CN117131393A - Bearing fault diagnosis method and system for improving initial clustering center and weight - Google Patents
Bearing fault diagnosis method and system for improving initial clustering center and weight Download PDFInfo
- Publication number
- CN117131393A CN117131393A CN202311156961.7A CN202311156961A CN117131393A CN 117131393 A CN117131393 A CN 117131393A CN 202311156961 A CN202311156961 A CN 202311156961A CN 117131393 A CN117131393 A CN 117131393A
- Authority
- CN
- China
- Prior art keywords
- data
- determining
- bearing
- dissimilarity
- data point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000003745 diagnosis Methods 0.000 title claims abstract description 48
- 230000009193 crawling Effects 0.000 claims description 63
- 238000004140 cleaning Methods 0.000 claims description 28
- 238000002372 labelling Methods 0.000 claims description 27
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000012423 maintenance Methods 0.000 claims description 18
- 238000001514 detection method Methods 0.000 claims description 17
- 238000001228 spectrum Methods 0.000 claims description 14
- 238000012512 characterization method Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 238000010183 spectrum analysis Methods 0.000 claims description 4
- 230000009286 beneficial effect Effects 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 8
- 238000000605 extraction Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 239000006185 dispersion Substances 0.000 description 4
- 230000004927 fusion Effects 0.000 description 4
- 230000003595 spectral effect Effects 0.000 description 4
- 238000012216 screening Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000000452 restraining effect Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application provides a bearing fault diagnosis method and a system for improving an initial clustering center and weight, wherein the method comprises the following steps: step 1: calculating a data dissimilarity between first data points in the bearing dataset; step 2: determining a second data point based on the data dissimilarity; step 3: determining attribute weights of the first data points; step 4: according to the attribute weight, calculating a weighted Euclidean distance, and dividing the first data point into target clusters; step 5: determining the median of a first data point in a target cluster, and determining a succession clustering center; step 6: outputting a clustering result according to the initial clustering center and the successive clustering center; step 7: and determining a bearing fault diagnosis result according to the clustering result. According to the bearing fault diagnosis method and system for improving the initial clustering center and the weight, the standard deviation coefficient method is introduced to assign the weight, so that the reserved information is more sufficient, and the classification accuracy is higher; the median is used for replacing mean value iteration, the influence of outliers is eliminated, and diagnosis is more accurate.
Description
Technical Field
The application relates to the technical field of bearing detection, in particular to a bearing fault diagnosis method and system for improving an initial clustering center and weight.
Background
Currently, with the development of big data technology, it is increasingly important to mine valuable information from complex data. Unsupervised cluster analysis is one of the important technologies for data mining, and the objective of clustering is to maximize the similarity of data objects in the same cluster, and minimize the similarity of data objects in different clusters, which is widely used for mechanical failure diagnosis.
The conventional K-means cluster determines the K value, i.e., the number of final classification clusters, in advance for a given dataset. And meanwhile, the initial class center is randomly determined, and the points are classified into clusters closest to each other by calculating Euclidean distance. And then, the center of each cluster is recalculated, the classification operation is repeated until the cluster center is not changed, and finally, the classification is finished. Therefore, if the selected initial clustering center is similar to the actual center, the clustering effect is better, and the iteration number required for reaching the specified precision is smaller; conversely, the worse the effect, the lower the efficiency. Meanwhile, the data of all indexes are analyzed by a traditional K-means clustering method based on the degree of difference between indexes in combination with actual conditions, the clustering result is possibly unsatisfactory, and if some unimportant indexes are removed, information can be lost.
The application number is: the application patent of CN202211264745.X discloses a bearing fault diagnosis method based on a fuzzy width learning model, wherein the method comprises the following steps: constructing an initial fuzzy width learning model based on a width learning system and a fuzzy system; training the output fuzzy width learning model through training set data to obtain a target fuzzy width learning model, wherein the training set data comprises a plurality of bearing vibration signal data with fault type labels; and calculating a membership value of the vibration signal data of the bearing to be tested by using the target fuzzy width learning model, and judging the fault type of the bearing to be tested according to the membership value. Said application not only greatly shortens learning time, but also utilizes said model to judge fault type of bearing to be tested, and its robustness is strong, diagnosis speed is quick and fault diagnosis accuracy is also high.
However, in the above prior art, when judging the fault type of the bearing to be tested according to the membership value, it is necessary to determine a membership value interval corresponding to the fault type in advance, and when determining the membership value interval, historical data is required to be used, but the historical data is not reliable, and an outlier may exist in the membership value interval according to the historical data, in this case, the accuracy of diagnosis is low.
In view of the foregoing, there is a need for a bearing failure diagnosis method and system that improves initial cluster centers and weights to address at least the above-mentioned shortcomings.
Disclosure of Invention
The application aims to provide a bearing fault diagnosis method for improving an initial clustering center and weights, which is characterized by calculating data dissimilarity between first data points of an obtained bearing data set, determining attribute weights of the first data points according to a standard deviation coefficient method, calculating weighted Euclidean distances according to the attribute weights, determining the initial clustering center according to the weighted Euclidean distances and second data points, introducing weights of the standard deviation coefficient method, and keeping high accuracy rate of classification under the condition of fully keeping information; in addition, the median in the first data point is determined to replace the mean value for iteration, so that the influence of outliers is eliminated, and the accuracy of diagnosis is further improved.
