CN108181105A - Logic-based returns and the rolling bearing fault of J divergences examines method and system in advance - Google Patents
Logic-based returns and the rolling bearing fault of J divergences examines method and system in advance Download PDFInfo
- Publication number
- CN108181105A CN108181105A CN201711216940.4A CN201711216940A CN108181105A CN 108181105 A CN108181105 A CN 108181105A CN 201711216940 A CN201711216940 A CN 201711216940A CN 108181105 A CN108181105 A CN 108181105A
- Authority
- CN
- China
- Prior art keywords
- bearing
- divergences
- value
- sample
- divergence
- 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
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
- G01M13/04—Bearings
Abstract
It is returned the invention discloses a kind of logic-based and the rolling bearing fault of J divergences examines method in advance, included the following steps:The failure sensing data of different faults position and the normal sensing data of normal condition bearing operation when acquiring bearing operation, respectively to failure sensing data and normal sensing data is pre-processed and feature extraction, abort situation feature samples and normal condition feature samples are established;Logic Regression Models are trained by the sample of established abort situation and the sample of normal condition, logical model parameter is obtained, establishes Logic Regression Models;The present invention carries out targetedly fault diagnosis according to the decline in health degree of bearing to bearing, simultaneously according to different operating mode difference objects, change the parameters such as training sample type, feature Value Types, healthy threshold value, model after training can adjust, and have many advantages, such as that real-time, data processing precision is high, core algorithm robustness is good, status assessment accuracy is high, diagnostic result accuracy is high.
Description
Technical field
The present invention relates to mechanical product quality reliability assessment and fault diagnosis technology fields more particularly to one kind to be based on
Logistic regression and the rolling bearing fault of J divergences examine method and system in advance.
Background technology
Rolling bearing is the important support part of rotating machinery, and real-time running state directly determines the reliable of mechanical equipment
Property, therefore, extremely important, assessment rolling bearing current operating conditions, positioning in real time is studied rolling bearing on-line fault diagnosis
Trouble location and quantization fault severity level, maintenance to mechanical equipment and are designed with extremely important directive significance.This patent
The online failure that a kind of logic-based recurrence and J divergences are realized for rolling bearing examines method in advance, can assess rolling in real time
The operating status of bearing, based on current operating conditions positioning failure position.
The research of existing on-line fault diagnosis method has very much, and ripe method has the judgement of ime domain virtual value peak value at present
Method, amplitude probability density analytic approach, Shock Pulse Method etc..These methods can only be out of order or normal two states, in fact,
Between the operating status of bearing most of the time is between failure and normally, neither complete failure, also non-fully normal, it can be by this
Kind intermediate state is known as decline in health state.Bearing decline in health state is different from common bearing fault, is one recessive, gradually
Process become, very long.Bearing in this state is being stablized the decline phase, and the performance of bearing, which is presented, to be stablized and slowly decline
Trend simultaneously forms initial failure, and from initial failure to final malfunction, bearing performance has decline faster, passes through monitoring
Technology and signal processing technology quantify entire degenerative process and domination, help preferably to manage the health status of bearing.
Obviously, the above method only to be out of order or normal two states data and lose most of decline in health status data, Wu Fazhun
Really assessment bearing operating status, there is higher False Rate and misdetection rate.
Invention content
The present invention provides one kind and is based on patrolling for the shortcomings that can not accurately assessing bearing operating status in the prior art
It collects to return and examines method and system in advance with the rolling bearing fault of J divergences.
In order to solve the above-mentioned technical problem, the present invention is addressed by following technical proposals:
A kind of logic-based returns and the rolling bearing fault of J divergences examines method in advance, includes the following steps:
The failure sensing data of different faults position and the normal sensing of normal condition bearing operation when acquiring bearing operation
Data respectively to failure sensing data and normal sensing data is pre-processed and feature extraction, establish abort situation feature sample
Sheet and normal condition feature samples;
Logic Regression Models are trained by the sample of established abort situation and the sample of normal condition, are obtained
Logical model parameter, establishes Logic Regression Models;
The real-time sensory data of bearing to be measured is acquired, the real-time sensory data is pre-processed and feature extraction, is obtained
To the real-time status characteristic index of bearing to be measured, the characteristic index is updated in the established Logic Regression Models,
By the health degree that bearing current time to be measured is calculated;
The health degree at the bearing current time to be measured being calculated and pre-set health degree threshold value are made comparisons, if
The health degree at bearing current time to be measured is then calculated to be measured respectively less than the threshold value of setting by the method for diagnosing faults of J divergences
The real-time status characteristic index of bearing and the J divergences of abort situation feature samples, the J divergences of normal condition feature samples, according to
The size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition judges the abort situation of bearing, and realization is treated
Survey the fault diagnosis of bearing.
