CN104792529A - Rolling bearing life prediction method based on state-space model - Google Patents
Rolling bearing life prediction method based on state-space model Download PDFInfo
- Publication number
- CN104792529A CN104792529A CN201510171299.1A CN201510171299A CN104792529A CN 104792529 A CN104792529 A CN 104792529A CN 201510171299 A CN201510171299 A CN 201510171299A CN 104792529 A CN104792529 A CN 104792529A
- Authority
- CN
- China
- Prior art keywords
- bearing
- numerical value
- value
- rolling bearing
- moment
- 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
Landscapes
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a rolling bearing life prediction method based on a state-space model. The rolling bearing life prediction method is characterized by including the steps of 1), acquiring rolling bearing operating data; 2), calculating characteristic values; 3), preprocessing the characteristic values; 4), learning parameters of the state-space model according to the rolling bearing operating data; 5), setting an initial bearing degradation threshold value and a later-period bearing degradation life threshold value Ak according to multiple actual fault cases; 6), subjecting particles to initial setting; 7) subjecting bearing data to circulating recursion by a particle filter algorithm and calculating a prediction change tendency of the bearing characteristic values; 8), estimating a numerical value N of remaining life of a bearing according to the particle filter algorithm. The rolling bearing life prediction method based on the state-space model has the advantages of real-time performance, convenience, high accuracy rate and the like, thereby being suitable for bearing life prediction in various industries.
Description
Technical field
The invention belongs to field of diagnosis about equipment fault, relate to the methods set up for the calculating of the character numerical value of conventional rolling bearing and bearing life forecast model.
Background technology
Rolling bearing is one of most important parts in the middle of plant equipment, is widely used, the health status of the fine or not decision device of its performance and life cycle.Rolling bearing locking may then can not run well by equipment, seriously even equipment explosion can be caused, very large threat is brought to safety in production, how effectively carrying out bearing condition monitoring and to predict that bearing is a large difficult point of research at present serviceable life, numerous staff and scientific research personnel have carried out a lot of exploration in this research.How effective current state monitoring system is widely applied, and monitoring system combined with actual rolling bearing residual life, is one to study bright spot greatly.
The reason that current bearing fault causes bearing life to shorten has a lot, and main cause is as follows: bearing causes fatigue flake by alternate load effect for a long time; Because of the wearing and tearing that relative sliding causes between bearing parts; Plastic yield is there is under external force and effect of environmental temperature; There is the corrosion that chemical reaction causes in the bearing parts metal surface caused due to surrounding medium; Bearing due to overload or defect cause the various faults such as sudden accident and retainer damage such as fracture.
Bearing life Forecasting Methodology has a variety of at present, mainly comprise: based on the bearing life Forecasting Methodology of Paris fatigue life prediction model, applied probability theory carries out life prediction, the research based on the bearing fatigue life of artificial intelligence technology and the Study on Fatigue Life etc. based on condition monitoring system of bearing.Paris fatigue life prediction model is not each the deterioration stage being applicable to bearing, and needs a point multiple stage to carry out, and the boundary in each stage can not judge very well; The bearing life prediction needs that applied probability theory is carried out are tested, and obtain a large amount of data; At present the artificial intelligence technology of main application mainly comprises support vector machine and neural network etc., and this is current stage more emerging technological means, and development space also has very large.
In recent years, along with the progress of science and technology, emerging artificial intelligence diagnosis's technology and condition monitoring system are progressively being developed and are being applied, this kind of method is applicable to solve the bearing life prediction case that physics law is complicated, uncertain influence factor is more, greatly enhances the intelligent of bearing life forecasting techniques and accuracy.The rolling bearing life Forecasting Methodology based on state-space model artificial intelligence technology and condition monitoring system combined is proposed herein according to foregoing.
Summary of the invention
The object of the invention is to apply modern advanced artificial intelligence technology to combine with Condition Monitoring Technology, a set of intelligence, real-time, rolling bearing life Forecasting Methodology is accurately provided.The method has can be applied Monitoring Data in real time and carry out rolling bearing life prediction in non-stop-machine situation, and accuracy rate is high, is applicable to the bearing life prediction of multiple occasion.
