CN108760266A - The virtual degeneration index building method of mechanical key component based on learning distance metric - Google Patents
The virtual degeneration index building method of mechanical key component based on learning distance metric Download PDFInfo
- Publication number
- CN108760266A CN108760266A CN201810548171.6A CN201810548171A CN108760266A CN 108760266 A CN108760266 A CN 108760266A CN 201810548171 A CN201810548171 A CN 201810548171A CN 108760266 A CN108760266 A CN 108760266A
- Authority
- CN
- China
- Prior art keywords
- degradation
- index
- mechanical key
- value
- indexes
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 22
- 230000007850 degeneration Effects 0.000 title abstract 2
- 230000015556 catabolic process Effects 0.000 claims abstract description 126
- 238000006731 degradation reaction Methods 0.000 claims abstract description 119
- 239000013598 vector Substances 0.000 claims abstract description 36
- 238000013528 artificial neural network Methods 0.000 claims abstract description 22
- 238000001228 spectrum Methods 0.000 claims abstract description 14
- 238000013139 quantization Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 5
- 238000005259 measurement Methods 0.000 claims description 28
- 238000011156 evaluation Methods 0.000 claims description 27
- 238000013507 mapping Methods 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000011161 development Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 3
- 230000002860 competitive effect Effects 0.000 claims description 3
- 230000007423 decrease Effects 0.000 claims description 3
- 239000006185 dispersion Substances 0.000 claims description 3
- 239000010902 straw Substances 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 230000004913 activation Effects 0.000 abstract description 3
- 230000002708 enhancing effect Effects 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000005096 rolling process Methods 0.000 description 11
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000036541 health Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012795 verification Methods 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
A kind of virtual degeneration index building method of the mechanical key component based on learning distance metric, the time domain of mechanical critical component vibration signal is extracted first, frequency domain and time and frequency domain characteristics, the degenerate state of mechanical key component is divided according to frequency spectrum and power spectrum simultaneously, secondly the correlation of each index of comprehensive assessment, monotonicity and predictability, selection performance is gone forward side by side row distance metric learning better than the feature vector of the index composition mechanical key component of root-mean-square value, then the self-organizing map neural network after the feature vector training optimization of normal phase is used, input the vibration signal data newly obtained, its feature vector is calculated at a distance from corresponding activation node weight vector, it virtually degenerates index to establish enhancing minimum quantization error；The present invention integrates multiple domain, a variety of physical degradation indexs, is capable of the degradation information of abundant excavating machinery equipment critical component, is conducive to the precision for improving predicting residual useful life.
Description
Technical Field
The invention belongs to the technical field of residual life prediction and health management of mechanical equipment, and particularly relates to a method for constructing a virtual degradation index of a mechanical key component based on distance metric learning.
Background
Mechanical equipment usually works in a complex and changeable environment, the faults of key parts of the mechanical equipment occur frequently, along with the development of modern science and technology, the coupling relation among the key parts of the mechanical equipment is tighter and tighter, once the parts in the mechanical equipment break down, the whole mechanical system breaks down or even breaks down, and serious economic loss or even casualties are caused. Therefore, the residual life prediction is carried out on the key parts of the machinery, so that the key parts of the machinery are preventively repaired before the faults occur, and the safety service of the mechanical equipment is urgently ensured.
The residual life prediction of the key parts of the mechanical equipment mainly comprises monitoring signal acquisition, physical degradation index extraction, index evaluation and optimization, virtual degradation index construction and residual life evaluation. The accuracy of the residual life prediction result is closely related to the used degradation index, besides being influenced by the selected prediction model. Good degradation indexes need to have good correlation, monotonicity and predictability, but are influenced by the quality of original signals and a signal processing method, and physical degradation indexes directly extracted from monitoring signals are often sensitive only to a certain stage of a degradation process and are difficult to keep a good trend in the whole degradation process. Meanwhile, the working environment of the mechanical equipment is complex and changeable, and the physical degradation index is greatly influenced by the working condition, so that the expression of the degradation information of the mechanical key part is not facilitated. The above disadvantages will result in early health monitoring of the mechanical critical components and a reduction in accuracy of the remaining life. Therefore, constructing a degradation index with excellent comprehensive performance is very important for the accuracy of the residual life prediction of the mechanical key components.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a method for constructing virtual degradation indexes of mechanical key components based on distance metric learning, which selects indexes with superior comprehensive performance from alternative physical degradation index sets through index evaluation, maps the optimal physical degradation indexes into a single virtual degradation index by means of distance metric learning and a self-organizing neural network algorithm, represents the degree of deviation of the mechanical key components from a normal state, and describes the degradation process of the mechanical key components.
