CN116977708B - Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view - Google Patents
Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view Download PDFInfo
- Publication number
- CN116977708B CN116977708B CN202310700517.0A CN202310700517A CN116977708B CN 116977708 B CN116977708 B CN 116977708B CN 202310700517 A CN202310700517 A CN 202310700517A CN 116977708 B CN116977708 B CN 116977708B
- Authority
- CN
- China
- Prior art keywords
- bearing
- data
- graph
- sub
- adaptive
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 60
- 238000000034 method Methods 0.000 title claims abstract description 53
- 230000000007 visual effect Effects 0.000 title claims abstract description 42
- 238000004220 aggregation Methods 0.000 title claims abstract description 34
- 230000002776 aggregation Effects 0.000 title claims abstract description 34
- 238000005096 rolling process Methods 0.000 claims abstract description 33
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 31
- 238000013507 mapping Methods 0.000 claims abstract description 27
- 238000013528 artificial neural network Methods 0.000 claims abstract description 22
- 238000010586 diagram Methods 0.000 claims abstract description 15
- 230000001133 acceleration Effects 0.000 claims abstract description 14
- 238000007635 classification algorithm Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims abstract description 4
- 239000013598 vector Substances 0.000 claims description 31
- 238000012545 processing Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 18
- 230000003044 adaptive effect Effects 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 238000012423 maintenance Methods 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 238000012800 visualization Methods 0.000 claims description 5
- 238000009826 distribution Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 3
- 238000002405 diagnostic procedure Methods 0.000 claims description 2
- 239000000203 mixture Substances 0.000 abstract description 6
- 230000014759 maintenance of location Effects 0.000 abstract 1
- 108091006146 Channels Proteins 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000000059 patterning Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000007788 roughening Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Acoustics & Sound (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to the field of mechanical fault diagnosis and discloses an intelligent bearing diagnosis method based on a self-adaptive aggregation visual view, which comprises the steps of developing a bearing typical fault vibration test and collecting vibration acceleration signals under a typical fault state; performing feature extraction and graph mapping on the acquired time sequence signals by adopting a self-adaptive aggregated visual algorithm to obtain graph data which is input into a graph neural network and consists of nodes and edges; establishing an end-to-end rolling bearing intelligent diagnosis framework based on the improved DiffPool diagram classification algorithm; inputting the graph data into a diagnosis framework, and optimizing an intelligent diagnosis model of the rolling bearing; and identifying and classifying different working conditions of the rolling bearing according to the optimized intelligent diagnosis frame of the rolling bearing. The method is based on a graph neural network, and adopts a self-adaptive algorithm, so that the problems that the traditional machine composition mode is inflexible, the mapping efficiency is low, and the original data retention characteristic cannot be maximized are solved.
Description
Technical Field
The invention relates to the field of mechanical fault diagnosis, in particular to an intelligent bearing diagnosis method based on self-adaptive aggregation visual images.
Background
Rolling bearings are one of the key components commonly used in mechanical devices, and their reliability and life have a significant impact on the performance and safety of the mechanical device. With the development of industrial automation and intellectualization, the intelligent diagnosis technology of the rolling bearing is paid attention to. The intelligent diagnosis technology can discover bearing faults in time, perform early warning and maintenance, effectively reduce the fault rate and the downtime of equipment, and improve the reliability and the safety of the equipment.
At present, the intelligent diagnosis technology of the rolling bearing mainly adopts methods such as signal analysis, pattern recognition, machine learning and the like. Among them, the machine learning method plays an important role in the failure diagnosis of the rolling bearing. The traditional machine learning method, such as a support vector machine, a decision tree, a neural network and the like, has been widely applied to the fault diagnosis of the rolling bearing, and has a good effect. However, these methods suffer from drawbacks such as the need to manually extract features, the inability to efficiently process large-scale and complex data, etc.
To solve these problems, a graph neural network (Graph Neural Network, GNN) has been developed. The graph neural network is a deep learning algorithm based on graph structure data, which can process and analyze unstructured data. Unlike conventional neural networks, the graph neural network can process data with complex topologies, such as vibration signals of rolling bearings, and the like. The graph neural network has achieved remarkable results in the fields of recommendation systems, social network analysis, bioinformatics, computer vision and the like. However, in the field of intelligent diagnosis of rolling bearings at present, algorithms based on graph neural networks have not been widely used. Therefore, a more efficient, flexible and accurate intelligent diagnosis algorithm for the rolling bearing is needed to meet the requirements of modern industrial production on the intelligent diagnosis technology for the rolling bearing.