The bearing fault diagnosis method for improving the initial clustering center and the weight provided by the embodiment of the application comprises the following steps:
step 1: acquiring a preset bearing data set, and calculating data dissimilarity among different first data points in the bearing data set;
step 2: determining a second data point serving as an initial clustering center according to the data dissimilarity;
step 3: determining attribute weights of the first data points according to a standard deviation coefficient method;
step 4: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point;
step 5: determining the median of the first data point in the target cluster, and determining a succession clustering center according to the median;
step 6: outputting a clustering result according to the initial clustering center and the successive clustering center;
step 7: and determining a bearing fault diagnosis result according to the clustering result.
Preferably, step 1: acquiring a preset bearing data set, and calculating data dissimilarity between different first data points in the bearing data set, wherein the method comprises the following steps of:
acquiring a dissimilarity matrix;
calculating the mean dissimilarity of the first data points and the overall dissimilarity of the bearing data sets according to the bearing data sets and the dissimilarity matrix;
the mean and overall dissimilarities were used together as data dissimilarities.
Preferably, step 2: determining a second data point as an initial cluster center based on the data dissimilarity, comprising:
step 2.1: analyzing the data dissimilarity to obtain the average dissimilarity;
step 2.2: acquiring a third data point with the maximum mean value dissimilarity, using the third data point as a first initial clustering sub-center, and updating by replacing the mean value dissimilarity of the third data point with 0 to acquire a fourth data point with the maximum mean value dissimilarity after updating;
step 2.3: judging whether the target dissimilarity between the third data point and the fourth data point is larger than a preset dissimilarity threshold, if so, taking the fourth data point as a second initial clustering sub-center, and replacing the average dissimilarity of the fourth data point with 0 for updating;
step 2.4: judging whether the center sum value of the first initial clustering sub-center and the second initial clustering sub-center is equal to the preset clustering number, if so, outputting the first initial clustering sub-center and the second initial clustering sub-center, and taking the first initial clustering sub-center and the second initial clustering sub-center as the initial clustering centers, otherwise, continuing to execute the steps 2.3 to 2.4 until the center sum value is equal to the clustering number.
Preferably, step 3: determining the attribute weight of the first data point according to a standard deviation coefficient method comprises the following steps:
acquiring the total number of the first data points, and simultaneously acquiring target data of the first data points of each attribute;
calculating an attribute mean value of target data of the same attribute of the first data point;
determining standard deviation of each attribute according to the total number of the first data points, the total number of the attributes of the first data points and the attribute mean value, wherein a calculation formula of the standard deviation is as follows:
wherein x is ik Is the kth attribute of the ith first data pointThe data of the object to be processed,is the attribute mean of the kth attribute, n is the total number of first data points, σ k Standard deviation for the kth attribute;
calculating attribute weights according to the standard deviation and the total number of the attributes of the first data points, wherein the attribute weights are as follows:
where m is the total number of attributes, ω, of the first data point k The attribute weight for the kth attribute.
Preferably, step 4: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point, wherein the method comprises the following steps:
and calculating a weighted Euclidean distance between each first data point and the initial clustering center according to the attribute weight, wherein the weighted Euclidean distance is expressed as follows:
wherein d ij Is x ik And x jk Weighted Euclidean distance between;
based on the principle of the nearest distance, determining the target cluster to which the first data point belongs according to the weighted Euclidean distance and the second data point.
Preferably, step 7: according to the clustering result, determining a bearing fault diagnosis result comprises the following steps:
constructing fault detection templates corresponding to different clustering results according to the clustering results;
acquiring the operation parameters of a target bearing needing to be subjected to fixed bearing fault diagnosis;
performing spectrum analysis on the operation parameters to obtain an observation spectrogram;
determining a target main frequency and a frequency position corresponding to the target main frequency in the observation spectrogram;
obtaining the frequency multiplication position of the target main frequency;
determining maintenance points on the target bearing according to the frequency position and the frequency multiplication position;
acquiring working conditions of a target bearing;
determining the maintenance type of the maintenance point location according to the target main frequency corresponding to the frequency position and the frequency doubling position based on the fault detection template and the working condition;
generating a model based on a preset overhaul strategy, generating an overhaul strategy according to the overhaul type, and carrying out corresponding overhaul according to the overhaul strategy.
Preferably, acquiring the frequency multiplication position of the target main frequency includes:
acquiring a plurality of bearing fault frequency spectrum labeling records;
training a frequency multiplication position marking model according to the bearing fault frequency spectrum marking record;
acquiring the bearing type of a target bearing;
determining a marked image according to the bearing type, the target main frequency, the observed spectrogram and the frequency multiplication position marked model;
and analyzing the marked image to obtain the frequency multiplication position.
Preferably, determining the labeling image according to the bearing type, the target main frequency, the observed spectrogram and the frequency multiplication position labeling model includes:
labeling the target main frequency on an observation spectrogram, and carrying out characterization based on a preset first characterization rule to obtain labeling characteristics;
inputting the labeling features into a frequency multiplication position labeling model to obtain a preprocessed image;
the bearing type is characterized based on a preset second characterization rule to obtain a cleaning characteristic;
obtaining a cleaning template according to the cleaning template generation rule and the cleaning characteristics;
and cleaning the preprocessed image by using a cleaning template to obtain the marked image.
The bearing fault diagnosis method for improving the initial clustering center and the weight provided by the embodiment of the application further comprises the following steps:
step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, and performing supplementary crawling if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed;
wherein, step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed, performing supplementary crawling, including:
acquiring a data type of a first data point;
according to the data type, determining a first classification cluster capable of carrying out data dissimilarity matching;
acquiring the data dimension of a first data point in each first classification cluster;
determining a second classification cluster capable of carrying out data dissimilarity matching according to the data dimension, wherein the second classification cluster is a sub-classification cluster of the first classification cluster;
determining first crawling data of all dimensions of a third classification cluster according to the first classification cluster;
determining second crawling data of the missing dimension of the first classification cluster according to the second classification cluster corresponding to the first classification cluster;
integrating the first crawling data and the second crawling data to obtain target crawling data;
and obtaining the crawling platform according to the classification information and the dimension information of the target crawling data, and sending a crawling request to the crawling platform to carry out supplementary crawling.
The bearing fault diagnosis system for improving the initial clustering center and the weight provided by the embodiment of the application comprises the following components:
the data dissimilarity calculating subsystem is used for acquiring a preset bearing data set and calculating data dissimilarity among different first data points in the bearing data set;
an initial cluster center determination subsystem for determining a second data point as an initial cluster center based on the data dissimilarity;
the attribute weight determining subsystem is used for determining the attribute weight of the first data point according to a standard deviation coefficient method;
the clustering subsystem is used for calculating a weighted Euclidean distance according to the attribute weight and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point;
the continuing cluster center determining subsystem is used for determining the median of the first data points in the target cluster and determining a continuing cluster center according to the median;
the clustering result output subsystem is used for outputting a clustering result according to the initial clustering center and the successive clustering center;
and the diagnosis result determining subsystem is used for determining a bearing fault diagnosis result according to the clustering result.
The beneficial effects of the application are as follows:
the method comprises the steps of calculating data dissimilarity between first data points of an obtained bearing data set, determining attribute weight of the first data points according to a standard deviation coefficient method, calculating weighted Euclidean distance according to the attribute weight, determining an initial clustering center according to the weighted Euclidean distance and a second data point, introducing a weight value assigned by the standard deviation coefficient method, and keeping high accuracy rate of classification under the condition of fully keeping information; in addition, the median in the first data point is determined to replace the mean value for iteration, so that the influence of outliers is eliminated, and the accuracy of diagnosis is further improved.
Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application. The objectives and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the application is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, serve to explain the application. In the drawings:
FIG. 1 is a schematic diagram of a bearing fault diagnosis method for improving initial cluster centers and weights according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a bearing fault diagnosis system with improved initial cluster centers and weights in accordance with an embodiment of the present application.
Detailed Description
The preferred embodiments of the present application will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present application only, and are not intended to limit the present application.
The embodiment of the application provides a bearing fault diagnosis method for improving an initial clustering center and weight, which is shown in fig. 1 and comprises the following steps:
step 1: acquiring a preset bearing data set, and calculating data dissimilarity among different first data points in the bearing data set; wherein the bearing dataset is: bearing data set of university of kesixi, usa; wherein the first data point is: individual data in the bearing dataset; the data dissimilarity is: differences or diversity between different first data points;
step 2: determining a second data point serving as an initial clustering center according to the data dissimilarity; wherein the second data point is: a first data point with the greatest data dissimilarity;
step 3: determining attribute weights of the first data points according to a standard deviation coefficient method; wherein, the attribute weight is: weights of the indexes determined according to a standard deviation coefficient method;
step 4: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point; wherein the weighted euclidean distance is: the distance between the first data point and the initial cluster center; the target cluster is as follows: a set of first data points partitioned together;
step 5: determining the median of the first data point in the target cluster, and determining a succession clustering center according to the median;
step 6: and outputting a clustering result according to the initial clustering center and the successive clustering center. When a clustering result is output according to the initial clustering center and the successive clustering center, judging whether the initial clustering center is consistent with the successive clustering center or not; if yes, outputting a clustering result, otherwise, replacing the initial clustering center in the step 4 by a continuous clustering center, and outputting the clustering result;
step 7: and determining a bearing fault diagnosis result according to the clustering result. The bearing fault diagnosis result is an integration result of fault types corresponding to the cluster data, and the fault types comprise: inner ring damage, outer ring damage, ball damage and normal.
The working principle and the beneficial effects of the technical scheme are as follows:
the method comprises the steps of calculating data dissimilarity between first data points of a bearing data set, determining attribute weight of the first data points according to a standard deviation coefficient method, calculating weighted Euclidean distance according to the attribute weight, determining an initial clustering center according to the weighted Euclidean distance and second data points, introducing a weight value of the standard deviation coefficient method, and under the condition of fully keeping information, keeping high accuracy of classification; in addition, the median in the first data point is determined to replace the mean value for iteration, so that the influence of outliers is eliminated, and the accuracy of diagnosis is further improved.
When the system is specifically applied, the system acquires the monitoring signal of the bearing which needs to be subjected to bearing fault detection, and then outputs a fault diagnosis result.
In one embodiment, step 1: acquiring a preset bearing data set, and calculating data dissimilarity between different first data points in the bearing data set, wherein the method comprises the following steps of:
acquiring a dissimilarity matrix; wherein the dissimilarity matrix stores the proximity between the plurality of data;
calculating the mean dissimilarity of the first data points and the overall dissimilarity of the bearing data sets according to the bearing data sets and the dissimilarity matrix;
the proximity degree is specifically:
the average value dissimilarity is specifically:
the general dissimilarity is specifically:
wherein dis (x) i ,x j ) Is x i And x j Degree of dissimilarity between, x i For the ith first data point, x j For the j-th first data point, x ik Target data, x, for the kth attribute of the ith first data point jk Target data for the kth attribute of the jth first data point, m being the total number of attributes of the first data point; adis (x) i ) Is x i Tdis is the overall dissimilarity, dis (x i -x j ) Is x i And x j The distance between n is the total number of first data points in the bearing data set;
the mean and overall dissimilarities were used together as data dissimilarities.
The working principle and the beneficial effects of the technical scheme are as follows:
according to the method, the dissimilarity matrix is introduced, the average dissimilarity and the overall dissimilarity are determined according to the bearing data set and the dissimilarity matrix, and the average dissimilarity and the overall dissimilarity are used together as the data dissimilarity, so that the comprehensiveness of dissimilarity judgment is improved.
In one embodiment, step 2: determining a second data point as an initial cluster center based on the data dissimilarity, comprising:
step 2.1: analyzing the data dissimilarity to obtain the average dissimilarity;
step 2.2: acquiring a third data point with the maximum mean value dissimilarity, using the third data point as a first initial clustering sub-center, and updating by replacing the mean value dissimilarity of the third data point with 0 to acquire a fourth data point with the maximum mean value dissimilarity after updating;
step 2.3: judging whether the target dissimilarity between the third data point and the fourth data point is larger than a preset dissimilarity threshold, if so, taking the fourth data point as a second initial clustering sub-center, and replacing the average dissimilarity of the fourth data point with 0 for updating; wherein the preset dissimilarity threshold is preset manually;
step 2.4: judging whether the center sum value of the first initial clustering sub-center and the second initial clustering sub-center is equal to the preset clustering number, if so, outputting the first initial clustering sub-center and the second initial clustering sub-center, and taking the first initial clustering sub-center and the second initial clustering sub-center as the initial clustering centers, otherwise, continuing to execute the steps 2.3 to 2.4 until the center sum value is equal to the clustering number. Wherein, the center sum value is: a summation result of the number of first initial cluster sub-centers and the number of second initial cluster sub-centers; the preset cluster number is input in advance manually, for example: 4.
the working principle and the beneficial effects of the technical scheme are as follows:
the method introduces the mean value dissimilarity, determines the third data point with the maximum mean value dissimilarity, updates the third data point, and acquires the initial cluster center with the maximum mean value dissimilarity obtained by updating, so that the acquisition process of the initial cluster center is more reasonable.
In one embodiment, step 3: determining the attribute weight of the first data point according to a standard deviation coefficient method comprises the following steps:
acquiring the total number of the first data points, and simultaneously acquiring target data of the first data points of each attribute;
calculating an attribute mean value of target data of the same attribute of the first data point;
determining standard deviation of each attribute according to the total number of the first data points, the total number of the attributes of the first data points and the attribute mean value, wherein a calculation formula of the standard deviation is as follows:
wherein x is ik Is the target data of the kth attribute of the ith first data point,is the attribute mean of the kth attribute, n is the total number of first data points, σ k Standard deviation for the kth attribute;
calculating attribute weights according to the standard deviation and the total number of the attributes of the first data points, wherein the attribute weights are as follows:
where m is the total number of attributes, ω, of the first data point k Attribute weight for the first data point that is the kth attribute.
The working principle and the beneficial effects of the technical scheme are as follows:
the attribute of the application is weighted by a standard deviation coefficient method, and the classification can also keep higher accuracy under the condition of fully preserving information.
In one embodiment, step 4: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point, wherein the method comprises the following steps:
and calculating a weighted Euclidean distance between each first data point and the initial clustering center according to the attribute weight, wherein the weighted Euclidean distance is expressed as follows:
wherein d ij Is x ik And x jk Weighted Euclidean distance between;
based on the principle of the nearest distance, determining the target cluster to which the first data point belongs according to the weighted Euclidean distance and the second data point.
The working principle and the beneficial effects of the technical scheme are as follows:
according to the method, a weighted Euclidean distance and nearest distance principle is introduced, the target cluster to which the first data point belongs is determined, and the classification accuracy and rationality are improved.
In one embodiment, step 7: according to the clustering result, determining a bearing fault diagnosis result comprises the following steps:
constructing fault detection templates corresponding to different clustering results according to the clustering results; the fault detection templates corresponding to different clustering results are as follows: fault detection templates corresponding to different fault types, such as: and restraining a fault detection template for detecting damage of the bearing inner ring only, for example: a fault detection template for restraining only ball damage detection;
acquiring the operation parameters of a target bearing needing to be subjected to fixed bearing fault diagnosis; wherein, the operation parameters are: vibration signals of bearing operation;
performing spectrum analysis on the operation parameters to obtain an observation spectrogram; wherein, the spectrum analysis is: a fast fourier transform; the observed spectrogram is: a spectrogram of an operating parameter;
determining a target main frequency and a frequency position corresponding to the target main frequency in the observation spectrogram;
obtaining the frequency multiplication position of the target main frequency; wherein, the frequency multiplication position is: observing the position of the peak value in the spectrogram, which is an integral multiple of the target dominant frequency;
determining maintenance points on the target bearing according to the frequency position and the frequency multiplication position; wherein, the servicing point is: a location on the target bearing where a fault may exist;
acquiring working conditions of a target bearing; wherein, operating conditions include: bearing type and operating conditions, etc.;
determining the maintenance type of the maintenance point location according to the target main frequency corresponding to the frequency position and the frequency doubling position based on the fault detection template and the working condition; when the maintenance type of the maintenance point location is determined, the target main frequency corresponding to the frequency position and the frequency multiplication position can be compared with the main frequency corresponding to the working condition listed in the fault detection template, and if the frequency match is met, the fault type corresponding to the fault detection template is judged to be possible to exist;
generating a model based on a preset overhaul strategy, generating an overhaul strategy according to the overhaul type, and carrying out corresponding overhaul according to the overhaul strategy. Wherein, the maintenance strategy is: what kind of maintenance tools need to be carried and how to maintain;
according to the frequency position and the frequency multiplication position, determining the maintenance point position on the target bearing comprises the following steps:
acquiring observation frequency spectrum characteristics according to the frequency position and the frequency multiplication position; wherein, the observed spectrum features are: spectral coefficients of the observed spectrogram of the frequency position and the frequency multiplication position, such as: fourier transform coefficients;
performing critical sorting on the processing results of the observed spectrum features processed by using the BN algorithm based on a preset critical sorting algorithm, and selecting the processing results based on sorting screening rules to obtain selected features; the key ordering algorithm is as follows: a random forest algorithm; the BN algorithm is a normalization algorithm; the sorting and screening rule is as follows: screening out the previous processing results;
determining feature dispersion in feature distribution according to the sorting order of the selected features in the processing result; the higher the sorting order, the lower the feature dispersion, and the denser the selection of the selected features;
determining feature extraction frequency according to the feature dispersion and the selected feature distribution; the feature extraction frequency is the extraction position of the complementary extraction in the observation spectrogram;
determining supplementary features according to the feature extraction frequency; wherein, supplementary characteristic is: observing spectral coefficients in the spectrogram corresponding to the extraction positions;
determining fusion characteristic distribution according to the observed spectrum characteristics and the supplementary characteristics; wherein, the fusion characteristic distribution is: a spectral feature distribution including frequency location information and a fault location of the spectral coefficients;
and determining maintenance points according to the fusion characteristic distribution.
The working principle and the beneficial effects of the technical scheme are as follows:
in general, a main frequency of a vibration signal can be obtained based on an operation parameter of a target bearing, the main frequency vibration signal corresponds to a fault type and a cause, and the main frequency can provide some clues, but the fault type and the cause cannot be directly determined. Determining the maintenance point position according to the frequency position and the frequency multiplication position, determining the observation frequency spectrum characteristic according to the frequency position and the frequency multiplication position, introducing a critical sorting algorithm and a BN algorithm to determine a processing result sequence of normalization processing, determining the characteristic dispersion in the characteristic distribution according to the criticality of the processing result, determining the characteristic extraction frequency of the supplementary characteristic to supplement and extract, and determining the maintenance point position according to the fusion characteristic distribution after supplementary sampling, thereby further improving the accuracy of the maintenance point position determination process. The working condition and the fault detection template are introduced to determine the overhaul type of the overhaul point location, so that the fault detection efficiency is higher and more accurate. And an overhaul strategy generation model is introduced to generate an overhaul strategy corresponding to the overhaul type, so that the overhaul suitability is improved.
In one embodiment, obtaining the frequency multiplication location of the target dominant frequency includes:
acquiring a plurality of bearing fault frequency spectrum labeling records; the bearing fault frequency spectrum marking record is as follows: manually analyzing the record of the frequency multiplication position corresponding to the frequency spectrum marking main frequency of the bearing fault signal;
training a frequency multiplication position marking model according to the bearing fault frequency spectrum marking record; the frequency multiplication position labeling model is an intelligent model for rapidly analyzing the frequency multiplication position of the main frequency according to the known main frequency;
acquiring the bearing type of a target bearing; wherein, the bearing type is: the type of bearing;
determining a marked image according to the bearing type, the target main frequency, the observed spectrogram and the frequency multiplication position marked model; the marked image is as follows: an observation spectrogram of the frequency multiplication position marked by the frequency multiplication position marking model;
and analyzing the marked image to obtain the frequency multiplication position.
The working principle and the beneficial effects of the technical scheme are as follows:
according to the method, a bearing fault frequency spectrum marking record training frequency multiplication position marking model is introduced, in addition, a bearing type condition variable is introduced, a marking image is determined according to the bearing type, the target main frequency, the observation frequency spectrogram and the frequency multiplication position marking model, the marking image is determined according to the bearing type, the target main frequency, the observation frequency spectrogram and the frequency multiplication position marking model by analyzing the marking image, and the frequency multiplication position determining process is more intelligent and has higher efficiency.