The invention discloses one
based on the rolling bearing life Forecasting Methodology of state-space model, concrete steps are:
Step (1) gathers rolling bearing service data: by monitoring pump on-line monitoring system, applied acceleration sensor gathers the real-time vibration data in rolling bearing deterioration stage;
Step (2) eigenwert is asked for: vibrating numerical is comparatively obvious with the change of bearing deterioration process, and application vibrating numerical can better predict bearing deterioration state as input data; Vibration data is brought into character numerical value formula, solve the input value that character numerical value is predicted as bearing life.
Character numerical value application following formula extracts:
F
a＝f(x
i,p(x
i),q)；
F
a: acceleration signature value;
X
i: acceleration signal value;
P (x
i): acceleration signal probability density function;
Q: acceleration signal formula power exponent;
F function is acceleration peak value, acceleration high frequency value, acceleration low frequency value or accelerated speed effective value; This prediction can adopt any in multiple acceleration signal.
Step (3) eigenwert pre-service: utilize smothing filtering algorithm to the smoothing pre-service of fault eigenvalue, and character numerical value is normalized.
Step (4) is according to rolling bearing character numerical value, study statespace model parameter: for Rolling Bearing Fault Character numerical value and the relation of rolling bearing running period, set up rolling bearing moving model, namely represent the running status in each moment of rolling bearing with state equation and observation equation.
xpart(k+1)＝xpart(k)+k1*Y
m+n+w(k+1)；
y(k+1)＝xpart(k+1)+v(k+1)；
Wherein xpart (k) the character numerical value size that is the bearing k moment; The character numerical value size that xpart (k+1) is the bearing k+1 moment, Y is the periodicity of bearing operation; The disturbance that w (k+1) shifts in k+1 moment state, for the process noise produced in bearing operation process is to numerical quantity; V (k+1) to make an uproar vector value for the observation produced in k moment bearing operation process; Y (k+1) is end-state predicted numerical value; M, n are material constants, and changing according to material behavior and experimental situation, is variable; K1 model compensation parameter;
In equation, the defining method of k1, m, n is:
(1) observe bearing running status according to monitoring system, choose data point according to the degradation of bearing, generally choose 200 ~ 500 data points after bearing starts deterioration;
(2) process data, and calculate character numerical value according to above-mentioned characteristic formula;
(3) going out the nonlinear relationship between character numerical value and periodicity according to least square fitting, namely when meeting least mean-square error, seeking out k1, m, the n in above-mentioned state equation; Declare, k1, m, n are the change real-time update with data herein.
Step (5) applied statistical method, with multiple physical fault case for the initial deterioration threshold A of foundation setting bearing
iand bearing deterioration remanent life threshold value A
k;
When character numerical value reaches initial threshold A
itime, bearing operation starts to break down; When character numerical value reaches threshold value A
ktime, bearing reaches expected value serviceable life, recommended replacement bearing;
Step (6) carries out Initialize installation to numerical value in particle filter, namely
(1) with reference to monitoring system, when rolling bearing works well, numerical value is basicly stable, fluctuates less, and along with rolling bearing degradation is deepened, vibrating numerical rises gradually; Choose data point according to the degradation of rolling bearing, after generally choosing deterioration, 200 ~ 500 data points are predicted;
(2) setting eigenwert initial value is first character numerical value that the Wave data collected calculates;
(3) according to formula z
k+1=z
k+ randn obtains N number of particle of the single time point of Gaussian distributed, and randn is 0 of Gaussian distributed
:random number within 1; z
kthe particle being the k moment is, z
k+1be k+1 moment particle.
(4) initial weight arranging particle is
(5) numerical value of setting up procedure noise w (k) and observation noise v (k) is respectively 0
:numerical value between 1, such as: 0.01,0.001.
Step (7) application particle filter algorithm carries out the circulation recursion of bearing data, asks for the Trend Forecast of bearing features value:
(1) according to the particle z in k moment
k, produce N number of k+1 moment particle z
k+1, be [z
1z
iz
n], wherein z
k+1=z
k(0, r), (0, r) for obedience average is 0, variance is the random number of the above-mentioned Gaussian distribution of r to R to+R.