In order to achieve the purpose, the invention adopts the technical scheme that:
a method for constructing a virtual degradation index of a mechanical key component based on distance metric learning comprises the following steps:
1) carrying out Fourier transform on vibration signals acquired in the whole life cycle of the mechanical key component in sequence to obtain the frequency spectrum and the power spectrum of the vibration signals, and dividing the degradation process of the mechanical key component into three stages of normal operation, fault development and serious degradation according to the change rule of the frequency component corresponding to the maximum amplitude in the spectrogram;
2) extracting physical degradation indexes of the vibration signals from a time domain, a frequency domain and a time-frequency domain respectively to form an alternative physical degradation index set; the method comprises the following specific steps:
2.1) extracting time domain degradation indexes of the vibration signals, which are sequentially a mean value, a standard deviation, a variance, a skewness, a kurtosis, a maximum value, a minimum value, a peak value, an average amplitude value, a root mean square value, a waveform index, a peak index, a pulse index, a margin index, a skewness index and a kurtosis index, and are recorded as F1-F16;
2.2) extracting the frequency domain degradation indexes of the vibration signals, sequentially comprising frequency domain average energy, frequency domain energy variance, mean frequency, root-mean-square frequency and frequency spectrum dispersion degree, and recording by F17-F21;
2.3) extracting time-frequency domain degradation indexes of the vibration signals, obtaining the first 8 intrinsic mode components of the vibration signals through empirical mode decomposition, sequentially calculating the energy of the 8 intrinsic mode components and the energy entropy of the empirical mode decomposition, and recording the energy entropy by F22-F30;
3) establishing a degradation index comprehensive evaluation method which comprises a correlation criterion, a monotonicity criterion and a predictive criterion, and quantitatively evaluating the performances of the 30 physical degradation indexes extracted in the step 2) according to formulas (1) to (4);
3.1) correlation criterion based on Spireman correlation coefficients
Wherein Y ═ Y1,y2,…,yL) And t ═ t (t)1,t2,…,tL) A degradation indicator sequence and a time sequence respectively, L is the length of the degradation indicator sequence,andthe average values of the degradation index sequence and the time sequence are respectively, the criterion describes the correlation between the degradation index and the time sequence, and the closer the value is to 1, the stronger the correlation between the degradation index and the time sequence is;
3.2) monotonicity criterion
Wherein Y ═ Y1,y2,…,yL) For the sequence of degradation indicators, L is the length of the sequence of degradation indicators, and ε (x) is a unit step function, the criterion being characterized by a monotonic increase or decrease in the degradation indicatorThe closer the value is to 1, the better the monotonicity of the degradation index is, and the closer the monotonicity is to the actual condition of the degradation of the mechanical key component;
3.3) predictive criterion
Wherein,the mean value of the degradation index at the initial moment and the failure moment, yfAs a degradation index value, σ (y)f) The standard deviation of the degradation index at the failure moment is described by the criterion, the variation range of the degradation index in the whole life cycle and the dispersity of the degradation index at the failure moment are described, the closer the value is to 1, the larger the variation range of the degradation index is and the smaller the standard deviation at the failure moment is, the more consistent the variation range and the failure threshold value of the index among different individuals is, and the more suitable the residual life prediction is;
3.4) general evaluation criteria
The evaluation criteria are combined by linear weighting, as shown in the following formula (4),
SM＝ω1Corr(Y)+ω2Mon(Y)+ω3Pro(Y) (4)
wherein SM is a comprehensive evaluation criterion, Y is each index sequence, and omega1+ω2+ω31, and ω1,ω2,ω3∈[0,1]Weights used to characterize the 3 evaluation criteria;
4) selecting degradation indexes with excellent performance according to the magnitude of the comprehensive evaluation criterion SM of the 30 physical degradation indexes in the step 3) to form a characteristic vector of a vibration signal of a mechanical key part;
5) the vibration signal characteristic vectors of the mechanical key components in the three stages have different labels, and distance measurement learning is carried out by utilizing the information to obtain a distance measurement matrix suitable for measuring state space similarity in the degradation process of the mechanical key components;
6) using the distance measurement matrix obtained by learning for similarity measurement in a competitive learning stage of the self-organizing mapping neural network to optimize the network;
7) training a self-organizing mapping neural network by using a vibration signal feature vector in a normal stage, and determining a neural network structure and parameters, wherein a Mexico straw hat function is used for weight adjustment;
8) in the testing stage, the feature vector of the newly acquired vibration signal is used as the input of the self-organizing mapping neural network, and the distance between the real-time feature vector and the corresponding activated node weight vector is calculated according to the formula (5) to obtain the minimum quantization error of the enhancement of the virtual degradation index of the mechanical key component;
the method comprises the steps of obtaining a vibration signal, obtaining a weight vector corresponding to an active node in a self-organizing mapping neural network, and obtaining a distance measurement matrix through distance measurement learning.