In one embodiment, determining the annotation image based on the bearing type, the target dominant frequency, the observed spectrogram, and the frequency doubling position annotation model comprises:
labeling the target main frequency on an observation spectrogram, and carrying out characterization based on a preset first characterization rule to obtain labeling characteristics; the preset first characterization rule is as follows: rules for characterizing images; the labeling features are as follows: observing where the target main frequency is marked on the spectrogram, and the number of the target main frequency is;
inputting the labeling features into a frequency multiplication position labeling model to obtain a preprocessed image; wherein, the preprocessing image is: the frequency multiplication position labeling model outputs images of frequency multiplication positions based on labeling features;
the bearing type is characterized based on a preset second characterization rule to obtain a cleaning characteristic; wherein the second characterization rule is: rules for semantically characterizing the semantics of the bearing type; the cleaning characteristic is as follows: what kind of semantics the bearing type corresponds to;
obtaining a cleaning template according to the cleaning template generation rule and the cleaning characteristics; the cleaning template generation rule is as follows: what cleaning templates are generated based on what cleaning features, such as: the semantics of which bearing type is constrained to generate a template for cleaning which frequency doubling data;
and cleaning the preprocessed image by using a cleaning template to obtain the marked image.
The working principle and the beneficial effects of the technical scheme are as follows:
according to the method, the labeling features are obtained, the frequency multiplication position labeling model is input, the labeling image is determined, meanwhile, the cleaning features and the cleaning template generation rules are introduced to determine the cleaning template, the cleaning template is used for cleaning the pretreatment image, the labeling image is obtained, and the labeling accuracy of the labeling image is further improved.
The embodiment of the application provides a bearing fault diagnosis method for improving an initial clustering center and weight, which further comprises the following steps:
step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, and performing supplementary crawling if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed;
wherein, step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed, performing supplementary crawling, including:
acquiring a data type of a first data point;
according to the data type, determining a first classification cluster capable of carrying out data dissimilarity matching; the first classification cluster capable of carrying out data dissimilarity matching is as follows: a set of first data point combinations of the same data type;
acquiring the data dimension of a first data point in each first classification cluster;
determining a second classification cluster capable of carrying out data dissimilarity matching according to the data dimension, wherein the second classification cluster is a sub-classification cluster of the first classification cluster; the second classification cluster capable of carrying out data dissimilarity matching is as follows: a set of first data points of the same data dimension in the first classification cluster;
determining first crawling data of all dimensions of a third classification cluster according to the first classification cluster; wherein, the first crawling data is: crawling data of all dimensions of what data type;
determining second crawling data of the missing dimension of the first classification cluster according to the second classification cluster corresponding to the first classification cluster; wherein, the second crawling data is: crawling data of a certain dimension of what data type;
integrating the first crawling data and the second crawling data to obtain target crawling data;
and obtaining the crawling platform according to the classification information and the dimension information of the target crawling data, and sending a crawling request to the crawling platform to carry out supplementary crawling. Wherein, the classification information is: information about what data type; the dimension information is: information about what data dimension; the crawling platform is as follows: large data platforms sharing bearing data.
The working principle and the beneficial effects of the technical scheme are as follows:
when data dissimilarity analysis is carried out according to a bearing data set, the analyzed data may be in an incomplete condition so as to influence subsequent clustering, so that the application introduces data types and data dimensions to carry out necessary analysis of first data point supplementary crawling, and when the analysis is carried out, all dimensional information of missing data types and missing dimensional information in a first classification cluster are determined, and a crawling platform is determined to carry out corresponding crawling according to the classification information and the dimensional information, thereby improving the pertinence of data crawling and also avoiding the problem of inaccurate subsequent diagnosis caused by incomplete comparison data of data dissimilarity.
The embodiment of the application provides a bearing fault diagnosis system for improving an initial clustering center and weight, which is shown in fig. 2 and comprises the following steps:
the data dissimilarity calculating subsystem 1 is used for acquiring a preset bearing data set and calculating data dissimilarity among different first data points in the bearing data set;
an initial cluster center determination subsystem 2 for determining a second data point as an initial cluster center based on the data dissimilarity;
the attribute weight determining subsystem 3 is used for determining the attribute weight of the first data point according to a standard deviation coefficient method;
the clustering subsystem 4 is used for calculating a weighted Euclidean distance according to the attribute weight and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point;
a continuing cluster center determining subsystem 5, configured to determine a median of the first data points in the target cluster, and determine a continuing cluster center according to the median;
the clustering result output subsystem 6 is used for outputting a clustering result according to the initial clustering center and the successive clustering center;
and the diagnosis result determination subsystem 7 is used for determining a bearing fault diagnosis result according to the clustering result.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. The bearing fault diagnosis method for improving the initial clustering center and the weight is characterized by comprising the following steps of:
step 1: acquiring a preset bearing data set, and calculating data dissimilarity among different first data points in the bearing data set;
step 2: determining a second data point serving as an initial clustering center according to the data dissimilarity;
step 3: determining attribute weights of the first data points according to a standard deviation coefficient method;
step 4: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point;
step 5: determining the median of the first data point in the target cluster, and determining a succession clustering center according to the median;
step 6: outputting a clustering result according to the initial clustering center and the successive clustering center;
step 7: and determining a bearing fault diagnosis result according to the clustering result.
2. The method for diagnosing bearing faults with improved initial cluster centers and weights as claimed in claim 1, wherein the method comprises the following steps: acquiring a preset bearing data set, and calculating data dissimilarity between different first data points in the bearing data set, wherein the method comprises the following steps of:
acquiring a dissimilarity matrix;
calculating the mean dissimilarity of the first data points and the overall dissimilarity of the bearing data sets according to the bearing data sets and the dissimilarity matrix;
the mean and overall dissimilarities were used together as data dissimilarities.
3. The method for diagnosing bearing faults with improved initial cluster centers and weights as claimed in claim 1, wherein the step 2 is as follows: determining a second data point as an initial cluster center based on the data dissimilarity, comprising:
step 2.1: analyzing the data dissimilarity to obtain the average dissimilarity;
step 2.2: acquiring a third data point with the maximum mean value dissimilarity, using the third data point as a first initial clustering sub-center, and updating by replacing the mean value dissimilarity of the third data point with 0 to acquire a fourth data point with the maximum mean value dissimilarity after updating;
step 2.3: judging whether the target dissimilarity between the third data point and the fourth data point is larger than a preset dissimilarity threshold, if so, taking the fourth data point as a second initial clustering sub-center, and replacing the average dissimilarity of the fourth data point with 0 for updating;
step 2.4: judging whether the center sum value of the first initial clustering sub-center and the second initial clustering sub-center is equal to the preset clustering number, if so, outputting the first initial clustering sub-center and the second initial clustering sub-center, and taking the first initial clustering sub-center and the second initial clustering sub-center as the initial clustering centers, otherwise, continuing to execute the steps 2.3 to 2.4 until the center sum value is equal to the clustering number.
4. The method for diagnosing bearing faults with improved initial cluster centers and weights as claimed in claim 1, wherein the step 3: determining the attribute weight of the first data point according to a standard deviation coefficient method comprises the following steps:
acquiring the total number of the first data points, and simultaneously acquiring target data of the first data points of each attribute;
calculating an attribute mean value of target data of the same attribute of the first data point;
determining standard deviation of each attribute according to the total number of the first data points, the total number of the attributes of the first data points and the attribute mean value, wherein a calculation formula of the standard deviation is as follows:
wherein x is ik Is the target data of the kth attribute of the ith first data point,is the attribute mean of the kth attribute, n is the total number of first data points, σ k Standard deviation for the kth attribute;
calculating attribute weights according to the standard deviation and the total number of the attributes of the first data points, wherein the attribute weights are as follows:
where m is the total number of attributes, ω, of the first data point k The attribute weight for the kth attribute.
5. The method for diagnosing bearing faults with improved initial cluster centers and weights as claimed in claim 1, wherein the step 4 is as follows: calculating a weighted Euclidean distance according to the attribute weight, and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point, wherein the method comprises the following steps:
and calculating a weighted Euclidean distance between each first data point and the initial clustering center according to the attribute weight, wherein the weighted Euclidean distance is expressed as follows:
wherein d ij Is x ik And x jk Weighted Euclidean distance between;
based on the principle of the nearest distance, determining the target cluster to which the first data point belongs according to the weighted Euclidean distance and the second data point.
6. The method for diagnosing bearing faults with improved initial cluster centers and weights as claimed in claim 1, wherein the step 7: according to the clustering result, determining a bearing fault diagnosis result comprises the following steps:
constructing fault detection templates corresponding to different clustering results according to the clustering results;
acquiring the operation parameters of a target bearing needing to be subjected to fixed bearing fault diagnosis;
performing spectrum analysis on the operation parameters to obtain an observation spectrogram;
determining a target main frequency and a frequency position corresponding to the target main frequency in the observation spectrogram;
obtaining the frequency multiplication position of the target main frequency;
determining maintenance points on the target bearing according to the frequency position and the frequency multiplication position;
acquiring working conditions of a target bearing;
determining the maintenance type of the maintenance point location according to the target main frequency corresponding to the frequency position and the frequency doubling position based on the fault detection template and the working condition;
generating a model based on a preset overhaul strategy, generating an overhaul strategy according to the overhaul type, and carrying out corresponding overhaul according to the overhaul strategy.
7. The method for bearing failure diagnosis for improving initial cluster centers and weights according to claim 6, wherein obtaining the frequency multiplication position of the target main frequency comprises:
acquiring a plurality of bearing fault frequency spectrum labeling records;
training a frequency multiplication position marking model according to the bearing fault frequency spectrum marking record;
acquiring the bearing type of a target bearing;
determining a marked image according to the bearing type, the target main frequency, the observed spectrogram and the frequency multiplication position marked model;
and analyzing the marked image to obtain the frequency multiplication position.
8. The method for bearing fault diagnosis for improved initial cluster centers and weights as claimed in claim 7, wherein determining the annotation image based on the bearing type, the target dominant frequency, the observed spectrogram and the frequency doubling position annotation model comprises:
labeling the target main frequency on an observation spectrogram, and carrying out characterization based on a preset first characterization rule to obtain labeling characteristics;
inputting the labeling features into a frequency multiplication position labeling model to obtain a preprocessed image;
the bearing type is characterized based on a preset second characterization rule to obtain a cleaning characteristic;
obtaining a cleaning template according to the cleaning template generation rule and the cleaning characteristics;
and cleaning the preprocessed image by using a cleaning template to obtain the marked image.
9. The bearing fault diagnosis method for improving initial cluster centers and weights as set forth in claim 1, further comprising:
step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, and performing supplementary crawling if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed;
wherein, step 8: performing supplementary crawling necessity analysis on the first data point for data dissimilarity calculation, if the analysis result of the supplementary crawling necessity analysis is that supplementary crawling is needed, performing supplementary crawling, including:
acquiring a data type of a first data point;
according to the data type, determining a first classification cluster capable of carrying out data dissimilarity matching;
acquiring the data dimension of a first data point in each first classification cluster;
determining a second classification cluster capable of carrying out data dissimilarity matching according to the data dimension, wherein the second classification cluster is a sub-classification cluster of the first classification cluster;
determining first crawling data of all dimensions of a third classification cluster according to the first classification cluster;
determining second crawling data of the missing dimension of the first classification cluster according to the second classification cluster corresponding to the first classification cluster;
integrating the first crawling data and the second crawling data to obtain target crawling data;
and obtaining the crawling platform according to the classification information and the dimension information of the target crawling data, and sending a crawling request to the crawling platform to carry out supplementary crawling.
10. A bearing fault diagnosis system for improving initial cluster centers and weights, comprising:
the data dissimilarity calculating subsystem is used for acquiring a preset bearing data set and calculating data dissimilarity among different first data points in the bearing data set;
an initial cluster center determination subsystem for determining a second data point as an initial cluster center based on the data dissimilarity;
the attribute weight determining subsystem is used for determining the attribute weight of the first data point according to a standard deviation coefficient method;
the clustering subsystem is used for calculating a weighted Euclidean distance according to the attribute weight and dividing the first data point into corresponding target clusters according to the weighted Euclidean distance and the second data point;
the continuing cluster center determining subsystem is used for determining the median of the first data points in the target cluster and determining a continuing cluster center according to the median;
the clustering result output subsystem is used for outputting a clustering result according to the initial clustering center and the successive clustering center;
and the diagnosis result determining subsystem is used for determining a bearing fault diagnosis result according to the clustering result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311156961.7A CN117131393A (en) | 2023-09-07 | 2023-09-07 | Bearing fault diagnosis method and system for improving initial clustering center and weight |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311156961.7A CN117131393A (en) | 2023-09-07 | 2023-09-07 | Bearing fault diagnosis method and system for improving initial clustering center and weight |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117131393A true CN117131393A (en) | 2023-11-28 |
Family
ID=88852645
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311156961.7A Pending CN117131393A (en) | 2023-09-07 | 2023-09-07 | Bearing fault diagnosis method and system for improving initial clustering center and weight |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117131393A (en) |
-
2023
- 2023-09-07 CN CN202311156961.7A patent/CN117131393A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Han et al. | An adaptive spatiotemporal feature learning approach for fault diagnosis in complex systems | |
| CN110888849B (en) | Online log analysis method and system and electronic terminal equipment thereof | |
| CN109781411B (en) | Bearing fault diagnosis method combining improved sparse filter and KELM | |
| CN110110768B (en) | Rolling bearing fault diagnosis method based on parallel feature learning and multiple classifiers | |
| CN101907681B (en) | Analog circuit dynamic online failure diagnosing method based on GSD-SVDD | |
| CN111562108A (en) | Rolling bearing intelligent fault diagnosis method based on CNN and FCMC | |
| CN112819059B (en) | Rolling bearing fault diagnosis method based on popular retention transfer learning | |
| CN112756759B (en) | Spot welding robot workstation fault judgment method | |
| CN113008559B (en) | Bearing fault diagnosis method and system based on sparse self-encoder and Softmax | |
| CN110633371A (en) | Log classification method and system | |
| CN108333468B (en) | The recognition methods of bad data and device under a kind of active power distribution network | |
| CN113609569B (en) | Distinguishing type generalized zero sample learning fault diagnosis method | |
| CN114564982A (en) | Automatic identification method for radar signal modulation type | |
| CN112785015B (en) | Equipment fault diagnosis method based on case reasoning | |
| CN114091504B (en) | Rotary machine small sample fault diagnosis method based on generation countermeasure network | |
| CN109034076A (en) | A kind of automatic clustering method and automatic cluster system of mechanical fault signals | |
| CN115343676B (en) | Feature optimization method for positioning technology of redundant substances in sealed electronic equipment | |
| CN112763215B (en) | Multi-working-condition online fault diagnosis method based on modular federal deep learning | |
| CN111860447A (en) | Method for accurately identifying telemetry time sequence data mode of random phase shift satellite | |
| CN111458145A (en) | Cable car rolling bearing fault diagnosis method based on road map characteristics | |
| CN112860183A (en) | Multisource distillation-migration mechanical fault intelligent diagnosis method based on high-order moment matching | |
| CN115859142A (en) | Small sample rolling bearing fault diagnosis method based on convolution transformer generation countermeasure network | |
| CN112685374B (en) | Log classification method and device and electronic equipment | |
| CN118427681A (en) | Cross-working condition open-set fault diagnosis method and equipment based on self-supervision contrast learning enhancement | |
| CN108805181A (en) | A kind of image classification device and sorting technique based on more disaggregated models |
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 |