CN108181105B - Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence - Google Patents
Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence Download PDFInfo
- Publication number
- CN108181105B CN108181105B CN201711216940.4A CN201711216940A CN108181105B CN 108181105 B CN108181105 B CN 108181105B CN 201711216940 A CN201711216940 A CN 201711216940A CN 108181105 B CN108181105 B CN 108181105B
- Authority
- CN
- China
- Prior art keywords
- fault
- divergence
- bearing
- sample
- characteristic
- 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.)
- Active
Links
Images
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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a rolling bearing fault pre-diagnosis method based on logistic regression and J divergence, which comprises the following steps of: acquiring fault sensing data of different fault positions of a bearing in operation and normal sensing data of the bearing in a normal state in operation, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample; training a logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters, and establishing the logistic regression model; according to the invention, the bearing is subjected to targeted fault diagnosis according to the health decline degree of the bearing, parameters such as training sample types, characteristic value types, health threshold values and the like are changed according to different objects under different working conditions, and the trained model can be adjusted, so that the method has the advantages of strong real-time performance, high data processing precision, good robustness of a core algorithm, high state evaluation accuracy, high accuracy of a diagnosis result and the like.
Description
Technical Field
The invention relates to the technical field of quality reliability assessment and fault diagnosis of mechanical products, in particular to a rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence.
Background
The rolling bearing is an important supporting piece of the rotary machine, and the real-time running state of the rolling bearing directly determines the reliability of mechanical equipment, so that the rolling bearing on-line fault diagnosis research is very important, the current running state of the rolling bearing, the fault position positioning and the fault severity quantification are evaluated in real time, and the rolling bearing on-line fault diagnosis method is of very important guiding significance to the maintenance and design of the mechanical equipment. The method realizes an online fault pre-diagnosis method based on logistic regression and J divergence aiming at the rolling bearing, can evaluate the running state of the rolling bearing in real time, and positions a fault part based on the current running state.
The existing online fault diagnosis methods are researched a lot, and the existing mature methods comprise a time domain effective value peak value judgment method, an amplitude probability density analysis method, an impact pulse method and the like. These methods only give a fault or normal condition, and in fact, the bearing is in operation for most of the time between the fault and the normal condition, neither complete fault nor complete normal condition, and this intermediate condition can be referred to as a degraded health condition. The state of health decline of the bearing is different from the common bearing fault and is a recessive, gradual change and long process. In the bearing in the state, in a stable decline period, the performance of the bearing shows a stable and slow decline trend and forms an early fault, the performance of the bearing is rapidly declined from the early fault to a final fault state, the whole decline process is quantified and dominated through a monitoring technology and a signal processing technology, and the health state of the bearing is favorably managed. Obviously, the method only gives out fault or normal state data and loses most health decline state data, the running state of the bearing cannot be accurately evaluated, and the method has high misjudgment rate and misjudgment rate.
Disclosure of Invention
The invention provides a rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence, aiming at the defect that the running state of a bearing cannot be accurately evaluated in the prior art.
In order to solve the technical problem, the invention is solved by the following technical scheme:
a rolling bearing fault pre-diagnosis method based on logistic regression and J divergence comprises the following steps:
acquiring fault sensing data of different fault positions of a bearing in operation and normal sensing data of the bearing in a normal state in operation, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
training a logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters, and establishing the logistic regression model;
acquiring real-time sensing data of a bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and comparing the calculated health degree of the bearing to be tested at the current moment with a preset health degree threshold, if the health degree of the bearing to be tested at the current moment is lower than the preset threshold, respectively calculating the J divergence of the real-time state characteristic index of the bearing to be tested, the fault position characteristic sample and the normal state characteristic sample according to the J divergence of the fault position sample and the J divergence of the normal state sample by a J divergence fault diagnosis method, and judging the fault position of the bearing to be tested according to the J divergence of the fault position sample and the J divergence of the normal state sample so as to realize fault diagnosis of the bearing to be tested.