The invention has the beneficial effects that:
the virtual degradation indexes of mechanical key components are constructed based on distance measurement learning and a self-organizing mapping neural network, state information of various physical degradation indexes is fused, degradation stages of the mechanical key components are scientifically divided, the physical degradation indexes with excellent performance are optimized, feature vectors of different degradation stages are obtained, a distance measurement matrix suitable for measuring state space similarity of the mechanical key components is obtained through distance measurement learning, and the distance between a real-time feature vector and a network activation node weight vector is calculated through the improved self-organizing mapping neural network, so that the minimum quantization error of the virtual degradation index enhancement of the mechanical key components is obtained. The index can well represent the degradation process of mechanical key components, and can be applied to residual life prediction and health management, so that the accuracy of a prediction result can be effectively improved.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
FIG. 2 shows a rolling bearing B according to an embodiment12 physical degradation indexes with optimal comprehensive performance.
FIG. 3 shows a rolling bearing B according to an embodiment12 physical degradation indexes with worst combination performance.
FIG. 4 shows a rolling bearing B according to an embodiment1The enhanced minimum quantization error index of (2) is compared with the minimum quantization error index.
Detailed Description
The invention is further elucidated with reference to the figures and embodiments.
Referring to fig. 1, a method for constructing a virtual degradation index of a mechanical key component based on distance metric learning includes the following steps:
1) carrying out Fourier transform on vibration signals acquired in the whole life cycle of the mechanical key component in sequence to obtain the frequency spectrum and the power spectrum of the vibration signals, and dividing the degradation process of the mechanical key component into three stages of normal operation, fault development and serious degradation according to the change rule of the frequency component corresponding to the maximum amplitude in the spectrogram;
2) and extracting the physical degradation indexes of the vibration signals from the time domain, the frequency domain and the time-frequency domain respectively to form an alternative physical degradation index set. The method comprises the following specific steps:
2.1) extracting time domain degradation indexes of the vibration signals, which are sequentially a mean value, a standard deviation, a variance, a skewness, a kurtosis, a maximum value, a minimum value, a peak value, an average amplitude value, a root mean square value, a waveform index, a peak index, a pulse index, a margin index, a skewness index and a kurtosis index, and are recorded as F1-F16;
2.2) extracting the frequency domain degradation indexes of the vibration signals, sequentially comprising frequency domain average energy, frequency domain energy variance, mean frequency, root-mean-square frequency and frequency spectrum dispersion degree, and recording by F17-F21;
2.3) extracting time-frequency domain degradation indexes of the vibration signals, obtaining the first 8 intrinsic mode components of the vibration signals through empirical mode decomposition, sequentially calculating the energy of the 8 intrinsic mode components and the energy entropy of the empirical mode decomposition, and recording the energy entropy by F22-F30;
3) establishing a degradation index comprehensive evaluation method which comprises a correlation criterion, a monotonicity criterion and a predictive criterion, and quantitatively evaluating the performances of the 30 physical degradation indexes extracted in the step 2) according to formulas (1) to (4);
3.1) correlation criterion based on Spireman correlation coefficients
Wherein Y ═ Y1,y2,…,yL) And t ═ t (t)1,t2,…,tL) A degradation indicator sequence and a time sequence respectively, L is the length of the degradation indicator sequence,andthe average values of the degradation index sequence and the time sequence are respectively, the criterion describes the correlation between the degradation index and the time sequence, and the closer the value is to 1, the stronger the correlation between the degradation index and the time sequence is;
3.2) monotonicity criterion