(2) by N number of Particle Circulation input in each moment of above-mentioned generation, namely
1. bring each particle of input into above-mentioned state equation, then obtain the result of observation equation according to state equation result;
2. the difference of chance for utilization character numerical value and observation value, asks for the weighted value of each particle;
3. the weighted value sum of all particles is sought out.
(3) weight is normalized;
(4) utilize system resampling technique, namely according to weighted value size, carry out data resampling, right of retention is heavily more than or equal to the particle of 0.7;
(5) predicted numerical value of mean values as this moment of all particles of synchronization after resampling is asked for.
(6) predicted numerical value in all moment is asked in circulation.
When the character numerical value of the bearing that step (8) application particle filter algorithm dopes reaches threshold value, circulation stops, and exports the residual life L of bearing, namely asks for residual life according to following formula:
L refers to residual life
D refers to prediction and counts
The input numerical value that d refers to data to be provided is counted
F
nrefer to sampling number
F
srefer to sample frequency
T
minrefer to interval and get a time
The present invention can carry out the real-time estimate of bearing, can carry out work under making bearing in working order, and applicability is strong, is applicable to multiple workplace;
A first aspect of the present invention, discloses the state equation based on the bearing life Forecasting Methodology of particle filter and observation equation, and wherein the data of state equation can real-time update be replaced, and the unknown parameter application principle of least square method in state equation is asked for;
A second aspect of the present invention, discloses the concrete round-robin algorithm flow process that the rolling bearing life based on particle filter is predicted.Mainly comprise: the production process of particle, the generation of particle weights, the normalization of particle weights, according to the resampling of weight, the determination of predicted numerical value.
A third aspect of the present invention, discloses the application parameter in rolling bearing life computing method.
A fourth aspect of the present invention, foundation real-time state monitoring data compare with predicted data, determine the current duty of bearing, according to existing sampling time, sampling number, sampling interval, determine final residual life.
Accompanying drawing explanation
fig. 1: characteristic value data variation tendency
figure
fig. 2: real data compares with predicted data
figure
fig. 3: the vibration signal of bearing
fig. 4: character numerical value variation tendency after filtering
figure
fig. 5: particle filter algorithm flow process
figureembodiment
Below in conjunction with
accompanying drawingconcrete life prediction flow process of the present invention is described further.
as Fig. 5shown in, idiographic flow of the present invention is as follows:
1, gather rolling bearing service data: by pump on-line monitoring system monitoring machine pump operation state, applied acceleration sensor, on harvester Test-bed for pump, corresponding real time data can obtain bearing deterioration stage real time data;
2, eigenwert is asked for: vibrating numerical is comparatively obvious with the change of bearing deterioration process, and application vibrating numerical can better predict bearing deterioration state as input data; Vibration data is brought into character numerical value formula, solve the input value that character numerical value is predicted as bearing life.
Character numerical value application following formula extracts:
F
a＝f(x
i,p(x
i),q)；
F
a: acceleration signature value;
X
i: acceleration signal value;
P (x
i): acceleration signal probability density function;
Q: acceleration signal formula power exponent;
F function is acceleration peak value, acceleration high frequency value, acceleration low frequency value or accelerated speed effective value; This prediction can adopt any in multiple acceleration signal.
3, eigenwert pre-service: utilize smothing filtering algorithm to the smoothing pre-service of fault eigenvalue, and character numerical value is normalized.
4, according to rolling bearing service data, the parameter of study statespace model: according to above-mentioned required bearing operation troubles eigenwert and bearing operation mechanical periodicity trend, the nonlinear fault model of what both foundation was associated comprise unknown parameter.Fault model can represent the state equation of the running status in each moment of bearing by one group and represent that the observation equation of each moment observation value forms, and both are as follows:
xpart(k+1)＝xpart(k)+k1*Y
m+n+w(k+1)；
y(k+1)＝xpart(k+1)+v(k+1)；
Wherein xpart (k) the character numerical value size that is the bearing k moment; The character numerical value size that xpart (k+1) is the bearing k+1 moment, Y is the periodicity of bearing operation; The disturbance that w (k+1) shifts in k+1 moment state, for the process noise produced in bearing operation process is to numerical quantity; V (k+1) to make an uproar vector value for the observation produced in k moment bearing operation process; Y (k+1) is end-state predicted numerical value; M, n are material constants, and changing according to material behavior and experimental situation, is variable; K1 model compensation parameter;
In equation, the defining method of k1, m, n is:
(1) observe bearing running status according to monitoring system, choose data point according to the degradation of bearing, generally choose 200 ~ 500 data points after bearing starts deterioration;
(2) that lists three kinds of parameters according to state equation asks for formula, is:
xpart(k+1)-xpart(k)＝k1*Y
m+n；
(3) data of each time point in above-mentioned data are chosen successively according to time series as input data;
(4) nonlinear relationship between character numerical value and bearing operation periodicity is simulated according to principle of least square method, namely, when meeting least mean-square error, above-mentioned EQUATION x part (k+1)-xpart (k)=k1*Y is sought out
munknown parameter k1, m, n in+n; Declare, input data pass peek backward according to time sequencing, so unknown parameter k1, m, n are the change real-time update with input data, state equation real-time change are upgraded, realistic data variation trend herein.
(5) make an uproar to numerical quantity v (k) according to Gaussian distribution assignment procedure noise vector numerical value w (k) and observation, numerical value can be set to the numerical value between 0 ~ 1 respectively, such as: 0.01, and 0.001.
(6) by above-mentioned 3 unknown parameter numerical value seeking out and set bring in state equation and observation equation according to the noise vector numerical value of Gaussian distribution, then can ask for the data of subsequent time according to current time data.
5, applied statistical method, with multiple physical fault case for the initial deterioration threshold A of foundation setting bearing
iand bearing deterioration remanent life threshold value A
k;
When character numerical value reaches initial threshold A
itime, bearing operation starts to break down; When character numerical value reaches threshold value A
ktime, bearing reaches expected value serviceable life, recommended replacement bearing;
6, Initialize installation is carried out to numerical value in particle filter, namely
(1) with reference to monitoring system, data point is chosen according to the degradation of bearing, 200 ~ 500 data points after generally choosing deterioration;
(2) setting eigenwert initial value is first character numerical value that the Wave data collected calculates;
(3) according to formula z
k+1=z
k+ randn obtains N number of particle of the single time point of Gaussian distributed, and randn is the random number within the 0:1 of Gaussian distributed; z
kthe particle being the k moment is, z
k+1be k+1 moment particle.
(4) initial weight arranging particle is
(5) numerical value of setting up procedure noise w (k) and observation noise v (k) is respectively the numerical value between 0 ~ 1, such as: 0.01, and 0.001.
7, apply particle filter algorithm carries out data circulation recursion according to time series, ask for the variation tendency of predicted numerical value according to the time of bearing features value:
According to time-varying sequence, at different k moment k=1,2,3 ..., according to the periodicity of not character numerical value and the bearing operation in the same time asked for, carry out according to
as figureparticle filter round-robin algorithm flow process shown in 5
figurecirculate:
(1) according to the particle z in k moment
k, produce N number of k+1 moment particle z
k+1, be [z
1z
iz
n], wherein z
k+1=z
k(0, r), (0, r) for obedience average is 0, variance is the random number of the above-mentioned Gaussian distribution of r to R to+R.
(2) by N number of particle in each moment of above-mentioned generation, circulate according to particle number, namely
1), by each particle of input bring above-mentioned state equation into according to genesis sequence, bring the state equation result of generation into observation equation;
2) character numerical value in k moment and the difference Δ V of predicted characteristics numerical value, is asked for;
3), above-mentioned difference is utilized, according to the following formula
Ask for the weighted value of particle;
Q (i): particle weights numerical value;
V (k): observation is made an uproar to numerical quantity;
The character numerical value in ▽ V:K moment and the difference of predicted characteristics numerical value;
4) weighted value of k+1 moment all particles, is sought out.
(3) the weight normalization carrying out k+1 moment all particles is arranged;
(4) carry out particle resampling according to particle weights numerical value, what resampling technique of the present invention was applied is system resampling technique, namely
1) by interval (0,1] be divided into the sub-range that M continuous print do not overlap:
2) be uniformly distributed first sub-range (0,1/M] in extract a sample U, and to calculate according to the following formula
3) define sequence number function D (.) following (wherein i, m=1,2 ... M):
D (u (i))=m, if
in gathering, each u (i) substitutes into above formula and calculates sequence number set
add up total number of the equal sequence number of each sequence number value, thus obtain
4) by each x
ki () copies N
isecondaryly in new particle set, form the particle assembly after resampling, the weight with seasonal each particle is 1/N.
(5) mean values of all particles of synchronization is asked for as last observation value.
(6) according to time series by not in the same time data input, ask for all moment character numerical value predicted values according to above-mentioned circle principle.
8, apply the character numerical value of bearing that particle filter algorithm dopes when reaching threshold value, circulation stops, and exports the residual life L of bearing, namely asks for residual life according to following formula:
L refers to residual life
D refers to prediction and counts
The input numerical value that d refers to data to be provided is counted
F
nrefer to sampling number
F
srefer to sample frequency
T
minrefer to interval and get a time
This cyclic process can be carried out by real-time online, namely synchronously can complete the life prediction of bearing in the course of work of bearing.
Claims (3)
1., based on the rolling bearing life Forecasting Methodology of state-space model, it is characterized in that comprising following step:
1) rolling bearing service data is gathered: by monitoring pump on-line monitoring system, applied acceleration sensor gathers the real-time vibration data in rolling bearing deterioration stage;
2) eigenwert is asked for: vibrating numerical is comparatively obvious with the change of bearing deterioration process, and application vibrating numerical can better predict bearing deterioration state as input data; Vibration data is brought into character numerical value formula, solve the input value that character numerical value is predicted as bearing life;
3) eigenwert pre-service: utilize smothing filtering algorithm to the smoothing pre-service of fault eigenvalue, and character numerical value is normalized;
4) according to rolling bearing character numerical value, study statespace model parameter: for Rolling Bearing Fault Character numerical value and the relation of rolling bearing running period, set up rolling bearing moving model, namely represent the running status in each moment of rolling bearing with state equation and observation equation;
5) applied statistical method, with multiple physical fault case for the initial deterioration threshold A of foundation setting bearing
iand bearing deterioration remanent life threshold value A
k;
6) Initialize installation is carried out to numerical value in particle filter, namely
(1) with reference to monitoring system, when rolling bearing works well, numerical value is basicly stable, fluctuates less, and along with rolling bearing degradation is deepened, vibrating numerical rises gradually; Choose data point according to the degradation of rolling bearing, after generally choosing deterioration, 200 ~ 500 data points are predicted;
(2) setting eigenwert initial value is first character numerical value that the Wave data collected calculates;
(3) according to formula z
k+1=z
k+ randn obtains N number of particle of single time point of Gaussian distributed, randn be Gaussian distributed 0 ~ 1 within random number; z
kthe particle being the k moment is, z
k+1be k+1 moment particle;
(4) initial weight arranging particle is
(5) numerical value of setting up procedure noise w (k) and observation noise v (k) is respectively the numerical value between 0:1;
7) apply the circulation recursion that particle filter algorithm carries out bearing data, ask for the Trend Forecast of bearing features value:
(1) according to the particle z in k moment
k, produce N number of k+1 moment particle z
k+1, be [z
1z
iz
n], wherein z
k+1=z
k(0, r), (0, r) for obedience average is 0, variance is the random number of the above-mentioned Gaussian distribution of r to R to+R;
(2) by N number of Particle Circulation input in each moment of above-mentioned generation, namely
1. bring each particle of input into above-mentioned state equation, then obtain the result of observation equation according to state equation result;
2. the difference of chance for utilization character numerical value and observation value, asks for the weighted value of each particle;
3. the weighted value sum of all particles is sought out;
(3) weight is normalized;
(4) utilize system resampling technique, namely according to weighted value size, carry out data resampling, right of retention is heavily more than or equal to the particle of 0.7;
(5) predicted numerical value of mean values as this moment of all particles of synchronization after resampling is asked for;
(6) predicted numerical value in all moment is asked in circulation;
8) apply the character numerical value of rolling bearing that particle filter algorithm dopes when reaching threshold value, circulation stops, and exports the residual life L of rolling bearing.2. method according to claim 1, is characterized in that: described step 2) in, character numerical value application following formula extracts:
F
a＝f(x
i,p(x
i),q)；
F
a: acceleration signature value;
X
i: acceleration signal value;
P (x
i): acceleration signal probability density function;
Q: acceleration signal formula power exponent;
F function is acceleration peak value, acceleration high frequency value, acceleration low frequency value or accelerated speed effective value.
2. method according to claim 1, it is characterized in that: described step 4) in, according to rolling bearing character numerical value, study statespace model parameter: for Rolling Bearing Fault Character numerical value and the relation of rolling bearing running period, set up rolling bearing moving model, namely represent the running status in each moment of rolling bearing with state equation and observation equation;
xpart(k+1)＝xpart(k)+k1*Y
m+n+w(k)；
y(k)＝xpart(k)+v(k)；
Wherein xpart (k) the character numerical value size that is the bearing k moment; The character numerical value size that xpart (k+1) is the bearing k+1 moment, Y is the periodicity of bearing operation; The disturbance that w (k) shifts in k moment state, for the process noise produced in bearing operation process is to numerical quantity; V (k) to make an uproar vector value for the observation produced in k moment bearing operation process; Y (k) is end-state predicted numerical value; M, n are material constants, and changing according to material behavior and experimental situation, is variable; K1 model compensation parameter;
In equation, the defining method of k1, m, n is:
(1) observe bearing running status according to monitoring system, get 200 ~ 500 data points after bearing starts deterioration;
(2) process data, and calculate character numerical value according to above-mentioned characteristic formula;
(3) going out the nonlinear relationship between character numerical value and periodicity according to least square fitting, namely when meeting least mean-square error, seeking out k1, m, the n in above-mentioned state equation; K1, m, n are the change real-time update with data.
3. method according to claim 1, is characterized in that: described step 8) in, when the character numerical value of the bearing that application particle filter algorithm dopes reaches threshold value, circulation stops, and exports the residual life L of bearing, namely asks for residual life according to following formula:
L refers to residual life
D refers to prediction and counts
The input numerical value that d refers to data to be provided is counted
F
nrefer to sampling number
F
srefer to sample frequency
T
minrefer to interval and get a time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510171299.1A CN104792529A (en) | 2015-04-12 | 2015-04-12 | Rolling bearing life prediction method based on state-space model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510171299.1A CN104792529A (en) | 2015-04-12 | 2015-04-12 | Rolling bearing life prediction method based on state-space model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104792529A true CN104792529A (en) | 2015-07-22 |
Family
ID=53557519
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510171299.1A Pending CN104792529A (en) | 2015-04-12 | 2015-04-12 | Rolling bearing life prediction method based on state-space model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104792529A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105653851A (en) * | 2015-12-27 | 2016-06-08 | 北京化工大学 | Residual life prediction method of antifriction bearing on the basis of staged physical model and particle filter |
| CN106296727A (en) * | 2016-07-26 | 2017-01-04 | 华北电力大学 | A kind of resampling particle filter algorithm based on Gauss disturbance |
| CN108760256A (en) * | 2018-03-23 | 2018-11-06 | 佛山科学技术学院 | A kind of prediction technique of axle box remaining life |
| CN109597700A (en) * | 2018-12-03 | 2019-04-09 | 郑州云海信息技术有限公司 | A kind of disk life-span prediction method and relevant apparatus |
| CN109670243A (en) * | 2018-12-20 | 2019-04-23 | 华中科技大学 | A kind of life-span prediction method based on lebesgue space model |
| CN110262450A (en) * | 2019-06-17 | 2019-09-20 | 浙江浙能嘉华发电有限公司 | The failure prediction method of various faults characteristic Cooperative Analysis towards steam turbine |
| CN110456199A (en) * | 2019-08-14 | 2019-11-15 | 四川大学 | A kind of method for predicting residual useful life of multisensor syste |
| CN111143990A (en) * | 2019-12-25 | 2020-05-12 | 西安交通大学 | Sensitive measuring point selection and fusion machine tool milling cutter residual life prediction method |
| CN111323663A (en) * | 2020-02-26 | 2020-06-23 | 中南大学 | Electromagnetic valve service life prediction method and device based on current feature extraction |
| CN111397900A (en) * | 2020-05-06 | 2020-07-10 | 贵州航天林泉电机有限公司 | Bearing life accelerated test device |
| CN111458143A (en) * | 2020-04-11 | 2020-07-28 | 湘潭大学 | Temperature fault diagnosis method for main bearing of wind turbine generator |
| CN112518425A (en) * | 2020-12-10 | 2021-03-19 | 南京航空航天大学 | Intelligent machining cutter wear prediction method based on multi-source sample migration reinforcement learning |
| CN113806874A (en) * | 2020-06-16 | 2021-12-17 | 罗克韦尔自动化技术公司 | Method and apparatus for electrical component life estimation with corrosion compensation |
| CN115879248A (en) * | 2023-03-03 | 2023-03-31 | 山东亿宁环保科技有限公司 | Full life cycle management method and system suitable for vacuum pump |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009030808A (en) * | 2008-10-06 | 2009-02-12 | Nsk Ltd | Rolling bearing |
| CN103955750A (en) * | 2014-04-04 | 2014-07-30 | 西安交通大学 | Rolling bearing remaining life prediction method based on feature fusion and particle filtering |
| DE102013110320B3 (en) * | 2013-09-19 | 2014-09-25 | AEMtec GmbH, Berlin | Sensor device for monitoring a lubricant state and method for manufacturing the sensor device |
| CN104156612A (en) * | 2014-08-25 | 2014-11-19 | 福建师范大学 | Fault forecasting method based on particle filter forward and reverse direction prediction errors |
-
2015
- 2015-04-12 CN CN201510171299.1A patent/CN104792529A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009030808A (en) * | 2008-10-06 | 2009-02-12 | Nsk Ltd | Rolling bearing |
| DE102013110320B3 (en) * | 2013-09-19 | 2014-09-25 | AEMtec GmbH, Berlin | Sensor device for monitoring a lubricant state and method for manufacturing the sensor device |
| CN103955750A (en) * | 2014-04-04 | 2014-07-30 | 西安交通大学 | Rolling bearing remaining life prediction method based on feature fusion and particle filtering |
| CN104156612A (en) * | 2014-08-25 | 2014-11-19 | 福建师范大学 | Fault forecasting method based on particle filter forward and reverse direction prediction errors |
Non-Patent Citations (1)
| Title |
|---|
| 刘晓平: "粒子滤波在轴承故障振动信号降噪中的应用", 《轴承》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105653851B (en) * | 2015-12-27 | 2018-09-21 | 北京化工大学 | Rolling bearing method for predicting residual useful life based on physical model stage by stage and particle filter |
| CN105653851A (en) * | 2015-12-27 | 2016-06-08 | 北京化工大学 | Residual life prediction method of antifriction bearing on the basis of staged physical model and particle filter |
| CN106296727A (en) * | 2016-07-26 | 2017-01-04 | 华北电力大学 | A kind of resampling particle filter algorithm based on Gauss disturbance |
| CN108760256A (en) * | 2018-03-23 | 2018-11-06 | 佛山科学技术学院 | A kind of prediction technique of axle box remaining life |
| CN109597700B (en) * | 2018-12-03 | 2021-10-29 | 郑州云海信息技术有限公司 | Disk life prediction method and related device |
| CN109597700A (en) * | 2018-12-03 | 2019-04-09 | 郑州云海信息技术有限公司 | A kind of disk life-span prediction method and relevant apparatus |
| CN109670243A (en) * | 2018-12-20 | 2019-04-23 | 华中科技大学 | A kind of life-span prediction method based on lebesgue space model |
| CN110262450A (en) * | 2019-06-17 | 2019-09-20 | 浙江浙能嘉华发电有限公司 | The failure prediction method of various faults characteristic Cooperative Analysis towards steam turbine |
| CN110262450B (en) * | 2019-06-17 | 2020-06-05 | 浙江浙能嘉华发电有限公司 | Fault prediction method for cooperative analysis of multiple fault characteristics of steam turbine |
| CN110456199A (en) * | 2019-08-14 | 2019-11-15 | 四川大学 | A kind of method for predicting residual useful life of multisensor syste |
| CN111143990A (en) * | 2019-12-25 | 2020-05-12 | 西安交通大学 | Sensitive measuring point selection and fusion machine tool milling cutter residual life prediction method |
| CN111323663A (en) * | 2020-02-26 | 2020-06-23 | 中南大学 | Electromagnetic valve service life prediction method and device based on current feature extraction |
| CN111458143A (en) * | 2020-04-11 | 2020-07-28 | 湘潭大学 | Temperature fault diagnosis method for main bearing of wind turbine generator |
| CN111397900A (en) * | 2020-05-06 | 2020-07-10 | 贵州航天林泉电机有限公司 | Bearing life accelerated test device |
| CN113806874A (en) * | 2020-06-16 | 2021-12-17 | 罗克韦尔自动化技术公司 | Method and apparatus for electrical component life estimation with corrosion compensation |
| CN113806874B (en) * | 2020-06-16 | 2024-03-08 | 罗克韦尔自动化技术公司 | Method and apparatus for electrical component life estimation using corrosion compensation |
| CN112518425A (en) * | 2020-12-10 | 2021-03-19 | 南京航空航天大学 | Intelligent machining cutter wear prediction method based on multi-source sample migration reinforcement learning |
| CN115879248A (en) * | 2023-03-03 | 2023-03-31 | 山东亿宁环保科技有限公司 | Full life cycle management method and system suitable for vacuum pump |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104792529A (en) | Rolling bearing life prediction method based on state-space model | |
| DE102017108169B4 (en) | Production system that defines a determination value of a variable in relation to a product deviation | |
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| DE3850347T2 (en) | Performance data processing system. | |
| CN104598734B (en) | Life prediction method of rolling bearing integrated expectation maximization and particle filter | |
| CN104614179B (en) | A kind of gearbox of wind turbine state monitoring method | |
| CN105300692B (en) | A kind of bearing failure diagnosis and Forecasting Methodology based on expanded Kalman filtration algorithm | |
| CN111582392B (en) | Multi-working-condition health state online monitoring method for key components of wind turbine generator | |
| CN103576050B (en) | A kind of running status appraisal procedure of capacitance type potential transformer | |
| CN106650122B (en) | A kind of equipment variable parameter operation methods of risk assessment | |
| CN105653851A (en) | Residual life prediction method of antifriction bearing on the basis of staged physical model and particle filter | |
| CN103115789B (en) | Second generation small-wave support vector machine assessment method for damage and remaining life of metal structure | |
| CN104976139B (en) | A kind of mechanical equipment state diagnostic method based on Gauss model | |
| CN106649755A (en) | Threshold self-adaption setting abnormity detection method for multi-dimensional real-time power transformation device data | |
| CN110417005B (en) | Transient stability serious fault screening method combining deep learning and simulation calculation | |
| CN106682159A (en) | Threshold configuration method | |
| CN105930629A (en) | On-line fault diagnosis method based on massive amounts of operating data | |
| CN102682180A (en) | Evaluation method for performance degradation of rotary mechanical equipment | |
| DE102015206515A1 (en) | Method for determining a remaining service life of a wind turbine | |
| CN103455658A (en) | Weighted grey target theory based fault-tolerant motor health status assessment method | |
| CN114925614A (en) | Method for predicting residual life of coal mining machine | |
| CN104318043A (en) | Rolling bearing vibration performance reliability variation process detection method and rolling bearing vibration performance reliability variation process detection device | |
| CN116304551A (en) | Motor bearing fault diagnosis and feature extraction method based on BCB model | |
| CN103308334A (en) | Nonlinear cumulative fatigue evaluation method for member | |
| CN117633690A (en) | Rotary machine health state monitoring method and equipment based on data driving |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150722 |