As an implementation mode, the specific steps of the logistic regression model obtaining method comprise,
assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
pi(yi=1/xi) Representing the probability (between 0 and 1) of the observed quantity relative to the occurrence of the ith event;
representing the regression intercept β1,β2,…βNRepresenting a regression coefficient; the regression intercept and the regression coefficient are obtained by a maximum likelihood estimation method.
As an implementation manner, the fault sensing data of different fault positions when the bearing operates comprise fault sensing data of one or more of inner ring faults, outer ring faults and rolling body faults of the bearing.
As an implementation manner, the feature samples of the fault positions include one or more fault feature samples of an inner ring fault feature sample, an outer ring fault feature sample and a rolling body fault feature sample.
As an implementation mode, when the characteristic samples of the fault position are an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample, the real-time state characteristic index of the bearing to be tested, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample are respectively calculated by the J divergence fault diagnosis method, the fault position of the bearing is judged according to the J divergence of the sample of the fault position and the J divergence of the sample of the normal state, the fault diagnosis of the bearing to be tested is realized, the process is as follows,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
A rolling bearing fault pre-diagnosis system based on logistic regression and J divergence comprises:
the system comprises an acquisition preprocessing module, a data processing module and a data processing module, wherein the acquisition preprocessing module is used for acquiring fault sensing data of different fault positions when a bearing runs and normal sensing data when the bearing runs in a normal state, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
the model establishing module is used for training the logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters and establish the logistic regression model;
the acquisition and calculation module is used for acquiring real-time sensing data of the bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and the judging module is used for comparing the calculated health degree of the bearing to be detected at the current moment with a preset health degree threshold, if the health degree of the bearing to be detected at the current moment is lower than the preset threshold, respectively calculating the real-time state characteristic index of the bearing to be detected, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing fault diagnosis of the bearing to be detected.
As an implementation manner, when the feature samples of the fault location are the inner ring fault feature sample, the outer ring fault feature sample and the rolling element fault feature sample, the model building module is configured to,
assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
pi(yi=1/xi) Representing the probability (between 0 and 1) of the observed quantity relative to the occurrence of the ith event;
representing the regression intercept β1,β2,…βNRepresenting a regression coefficient; the regression intercept and the regression coefficient are obtained by a maximum likelihood estimation method.
As an implementation manner, the acquisition preprocessing module is configured to set the fault sensing data of different fault positions of the bearing during operation to include one or more fault sensing data of an inner ring fault, an outer ring fault and a rolling body fault of the bearing.
As an implementation manner, the acquisition preprocessing module is configured to set the feature samples of the fault location to include one or more fault feature samples of an inner ring fault feature sample, an outer ring fault feature sample, and a rolling element fault feature sample.
In one embodiment, the determining module is configured to,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
Due to the adoption of the technical scheme, the invention has the remarkable technical effects that:
according to the invention, the bearing is subjected to targeted fault diagnosis according to the health decline degree of the bearing, parameters such as training sample types, characteristic value types, health threshold values and the like are changed according to different objects under different working conditions, and the trained model can be adjusted, so that the method has the advantages of strong real-time performance, high data processing precision, good robustness of a core algorithm, high state evaluation accuracy, high accuracy of a diagnosis result and the like.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a schematic flow diagram of an exemplary method of the present invention;
FIG. 2 is a schematic flow chart of the present invention;
FIG. 3 is a schematic view of an experimental set-up according to the present invention;
FIG. 4 is a graphical representation of an evaluation index of a full life cycle plotted in accordance with the present invention;
fig. 5 is a schematic diagram of the system architecture of the present invention.
Detailed Description
The present invention will be described in further detail with reference to examples, which are illustrative of the present invention and are not to be construed as being limited thereto.