Wherein Y ═ Y1,y2,…,yL) The characteristic that the degradation index monotonically increases or monotonically decreases is described as a degradation index sequence, L is the length of the degradation index sequence, and epsilon (x) is a unit step function, and the more the value is close to 1, the better the monotonicity of the degradation index is represented, and the closer the monotonicity is to the actual condition of the degradation of the mechanical key component is;
3.3) predictive criterion
Wherein,the mean value of the degradation index at the initial moment and the failure moment, yfAs a degradation index value, σ (y)f) The standard deviation of the degradation index at the failure moment is described by the criterion, the variation range of the degradation index in the whole life cycle and the dispersity of the degradation index at the failure moment are described, the closer the value is to 1, the larger the variation range of the degradation index is and the smaller the standard deviation at the failure moment is, the more consistent the variation range and the failure threshold value of the index among different individuals is, and the more suitable the residual life prediction is;
3.4) general evaluation criteria
The evaluation criteria are combined by linear weighting, as shown in the following formula (4),
SM＝ω1Corr(Y)+ω2Mon(Y)+ω3Pro(Y) (4)
wherein SM is a comprehensive evaluation criterion, Y is each index sequence, and omega1+ω2+ω31, and ω1,ω2,ω3∈[0,1]Is used for characterizing 3A weight of the evaluation criterion;
4) selecting degradation indexes with excellent performance according to the magnitude of the comprehensive evaluation criterion SM of the 30 physical degradation indexes in 3), and forming a characteristic vector of a vibration signal of a mechanical key part;
5) the vibration signal characteristic vectors of the mechanical key components in the three stages have different labels, and distance measurement learning is carried out by utilizing the information to obtain a distance measurement matrix suitable for measuring state space similarity in the degradation process of the mechanical key components;
6) using the distance measurement matrix obtained by learning for similarity measurement in a competitive learning stage of the self-organizing mapping neural network to optimize the network;
7) training a self-organizing mapping neural network by using a vibration signal feature vector in a normal stage, and determining a neural network structure and parameters, wherein a Mexico straw hat function is used for weight adjustment;
8) in the testing stage, the feature vector of the newly acquired vibration signal is used as the input of the self-organizing mapping neural network, and the distance between the real-time feature vector and the corresponding activated node weight vector is calculated according to the formula (5) to obtain the minimum quantization error of the enhancement of the virtual degradation index of the mechanical key component;
the method comprises the steps of obtaining a vibration signal, obtaining a weight vector corresponding to an active node in a self-organizing mapping neural network, and obtaining a distance measurement matrix through distance measurement learning.
The rolling bearing is used as a common mechanical key component, and in order to further prove the effectiveness of the virtual degradation index construction method of the mechanical key component based on distance measurement learning, the whole life cycle data of the rolling bearing, which is acquired by an accelerated life experiment through a PRONOSTIA experiment table, is used for verification.The experiment table mainly comprises three parts of rotation, loading and test, wherein the rotation part comprises an asynchronous motor, a gear box and a rotating shaft, the loading part accelerates the degradation of a bearing by applying radial force on a test bearing, and the test part acquires a vibration acceleration signal, wherein the sampling frequency f of the vibration acceleration signals25600Hz, 2560 samples, each sample duration is 0.1s, and the time interval between two adjacent samples is 10 s. Vibration signals of 7 groups of rolling bearings are collected under the working conditions of 1800rpm of rotating speed and 4000N of load and are respectively marked as B1～B7。
Firstly, extracting time domain, frequency domain and time-frequency domain degradation indexes of the rolling bearing, and dividing the degradation stage of the bearing into three types of normal, fault development and serious degradation according to a frequency spectrum and a power spectrum; next, the correlation, monotonicity and predictability of each index were comprehensively evaluated using the formula (4), and set in the present inventionThe extracted physical index properties were quantified, and the evaluation results are shown in table 1.
