aspect, based on embodiment 1, the present embodiment provides an apparatus comprising,
a memory;
a processor;
and
A computer program;
wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method of embodiment 1.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is only a preferred embodiment of the invention, it being noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the invention.
Claims (10)
1. The composite fault diagnosis method for the variable rotating speed of the gear box is characterized by comprising the following steps of:
Acquiring an original vibration signal of the fault simulation of the gear box;
preprocessing the original vibration signal by adopting a keyless phase order analysis method based on improved synchronous extrusion wavelet transformation to obtain a reconstruction signal of a variable rotation speed signal;
decomposing the reconstruction signal by adopting a composite fault decoupling method based on self-adaptive multipoint optimal minimum entropy deconvolution, and extracting a fault impact component from the reconstruction signal;
extracting multi-angle features of the fault impact components, and selecting effective features in the multi-angle features; and carrying out multi-sensor fusion diagnosis by utilizing a multi-sensor fusion diagnosis model according to the local diagnosis results of the plurality of single sensors to obtain a composite fault diagnosis result.
2. The method of claim 1, wherein the raw vibration signal comprises a normal signal, a bearing outer race fault signal, a bearing inner race fault signal, a bearing roller fault signal, a bull gear fault signal, a bearing outer race and bull gear composite fault signal, a bearing inner race and bull gear composite fault signal, a bearing roller and bull gear composite fault signal in a gearbox constant speed operating state, a normal signal, a bearing outer race fault signal, a bearing outer race and bull gear composite fault signal, a bearing inner race and bull gear composite fault signal, a bearing roller and bull gear composite fault signal in a gearbox variable speed operating state.
3. The method of claim 1, wherein preprocessing the original vibration signal using a keyless phase order analysis method based on an improved synchronous extrusion wavelet transform to obtain a reconstructed signal of a variable rotational speed signal comprises:
(21) The method comprises the steps of performing enhancement and noise reduction treatment on the original vibration signal frequency domain part by using an improved synchronous extrusion wavelet transformation SAM-SWT algorithm based on SAM, and extracting high-precision instantaneous frequency conversion data, wherein the method specifically comprises the following steps:
the method comprises the steps that firstly, fourier transformation is conducted on an original vibration signal to obtain a signal amplitude;
performing power operation on the signal amplitude, and correcting the phase of the original vibration signal and the signal amplitude subjected to power operation through Fourier inversion conversion to obtain a corrected signal;
thirdly, carrying out continuous wavelet transformation on the correction signal to obtain a series of wavelet coefficients;
step four, performing bias guide on the wavelet coefficient to obtain the instantaneous frequency of the signal, performing extrusion rearrangement in a time-frequency domain to obtain a SAM-SWT time-frequency diagram, and obtaining the instantaneous frequency conversion of the vibration signal;
(22) Processing the instantaneous frequency conversion data by using an interpolation algorithm, and then fitting to obtain an instantaneous frequency conversion curve;
(23) And performing equal-angle resampling on the instantaneous frequency conversion curve by using an equal-angle resampling algorithm to obtain a reconstructed signal of the variable-speed signal.
4. The method according to claim 1, characterized in that the instantaneous frequency conversion data is processed using a four-segment Hermite interpolation algorithm, in particular comprising:
the instantaneous frequency-converted data was divided into four equal parts by time, and each part of instantaneous frequency-converted data was individually subjected to a piecewise three-dimensional Hermite interpolation method.
5. The method of claim 1, wherein decomposing the reconstructed signal using a composite fault decoupling method based on adaptive multipoint optimal minimum entropy deconvolution, extracting a fault impact component from the reconstructed signal, comprises:
(31) Calculating pulse period Tg of the gear and the bearing according to the inherent characteristic frequency of the gear and the bearing;
(32) Setting grid parameters of a large step length by using a grid optimization algorithm, wherein the grid parameters comprise a filter length L, a period T and a search step length, and constructing a two-dimensional grid on a coordinate system by taking the filter length L and the period T as an abscissa;
(33) Decomposing the reconstructed signal by a self-adaptive multipoint optimal minimum entropy deconvolution MOMEDA method, and comparing the inverse of the SPSE value to obtain the large-step optimal parameter, wherein the filter order is Ls and the period is Ts; wherein the SPSE value is the ratio of the kurtosis to the power spectral entropy;
(34) Judging whether the obtained value meets the optimal solution, if so, outputting the step (35), and if not, returning to the step (32); wherein the optimal solution is determined by comparing a maximum or minimum of an objective function between grid points;
(35) Setting grid parameters of a small step length, updating and optimizing a network, and updating a two-dimensional grid by taking the updated filter length L and the updated filter period T as the abscissa and the ordinate;
(36) Judging whether the obtained value meets the optimal solution, if so, decomposing out a signal containing an obvious fault pulse period to realize composite fault decoupling, and if not, continuing the step (35).
6. The method of claim 1, wherein multi-angle features of the fault impact component are extracted, and valid ones of the multi-angle features are selected; according to the local diagnosis results of a plurality of single sensors, performing multi-sensor fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result, comprising:
(41) Extracting multi-angle features of the fault impact component by using a multi-angle feature extraction model;
(42) Selecting, for each sensor, a valid feature from the multi-angle features using a modified ReliefF method; performing single-sensor local diagnosis according to the effective characteristics to obtain a single-sensor local diagnosis result;
(43) Correcting a basic probability assignment function of the D-S evidence theory, and constructing a multi-sensor fusion diagnosis model; and according to the local diagnosis results of the plurality of single sensors, performing fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result.
7. The method of claim 6, wherein the multi-angle features include absolute mean, root mean square, standard deviation, permutation entropy, and center of gravity frequency; selecting valid features from the multi-angle features using a modified ReliefF method, comprising: improving the ReliefF based on cosine similarity to obtain a CRelieff algorithm, wherein the CRelieff algorithm constructs an evaluation function of a feature set through feature weights and redundancy metrics, searches and generates subsets by using sequence forward search, and performs overall evaluation on the feature subsets, wherein a scoring function L (F) is defined as follows:
wherein Σw is the sum of feature weights and Y (F) is the redundancy of feature set F.
8. The method of claim 1, wherein the single sensor local diagnosis uses a wavelet neural network with a learning rate of 0.001, a weight of 0.01, and a number of iterations of 100.
9. The method of claim 1, wherein the basic probability assignment function of the D-S evidence theory is modified to construct a multisensor fusion diagnostic model; according to the local diagnosis results of a plurality of single sensors, performing fusion diagnosis by using a multi-sensor fusion diagnosis model to obtain a composite fault diagnosis result comprises the following steps:
Basic probability assignment function based on pearson correlation coefficient correction D-S evidence theory, pearson correlation coefficient r XY The expression is as follows:
wherein X, Y is a variable, the number of elements is n, x i 、y i The ith variable representing X and Y,and->Respectively representing sample average values of the two;
first defining a similarity r between two evidences using pearson correlation coefficients ij :
Wherein R is the number of focal elements, m 1 ,m 2 ,…,m N Assigning functions to the basic probabilities corresponding to the N evidence bodies, wherein the basic probability assigning function of the r focus element corresponding to the N evidence bodies is expressed as m n (A r ),E i ={m i (A 1 ),m i (A 2 ),…,m i (A R )}、E j ={m j (A 1 ),m j (A 2 ),…,m j (A R ) Assigning a function to the base probabilities corresponding to any two evidence volumes,assigning a mean value of the function to the basic probabilities corresponding to the evidence body respectively; then define all evidence pairs evidence body E i Support degree s (E) i ) And normalizing the obtained product:
wherein,r in for E i And E is n Pearson correlation coefficient of (b);
finally, correcting the original basic probability distribution function to obtain corrected probability
Wherein m is i (A) Is an original basic probability distribution function based on WNN error distance, and
wherein the multiple sensor placement locations of the multiple sensor fusion diagnostic model are selected from at least above the large gear, at the large gear bearing, and above the small gear.
10. The gear box variable rotation speed compound fault diagnosis device is characterized by comprising a processor and a storage medium;
the storage medium is used for storing instructions;
the processor is operative according to the instructions to perform the method according to any one of claims 1 to 9.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311196564.2A CN117309377A (en) | 2023-09-18 | 2023-09-18 | Gear box variable rotation speed compound fault diagnosis method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311196564.2A CN117309377A (en) | 2023-09-18 | 2023-09-18 | Gear box variable rotation speed compound fault diagnosis method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117309377A true CN117309377A (en) | 2023-12-29 |
Family
ID=89236403
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311196564.2A Pending CN117309377A (en) | 2023-09-18 | 2023-09-18 | Gear box variable rotation speed compound fault diagnosis method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117309377A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118114101A (en) * | 2024-04-30 | 2024-05-31 | 武汉理工大学 | Fuel cell fault diagnosis method and system |
-
2023
- 2023-09-18 CN CN202311196564.2A patent/CN117309377A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118114101A (en) * | 2024-04-30 | 2024-05-31 | 武汉理工大学 | Fuel cell fault diagnosis method and system |
CN118114101B (en) * | 2024-04-30 | 2024-07-19 | 武汉理工大学 | Fuel cell fault diagnosis method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Chen et al. | Fault feature extraction and diagnosis of rolling bearings based on wavelet thresholding denoising with CEEMDAN energy entropy and PSO-LSSVM | |
Jiao et al. | Deep coupled dense convolutional network with complementary data for intelligent fault diagnosis | |
CN114357663B (en) | Method for training gearbox fault diagnosis model and gearbox fault diagnosis method | |
CN110309603B (en) | Short-term wind speed prediction method and system based on wind speed characteristics | |
CN109827777B (en) | Rolling bearing fault prediction method based on partial least square method extreme learning machine | |
CN107657250B (en) | Bearing fault detection and positioning method and detection and positioning model implementation system and method | |
Cui et al. | Improved fault size estimation method for rolling element bearings based on concatenation dictionary | |
CN111413089A (en) | Gear fault diagnosis method based on combination of VMD entropy method and VPMCD | |
CN111079690B (en) | Main shaft and workpiece vibration prediction method based on stack sparse automatic coding network | |
Dai et al. | Fault diagnosis of rotating machinery based on deep reinforcement learning and reciprocal of smoothness index | |
Kumar et al. | Least square fitting for adaptive wavelet generation and automatic prediction of defect size in the bearing using Levenberg–Marquardt backpropagation | |
CN113567131B (en) | Bearing fault diagnosis method based on S transformation and miniature convolutional neural network model | |
CN117309377A (en) | Gear box variable rotation speed compound fault diagnosis method | |
CN105424366A (en) | Bearing fault diagnosis method based on EEMD adaptive denoising | |
CN110348468B (en) | Bearing sub-health identification method for optimized strong reconstruction edge noise reduction automatic encoder of correlation vector machine | |
Xu et al. | Hierarchical multiscale dense networks for intelligent fault diagnosis of electromechanical systems | |
Shang et al. | An intelligent fault diagnosis system for newly assembled transmission | |
CN116358871B (en) | Rolling bearing weak signal composite fault diagnosis method based on graph rolling network | |
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 | |
Xiong et al. | A novel deep convolutional image-denoiser network for structural vibration signal denoising | |
Zhao et al. | A novel deep fuzzy clustering neural network model and its application in rolling bearing fault recognition | |
Meng et al. | A gear fault diagnosis method based on improved accommodative random weighting algorithm and BB-1D-TP | |
Zhao et al. | Hybrid pre-training strategy for deep denoising neural networks and its application in machine fault diagnosis | |
Li et al. | A Classification Algorithm of Fault Modes‐Integrated LSSVM and PSO with Parameters’ Optimization of VMD | |
Abdul et al. | Highly accurate gear fault diagnosis based on support vector machine |
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 |