An exemplary method:
a rolling bearing fault pre-diagnosis method based on logistic regression and J divergence is disclosed, as shown in figure 1, and comprises the following steps:
s1: acquiring fault sensing data of different fault positions of a bearing in operation and normal sensing data of the bearing in a normal state in operation, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
s2: training a logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters, and establishing the logistic regression model;
s3: acquiring real-time sensing data of a bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
s4: and comparing the calculated health degree of the bearing to be tested at the current moment with a preset health degree threshold, if the health degree of the bearing to be tested at the current moment is lower than the preset threshold, respectively calculating the J divergence of the real-time state characteristic index of the bearing to be tested, the fault position characteristic sample and the normal state characteristic sample according to the J divergence of the fault position sample and the J divergence of the normal state sample by a J divergence fault diagnosis method, and judging the fault position of the bearing to be tested according to the J divergence of the fault position sample and the J divergence of the normal state sample so as to realize fault diagnosis of the bearing to be tested.
In order to better explain fault diagnosis method data through J divergence, fault sensing data of different fault positions of a bearing during operation comprise one or more fault sensing data of inner ring faults, outer ring faults and rolling body faults of the bearing, a fault position feature sample is obtained after preprocessing the obtained fault sensing data of the different fault positions, the fault position feature sample corresponding to the fault sensing data comprises an inner ring fault feature sample, an outer ring fault feature sample and a rolling body fault feature sample, namely, if one or more fault sensing data are selected, one or more corresponding fault position feature samples are selected, and the fault position feature samples can be interpreted as follows: in order to adapt to different types of bearings and different working conditions, different feature types can be selected and extracted to adapt to and meet different analysis requirements, and meanwhile, the fault position can be adaptively adjusted according to the actual application condition, for example, the fault occurrence frequency of the retainer is high under certain conditions, the data samples of the retainer faults can be increased, and the feature extraction is carried out; or under some working conditions, the outer ring fault hardly occurs, and the data sample of the outer ring fault can be deleted.
If the fault sensing data of different fault positions of the bearing during operation simultaneously comprises three fault sensing data of an inner ring fault, an outer ring fault and a rolling body fault of the bearing, respectively calculating the J divergence of a real-time state characteristic index, a fault position characteristic sample and a normal state characteristic sample of the bearing to be tested by a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing the fault diagnosis of the bearing to be tested, wherein the process comprises the following steps,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
Example 2:
the method of the present invention is explained with reference to specific data and accompanying drawings, in this embodiment, three kinds of fault sensing data are taken as an example for explanation, a specific device diagram is shown in fig. 2, and a rolling bearing fault pre-diagnosis method based on logistic regression and J divergence includes the following steps:
step 1: acquiring three fault sensing data of different fault positions of a bearing in operation and normal sensing data of the bearing in a normal state, respectively preprocessing and extracting characteristics of the three fault sensing data and the normal sensing data, and establishing three fault position characteristic samples and normal state characteristic samples;
step 2: training a logistic regression model through the established samples of the three fault positions and the samples of the normal state to obtain logistic model parameters, and establishing the logistic regression model;
and step 3: acquiring real-time sensing data of a bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and 4, step 4: and comparing the health degree of the bearing to be tested at the current moment obtained by calculation with a preset health degree threshold, if the health degree of the bearing to be tested at the current moment is lower than the preset threshold, respectively calculating the real-time state characteristic index of the bearing to be tested, the J divergence of the three fault position characteristic samples and the J divergence of the normal state characteristic samples by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the three fault position samples and the J divergence of the normal state samples, and realizing fault diagnosis of the bearing to be tested.
According to the method, only when the health degree of the bearing at the current moment is lower than a set threshold value, the J divergence of the bearing state at the current moment, the J divergence of three fault position characteristic samples and the J divergence of normal state characteristic samples are respectively calculated through a J divergence fault diagnosis method, and the fault position of the bearing is judged according to the J divergence of the fault position samples and the J divergence of the normal state samples, namely, if the health degree of the bearing at the current moment is not lower than the set threshold value, the bearing does not have faults, and fault diagnosis is not needed.