TABLE 1
Based on the results of the comprehensive evaluation, FIG. 2 shows a bearing B1The first 2 physical degradation indexes with the optimal comprehensive evaluation criterion SM are respectively the 3 rd intrinsic mode component energy index and the 4 th intrinsic mode component energy index, and fig. 3 shows a bearing B1The 2 physical degradation indexes with the worst comprehensive evaluation criteria are respectively a mean frequency index and a mean index. According to the principle that the comprehensive performance is not lower than the root mean square value index, determining the characteristic vector X of the rolling bearing state space formed by 9 physical degradation indexest＝(xt,1,xt,2,…,xt,9) Wherein x ist,1,xt,2,…,xt,9Respectively represents the 3 rd eigenmode of the rolling bearing at the time tThe method comprises the following steps of component energy index, 4 th intrinsic mode component energy index, root mean square frequency index, frequency spectrum dispersity index, empirical mode decomposition energy entropy index, peak-to-peak value index, minimum value index, maximum value index and root mean square value index. Then, distance measurement learning is carried out by utilizing the feature vectors of three different stages to obtain a distance measurement matrix A with the dimension of 9 multiplied by 9, and the distance measurement matrix A is used for similarity measurement of the input vector and the weight vector in the self-organizing mapping neural network. And finally, training a self-organizing mapping neural network by using the feature vector of the normal stage, wherein the trained neural network represents the state space of the rolling bearing in the normal stage, using the newly acquired feature vector of the rolling bearing vibration signal as the input of the network, calculating the distance between the input vector and the corresponding activation node weight vector, and constructing a virtual degradation index to enhance the minimum quantization error. FIG. 4 shows a pair of bearings B1And using the Euclidean distance to measure the contrast graph of the constructed minimum quantization error index and the enhanced minimum quantization error index of the similarity. The evaluation results of the two indexes are shown in table 2, and it can be seen that the correlation, monotonicity and comprehensive evaluation criteria for enhancing the minimum quantization error index are all obviously better than the minimum quantization error index, but are slightly inferior to the minimum quantization error index in the aspect of predictive criteria, possibly being the reason for the small number of bearings. In conclusion, the method provided by the invention can better reflect the degradation process of the mechanical key components, and is beneficial to improving the precision of the residual life prediction.
TABLE 2
The method for constructing the virtual degradation index of the mechanical key component based on distance measurement learning can be suitable for constructing the virtual degradation index of various key components of mechanical equipment. In practical application, an implementer can specifically extract corresponding physical degradation indexes to form a corresponding alternative physical degradation index set according to degradation characteristics of key components of various types of mechanical equipment. The method provided by the invention is beneficial to improving the accuracy of the residual life prediction of the key parts of the mechanical equipment. It should be noted that modifications and variations to the method described herein are possible without departing from the inventive concept.
Claims (1)
1. A method for constructing virtual degradation indexes of mechanical key components based on distance measurement learning is characterized in that the method is suitable for constructing virtual degradation indexes of various key components of mechanical equipment, and in practical application, an implementer purposefully extracts corresponding physical degradation indexes to form an alternative physical degradation index set according to specific degradation characteristics of various key components, and the method comprises the following steps:
1) carrying out Fourier transform on vibration signals acquired in the whole life cycle of the mechanical key component in sequence to obtain the frequency spectrum and the power spectrum of the vibration signals, and dividing the degradation process of the mechanical key component into three stages of normal operation, fault development and serious degradation according to the change rule of the frequency component corresponding to the maximum amplitude in the spectrogram;
2) extracting physical degradation indexes of the vibration signals from a time domain, a frequency domain and a time-frequency domain respectively to form an alternative physical degradation index set; the method comprises the following specific steps:
2.1) extracting time domain degradation indexes of the vibration signals, which are sequentially a mean value, a standard deviation, a variance, a skewness, a kurtosis, a maximum value, a minimum value, a peak value, an average amplitude value, a root mean square value, a waveform index, a peak index, a pulse index, a margin index, a skewness index and a kurtosis index, and are recorded as F1-F16;