The conversion of the acquired signals into graph data identified by the graph neural network is an important step. The invention provides a self-adaptive composition mode, local and global characteristics of original information can be extracted through control of a convolution kernel size parameter m, meanwhile, the composition mode is flexible, the mapping efficiency is high, and characteristics of original data can be reserved maximally. Compared with the existing classification algorithm, the AcvgNet algorithm provided by the method has higher classification precision, can effectively solve the problems of the existing algorithm, and achieves good effect in practical application.
Disclosure of Invention
This section is intended to outline some aspects of embodiments of the invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section as well as in the description summary and in the title of the application, to avoid obscuring the purpose of this section, the description summary and the title of the invention, which should not be used to limit the scope of the invention.
The present invention has been made in view of the above-described problems occurring in the prior art.
Therefore, the invention provides an intelligent bearing diagnosis method based on self-adaptive aggregated visual views, which can solve the problems that the traditional machine learning method needs to manually extract features and cannot effectively process large-scale and complex data.
In order to solve the technical problems, the invention provides the following technical scheme, namely an intelligent bearing diagnosis method based on self-adaptive aggregation visual view, which comprises the following steps:
carrying out a bearing typical fault vibration test and collecting vibration acceleration signals under a typical fault state;
performing feature extraction and graph mapping on the acquired time sequence signals by adopting a self-adaptive aggregated visual algorithm to obtain graph data which is input into a graph neural network and consists of nodes and edges;
establishing an end-to-end rolling bearing intelligent diagnosis framework based on an improved DiffPool diagram classification algorithm; inputting the graph data into a diagnosis framework, and optimizing an intelligent diagnosis model of the rolling bearing;
and identifying and classifying different working conditions of the rolling bearing according to the optimized intelligent diagnosis frame of the rolling bearing.
As a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the vibration acceleration signals include vibration acceleration signals in typical fault conditions of a normal bearing, an inner ring fault bearing, an outer ring fault bearing, and a rolling body fault bearing.
As a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the feature extraction includes dividing original time series data Q with length N into sub-samples with length d, and allocating corresponding labels to each sub-sample, wherein there is no overlap between every two sub-samples, and the sub-sample set is expressed as:
wherein x represents a sub-sample y represents a label corresponding to the sub-sample, S is an obtained sub-sample set, and n represents the number of sub-samples;
and (3) performing an aggregation operation on the obtained sub-sample set S:
f Φ (S)＝Φ(x 1 ，x 2 ，…，x m )
wherein x is 1 ,x 2 ,...,x n Representing data points in subsamples, Φ representing an aggregation function;
the sub-sample set T is represented by the aggregate operation as:
T＝f Φ (S 1 ),f Φ (S 2 ),…,f Φ (S n )＝[T 1 ，T 2 ，…，T n ]
wherein f Φ (S i ) For subsamples S i Using the result obtained after the aggregation function Φ, T n A signal value corresponding to the sampling point n;
with m-1 different kernel lengths s.epsilon.2, m]One-dimensional convolution operator Conv of (1) s The feature sequence of the sub-sample set T, T obtained by processing is calculated by the following formula:
wherein Conv s (g) A one-dimensional convolution layer with a convolution kernel length s and a step length of 1 is represented, and m is a super parameter for controlling the distance between two time sampling points;
using non-active function ReLU pairsProcessing is performed, and the process is expressed as follows:
wherein,representing the weight of the edge between nodes i and j.
As a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the map mapping comprises the steps of constructing an n multiplied by n feature matrix M by arranging feature sequences obtained by one-dimensional convolution in a diagonal parallel direction T Mapping a given time series data Q to g= (V, E),
as a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the improved DiffPool graph classification algorithm comprises taking DiffPool as a distinguishable graph pooling module, learning a differentiable soft cluster allocation for each layer of nodes of deep GNNs, mapping the nodes into a group of clusters and retaining important node information to form coarsening input of the next GNN layer, forming layered representation of the graph, and combining with various graph neural network architectures in an end-to-end mode.
As a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the identification of different working conditions of the rolling bearing comprises multi-channel data splicing, and for a multi-channel data sequence, two channels are respectively expressed as:
P＝[P 1 ,P 2 ,L,P j ,LP n ]
Q＝[Q 1 ,Q 2 ,L,Q j ,LQ n ]
mapping the time data sequence of each channel into a corresponding graph G through an AcvGraph algorithm p ＝(V p ,E p ) And G Q ＝(V Q ,E Q ) Treatment of G with DiffPool without full-connection layer P And G Q To obtain respective feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ :
The final node feature vector is input into the connection layer for classification.
As a preferable scheme of the bearing intelligent diagnosis method based on the self-adaptive aggregation visual view, the invention comprises the following steps: the AcvgNet antifriction bearing intelligent diagnosis frame comprises that an acceleration sensor is adopted to collect vibration signals in the operation process of mechanical equipment, and the vibration signals are divided into two types: single channel sequence data and multi-channel sequence data;
when the input data are multi-channel data P and Q, the input data are mapped into G by an AcvGraph algorithm P And G Q Graph G was run separately using the modified DiffPool model P And G Q Processing to obtain respective corresponding feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ The labels with different types are obtained through inputting the labels into a full-connection layer for processing, so that the classification target is achieved;
when the input data is single-channel data, the processing mode removes splicing operation, the remaining operation is consistent with the double-channel data processing mode, time series data are still converted into image data through an AcvGraph algorithm, corresponding feature vectors are obtained through DiffPool, and the feature vectors are input into a full-connection layer, so that the classification effect is achieved.
The invention further aims to provide a bearing intelligent diagnosis system based on the self-adaptive aggregation visual view, which can solve the problems of inflexible composition mode and low mapping efficiency of the conventional algorithm through a self-adaptive composition algorithm and an end-to-end bearing intelligent diagnosis framework.
A system of a bearing intelligent diagnosis method based on self-adaptive aggregation visual view is characterized in that: the system comprises a sensor module, a state evaluation module, a visualization module, an alarm module and a recording module;
the sensor module is used for collecting vibration, temperature and current signals during the operation of the shaft;
the state evaluation module is used for evaluating the state of the bearing by using the self-adaptive aggregated visual neural network, judging whether the bearing has a fault or not and identifying the fault type;
the visualization module presents the state evaluation result to the user in a visual form, so that the user can observe and analyze conveniently;
the alarm module is used for giving an alarm when the bearing is detected to have a fault, and reminding a user to take maintenance measures in time;
the recording module is used for recording the historical state and fault information of the bearing, and is convenient for subsequent analysis and maintenance.
A computer device comprising a memory and a processor, said memory storing a computer program, characterized in that said processor, when executing said computer program, implements the steps of a bearing intelligent diagnostic method based on an adaptive aggregated visual view.
A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of a bearing intelligent diagnosis method based on an adaptive aggregated visual view.
The invention has the beneficial effects that: the method is based on a self-adaptive aggregation strategy, parameters and structures of the model are automatically adjusted according to characteristics of input data, generalization capability and adaptability of the model are improved, a large amount of data are processed by utilizing a visual neural network, diagnosis efficiency is improved, rapid prediction and decision are realized, the model is expanded and optimized by adding training data, adjusting a network structure or replacing an algorithm and the like, and requirements of different environments and application scenes are met.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:
FIG. 1 is a schematic flow chart of a bearing intelligent diagnosis method based on adaptive aggregation visual view according to an embodiment of the present invention;
fig. 2 is a schematic diagram of an acvigraph algorithm mapping process of a bearing intelligent diagnosis method based on adaptive aggregation visual views according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an AcvgNet classification framework of a bearing intelligent diagnosis method based on adaptive aggregation visual views according to an embodiment of the present invention;
FIG. 4 is a schematic view of super-parameter optimization of a bearing intelligent diagnosis method based on adaptive aggregation visual view according to an embodiment of the present invention;
FIG. 5 is a schematic diagram showing a comparison of mapping time and single epoch time of three map mapping methods of a bearing intelligent diagnosis method based on an adaptive aggregated visual map according to an embodiment of the present invention;
fig. 6 is a schematic diagram of classification accuracy of an AcvgNet algorithm on different data sets based on an adaptive aggregated visual bearing intelligent diagnosis method according to an embodiment of the present invention;
fig. 7 is a schematic diagram of mapping AcvgNet under different working conditions according to an intelligent bearing diagnosis method based on adaptive aggregation visual view according to an embodiment of the present invention;
fig. 8 is a schematic workflow diagram of a network frame periodic intelligent diagnosis analysis system of an on-line system of a distribution network according to an embodiment of the present invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
While the embodiments of the present invention have been illustrated and described in detail in the drawings, the cross-sectional view of the device structure is not to scale in the general sense for ease of illustration, and the drawings are merely exemplary and should not be construed as limiting the scope of the invention. In addition, the three-dimensional dimensions of length, width and depth should be included in actual fabrication.
Also in the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "upper, lower, inner and outer", etc. are based on the orientation or positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The terms "mounted, connected, and coupled" should be construed broadly in this disclosure unless otherwise specifically indicated and defined, such as: can be fixed connection, detachable connection or integral connection; it may also be a mechanical connection, an electrical connection, or a direct connection, or may be indirectly connected through an intermediate medium, or may be a communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
Example 1
Referring to fig. 1-3, a first embodiment of the present invention provides a bearing intelligent diagnosis method based on adaptive aggregation visual view, including:
s1: and carrying out a bearing typical fault vibration test, and collecting vibration acceleration signals under a typical fault state.
Still further, the vibration acceleration signals include vibration acceleration signals in four typical fault states including a normal bearing (NC), an inner ring fault bearing (IF), an outer ring fault bearing (OF), and a rolling element fault Bearing (BF).
S2: and performing feature extraction and graph mapping on the acquired time sequence signals by adopting a self-adaptive aggregated visual algorithm to obtain graph data input by nodes and edges in the graph neural network.
Still further, the feature extraction includes dividing the original time-series data Q with a length N into sub-samples with a length d, and assigning a corresponding label to each sub-sample, where there is no overlap between every two sub-samples, and the sub-sample set can be expressed as:
wherein x represents the sub-sample y represents the label corresponding to the sub-sample, S is the obtained sub-sample set, and n represents the number of sub-samples, namely the number of sampling points;
further, the resulting sub-sample set S is subjected to an aggregation operation:
f Φ (S)＝Φ(x 1 ，x 2 ，…，x n )
wherein x is 1 ,x 2 ,...,x n Representing data points in subsamples S, Φ represents the aggregate function is any one of max, min, mean operations;
still further, the sub-sample set is represented by the aggregate operation as:
T＝f Φ (S 1 ),f Φ (S 2 ),…,f Φ (S n )＝[T 1 ，T 2 ，…，T n ]
wherein f Φ (S i ) For subsamples S i Using the result obtained after the aggregation function Φ, T n A signal value corresponding to the sampling point n;
with m-1 different kernel lengths s.epsilon.2, m]One-dimensional convolution operator Conv of (1) s The feature sequence of the sub-sample set T, T obtained by processing is calculated by the following formula:
wherein Conv s (g) A one-dimensional convolution layer with a convolution kernel length s and a step length of 1 is represented, and m is a super parameter for controlling the distance between two time sampling points;
using non-active function ReLU pairsProcessing is performed, and the above process can be expressed as:
wherein,weights representing edges between nodes i and j
It should be noted that the map mapping includes constructing an n×n feature matrix M by arranging feature sequences obtained by one-dimensional convolution in a diagonal parallel direction T Mapping a given time series data Q to g= (V, E),
further, the specific flow of the adaptive patterning algorithm (AcvGraph) is as follows: collecting vibration acceleration signals under a typical fault state, cutting off time windows defined by the collected signals, adopting aggregation operation for each time window, wherein an aggregated result is a characteristic vector of the time window, adopting one-dimensional convolution operators with different kernel function lengths to carry out convolution operation on the aggregated result, extracting characteristic information, and adopting an inactive function ReLU to carry out noise reduction treatment on the obtained characteristic sequence for filtering noise influence and improving model accuracy and performance. And arranging the noise-reduced characteristic sequences in a diagonal parallel direction to construct an adjacency matrix with weights, so that the signal data are mapped into graph data.
S3: establishing an end-to-end rolling bearing intelligent diagnosis framework based on an improved DiffPool diagram classification algorithm; and inputting the graph data into a diagnosis framework, and optimizing the intelligent diagnosis model of the rolling bearing.
Still further, the improved DiffPool graph classification algorithm includes, as a differentiable graph pooling module, mapping nodes into a set of clusters and retaining important node information by learning a differentiable soft cluster allocation for each layer of nodes of the deep GNNs, forming a coarsened input for the next GNN layer, forming a hierarchical representation of the graph, and combining with various graph neural network architectures in an end-to-end mode.
Further, assume that the cluster distribution matrix learned by layer i isWherein C is l Each row represents l layers n l One of the nodes or clusters, C l Each column corresponds to the next level of l+1 nodes n l+1 One of the clusters. Popular speaking, C l And soft allocation is carried out on each node of the layer I as a cluster of the layer I+1, and each node can be allocated to a plurality of clusters with different nodes by the soft allocation, so that the graph structure is more flexibly and finely presented. The input defining DiffPool is (a l ,Z l ) Wherein A is l Representing a layer l adjacency matrix, Z l For the nodes of this layer, a matrix is embedded, by DiffPool layer (a (l+1) ,X (l+1) )＝Diffpool(A (l) ,Z (l) ) Coarsening the input graph, generating a new coarsened adjacent matrix A for each node or cluster in the coarsened graph l+1 Embedding matrix X l+1 . Specifically, the following two formulas are adopted:
it should be noted that matrix C is allocated by cluster l Aggregation node embedding matrix Z l Is n l+1 Generating a new embedding, i.e. performing a fusion operation on the intra-cluster information; wherein the adjacency matrix A l And generating a coarsened adjacent matrix representing the connection strength between clusters, namely calculating the adjacent matrix between clusters. The formula coarsens the input diagram, namely: next layer roughening adjacent matrix a l+1 Is represented by n l+1 Coarsening graphs of individual nodes or clusters, wherein the nodes or clusters in each new coarsening graph correspond to the cluster node groups of the l layers andrepresenting the magnitude of the correlation between node i and node j.
It should be noted that in the second last layer of DiffPool, i.e., the l-1 layer, clusters are assigned to matrix C l-1 The vector set to 1 assigns all nodes of the last layer to one cluster and generates the final embedded vector corresponding to the whole graph, and the global representation of the vector, i.e. the graph, is input as the characteristic of a classifier (softmax) to realize the task of classifying the graph.
Still further, the multi-channel data concatenation includes, for a multi-channel data sequence, two channels respectively denoted as:
P＝[P 1 ,P 2 ,L,P j ,L P n ]
Q＝[Q 1 ,Q 2 ,L,Q j ,LQ n ]
it should be noted that, mapping the time data sequence of each channel into a corresponding graph G by the AcvGraph algorithm p ＝(V p ,E p ) And G Q ＝(V Q ,E Q ) Treatment of G with DiffPool without full-connection layer P And G Q To obtain respective feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ :
The final node feature vector is input into the connection layer for classification.
S4: and identifying and classifying different working conditions of the rolling bearing according to the optimized intelligent diagnosis frame of the rolling bearing.
Furthermore, the AcvgNet rolling bearing intelligent diagnosis framework comprises that an acceleration sensor is adopted to collect vibration signals in the operation process of mechanical equipment, and the vibration signals are divided into two types: single channel sequence data and multi-channel sequence data;
when the input data are multi-channel data P and Q, the input data are mapped into G by an AcvGraph algorithm P And G Q Graph G was run separately using the modified DiffPool model P And G Q Processing to obtain respective corresponding feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ The labels with different types are obtained through inputting the labels into a full-connection layer for processing, so that the classification target is achieved;
when the input data is single-channel data, the processing mode is identical with the two-channel data processing mode except splicing operation, time sequence data is converted into graph data through an AcvGraph algorithm, corresponding feature vectors are obtained through DiffPool, and the feature vectors are input into a full-connection layer, so that the classification effect is realized.
Example 2
Referring to fig. 4-7, for one embodiment of the present invention, an intelligent bearing diagnosis method based on adaptive aggregation visual is provided, and in order to verify the beneficial effects of the present invention, scientific demonstration is performed through experiments.
The relevant experiments are carried out on the rolling bearing data set self-picked in a laboratory, and the result is shown in fig. 4, and the classification accuracy of the AcvgNet on the data set is highest when the super parameter m is equal to 7 and the aggregator selects the maximum value. Meanwhile, the AcvgNet algorithm can adapt to different types of time sequence data by adjusting different values of m, namely controlling the number of neighbors of nodes in the graph, so that related features are captured more effectively, classification performance is improved, and time sequence data are analyzed more accurately. The model can achieve a relatively high classification accuracy when max is chosen as the aggregator because the maxima in the signal data can better reflect the peaks of the signal, which peaks have a significant meaning for the classification of faults, while the maximum aggregator can resist the influence of certain noise types because the peaks of the signal are still present and can be detected even if noise is present.
Secondly, the time complexity problem of three patterning modes VGraph, LPVGraph and AcvGraph in the visual field. Assuming that n nodes exist in one graph, acvGraph introduces super parameter m, and m determines the adjacent of each nodeThe number of counts, therefore, the time complexity of AcvGraph is O (n, m). In the process of the VGgraph, each node needs to be calculated with n nodes to judge whether edges exist between the two nodes, so the time complexity of the Acvgraph is O (n) 2 ) Meanwhile, the LPVGgraph adds the limit penetrable distance lambda on the basis of VGgraph, so that the time complexity is far greater than that of VGgraph. It is worth noting that vggraph and lpvggraph need to infer whether an edge relationship exists between nodes through a determining statement, and acvggraph is that matrix operation can be carried out to load the nodes on a GPU for running, so that mapping efficiency of a graph is improved, and mapping time is shortened. We have experimentally verified that using the same dataset as input, record the time for three patterning schemes to map out 10 maps and input the map data to the GCN algorithm and iterate a single epoch, the results are shown in fig. 5. The map mapping time and the iterative single epoch time of the LPVGgraph algorithm are longest, and are far smaller than VGraph, LPVGraph and only 3.17s and 1.2s due to the flexibility of the Acvgraph composition mode. The method is beneficial to the fact that the AcvgNet adopts an improved DiffPool algorithm, each layer generates a soft cluster distribution for the next layer, the nodes are mapped into a group of clusters, important node information is reserved, coarsening input of the next GNN layer is formed, layered representation of the graph is formed, and model training and mapping time is greatly shortened.
Then, a conventional experiment is carried out, the generalization of the AcvgNet algorithm is verified, and a public data set noted in four fault diagnosis fields and a self-collected data set of the invention are selected for experiment, wherein the classification precision is shown in figure 6. From the results, acvgNet can be seen to exhibit strong classification advantages on different data sets, indicating that the invention has good generalization capability.
Finally, deep learning today makes a significant breakthrough in various fields, but its black box nature limits its application in certain fields. Deep-learned neural network models are typically composed of many complex nonlinear functions, which make it difficult to interpret the decision process and the prediction results of the model. In this context, in order to increase the interpretability of the AcvgNet algorithm, the map into which the time data is mapped is visualized, and differences between maps formed by different working conditions are found, so that the interpretability of the AcvgNet is increased by way of illustration.
Specifically, the obtained weighted adjacency matrix is input into Gephi, and the layout is selected to open the drawing of the graph. In order to make the graph distinction of different working conditions more obvious, the edges with particularly small weight in each graph are deleted, and the weights of different edges are distinguished by adopting gradual change, wherein the darker the color is, the greater the edge weight value is, namely the closer the relationship between two nodes is, as shown in fig. 7. By observing the connection relation between the structure of the graph and the nodes, the transverse comparison can find that different working conditions under the same data set present different graph structures, which is helpful for identifying and classifying fault working conditions.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered by the scope of the claims of the present invention.
Example 3
A third embodiment of the present invention, which is different from the first two embodiments, is:
the functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Example 4
Referring to fig. 8, for one embodiment of the present invention, a system for a bearing intelligent diagnosis method based on adaptive aggregated visual view is provided, which is characterized in that: the system comprises a sensor module, a state evaluation module, a visualization module, an alarm module and a recording module;
the sensor module is used for collecting vibration, temperature and current signals during the operation of the shaft, and the subsequent analysis and processing provide raw data.
The state evaluation module carries out self-adaptive aggregation on the preprocessed data, improves the accuracy and stability of the data, evaluates the state of the bearing through a self-adaptive aggregation visual neural network, judges whether the bearing has faults, identifies the fault type and improves the quality and accuracy of the data.
The visual module models and learns the aggregated data by using a visual neural network, generates a prediction result, identifies and classifies the state of the rolling bearing, and presents the state evaluation result to a user in a visual form, so that the user can observe and analyze conveniently.
And the alarm module is used for giving an alarm when the bearing is detected to have a fault, and reminding a user to take maintenance measures in time.
The recording module is used for generating a diagnosis report according to the result of the abnormality detection module, recording the historical state and fault information of the bearing, providing fault information and suggested treatment measures for operators, helping the operators to quickly locate the problem and solve the problem, and facilitating subsequent analysis and maintenance.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered by the scope of the claims of the present invention.
Claims (5)
1. An intelligent bearing diagnosis method based on self-adaptive aggregation visual view is characterized by comprising the following steps of: comprising the steps of (a) a step of,
carrying out a bearing typical fault vibration test and collecting vibration acceleration signals under a typical fault state;
performing feature extraction and graph mapping on the acquired time sequence signals by adopting a self-adaptive aggregated visual algorithm to obtain graph data which is input into a graph neural network and consists of nodes and edges;
establishing an end-to-end rolling bearing intelligent diagnosis framework based on an improved DiffPool diagram classification algorithm; inputting the graph data into a diagnosis framework, and optimizing an intelligent diagnosis model of the rolling bearing;
identifying and classifying different working conditions of the rolling bearing according to the optimized intelligent diagnosis frame of the rolling bearing;
the feature extraction includes dividing original time series data Q with length N into sub-samples with length d, and allocating corresponding labels to each sub-sample, wherein there is no overlap between every two sub-samples, and the sub-sample set is expressed as:
wherein x represents a sub-sample y represents a label corresponding to the sub-sample, S is an obtained sub-sample set, and n represents the number of sub-samples; and (3) performing an aggregation operation on the obtained sub-sample set S:
f Φ (S)＝Φ(x 1 ，x 2 ，…，x n )
wherein x is 1 ,x 2 ,...,x n Representing data points in subsamples, Φ representing an aggregation function;
the sub-sample set T is represented by the aggregate operation as:
T＝f Φ (S 1 ),f Φ (S 2 ),…,f Φ (S n )＝[T 1 ，T 2 ，…，T n ]
wherein the method comprises the steps off Φ (S i ) For subsamples S i Using the result obtained after the aggregation function Φ, T n A signal value corresponding to the sampling point n; with m-1 different kernel lengths s.epsilon.2, m]One-dimensional convolution operator Conv of (1) s The feature sequence of the sub-sample set T, T obtained by processing is calculated by the following formula:
wherein Conv s (. Cndot.) represents a one-dimensional convolution layer with a convolution kernel length s and a step length 1, m being a super parameter controlling the distance between two time sampling points; using non-active function ReLU pairsProcessing is performed, and the process is expressed as follows:
wherein,representing the weight of the edge between nodes i and j;
the map mapping comprises the steps of constructing an n multiplied by n feature matrix M by arranging feature sequences obtained by one-dimensional convolution in a diagonal parallel direction T Mapping a given time-series data Q to g= (V, E);
the improved DiffPool diagram classification algorithm comprises the steps of taking DiffPool as a distinguishable diagram pooling module, learning a differentiable soft cluster distribution for each layer of nodes of deep GNN, mapping the nodes into a group of clusters, reserving important node information, forming coarsening input of the next GNN layer, forming layered representation of the diagram, and combining the layered representation with various diagram neural network architectures in an end-to-end mode;
the identification of different working conditions of the rolling bearing comprises multi-channel data splicing, and for a multi-channel data sequence, two channels are respectively expressed as:
P＝[P 1 ,P 2 ,…,P j ,…P n ]
Q＝[Q 1 ,Q 2 ,…,Q j ,…Q n ]
mapping the time data sequence of each channel into a corresponding graph G through an AcvGraph algorithm p ＝(V p ,E p ) And G Q ＝(V Q ,E Q ) Treatment of G with DiffPool without full-connection layer P And G Q To obtain respective feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ :
Inputting the final node feature vector into a connection layer for classification;
an acceleration sensor is adopted to collect vibration signals in the operation process of mechanical equipment, and when input data are multichannel data P and Q, the multichannel data are mapped into G through an AcvGraph algorithm P And G Q Graph G was run separately using the modified DiffPool model P And G Q Processing to obtain respective corresponding feature vectors Z P And Z Q The feature vector Z P And Z Q Splicing to obtain a final node characteristic vector Z PQ The labels with different types are obtained through inputting the labels into a full-connection layer for processing, so that the classification target is achieved;
when the input data is single-channel data, the processing mode removes splicing operation, the remaining operation is consistent with the double-channel data processing mode, time series data are still converted into image data through an AcvGraph algorithm, corresponding feature vectors are obtained through DiffPool, and the feature vectors are input into a full-connection layer, so that the classification effect is achieved.
2. The intelligent bearing diagnosis method based on the adaptive aggregate visual view as claimed in claim 1, wherein the method comprises the following steps: the vibration acceleration signals include vibration acceleration signals in a typical failure state of a normal bearing, an inner ring failure bearing, an outer ring failure bearing, and a rolling body failure bearing.
3. A system employing an adaptive aggregated view based bearing intelligent diagnostic method according to any one of claims 1-2, characterized in that: the system comprises a sensor module, a state evaluation module, a visualization module, an alarm module and a recording module;
the sensor module is used for collecting vibration, temperature and current signals during the operation of the shaft;
the state evaluation module is used for evaluating the state of the bearing by using the self-adaptive aggregated visual neural network, judging whether the bearing has a fault or not and identifying the fault type;
the visualization module presents the state evaluation result to the user in a visual form, so that the user can observe and analyze conveniently;
the alarm module is used for giving an alarm when the bearing is detected to have a fault, and reminding a user to take maintenance measures in time;
the recording module is used for recording the historical state and fault information of the bearing, and is convenient for subsequent analysis and maintenance.
4. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 3 when the computer program is executed.
5. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310700517.0A CN116977708B (en) | 2023-06-14 | 2023-06-14 | Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310700517.0A CN116977708B (en) | 2023-06-14 | 2023-06-14 | Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116977708A CN116977708A (en) | 2023-10-31 |
| CN116977708B true CN116977708B (en) | 2024-04-12 |
Family
ID=88483986
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310700517.0A Active CN116977708B (en) | 2023-06-14 | 2023-06-14 | Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116977708B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117636253A (en) * | 2023-11-30 | 2024-03-01 | 江苏圣创半导体科技有限公司 | Intelligent lamp fault identification method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112380959A (en) * | 2020-11-11 | 2021-02-19 | 浙江工业大学 | Univariate time series classification method based on graph neural network |
| CN112508181A (en) * | 2020-12-18 | 2021-03-16 | 南京航空航天大学 | Graph pooling method based on multi-channel mechanism |
| CN115062648A (en) * | 2022-06-02 | 2022-09-16 | 华南理工大学 | Fault diagnosis method, system and device for rotary machine and storage medium |
| CN115290326A (en) * | 2021-12-20 | 2022-11-04 | 兰州理工大学 | Rolling bearing fault intelligent diagnosis method |
| CN116026593A (en) * | 2022-12-19 | 2023-04-28 | 山东大学 | Cross-working-condition rolling bearing fault targeted migration diagnosis method and system |
| WO2023078009A1 (en) * | 2021-11-05 | 2023-05-11 | 华为云计算技术有限公司 | Model weight acquisition method and related system |
| CN116187200A (en) * | 2023-03-20 | 2023-05-30 | 杭州电子科技大学 | Lightweight three-dimensional CAD model classification and retrieval method based on graph convolution network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11238277B2 (en) * | 2019-06-16 | 2022-02-01 | Way2Vat Ltd. | Systems and methods for document image analysis with cardinal graph convolutional networks |
-
2023
- 2023-06-14 CN CN202310700517.0A patent/CN116977708B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112380959A (en) * | 2020-11-11 | 2021-02-19 | 浙江工业大学 | Univariate time series classification method based on graph neural network |
| CN112508181A (en) * | 2020-12-18 | 2021-03-16 | 南京航空航天大学 | Graph pooling method based on multi-channel mechanism |
| WO2023078009A1 (en) * | 2021-11-05 | 2023-05-11 | 华为云计算技术有限公司 | Model weight acquisition method and related system |
| CN115290326A (en) * | 2021-12-20 | 2022-11-04 | 兰州理工大学 | Rolling bearing fault intelligent diagnosis method |
| CN115062648A (en) * | 2022-06-02 | 2022-09-16 | 华南理工大学 | Fault diagnosis method, system and device for rotary machine and storage medium |
| CN116026593A (en) * | 2022-12-19 | 2023-04-28 | 山东大学 | Cross-working-condition rolling bearing fault targeted migration diagnosis method and system |
| CN116187200A (en) * | 2023-03-20 | 2023-05-30 | 杭州电子科技大学 | Lightweight three-dimensional CAD model classification and retrieval method based on graph convolution network |
Non-Patent Citations (2)
| Title |
|---|
| 基于同步挤压S变换和集成深层脊波自编码器的轴承故障诊断;杜小磊;陈志刚;王衍学;;振动与冲击;20200728(第14期);全文 * |
| 基于改进卷积神经网络的滚动轴承智能故障诊断研究;宫文峰;陈辉;张泽辉;张美玲;管聪;王鑫;;振动工程学报;20200415(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116977708A (en) | 2023-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109886273B (en) | CMR image segmentation and classification system | |
| Zhao et al. | Improved generative adversarial network for vibration-based fault diagnosis with imbalanced data | |
| CN113344295B (en) | Method, system and medium for predicting residual life of equipment based on industrial big data | |
| CN111967486A (en) | Complex equipment fault diagnosis method based on multi-sensor fusion | |
| CN113673346B (en) | Motor vibration data processing and state identification method based on multiscale SE-Resnet | |
| CN113158814B (en) | Bearing health state monitoring method based on convolution self-encoder | |
| CN109946080B (en) | Mechanical equipment health state identification method based on embedded circulation network | |
| CN116977708B (en) | Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view | |
| CN111224805A (en) | Network fault root cause detection method, system and storage medium | |
| CN113505655A (en) | Bearing fault intelligent diagnosis method for digital twin system | |
| CN115909006B (en) | Mammary tissue image classification method and system based on convolution transducer | |
| CN114970715A (en) | Variable working condition fault diagnosis method and system under small sample and unbalanced data constraint | |
| US20230108916A1 (en) | Method and system for forecasting non-stationary time-series | |
| CN114926825A (en) | Vehicle driving behavior detection method based on space-time feature fusion | |
| CN116030538B (en) | Weak supervision action detection method, system, equipment and storage medium | |
| CN115358259A (en) | Self-learning-based unsupervised cross-working-condition bearing fault diagnosis method | |
| CN110728377B (en) | Intelligent fault diagnosis method and system for electromechanical equipment | |
| CN114841250A (en) | Industrial system production abnormity detection and diagnosis method based on multi-dimensional sensing data | |
| CN116842379A (en) | Mechanical bearing residual service life prediction method based on DRSN-CS and BiGRU+MLP models | |
| CN116223020A (en) | Intelligent fault diagnosis method for gearbox based on Li-MDCAN | |
| CN116680643A (en) | Abnormality detection method combining deep contrast learning and density clustering | |
| CN114492534A (en) | Construction method and application of cross-size motor bearing fault diagnosis model | |
| CN113092044B (en) | Rotary machine fault diagnosis method based on weighted level visible graph | |
| CN116124460B (en) | Bearing life prediction method and system based on health index construction | |
| CN117312948A (en) | Rotary machine fault diagnosis method based on multi-sensor fusion and ECA-CNN |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20231031 Assignee: Wuhan Moming Technology Co.,Ltd. Assignor: Beijing University of Civil Engineering and Architecture Contract record no.: X2024980007601 Denomination of invention: A bearing intelligent diagnosis method and system based on adaptive aggregation and viewability Granted publication date: 20240412 License type: Common License Record date: 20240620 |