In this embodiment, the three types of fault sensing data are fault sensing data of an inner ring fault, an outer ring fault and a rolling element fault of the bearing, the obtained fault sensing data of different fault positions are preprocessed to obtain feature samples of the fault positions, and the three corresponding feature samples of the fault positions include an inner ring fault feature sample, an outer ring fault feature sample and a rolling element fault feature sample.
In this embodiment, the logistic regression model is established according to the following theory: assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
pi(yi=1/xi) Representing the probability (between 0 and 1) of the observed quantity relative to the occurrence of the ith event;
representing the regression intercept β1,β2,…βNRepresenting a regression coefficient; the regression intercept and the regression coefficient are obtained by a maximum likelihood estimation method.
When the characteristic samples of the fault position are an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample, respectively calculating the real-time state characteristic index of the bearing to be tested, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing the fault diagnosis of the bearing to be tested, wherein the process comprises the following steps,
by the formula J divergence formula:respectively calculating J divergence between the characteristic index of the real-time bearing state and the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, and respectively calculating the J divergence valuesComprises the following steps: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
In terms of practical conditions, the Rexnord ZA-2115 bearing is taken as an example in the invention, 4 bearings of the type are installed on a shaft, the rotating speed of an alternating current motor is 2000rpm, a 27kN radial load is applied, a PCB353B33 vibration sensor is used for acquiring vibration signals of the bearings in real time, and the experimental device is shown in figure 3.
Calculating the health degree of the bearings until one of the bearings completely fails, and drawing a full-life health evaluation curve of the failed bearing, wherein the health degree changes are shown in the attached figure 4: the health degree threshold is set to 0.3 (preset health degree threshold), and as can be seen from fig. 3, the CV value of the bearing state evaluation index varies between 0 and 1, where the physical meaning of the CV value is the health degree of the current health state of the bearing. The CV value of 1 indicates that the rolling bearing completely belongs to a normal performance state, the CV value of 0 indicates that the rolling bearing is in a worst performance state, and the two states and the CV value thereof can be set in a self-defined mode in order to adapt to different working conditions and bearing models. For example, if the bearing performance requirements are strict by applying engineering, the CV value of a sample in the middle stage of health decline can be selected to be set as 0, and a logistic regression model is adopted for training to obtain a health degree curve under the strict performance requirements, wherein the CV value is basically in the range of [0.7, 1] from the first day to the twenty-fifth day, the numerical value is not changed greatly, and the CV value is defined as the bearing health period. As the bearing continues to operate, the CV value of the bearing begins to taper as the bearing operates to day 25, and falls below a pre-set health threshold for the health period, in the range of [0.3, 0.7], defined as the health decline period. When the bearing works to the 27 th day, the health degree of the bearing is sharply reduced, the CV value is less than 0.3, and the health degree of the bearing reaches the fault threshold value of the bearing, fault diagnosis is started, and the fault type is judged. In the example, the health threshold for starting fault diagnosis is set to be 0.3, the health threshold can be adjusted according to the actual use condition of the bearing to adapt to different analysis requirements, and the health threshold can be increased in an application scene with high reliability requirement; in an application scenario where high economy is pursued and the unexpected shutdown loss is not great, the threshold value can be lowered.
When the health degree of the bearing reaches a bearing fault threshold value, starting fault diagnosis and judging the fault type, calculating the J divergence between the characteristic indexes of the real-time bearing state and the normal state, and the characteristic indexes of the inner ring fault, the outer ring fault and the rolling element fault, and obtaining four J divergence values in total: divergence value J1Divergence value J2Divergence value J3Divergence value J4As shown in table 1.
TABLE 1J divergence between real-time status and Standard Fault samples
As can be seen from Table 1, the bearing states can be clearly distinguished through the J divergence, and the J divergence value difference between the real-time bearing state and the three standard states is obvious, wherein the J divergence value J between the real-time bearing state and the inner ring fault state2The smaller the bearing is, the closest the current state of the bearing to the fault state of the inner ring is represented. Through multiple times of experimental calculation verification, the method can clearly distinguish different states of the bearing, effectively diagnoses the bearing fault, and has higher accuracy and sensitivity.
Example 2:
a rolling bearing fault pre-diagnosis system based on logistic regression and J divergence, as shown in fig. 5, comprising:
the system comprises an acquisition preprocessing module 1, a data processing module and a data processing module, wherein the acquisition preprocessing module 1 is used for acquiring fault sensing data of different fault positions when a bearing runs and normal sensing data of the bearing in a normal state, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
the model establishing module 2 is used for training the logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters and establish the logistic regression model;
the acquisition and calculation module 3 is used for acquiring real-time sensing data of the bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and the judging module 4 is used for comparing the calculated health degree of the bearing to be detected at the current moment with a preset health degree threshold, if the health degree of the bearing to be detected at the current moment is lower than the preset threshold, respectively calculating the real-time state characteristic index of the bearing to be detected, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing fault diagnosis of the bearing to be detected.
In particular, the model building block 2 is arranged to,
assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
pi(yi=1/xi) Representing the probability (between 0 and 1) of the observed quantity relative to the occurrence of the ith event;
representing the regression intercept β1,β2,…βNRepresenting a regression coefficient; the regression intercept and the regression coefficient are obtained by a maximum likelihood estimation method.
Furthermore, the acquisition preprocessing module 1 is configured to set the fault sensing data of different fault positions of the bearing during operation to include one or more fault sensing data of an inner ring fault, an outer ring fault and a rolling element fault of the bearing. The collection preprocessing module is set to be in a state that the characteristic samples of the fault positions comprise one or more of inner ring fault characteristic samples, outer ring fault characteristic samples and rolling body fault characteristic samples.
Further, when the characteristic samples of the fault position are an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample, the judging module is configured to,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
In addition, it should be noted that the specific embodiments described in the present specification may differ in the shape of the components, the names of the components, and the like. All equivalent or simple changes of the structure, the characteristics and the principle of the invention which are described in the patent conception of the invention are included in the protection scope of the patent of the invention. Various modifications, additions and substitutions for the specific embodiments described may be made by those skilled in the art without departing from the scope of the invention as defined in the accompanying claims.
Claims (10)
1. A rolling bearing fault pre-diagnosis method based on logistic regression and J divergence is characterized by comprising the following steps:
acquiring fault sensing data of different fault positions of a bearing in operation and normal sensing data of the bearing in a normal state in operation, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
training a logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters, and establishing the logistic regression model;
acquiring real-time sensing data of a bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and comparing the calculated health degree of the bearing to be tested at the current moment with a preset health degree threshold, if the health degree of the bearing to be tested at the current moment is lower than the preset threshold, respectively calculating the J divergence of the real-time state characteristic index of the bearing to be tested, the fault position characteristic sample and the normal state characteristic sample according to the J divergence of the fault position sample and the J divergence of the normal state sample by a J divergence fault diagnosis method, and judging the fault position of the bearing to be tested according to the J divergence of the fault position sample and the J divergence of the normal state sample so as to realize fault diagnosis of the bearing to be tested.
2. The rolling bearing fault pre-diagnosis method based on logistic regression and J divergence according to claim 1, characterized in that: the method for obtaining the logistic regression model comprises the following specific steps,
assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
yi1 indicates an event occurred; y isi0 means that the event did not occur;
pi(yi=1/xi) Representing the probability of occurrence of the observed quantity relative to the ith event, wherein the value of the probability of occurrence is between 0 and 1;
3. The rolling bearing fault pre-diagnosis method based on logistic regression and J divergence according to claim 1, characterized in that: the fault sensing data of different fault positions of the bearing during operation comprise one or more fault sensing data of inner ring faults, outer ring faults and rolling body faults of the bearing.