2.2) extracting the frequency domain degradation indexes of the vibration signals, sequentially comprising frequency domain average energy, frequency domain energy variance, mean frequency, root-mean-square frequency and frequency spectrum dispersion degree, and recording by F17-F21;
2.3) extracting time-frequency domain degradation indexes of the vibration signals, obtaining the first 8 intrinsic mode components of the vibration signals through empirical mode decomposition, sequentially calculating the energy of the 8 intrinsic mode components and the energy entropy of the empirical mode decomposition, and recording the energy entropy by F22-F30;
3) establishing a degradation index comprehensive evaluation method which comprises a correlation criterion, a monotonicity criterion and a predictive criterion, and quantitatively evaluating the performances of the 30 physical degradation indexes extracted in the step 2) according to formulas (1) to (4);
3.1) correlation criterion based on Spireman correlation coefficients
Wherein Y ═ Y1,y2,…,yL) And t ═ t (t)1,t2,…,tL) A degradation indicator sequence and a time sequence respectively, L is the length of the degradation indicator sequence,andthe average values of the degradation index sequence and the time sequence are respectively, the criterion describes the correlation between the degradation index and the time sequence, and the closer the value is to 1, the stronger the correlation between the degradation index and the time sequence is;
3.2) monotonicity criterion
Wherein Y ═ Y1,y2,…,yL) The characteristic that the degradation index monotonically increases or monotonically decreases is described as a degradation index sequence, L is the length of the degradation index sequence, and epsilon (x) is a unit step function, and the more the value is close to 1, the better the monotonicity of the degradation index is represented, and the closer the monotonicity is to the actual condition of the degradation of the mechanical key component is;
3.3) predictive criterion
Wherein, the mean value of the degradation index at the initial moment and the failure moment, yfAs a degradation index value, σ (y)f) The standard deviation of the degradation index at the failure moment is described by the criterion, the variation range of the degradation index in the whole life cycle and the dispersity of the degradation index at the failure moment are described, the closer the value is to 1, the larger the variation range of the degradation index is and the smaller the standard deviation at the failure moment is, the more consistent the variation range and the failure threshold value of the index among different individuals is, and the more suitable the residual life prediction is;
3.4) general evaluation criteria
The evaluation criteria are combined by linear weighting, as shown in the following formula (4),
SM＝ω1Corr(Y)+ω2Mon(Y)+ω3Pro(Y) (4)
wherein SM is a comprehensive evaluation criterion, Y is each index sequence, and omega1+ω2+ω31, and ω1,ω2,ω3∈[0,1]Weights used to characterize the 3 evaluation criteria;
4) selecting degradation indexes with excellent performance according to the magnitude of the comprehensive evaluation criterion SM of the 30 physical degradation indexes in the step 3) to form a characteristic vector of a vibration signal of a mechanical key part;
5) the vibration signal characteristic vectors of the mechanical key components in the three stages have different labels, and distance measurement learning is carried out by utilizing the information to obtain a distance measurement matrix suitable for measuring state space similarity in the degradation process of the mechanical key components;
6) using the distance measurement matrix obtained by learning for similarity measurement in a competitive learning stage of the self-organizing mapping neural network to optimize the network;
7) training a self-organizing mapping neural network by using a vibration signal feature vector in a normal stage, and determining a neural network structure and parameters, wherein a Mexico straw hat function is used for weight adjustment;
8) in the testing stage, the feature vector of the newly acquired vibration signal is used as the input of the self-organizing mapping neural network, and the distance between the real-time feature vector and the corresponding activated node weight vector is calculated according to the formula (5) to obtain the minimum quantization error of the enhancement of the virtual degradation index of the mechanical key component;
the method comprises the steps of obtaining a vibration signal, obtaining a weight vector corresponding to an active node in a self-organizing mapping neural network, and obtaining a distance measurement matrix through distance measurement learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810548171.6A CN108760266B (en) | 2018-05-31 | 2018-05-31 | The virtual degeneration index building method of mechanical key component based on learning distance metric |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810548171.6A CN108760266B (en) | 2018-05-31 | 2018-05-31 | The virtual degeneration index building method of mechanical key component based on learning distance metric |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108760266A true CN108760266A (en) | 2018-11-06 |