As a kind of embodiment, the specific steps of the Logic Regression Models acquisition methods include,
Assuming that the vector of N number of independent variable, XT=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, Logic Regression Models
Mathematic(al) representation be:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient is estimated by maximum likelihood
Meter method is asked for.
As a kind of embodiment, the failure sensing data of different faults position includes bearing when the bearing is run
One or several kinds of failure sensing datas in inner ring failure, outer ring failure and rolling element failure.
As a kind of embodiment, the feature samples of the abort situation include inner ring fault signature sample, outer ring event
Hinder one or several kinds of fault signature samples in feature samples and rolling element fault signature sample.
As a kind of embodiment, when the feature samples of abort situation are inner ring fault signature sample, outer ring failure spy
When levying sample and rolling element fault signature sample, the method for diagnosing faults by J divergences calculates the reality of bearing to be measured respectively
When state characteristic index and abort situation feature samples J divergences, the J divergences of normal condition feature samples, according to abort situation
The J divergences of sample and the size of J divergences of sample of normal condition judge the abort situation of bearing, realize to bearing to be measured
Fault diagnosis, process is as follows,
Pass through formula J divergence formula:The spy of real-time bearing state is calculated respectively
Levy index and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
Between J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergences public affairs
In formula, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two,
N is the number of signal characteristic value, and i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, is dissipated
Angle value is smaller, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, leads to
The size of divergence value is crossed to determine the fault type of bearing.
A kind of logic-based returns and the rolling bearing fault of J divergences examines system in advance, including:
Preprocessing module is acquired, for acquiring the failure sensing data and normal condition of different faults position when bearing is run
The normal sensing data of bearing operation, respectively to failure sensing data and normal sensing data is pre-processed and feature extraction,
Establish abort situation feature samples and normal condition feature samples;
Model building module, for by the sample of established abort situation and the sample of normal condition to logistic regression
Model is trained, and is obtained logical model parameter, is established Logic Regression Models;
Computing module is acquired, for acquiring the real-time sensory data of bearing to be measured, the real-time sensory data is carried out pre-
Processing and feature extraction, obtain the real-time status characteristic index of bearing to be measured, the characteristic index are updated to established institute
It states in Logic Regression Models, by the health degree that bearing current time to be measured is calculated;
Judgment module, for by the health degree at the bearing current time to be measured being calculated and pre-set health degree
Threshold value is made comparisons, if the health degree at bearing current time to be measured passes through the method for diagnosing faults of J divergences less than the threshold value of setting
The real-time status characteristic index of bearing to be measured and the J divergences of abort situation feature samples, normal condition feature samples are calculated respectively
J divergences, the failure of bearing is judged according to the size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition
The fault diagnosis to bearing to be measured is realized in position.
As a kind of embodiment, when the feature samples of abort situation are inner ring fault signature sample, outer ring failure spy
When levying sample and rolling element fault signature sample, model building module is arranged to,
Assuming that the vector of N number of independent variable, XT=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, Logic Regression Models
Mathematic(al) representation be:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient is estimated by maximum likelihood
Meter method is asked for.
As a kind of embodiment, the acquisition preprocessing module is arranged to, the different faults when bearing is run
The failure sensing data of position includes one or several kinds of failures in the inner ring failure, outer ring failure and rolling element failure of bearing
Sensing data.
As a kind of embodiment, the acquisition preprocessing module is arranged to, the feature samples of the abort situation
Including one or several kinds of failures in inner ring fault signature sample, outer ring fault signature sample and rolling element fault signature sample
Feature samples.
As a kind of embodiment, the judgment module is arranged to,
Pass through formula J divergence formula:The spy of real-time bearing state is calculated respectively
Levy index and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
Between J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergences public affairs
In formula, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two,
N is the number of signal characteristic value, and i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, is dissipated
Angle value is smaller, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, leads to
The size of divergence value is crossed to determine the fault type of bearing.
The present invention has significant technique effect as a result of above technical scheme:
The present invention carries out targetedly fault diagnosis, while according to difference according to the decline in health degree of bearing to bearing
Operating mode difference object, changing parameters, the models after training such as training sample type, feature Value Types, healthy threshold value can adjust,
With it is real-time, data processing precision is high, core algorithm robustness is good, status assessment accuracy is high, diagnostic result accuracy
The advantages that high.
Description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention, for those of ordinary skill in the art, without having to pay creative labor, may be used also
To obtain other attached drawings according to these attached drawings.
Fig. 1 is the flow diagram of the illustrative methods of the present invention;
Fig. 2 is the idiographic flow schematic diagram of the present invention;
Fig. 3 is the experimental provision schematic diagram of the present invention;
Fig. 4 is the assessment index schematic diagram for the life cycle management drawn in the present invention;
Fig. 5 is the system structure diagram of the present invention.
Specific embodiment
With reference to embodiment, the present invention is described in further detail, following embodiment be explanation of the invention and
The invention is not limited in following embodiments.
Illustrative methods:
A kind of logic-based returns and the rolling bearing fault of J divergences examines method in advance, as shown in Figure 1, including the following steps:
S1:The failure sensing data of different faults position and the normal biography of normal condition bearing operation when acquiring bearing operation
Feel data, respectively to failure sensing data and normal sensing data is pre-processed and feature extraction, establish abort situation feature
Sample and normal condition feature samples;
S2:Logic Regression Models are trained by the sample of established abort situation and the sample of normal condition,
Logical model parameter is obtained, establishes Logic Regression Models;
S3:The real-time sensory data of bearing to be measured is acquired, the real-time sensory data is pre-processed and feature extraction,
The real-time status characteristic index of bearing to be measured is obtained, the characteristic index is updated to the established Logic Regression Models
In, by the health degree that bearing current time to be measured is calculated;
S4:The health degree at the bearing current time to be measured being calculated and pre-set health degree threshold value are made into ratio
Compared with if the health degree at bearing current time to be measured is calculated respectively less than the threshold value of setting by the method for diagnosing faults of J divergences
The real-time status characteristic index of bearing to be measured and the J divergences of abort situation feature samples, the J divergences of normal condition feature samples,
The abort situation of bearing is judged according to the size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition, is realized
To the fault diagnosis of bearing to be measured.
In order to better illustrate the method for diagnosing faults data by J divergences, different faults position when the bearing is run
One or several kinds of failures that the failure sensing data put is included in the inner ring failure, outer ring failure and rolling element failure of bearing pass
Feel data, the feature sample of abort situation is obtained after pretreatment by the failure sensing data of the different faults position of acquisition
This, the feature samples of abort situation corresponding with one or several kinds of failure sensing datas include inner ring fault signature sample, outer
Enclose fault signature sample and rolling element fault signature sample, that is to say, that if choosing one or several kinds of failure sensing datas,
So, the feature samples of one or several kinds of corresponding abort situation are just selected, here, can be construed as:To adapt to not
Same type bearing, different operating modes, may be selected the different characteristic type of extraction, to adapt to and meet different analysis demands, simultaneously
For the situation of practical application, abort situation can adaptively be adjusted, such as the failure of retainer in some cases
Occurrence frequency is higher, can increase the data sample of retainer failure, carries out feature extraction;Or under certain operating modes, outer ring
Failure hardly occurs, and can delete the data sample of outer ring failure.
If the failure sensing data of different faults position includes inner ring failure, the outer ring of bearing simultaneously when bearing is run
When failure and these three failure sensing datas of rolling element failure, the method for diagnosing faults by J divergences calculates to be measured respectively
The real-time status characteristic index of bearing and the J divergences of abort situation feature samples, the J divergences of normal condition feature samples, according to
The size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition judges the abort situation of bearing, and realization is treated
The fault diagnosis of bearing is surveyed, process is as follows,
Pass through formula J divergence formula:The spy of real-time bearing state is calculated respectively
Levy index and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
Between J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergences public affairs
In formula, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two,
N is the number of signal characteristic value, and i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, is dissipated
Angle value is smaller, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, leads to
The size of divergence value is crossed to determine the fault type of bearing.
Embodiment 2:
The method that the present invention is explained with reference to specific data and attached drawing is to sense number with three kinds of failures in the present embodiment
Illustrate for, as shown in Figure 2, a kind of logic-based returns specific device figure and the rolling bearing fault of J divergences is pre-
Method is examined, is included the following steps:
Step 1:Three kinds of failure sensing datas of different faults position and the operation of normal condition bearing when acquiring bearing operation
Normal sensing data, respectively to three kinds of failure sensing datas and normal sensing data is pre-processed and feature extraction, establish
Three kinds of abort situation feature samples and normal condition feature samples;
Step 2:By the sample of established three kinds of abort situation and the sample of normal condition to Logic Regression Models into
Row training, obtains logical model parameter, establishes Logic Regression Models;
Step 3:The real-time sensory data of bearing to be measured is acquired, the real-time sensory data is pre-processed and feature carries
It takes, obtains the real-time status characteristic index of bearing to be measured, the characteristic index is updated to the established logistic regression mould
In type, by the health degree that bearing current time to be measured is calculated;
Step 4:The health degree at the bearing current time to be measured being calculated and pre-set health degree threshold value are made
Compare, if the health degree at bearing current time to be measured is counted respectively less than the threshold value of setting by the method for diagnosing faults of J divergences
Calculate the real-time status characteristic index of bearing to be measured and the J divergences of three kinds of abort situation feature samples, the J of normal condition feature samples
Divergence judges the failure of bearing according to the size of the J divergences of the sample of three kinds of abort situation and the J divergences of the sample of normal condition
The fault diagnosis to bearing to be measured is realized in position.
This method only in threshold value of the health degree at bearing current time less than setting, just passes through the fault diagnosis of J divergences
Method calculates the bearing state at current time and J divergences, the normal condition feature samples of three kinds of abort situation feature samples respectively
J divergences, the failure of bearing is judged according to the size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition
Position, that is to say, that if the health degree at bearing current time is not less than the threshold value set, illustrating bearing, there is no events
Barrier does not need to carry out fault diagnosis.
In the present embodiment, three kinds of failure sensing datas are the inner ring failure of bearing, outer ring failure and rolling element failure
Failure sensing data obtains abort situation by the failure sensing data of the different faults position of acquisition after pretreatment
Feature samples, the feature samples of abort situation corresponding with three kinds include inner ring fault signature sample, outer ring fault signature sample
With rolling element fault signature sample.
In the present embodiment, the foundation of Logic Regression Models is according to following theory:Assuming that the vector of N number of independent variable, XT
=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, and the mathematic(al) representation of Logic Regression Models is:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient is estimated by maximum likelihood
Meter method is asked for.
When the feature samples of abort situation are that inner ring fault signature sample, outer ring fault signature sample and rolling element failure are special
When levying sample, the method for diagnosing faults by J divergences calculates the real-time status characteristic index and failure of bearing to be measured respectively
The J divergences of position feature sample, the J divergences of normal condition feature samples, according to the J divergences of the sample of abort situation and normal shape
The size of the J divergences of the sample of state judges the abort situation of bearing, realizes the fault diagnosis to bearing to be measured, and process is as follows,
Pass through formula J divergence formula:The spy of real-time bearing state is calculated respectively
Levy index and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
Between J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergences public affairs
In formula, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two,
N is the number of signal characteristic value, and i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, is dissipated
Angle value is smaller, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, leads to
The size of divergence value is crossed to determine the fault type of bearing.
For actual conditions, in the present invention by taking Rexnord ZA-2115 bearings as an example, by the axis of 4 models
It holds and is mounted on an axle, alternating current generator rotating speed is 2000rpm, applies 27kN radial loads, utilizes PCB353B33 vibrating sensors reality
When acquire bearing vibration signal, experimental provision is as shown in Figure 3.
The health degree of bearing is calculated, until a certain bearing therein is entirely ineffective, draws the full life and health of failing bearings
Curve is assessed, health degree changes as shown in Figure 4:Health degree threshold value is set as 0.3 (preset health degree threshold value), from attached
As can be seen that bearing state evaluation index CV values change between 0-1 in Fig. 3, here, the physical significance of CV values is current for bearing
The health degree of health status.CV values illustrate that rolling bearing fully belongs to normal performance state for 1, and CV values illustrate to roll for 0
Bearing is in performance worst state, and to adapt to different operating modes, bearing designation, above two state and its CV values are can to do certainly
Definition setting.For example, if application project is stringenter to bearing performance requirement, the mid-term stage of decline in health can be chosen
The CV values of sample are set as 0, and are trained using Logic Regression Models, obtain the health degree curve under stringent performance requirements,
For CV values in the range of first day was substantially at [0.7,1] by the 25th day, numerical value change is little, is defined as bearing health
Phase.As bearing works on, when bearing working was to the 25th day, the CV values of bearing start to taper into, and less than advance
The health degree threshold value of the healthy phase of setting in the range of [0.3,0.7], is defined as the decline in health phase.When bearing working extremely
At the 27th day, the health degree of bearing drastically declines, and CV values are less than 0.3, and the health degree of bearing has reached bearing fault threshold value, then opens
Begin to carry out fault diagnosis and fault type is judged.In this example, the health degree threshold value of startup separator diagnosis is set
Into 0.3, situation is actually used according to bearing, which can be adjusted, will in high reliability to adapt to different analysis demands
Under the application scenarios asked, threshold value can be improved;High economy is being pursued, and hang-up is lost under little application scenarios,
Threshold value can be reduced.
When the health degree of bearing has reached bearing fault threshold value, then proceed by fault diagnosis and fault type is sentenced
It is disconnected, at this point, calculate the characteristic index and normal condition of real-time bearing state, inner ring failure, outer ring failure, rolling element failure spy
The J divergences between index are levied, four J divergence values are obtained:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, such as table 1
It is shown.
J divergences between 1 real-time status of table and standard failure sample
As shown in Table 1, the state of bearing, real-time bearing state and three kinds of standard shapes can be clearly distinguished by J divergences
J divergence value differences between state are apparent, wherein real-time bearing state and the J divergence values J of inner ring malfunction2It is smaller, represent axis
It holds current state and inner ring malfunction is most close.It calculates and verifies by many experiments, this method can clearly distinguish bearing
Different states, effectively diagnoses bearing fault, and accuracy rate and sensitivity are higher.
Embodiment 2:
A kind of logic-based returns and the rolling bearing fault of J divergences examines system in advance, as shown in figure 5, including:
Preprocessing module 1 is acquired, for acquiring the failure sensing data of different faults position and normal shape when bearing is run
The normal sensing data of state bearing operation, respectively pre-processes failure sensing data and normal sensing data and feature carries
It takes, establishes abort situation feature samples and normal condition feature samples;
Model building module 2, for being returned by the sample of established abort situation and the sample of normal condition to logic
Model is returned to be trained, logical model parameter is obtained, establishes Logic Regression Models;
Computing module 3 is acquired, for acquiring the real-time sensory data of bearing to be measured, the real-time sensory data is carried out pre-
Processing and feature extraction, obtain the real-time status characteristic index of bearing to be measured, the characteristic index are updated to established institute
It states in Logic Regression Models, by the health degree that bearing current time to be measured is calculated;
Judgment module 4, for by the health degree at the bearing current time to be measured being calculated and pre-set health
Degree threshold value is made comparisons, if the health degree at bearing current time to be measured passes through the fault diagnosis side of J divergences less than the threshold value of setting
Method calculates the real-time status characteristic index of bearing to be measured and the J divergences of abort situation feature samples, normal condition feature sample respectively
This J divergences judge the event of bearing according to the size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition
Hinder position, realize the fault diagnosis to bearing to be measured.
Specifically, model building module 2 is arranged to,
Assuming that the vector of N number of independent variable, XT=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, Logic Regression Models
Mathematic(al) representation be:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient is estimated by maximum likelihood
Meter method is asked for.
Further, the acquisition preprocessing module 1 is arranged to, the event of different faults position when the bearing is run
Hinder one or several kinds of failures that sensing data is included in the inner ring failure, outer ring failure and rolling element failure of bearing and sense number
According to.The acquisition preprocessing module is arranged to, and the feature samples of the abort situation include inner ring fault signature sample, outer ring
One or several kinds in fault signature sample and rolling element fault signature sample.
For further, when the feature samples of abort situation are inner ring fault signature sample, outer ring fault signature sample
During with rolling element fault signature sample, the judgment module is arranged to,
Pass through formula J divergence formula:The spy of real-time bearing state is calculated respectively
Levy index and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
Between J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergences public affairs
In formula, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two,
N is the number of signal characteristic value, and i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, is dissipated
Angle value is smaller, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, leads to
The size of divergence value is crossed to determine the fault type of bearing.
Furthermore, it is necessary to illustrate, the specific embodiment described in this specification, the shape of parts and components is named
Title etc. can be different.The equivalent or simple change that all construction, feature and principles according to described in inventional idea of the present invention are done, is wrapped
It includes in the protection domain of patent of the present invention.Those skilled in the art can be to described specific implementation
Example is done various modifications or additions or is substituted in a similar way, without departing from structure of the invention or surmounts this
Range as defined in the claims, is within the scope of protection of the invention.
Claims (10)
1. a kind of logic-based returns and the rolling bearing fault of J divergences examines method in advance, it is characterised in that includes the following steps:
The failure sensing data of different faults position and the normal sensing data of normal condition bearing operation when acquiring bearing operation,
Respectively to failure sensing data and normal sensing data is pre-processed and feature extraction, abort situation feature samples and just are established
Normal state feature samples;
Logic Regression Models are trained by the sample of established abort situation and the sample of normal condition, obtain logic
Model parameter establishes Logic Regression Models;
The real-time sensory data of bearing to be measured is acquired, the real-time sensory data is pre-processed and feature extraction, is treated
The real-time status characteristic index of bearing is surveyed, the characteristic index is updated in the established Logic Regression Models, is passed through
The health degree at bearing current time to be measured is calculated;
The health degree at the bearing current time to be measured being calculated and pre-set health degree threshold value are made comparisons, if to be measured
The health degree at bearing current time then calculates bearing to be measured respectively less than the threshold value of setting by the method for diagnosing faults of J divergences
Real-time status characteristic index and abort situation feature samples J divergences, the J divergences of normal condition feature samples, according to failure
The size of the J divergences of the sample of position and the J divergences of the sample of normal condition judges the abort situation of bearing, realizes to axis to be measured
The fault diagnosis held.
2. logic-based according to claim 1 returns and the rolling bearing fault of J divergences examines method in advance, feature exists
In:The specific steps of the Logic Regression Models acquisition methods include,
Assuming that the vector of N number of independent variable, XT=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, the number of Logic Regression Models
Learning expression formula is:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient passes through Maximum Likelihood Estimation Method
To ask for.
3. logic-based according to claim 1 returns and the rolling bearing fault of J divergences examines method in advance, feature exists
In:The failure sensing data of different faults position includes inner ring failure, outer ring failure and the rolling of bearing during the bearing operation
One or several kinds of failure sensing datas in body failure.
4. logic-based according to claim 3 returns and the rolling bearing fault of J divergences examines method in advance, feature exists
In:It is special that the feature samples of the abort situation include inner ring fault signature sample, outer ring fault signature sample and rolling element failure
Levy one or several kinds of fault signature samples in sample.
5. logic-based according to claim 4 returns and the rolling bearing fault of J divergences examines method in advance, feature exists
In:When the feature samples of abort situation are inner ring fault signature sample, outer ring fault signature sample and rolling element fault signature sample
This when, the method for diagnosing faults by J divergences calculate the real-time status characteristic index and abort situation of bearing to be measured respectively
The J divergences of feature samples, the J divergences of normal condition feature samples, according to the J divergences of the sample of abort situation and normal condition
The size of the J divergences of sample judges the abort situation of bearing, realizes the fault diagnosis to bearing to be measured, and process is as follows,
Pass through formula J divergence formula:The feature for calculating real-time bearing state respectively refers to
Between mark and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergence formula
In, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two, N
For the number of signal characteristic value, i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, divergence value is got over
Small, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, passes through divergence
The size of value determines the fault type of bearing.
6. a kind of logic-based returns and the rolling bearing fault of J divergences examines system in advance, which is characterized in that including:
Preprocessing module is acquired, for acquiring the failure sensing data of different faults position and normal condition bearing when bearing is run
The normal sensing data of operation respectively to failure sensing data and normal sensing data is pre-processed and feature extraction, is established
Abort situation feature samples and normal condition feature samples;
Model building module, for by the sample of established abort situation and the sample of normal condition to Logic Regression Models
It is trained, obtains logical model parameter, establish Logic Regression Models;
Computing module is acquired, for acquiring the real-time sensory data of bearing to be measured, the real-time sensory data is pre-processed
And feature extraction, the real-time status characteristic index of bearing to be measured is obtained, the characteristic index is updated to established described patrol
It collects in regression model, by the health degree that bearing current time to be measured is calculated;
Judgment module, for by the health degree at the bearing current time to be measured being calculated and pre-set health degree threshold value
It makes comparisons, if the health degree at bearing current time to be measured is distinguished less than the threshold value of setting by the method for diagnosing faults of J divergences
It calculates the real-time status characteristic index of bearing to be measured and the J divergences of abort situation feature samples, the J of normal condition feature samples dissipates
Degree judges the abort situation of bearing according to the size of the J divergences of the sample of abort situation and the J divergences of the sample of normal condition,
Realize the fault diagnosis to bearing to be measured.
7. logic-based according to claim 6 returns and the rolling bearing fault of J divergences examines system in advance, feature exists
In:Model building module is arranged to,
Assuming that the vector of N number of independent variable, XT=(x1,x2,…,xN),yi∈ { 0,1 }, y are 0 or 1, the number of Logic Regression Models
Learning expression formula is:
yi=1 represents event;yi=0 expression event does not occur;
pi(yi=1/xi) represent the probability that observed quantity occurs relative to i-th of event (value is between 0 to 1);
It represents to return intercept;β1, β2... βNRepresent regression coefficient;It returns intercept and regression coefficient passes through Maximum Likelihood Estimation Method
To ask for.
8. logic-based according to claim 6 returns and the rolling bearing fault of J divergences examines system in advance, feature exists
In:The acquisition preprocessing module is arranged to, and the failure sensing data of different faults position includes axis when the bearing is run
One or several kinds of failure sensing datas in inner ring failure, outer ring failure and the rolling element failure held.
9. logic-based according to claim 8 returns and the rolling bearing fault of J divergences examines system in advance, feature exists
In:The acquisition preprocessing module is arranged to, and the feature samples of the abort situation include inner ring fault signature sample, outer ring
One or several kinds of fault signature samples in fault signature sample and rolling element fault signature sample.
10. logic-based according to claim 9 returns and the rolling bearing fault of J divergences examines system in advance, feature exists
In:When the feature samples of abort situation are inner ring fault signature sample, outer ring fault signature sample and rolling element fault signature sample
This when, the judgment module are arranged to,
Pass through formula J divergence formula:The feature for calculating real-time bearing state respectively refers to
Between mark and inner ring fault signature sample, outer ring fault signature sample, rolling element fault signature sample, normal condition feature samples
J divergences, obtain four J divergence values, respectively:Divergence value J1, divergence value J2, divergence value J3, divergence value J4, J divergence formula
In, S is the characteristic value of normal state signal;τ is the characteristic value of unknown state signal;J (s, τ) is J divergences between the two, N
For the number of signal characteristic value, i is the sequence of signal characteristic value;
Compare divergence value J1, divergence value J2, divergence value J3With divergence value J4, the minimum value in four J divergence values is found out, divergence value is got over
Small, then bearing current time state state corresponding with preset pre-set health degree threshold value is more close, passes through divergence
The size of value determines the fault type of bearing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711216940.4A CN108181105B (en) | 2017-11-28 | 2017-11-28 | Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711216940.4A CN108181105B (en) | 2017-11-28 | 2017-11-28 | Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108181105A true CN108181105A (en) | 2018-06-19 |
CN108181105B CN108181105B (en) | 2020-08-07 |
Family
ID=62545456
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711216940.4A Active CN108181105B (en) | 2017-11-28 | 2017-11-28 | Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108181105B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110987433A (en) * | 2019-12-13 | 2020-04-10 | 华中科技大学 | Bearing fault early warning method based on high-frequency signal characteristic amplitude |
CN111076962A (en) * | 2020-01-03 | 2020-04-28 | 中国电建集团华东勘测设计研究院有限公司 | Electromechanical equipment fault diagnosis method for intelligent hydraulic power plant |
CN111260822A (en) * | 2019-12-31 | 2020-06-09 | 杭州安脉盛智能技术有限公司 | Rail transit vehicle health state analysis method and terminal based on big data |
CN111651933A (en) * | 2020-05-22 | 2020-09-11 | 宁波诺丁汉新材料研究院有限公司 | Industrial boiler fault early warning method and system based on statistical inference |
CN112396344A (en) * | 2020-11-30 | 2021-02-23 | 天津大学 | Chemical process reliability online evaluation method based on product quality |
WO2021042748A1 (en) * | 2019-09-02 | 2021-03-11 | 佛山科学技术学院 | Estimation method of bearing operation condition stability employing probability distribution, and device |
CN112614342A (en) * | 2020-12-10 | 2021-04-06 | 大唐高鸿数据网络技术股份有限公司 | Early warning method for road abnormal event, vehicle-mounted equipment and road side equipment |
CN112948223A (en) * | 2019-11-26 | 2021-06-11 | 北京沃东天骏信息技术有限公司 | Method and device for monitoring operation condition |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102682180A (en) * | 2011-03-07 | 2012-09-19 | 上海宝信软件股份有限公司 | Evaluation method for performance degradation of rotary mechanical equipment |
CN103679378A (en) * | 2013-12-20 | 2014-03-26 | 北京航天测控技术有限公司 | Method and device for evaluating heath state of spacecraft on basis of telemeasuring data |
CN103868692A (en) * | 2014-03-18 | 2014-06-18 | 电子科技大学 | Rotary machine fault diagnosis method based on kernel density estimation and K-L divergence |
CN104392082A (en) * | 2014-07-10 | 2015-03-04 | 中山火炬职业技术学院 | Diagnosis method for initial failure of gearbox of wind generating set based on vibration monitoring |
CN105241680A (en) * | 2015-08-26 | 2016-01-13 | 电子科技大学 | Health state assessment method for rotary machine based on probability density function |
CN105510023A (en) * | 2015-11-24 | 2016-04-20 | 国网内蒙古东部电力有限公司电力科学研究院 | Divergence-index-based fault diagnosis method of variable-working-condition wind-power planetary gearbox |
CN106053066A (en) * | 2016-05-23 | 2016-10-26 | 华东交通大学 | Antifriction bearing performance degradation assessment method based on empirical mode decomposition and logistic regression |
-
2017
- 2017-11-28 CN CN201711216940.4A patent/CN108181105B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102682180A (en) * | 2011-03-07 | 2012-09-19 | 上海宝信软件股份有限公司 | Evaluation method for performance degradation of rotary mechanical equipment |
CN103679378A (en) * | 2013-12-20 | 2014-03-26 | 北京航天测控技术有限公司 | Method and device for evaluating heath state of spacecraft on basis of telemeasuring data |
CN103868692A (en) * | 2014-03-18 | 2014-06-18 | 电子科技大学 | Rotary machine fault diagnosis method based on kernel density estimation and K-L divergence |
CN104392082A (en) * | 2014-07-10 | 2015-03-04 | 中山火炬职业技术学院 | Diagnosis method for initial failure of gearbox of wind generating set based on vibration monitoring |
CN105241680A (en) * | 2015-08-26 | 2016-01-13 | 电子科技大学 | Health state assessment method for rotary machine based on probability density function |
CN105510023A (en) * | 2015-11-24 | 2016-04-20 | 国网内蒙古东部电力有限公司电力科学研究院 | Divergence-index-based fault diagnosis method of variable-working-condition wind-power planetary gearbox |
CN106053066A (en) * | 2016-05-23 | 2016-10-26 | 华东交通大学 | Antifriction bearing performance degradation assessment method based on empirical mode decomposition and logistic regression |
Non-Patent Citations (2)
Title |
---|
郭艳平 等: ""基于经验模态分解和散度指标的风力发电机滚动轴承故障诊断方法"", 《电力系统保护与控制》 * |
钟鑫 等: ""基于逻辑回归的滚动轴承性能退化评估"", 《科技信息》 * |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021042748A1 (en) * | 2019-09-02 | 2021-03-11 | 佛山科学技术学院 | Estimation method of bearing operation condition stability employing probability distribution, and device |
CN112948223A (en) * | 2019-11-26 | 2021-06-11 | 北京沃东天骏信息技术有限公司 | Method and device for monitoring operation condition |
CN110987433A (en) * | 2019-12-13 | 2020-04-10 | 华中科技大学 | Bearing fault early warning method based on high-frequency signal characteristic amplitude |
CN110987433B (en) * | 2019-12-13 | 2021-05-18 | 华中科技大学 | Bearing fault early warning method based on high-frequency signal characteristic amplitude |
CN111260822A (en) * | 2019-12-31 | 2020-06-09 | 杭州安脉盛智能技术有限公司 | Rail transit vehicle health state analysis method and terminal based on big data |
CN111260822B (en) * | 2019-12-31 | 2022-07-26 | 杭州安脉盛智能技术有限公司 | Rail transit vehicle health state analysis method and terminal based on big data |
CN111076962A (en) * | 2020-01-03 | 2020-04-28 | 中国电建集团华东勘测设计研究院有限公司 | Electromechanical equipment fault diagnosis method for intelligent hydraulic power plant |
CN111076962B (en) * | 2020-01-03 | 2022-03-29 | 中国电建集团华东勘测设计研究院有限公司 | Electromechanical equipment fault diagnosis method for intelligent hydraulic power plant |
CN111651933A (en) * | 2020-05-22 | 2020-09-11 | 宁波诺丁汉新材料研究院有限公司 | Industrial boiler fault early warning method and system based on statistical inference |
CN111651933B (en) * | 2020-05-22 | 2023-09-26 | 宁波诺丁汉新材料研究院有限公司 | Industrial boiler fault early warning method and system based on statistical inference |
CN112396344A (en) * | 2020-11-30 | 2021-02-23 | 天津大学 | Chemical process reliability online evaluation method based on product quality |
CN112614342A (en) * | 2020-12-10 | 2021-04-06 | 大唐高鸿数据网络技术股份有限公司 | Early warning method for road abnormal event, vehicle-mounted equipment and road side equipment |
Also Published As
Publication number | Publication date |
---|---|
CN108181105B (en) | 2020-08-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108181105A (en) | Logic-based returns and the rolling bearing fault of J divergences examines method and system in advance | |
CN110823576B (en) | Mechanical anomaly detection method based on generation of countermeasure network | |
CN110687896A (en) | Fault diagnosis method, device, equipment and readable medium | |
CN108896299A (en) | A kind of gearbox fault detection method | |
CN109213121A (en) | A kind of fan braking system card cylinder method for diagnosing faults | |
CN106429689B (en) | Elevator maintenance system based on the support of Internet of Things big data | |
CN110060368A (en) | Mechanical method for detecting abnormality based on potential feature coding | |
CN111665066B (en) | Equipment fault self-adaptive upper and lower early warning boundary generation method based on convolutional neural network | |
CN108267312A (en) | A kind of subway train bearing intelligent diagnostic method based on fast search algorithm | |
CN111611294B (en) | Star sensor data anomaly detection method | |
CN105487009A (en) | Motor fault diagnosis method based on k-means RBF neural network algorithm | |
CN112729834B (en) | Bearing fault diagnosis method, device and system | |
CN108398265A (en) | A kind of online fault detection method of rolling bearing | |
CN113408068A (en) | Random forest classification machine pump fault diagnosis method and device | |
CN109855873A (en) | The method for diagnosing faults and device of steam turbine main shaft | |
CN107977679A (en) | Method based on frequency response function and operation response characteristic diagnosis of complex device initial failure | |
CN109990803A (en) | The method, apparatus of method, apparatus and the sensor processing of detection system exception | |
CN116611184A (en) | Fault detection method, device and medium for gear box | |
CN108414228B (en) | Based on averagely more granularity decision rough sets and NNBC Method for Bearing Fault Diagnosis | |
CN113487141B (en) | Bearing state evaluation method based on multisource information clustering fusion | |
CN109894476B (en) | Fault diagnosis method and device for hydraulic equipment of cold-rolled silicon steel production line | |
Chen et al. | Dynamic Vision Enabled Contactless Cross-Domain Machine Fault Diagnosis with Neuromorphic Computing | |
Zhang et al. | A health indicator construction method based on the information entropy of specific frequency band and GGRBM | |
CN113377564A (en) | Fault diagnosis method and device, computer equipment and storage medium | |
Kolar et al. | Identification of inability states of rotating machinery subsystems using industrial IoT and convolutional neural network–initial research |
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 |