4. The rolling bearing fault pre-diagnosis method based on logistic regression and J divergence according to claim 3, characterized in that: the fault position characteristic sample comprises one or more fault characteristic samples of an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample.
5. The rolling bearing fault pre-diagnosis method based on logistic regression and J divergence according to claim 4, characterized in that: when the characteristic samples of the fault position are an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample, respectively calculating the real-time state characteristic index of the bearing to be tested, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing the fault diagnosis of the bearing to be tested, wherein the process comprises the following steps,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Divergence value J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
6. A rolling bearing fault pre-diagnosis system based on logistic regression and J divergence is characterized by comprising:
the system comprises an acquisition preprocessing module, a data processing module and a data processing module, wherein the acquisition preprocessing module is used for acquiring fault sensing data of different fault positions when a bearing runs and normal sensing data when the bearing runs in a normal state, respectively preprocessing and extracting characteristics of the fault sensing data and the normal sensing data, and establishing a fault position characteristic sample and a normal state characteristic sample;
the model establishing module is used for training the logistic regression model through the established fault position sample and the normal state sample to obtain logistic model parameters and establish the logistic regression model;
the acquisition and calculation module is used for acquiring real-time sensing data of the bearing to be detected, preprocessing and extracting characteristics of the real-time sensing data to obtain a real-time state characteristic index of the bearing to be detected, substituting the characteristic index into the established logistic regression model, and calculating to obtain the current health degree of the bearing to be detected;
and the judging module is used for comparing the calculated health degree of the bearing to be detected at the current moment with a preset health degree threshold, if the health degree of the bearing to be detected at the current moment is lower than the preset threshold, respectively calculating the real-time state characteristic index of the bearing to be detected, the J divergence of the fault position characteristic sample and the J divergence of the normal state characteristic sample by using a J divergence fault diagnosis method, judging the fault position of the bearing according to the J divergence of the fault position sample and the J divergence of the normal state sample, and realizing fault diagnosis of the bearing to be detected.
7. The rolling bearing fault pre-diagnosis system based on logistic regression and J divergence according to claim 6, characterized in that: the model-building module is arranged to set up,
assume a vector of N independent variables, XT=(x1,x2,…,xN),yi∈ {0,1}, y is 0 or 1, the mathematical expression of the logistic regression model is:
yi1 indicates an event occurred; y isi0 means that the event did not occur;
pi(yi=1/xi) Representing observed quantitiesThe occurrence probability value is between 0 and 1 relative to the occurrence probability of the ith event;
8. The rolling bearing fault pre-diagnosis system based on logistic regression and J divergence according to claim 6, characterized in that: the acquisition preprocessing module is set to be that the fault sensing data of different fault positions of the bearing during operation comprise one or more fault sensing data of inner ring faults, outer ring faults and rolling body faults of the bearing.
9. The rolling bearing fault pre-diagnosis system based on logistic regression and J divergence according to claim 8, characterized in that: the collection preprocessing module is set to be in a state that the fault position characteristic sample comprises one or more fault characteristic samples of an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample.
10. The rolling bearing fault pre-diagnosis system based on logistic regression and J divergence according to claim 9, characterized in that: when the characteristic samples of the fault position are an inner ring fault characteristic sample, an outer ring fault characteristic sample and a rolling body fault characteristic sample, the judging module is set to be,
by the formula J divergence formula:respectively calculating J divergence among the characteristic indexes of the real-time bearing state, the inner ring fault characteristic sample, the outer ring fault characteristic sample, the rolling body fault characteristic sample and the normal state characteristic sample to obtain four J divergence values, which are respectively: divergence value J1Divergence value J2Powder medicineValue J3Divergence value J4In the formula of J divergence, S is a characteristic value of a normal state signal; tau is the characteristic value of the unknown state signal; j (s, tau) is the J divergence between the two, N is the number of signal eigenvalues, and i is the sequence of signal eigenvalues;
contrast divergence value J1Divergence value J2Divergence value J3And divergence value J4And finding out the minimum value of the four J divergence values, wherein the smaller the divergence value is, the closer the current moment state of the bearing is to the state corresponding to the preset health threshold value, and determining the fault type of the bearing according to the magnitude of the divergence value.