| CN108760266B CN108760266B (en) | 2019-11-26 |
Family
ID=64001019
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810548171.6A Active CN108760266B (en) | 2018-05-31 | 2018-05-31 | The virtual degeneration index building method of mechanical key component based on learning distance metric |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108760266B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210526A (en) * | 2019-05-14 | 2019-09-06 | 广州虎牙信息科技有限公司 | Predict method, apparatus, equipment and the storage medium of the key point of measurand |
| CN111243121A (en) * | 2020-01-14 | 2020-06-05 | 广东寰球智能科技有限公司 | Quality monitoring method and device for metal cover and computer readable storage medium |
| CN111289250A (en) * | 2020-02-24 | 2020-06-16 | 湖南大学 | Method for predicting residual service life of rolling bearing of servo motor |
| CN111597722A (en) * | 2020-05-20 | 2020-08-28 | 北京航空航天大学 | Method for predicting equipment precision retention time by using running state information |
| CN111985380A (en) * | 2020-08-13 | 2020-11-24 | 山东大学 | Bearing degradation process state monitoring method, system, equipment and storage medium |
| CN112067289A (en) * | 2020-08-21 | 2020-12-11 | 天津电气科学研究院有限公司 | Motor shaft and transmission shaft abnormal vibration early warning algorithm based on neural network |
| CN112101142A (en) * | 2020-08-27 | 2020-12-18 | 深圳市行健自动化股份有限公司 | Slurry pump running state evaluation method, monitoring terminal and computer readable storage medium |
| CN112380932A (en) * | 2020-11-02 | 2021-02-19 | 上海三菱电梯有限公司 | Vibration signal characteristic value selection method and elevator health state evaluation or fault diagnosis method |
| CN113743010A (en) * | 2021-08-31 | 2021-12-03 | 三峡大学 | Rolling bearing running state evaluation method based on EEMD energy entropy |
| CN113848045A (en) * | 2020-06-25 | 2021-12-28 | 大众汽车股份公司 | Estimating mechanical degradation of a machine |
| CN113934982A (en) * | 2021-10-18 | 2022-01-14 | 河北工业大学 | Method for predicting mechanical life of circuit breaker operating mechanism based on vibration-electric signal fusion |
| CN114647909A (en) * | 2022-04-02 | 2022-06-21 | 江苏科技大学 | Method for determining rolling bearing degradation point based on maximum value of spectral kurtosis characteristic |
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4584085B2 (en) * | 2005-09-07 | 2010-11-17 | トライボテックス株式会社 | Degradation evaluation method for rolling bearings |
| CN103632035A (en) * | 2013-11-07 | 2014-03-12 | 中国兵器工业集团第七〇研究所 | Method for determining lifetime probability distribution and average lifetime of mechanical parts |
| CN103729548A (en) * | 2013-12-18 | 2014-04-16 | 西安交通大学 | Method for indirectly evaluating potential performance degradation of mechanical system |
| CN104834977A (en) * | 2015-05-15 | 2015-08-12 | 浙江银江研究院有限公司 | Traffic alarm condition level predication method based on distance metric learning |
| JP2016148393A (en) * | 2015-02-12 | 2016-08-18 | 日本精工株式会社 | Rolling bearing and method of evaluating impression resistance and acoustic deterioration level of rolling bearing |
| CN106934126A (en) * | 2017-02-28 | 2017-07-07 | 西安交通大学 | Component of machine health indicator building method based on Recognition with Recurrent Neural Network fusion |
-
2018
- 2018-05-31 CN CN201810548171.6A patent/CN108760266B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4584085B2 (en) * | 2005-09-07 | 2010-11-17 | トライボテックス株式会社 | Degradation evaluation method for rolling bearings |
| CN103632035A (en) * | 2013-11-07 | 2014-03-12 | 中国兵器工业集团第七〇研究所 | Method for determining lifetime probability distribution and average lifetime of mechanical parts |
| CN103729548A (en) * | 2013-12-18 | 2014-04-16 | 西安交通大学 | Method for indirectly evaluating potential performance degradation of mechanical system |
| JP2016148393A (en) * | 2015-02-12 | 2016-08-18 | 日本精工株式会社 | Rolling bearing and method of evaluating impression resistance and acoustic deterioration level of rolling bearing |
| CN104834977A (en) * | 2015-05-15 | 2015-08-12 | 浙江银江研究院有限公司 | Traffic alarm condition level predication method based on distance metric learning |
| CN106934126A (en) * | 2017-02-28 | 2017-07-07 | 西安交通大学 | Component of machine health indicator building method based on Recognition with Recurrent Neural Network fusion |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210526A (en) * | 2019-05-14 | 2019-09-06 | 广州虎牙信息科技有限公司 | Predict method, apparatus, equipment and the storage medium of the key point of measurand |
| CN111243121A (en) * | 2020-01-14 | 2020-06-05 | 广东寰球智能科技有限公司 | Quality monitoring method and device for metal cover and computer readable storage medium |
| CN111289250A (en) * | 2020-02-24 | 2020-06-16 | 湖南大学 | Method for predicting residual service life of rolling bearing of servo motor |
| CN111597722A (en) * | 2020-05-20 | 2020-08-28 | 北京航空航天大学 | Method for predicting equipment precision retention time by using running state information |
| CN111597722B (en) * | 2020-05-20 | 2023-11-10 | 北京航空航天大学 | Method for predicting equipment precision holding time by using running state information |