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 CN108181105A (en) | 2018-06-19 |
CN108181105B true 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) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110672325A (en) * | 2019-09-02 | 2020-01-10 | 佛山科学技术学院 | Bearing working condition stability evaluation method and device based on probability distribution |
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 |
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 |
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 |
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 |
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 |
CN112614342B (en) * | 2020-12-10 | 2022-08-30 | 大唐高鸿智联科技(重庆)有限公司 | Early warning method for road abnormal event, vehicle-mounted equipment and road side equipment |
Family Cites Families (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 |
CN103679378B (en) * | 2013-12-20 | 2016-08-31 | 北京航天测控技术有限公司 | Method and device based on telemetry assessment heath state of spacecraft |
CN103868692B (en) * | 2014-03-18 | 2016-02-10 | 电子科技大学 | Based on the rotary machinery fault diagnosis method of Density Estimator 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 |
CN105241680B (en) * | 2015-08-26 | 2018-01-02 | 电子科技大学 | A kind of rotating machinery health state evaluation method based on probability density function |
CN105510023B (en) * | 2015-11-24 | 2019-10-11 | 国网内蒙古东部电力有限公司电力科学研究院 | Variable working condition wind power planetary gear box fault diagnosis method based on divergence index |
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
Also Published As
Publication number | Publication date |
---|---|
CN108181105A (en) | 2018-06-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108181105B (en) | Rolling bearing fault pre-diagnosis method and system based on logistic regression and J divergence | |
Li et al. | An improved exponential model for predicting remaining useful life of rolling element bearings | |
Shi et al. | Rolling bearing initial fault detection using long short-term memory recurrent network | |
CN107144430B (en) | A kind of Method for Bearing Fault Diagnosis based on incremental learning | |
CN110135492B (en) | Equipment fault diagnosis and abnormality detection method and system based on multiple Gaussian models | |
CN110823576B (en) | Mechanical anomaly detection method based on generation of countermeasure network | |
CN109397703B (en) | Fault detection method and device | |
Shakya et al. | Bearing diagnosis based on Mahalanobis–Taguchi–Gram–Schmidt method | |
CN105004523A (en) | Rolling bearing state monitoring method based on weighted similarity measure | |
CN108267312B (en) | A kind of subway train bearing intelligent diagnostic method based on fast search algorithm | |
CN105067262A (en) | Rolling bearing state monitoring method | |
CN106441896A (en) | Characteristic vector extraction method for rolling bearing fault mode identification and state monitoring | |
CN109141625B (en) | Online state monitoring method for ball screw pair | |
CN112729834B (en) | Bearing fault diagnosis method, device and system | |
CN108398265A (en) | A kind of online fault detection method of rolling bearing | |
CN110060368A (en) | Mechanical method for detecting abnormality based on potential feature coding | |
CN111678699B (en) | Early fault monitoring and diagnosing method and system for rolling bearing | |
CN112131797B (en) | Main shaft bearing life prediction and reliability assessment method based on stress analysis | |
CN115034137A (en) | RVM and degradation model-based two-stage hybrid prediction method for residual life of bearing | |
CN112711850A (en) | Unit online monitoring method based on big data | |
CN115046764B (en) | Early fault diagnosis method for rolling bearing | |
Lu et al. | Bearing health assessment based on chaotic characteristics | |
KR20210006832A (en) | Method and apparatus for machine fault diagnosis | |
CN112964470A (en) | Method for detecting early failure of rolling bearing based on bearing health index | |
CN108982106B (en) | Effective method for rapidly detecting kinetic mutation of complex system |
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 |