| CN113848045A (en) * | 2020-06-25 | 2021-12-28 | 大众汽车股份公司 | Estimating mechanical degradation of a machine |
| US12001201B2 (en) | 2020-06-25 | 2024-06-04 | Volkswagen Aktiengesellschaft | Estimating a mechanical degradation of a machine |
| CN111985380A (en) * | 2020-08-13 | 2020-11-24 | 山东大学 | Bearing degradation process state monitoring method, system, equipment and storage medium |
| CN111985380B (en) * | 2020-08-13 | 2022-07-05 | 山东大学 | Bearing degradation process state monitoring method, system, equipment and storage medium |
| CN112067289A (en) * | 2020-08-21 | 2020-12-11 | 天津电气科学研究院有限公司 | Motor shaft and transmission shaft abnormal vibration early warning algorithm based on neural network |
| CN112101142A (en) * | 2020-08-27 | 2020-12-18 | 深圳市行健自动化股份有限公司 | Slurry pump running state evaluation method, monitoring terminal and computer readable storage medium |
| CN112380932A (en) * | 2020-11-02 | 2021-02-19 | 上海三菱电梯有限公司 | Vibration signal characteristic value selection method and elevator health state evaluation or fault diagnosis method |
| CN112380932B (en) * | 2020-11-02 | 2022-10-14 | 上海三菱电梯有限公司 | Vibration signal characteristic value selection method and elevator health state evaluation or fault diagnosis method |
| CN113743010A (en) * | 2021-08-31 | 2021-12-03 | 三峡大学 | Rolling bearing running state evaluation method based on EEMD energy entropy |
| CN113743010B (en) * | 2021-08-31 | 2023-11-07 | 三峡大学 | Rolling bearing running state evaluation method based on EEMD energy entropy |
| CN113934982A (en) * | 2021-10-18 | 2022-01-14 | 河北工业大学 | Method for predicting mechanical life of circuit breaker operating mechanism based on vibration-electric signal fusion |
| CN113934982B (en) * | 2021-10-18 | 2024-04-19 | 河北工业大学 | Method for predicting mechanical life of breaker operating mechanism based on vibration-electric signal fusion |
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
| CN114647909B (en) * | 2022-04-02 | 2023-04-21 | 江苏科技大学 | Method for determining degradation point of rolling bearing based on spectral kurtosis characteristic maximum value |
| CN114647909A (en) * | 2022-04-02 | 2022-06-21 | 江苏科技大学 | Method for determining rolling bearing degradation point based on maximum value of spectral kurtosis characteristic |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108760266B (en) | 2019-11-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108760266B (en) | The virtual degeneration index building method of mechanical key component based on learning distance metric | |
| Yang et al. | Bearing fault automatic classification based on deep learning | |
| JP5101396B2 (en) | Soundness diagnosis method and program, and windmill soundness diagnosis device | |
| CN115238753B (en) | Self-adaptive SHM data cleaning method based on local outlier factor | |
| JP7462400B2 (en) | Artificial intelligence device and method for pre-processing noise data to identify problem noise sources | |
| CN104614179B (en) | A kind of gearbox of wind turbine state monitoring method | |
| CN106934126B (en) | Mechanical part health index construction method based on recurrent neural network fusion | |
| CN106441896A (en) | Characteristic vector extraction method for rolling bearing fault mode identification and state monitoring | |
| CN108304348B (en) | Bearing residual life prediction method based on binary wiener process | |
| Hoseinzadeh et al. | Modifying the Hilbert-Huang transform using the nonlinear entropy-based features for early fault detection of ball bearings | |
| CN109932179A (en) | A kind of rolling bearing fault testing method based on the reconstruct of DS Adaptive spectra | |
| Pestana-Viana et al. | The influence of feature vector on the classification of mechanical faults using neural networks | |
| CN114705432B (en) | Method and system for evaluating health state of explosion-proof motor bearing | |
| CN114755017B (en) | Variable-speed bearing fault diagnosis method of cross-domain data driving unsupervised field shared network | |
| Huang et al. | A Fault Diagnosis Approach for Rolling Bearing Based on Wavelet Packet Decomposition and GMM-HMM. | |
| CN111504647A (en) | AR-MSET-based performance degradation evaluation method for rolling bearing | |
| CN112146882B (en) | Bearing fault diagnosis method based on transfer learning vibration signal image recognition | |
| CN116168720A (en) | Motor sound abnormality fault diagnosis method, system and storable medium | |
| CN110864902A (en) | Rolling bearing early fault detection method based on fractional-order step entropy | |
| Sun et al. | A novel rolling bearing vibration impulsive signals detection approach based on dictionary learning | |
| CN114263621A (en) | Test method and system for diagnosing and simulating cavitation fault of centrifugal pump | |
| CN110887664B (en) | Method and device for establishing bearing fault identification model | |
| Yuhang et al. | Prediction of bearing degradation trend based on LSTM | |
| CN117451360A (en) | Variable-working-condition small-sample rolling bearing fault diagnosis method and system based on self-attention mechanism | |
| Wan et al. | Adaptive asymmetric real Laplace wavelet filtering and its application on rolling bearing early fault diagnosis |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |