CN116183231B - Bearing fault diagnosis method based on digital twin - Google Patents
Bearing fault diagnosis method based on digital twin Download PDFInfo
- Publication number
- CN116183231B CN116183231B CN202310472921.7A CN202310472921A CN116183231B CN 116183231 B CN116183231 B CN 116183231B CN 202310472921 A CN202310472921 A CN 202310472921A CN 116183231 B CN116183231 B CN 116183231B
- Authority
- CN
- China
- Prior art keywords
- bearing
- rolling bearing
- rolling
- digital twin
- fault
- 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
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000003745 diagnosis Methods 0.000 title claims abstract description 65
- 238000005096 rolling process Methods 0.000 claims abstract description 265
- 238000004088 simulation Methods 0.000 claims abstract description 29
- 238000012360 testing method Methods 0.000 claims abstract description 23
- 238000013016 damping Methods 0.000 claims description 57
- 238000012549 training Methods 0.000 claims description 38
- 238000013528 artificial neural network Methods 0.000 claims description 37
- 238000006073 displacement reaction Methods 0.000 claims description 31
- 238000003062 neural network model Methods 0.000 claims description 30
- 230000004044 response Effects 0.000 claims description 29
- 238000001228 spectrum Methods 0.000 claims description 28
- 230000005284 excitation Effects 0.000 claims description 21
- 230000001133 acceleration Effects 0.000 claims description 15
- 230000000694 effects Effects 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000010200 validation analysis Methods 0.000 claims 1
- 238000012423 maintenance Methods 0.000 abstract description 5
- 238000013461 design Methods 0.000 abstract description 4
- 238000012544 monitoring process Methods 0.000 abstract description 4
- 238000004901 spalling Methods 0.000 description 12
- 230000007547 defect Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000005299 abrasion Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Acoustics & Sound (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention provides a digital twin-based bearing fault diagnosis method, which solves the problems that an industrial equipment bearing lacks historical fault data and cannot perform fault diagnosis through historical data information; by constructing a rolling bearing digital twin body, performing a multi-working-condition simulation test in the digital twin body, simulating fault data which are difficult to obtain in actual operation, thereby realizing real-time monitoring and predictive maintenance and finding out potential problems which are not considered in design; by updating parameters of the digital twin body in real time, a high-fidelity vibration signal is obtained, and precise identification of the fault width of the bearing under the variable speed working condition is realized.
Description
Technical Field
The invention relates to the technical field of fault diagnosis, in particular to a bearing fault diagnosis method based on digital twinning.
Background
Bearings are key parts of modern equipment such as aeroengines, gas turbines, industrial robots and the like, and the running state of the bearings influences the performance and reliability of the mechanical equipment. The outer ring raceway surface of the rolling bearing may generate a local peeling defect due to the influence of the weight and impact force of the bearing carrying the rotating machine equipment, resulting in abnormal operation of the rolling bearing, and further, a serious accident may be caused. Therefore, the bearing fault diagnosis method is constructed, the running state of the production line equipment is reflected in real time, production personnel are helped to master the production running condition, the efficient utilization of the equipment is realized, and the normal running of the equipment and the improvement of the production efficiency are ensured.
At present, the traditional bearing fault diagnosis method is mainly based on analysis of vibration signals and sound signals, and features are required to be manually extracted and judged. This method has the following disadvantages: firstly, the artificial feature extraction process is easily influenced by subjective factors of personnel, so that the reliability of a diagnosis result is not high; secondly, the method can only detect some obvious characteristics of the bearing fault, but can not detect the hidden characteristics of the bearing fault; finally, for some new devices lacking historical fault data, fault diagnosis through historical data information cannot be performed. Therefore, providing a bearing fault diagnosis method for updating historical fault data in real time to effectively improve the reliability of the diagnosis result is a problem to be solved in the prior art.
Disclosure of Invention
The invention aims to provide a bearing fault diagnosis method based on digital twinning. The method solves the problem that the bearing of the industrial equipment lacks historical fault data and cannot carry out fault diagnosis through the historical data information; by constructing a rolling bearing digital twin body, performing a multi-working-condition simulation test in the digital twin body, simulating fault data which are difficult to obtain in actual operation, thereby realizing real-time monitoring and predictive maintenance and finding out potential problems which are not considered in design; by updating parameters of the digital twin body in real time, a high-fidelity vibration signal is obtained, and precise identification of the fault width of the bearing under the variable speed working condition is realized.
In order to achieve the above object, the present invention provides the following technical solutions:
the invention provides a digital twin-based bearing fault diagnosis method, which is based on the construction of a rolling bearing digital twin body and the real-time updating of parameters of the rolling bearing digital twin body, and comprises the following steps:
(1) Preprocessing the collected real-time vibration signals of the rolling bearing to obtain root mean square and envelope spectrum of the real-time vibration signals;
(2) According to the physical parameters of different rolling bearings, including the supporting rigidity of the outer ring and the bearing seatk h Two variable parameters of the bearing damping c, and establishing a rolling bearing digital twin body;
(3) Utilizing the rolling bearing digital twin body simulation obtained in the step (2) to obtain the vibration response of the rolling bearing under the normal state and the fault state, and then simulating to obtain the fault width of the rolling bearing and the root mean square of vibration signals, and taking the root mean square as a sample set;
(4) Inputting the sample set obtained in the step (3) into an MATLAB regression learner to select a wide neural network for training (the wide neural network is a neural network in the MATLAB regression learner), and obtaining a trained wide neural network model;
(5) Performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the wide neural network model obtained in the step (4) to obtain a fault state of the rolling bearing to be diagnosed;
updating the supporting rigidity of the rolling bearing digital twin body obtained in the step (2) comprising the bearing outer ring and the bearing seat in real timek h And the physical parameters of the bearing damping c until the difference value between the envelope spectrum main peak amplitude of the vibration response obtained by the real-time updated rolling bearing digital twin body simulation and the envelope spectrum main peak amplitude of the real-time vibration signal in the step (1) is the minimum value;
repeating the steps (3) - (5) by using the rolling bearing digital twin body updated in real time.
Preferably, the real-time vibration signal of the rolling bearing in the step (1) is acquired by a vibration acceleration sensor; the pretreatment comprises the following steps: and carrying out band-pass filtering treatment on the real-time vibration signal of the rolling bearing, removing low-frequency noise, calculating to obtain a root mean square value of the vibration signal of the bearing, and drawing an envelope spectrum through Hilbert transformation.
Preferably, the establishing of the rolling bearing digital twin in step (2) comprises the steps of: selecting the model of the rolling bearing, and obtaining different selected physical parameters of the rolling bearing, wherein the physical parameters comprise the supporting rigidity of the bearing outer ring and the bearing seatk h And bearing damping c; establishing a four-degree-of-freedom dynamic model of the bearing system; calculating the rigidity and damping of the bearing system; establishing a peeling fault time-varying displacement excitation model and a timely displacement excitation function; establishing a multi-degree-of-freedom dynamic differential equation; and solving a multi-degree-of-freedom dynamic differential equation by using a fixed-step length four-order Dragon-Gregory tower method to obtain the vibration response of the rolling bearing with the peeling fault.
Preferably, the four-degree-of-freedom dynamic model of the bearing system is built, comprising the following steps: the inner ring and the shaft of each rolling bearing are regarded as a whole, and the total mass of the inner ring and the shaft is m i A representation; the outer ring of the rolling bearing is fixed on the bearing seat, and the total mass of the rolling bearing is m o A representation; elastic support with rigidity and damping exists between the shaft and the bearing seat, and rolling bodies of the rolling bearing, the outer ring and the inner ring of the racewayThe contact of the ring roller path also has elastic support of rigidity and damping, the rolling body of the rolling bearing is also expressed by a spring-damping model, and a plane coordinate system, namely a vertical X direction and a horizontal Y direction, is established; based on the plane motion of the outer ring and the inner ring of the rolling bearing in the radial direction, 4 degrees of freedom are taken as a total, and a four-degree-of-freedom dynamic model of the bearing system is built.
Preferably, the multiple degree of freedom kinetic differential equation comprises an inner ring kinetic differential equation of the rolling bearing;
the formula of the inner ring kinematic differential equation of the rolling bearing is as follows:
in the method, in the process of the invention,m i for the total mass of the bearing inner race and the shaft,c s is the damping of the inner ring of the bearing and the shaft,k s for the bearing inner race and the supporting rigidity of the shaft,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、And->Respectively is provided with an inner ring atXAnd (3) withYVibration in directionAcceleration, velocity and displacement, +.>And->Is arranged at the outer ringXAnd (3) withYSpeed in the direction.
Preferably, the multiple degree of freedom kinetic differential equation comprises an outer ring kinetic differential equation of the rolling bearing;
the formula of the outer ring kinematic differential equation of the rolling bearing is as follows:
in the method, in the process of the invention,m o is the total mass of the outer ring of the bearing and the bearing seat,c h is the damping of the outer ring of the bearing and the bearing seat,k h the support rigidity of the bearing outer ring and the bearing seat is adopted,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and (3) with、/>And->、/>And->Respectively is at the outer ringXAnd (3) withYIn the direction ofVibration acceleration, velocity and displacement of +.>And->Is arranged at the inner ringXAnd (3) withYSpeed in the direction.
Preferably, in the step (4), the sample set is input into a regression learner for training to obtain an optimal neural network, and the method comprises the following steps: dividing a sample set into a test set and a training set according to a preset proportion; inputting the training set into a MATLAB regression learner to select a wide neural network for training, and obtaining a trained wide neural network model; and verifying the validity of the fault diagnosis effect of the trained wide neural network model by using the test set.
Preferably, verifying the validity of the fault diagnosis effect of the trained wide neural network model by using the test set comprises the following steps: and inputting the test set into a wide neural network model for training, and verifying the effectiveness of the fault diagnosis effect of the trained wide neural network.
Preferably, the support rigidity of the bearing outer ring and the bearing seat is updated in real time in the step (5)k h Bearing damping cTwo pairs of Personal (S)Repeating the steps (3) - (5) of the digital twin body of the rolling bearing after parameters, wherein the method comprises the following steps:
(S-1) updating the supporting rigidity k of the bearing outer race and the bearing housing in real time h The rolling bearing digital twin body simulation after the two parameters of the bearing damping c obtain the vibration response of the rolling bearing under the normal state and the fault state, and then the simulation obtains the fault width of the rolling bearing and the root mean square of vibration signals, and the root mean square is used as an updated sample set;
(S-2) inputting the updated sample set obtained in the step (S-1) into a regression learner to select a wide neural network for training, so as to obtain an updated wide neural network model;
and (S-3) performing fault diagnosis on the root mean square of the vibration signal of the rolling bearing to be diagnosed by using the updated wide neural network model obtained in the step (S-2) to obtain the fault state of the rolling bearing to be diagnosed.
The invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes any digital twin-based bearing fault diagnosis method in the technical scheme when executing the computer program.
The invention provides a digital twin-based bearing fault diagnosis method, which belongs to digital twin fault diagnosis technology, is a novel fault diagnosis method combining dynamic modeling and a neural network, and can fully utilize a rolling bearing digital twin body to realize fault diagnosis in a virtual space by combining the method with the neural network, so that single fault and compound fault of a rolling bearing can be diagnosed; the rolling bearing digital twin body can dynamically simulate the rolling bearing, simulate the abrasion and faults of the rolling bearing, thereby realizing fault diagnosis of the rolling bearing, and can perform multi-working-condition simulation test in the rolling bearing digital twin body, simulate fault data which are difficult to obtain in actual operation, and find out potential problems which are not considered in design. According to the invention, the trained neural network is migrated from the virtual space to the physical space, and the relation between the vibration signal and the fault state is revealed, so that real-time monitoring and predictive maintenance are realized, and the reliability of the fault diagnosis result is effectively improved.
Compared with the prior art, the method provided by the invention has the advantages that:
(1) The invention solves the problems that the bearing of the industrial equipment lacks historical fault data and cannot carry out fault diagnosis through the information of the historical data;
(2) According to the invention, by constructing the rolling bearing digital twin body, a multi-working-condition simulation test is carried out, and fault data which is difficult to obtain in actual operation is simulated, so that real-time monitoring and predictive maintenance are realized, and potential problems which are not considered in the design of the rolling bearing are found;
(3) According to the invention, the parameters of the digital twin body of the rolling bearing are updated in real time, the high-fidelity vibration response signal is obtained through simulation, the accurate identification of the fault width of the rolling bearing under the variable speed working condition is realized, and the reliability of the fault diagnosis result is effectively improved.
Drawings
FIG. 1 is a block diagram of a digital twin body of a rolling bearing of the present invention;
FIG. 2 is a diagram of a four degree-of-freedom dynamics model of the bearing system of the present invention;
FIG. 3 is a graph of a time-varying displacement excitation model of bearing spalling failure according to the present invention;
FIG. 4 is a flowchart of a method for diagnosing a rolling bearing failure in an embodiment of the present invention;
FIG. 5 is a graph showing the contrast of the envelope spectrum image before updating the parameters of the bearing 1 in the embodiment 1 of the present invention;
FIG. 6 is a graph showing the contrast of the envelope spectrum image after updating the parameters of the bearing 1 in example 1 of the present invention;
fig. 7 is a comparison chart of actual fault widths of the rolling bearings 1 to 4 and fault widths of the rolling bearings 1 to 4 obtained before and after updating parameters of digital twin bodies of the rolling bearings in real time in the embodiment of the present invention.
Detailed Description
The invention provides a digital twin-based bearing fault diagnosis method, which is based on the construction of a rolling bearing digital twin body and the real-time updating of parameters of the rolling bearing digital twin body, and comprises the following steps:
(1) Preprocessing the acquired real-time vibration signals of the rolling bearing to be diagnosed to obtain root mean square and envelope spectrum of the real-time vibration signals;
(2) According to the physical parameters of different rolling bearings, including the supporting rigidity of the outer ring and the bearing seatk h The method comprises the steps of (1) establishing a rolling bearing digital twin body by two variable parameters of bearing damping c;
(3) Utilizing the rolling bearing digital twin body simulation obtained in the step (2) to obtain the vibration response of the rolling bearing under the normal state and the fault state, and then simulating to obtain the fault width of the rolling bearing and the root mean square of vibration signals, and taking the root mean square as a sample set;
(4) Inputting the sample set obtained in the step (3) into an MATLAB regression learner to select a wide neural network for training, and obtaining a trained wide neural network model;
(5) Performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the wide neural network model obtained in the step (4) to obtain a fault state of the rolling bearing to be diagnosed;
updating the supporting rigidity of the bearing outer ring and the bearing seat of the rolling bearing digital twin body obtained in the step (2) in real timek h And bearing damping c physical parameters until the difference between the envelope spectrum main peak amplitude of the vibration response obtained by the real-time updated rolling bearing digital twin body simulation and the envelope spectrum main peak amplitude of the real-time vibration signal in the step (1) is the minimum value;
repeating the steps (3) - (5) by using the rolling bearing digital twin body updated in real time.
In the present invention, the real-time vibration signal of the rolling bearing to be diagnosed is preferably the outer ring acceleration of the rolling bearing to be diagnosed. In the invention, the real-time vibration signal of the rolling bearing to be diagnosed is preferably acquired by a vibration acceleration sensor; the pretreatment preferably comprises the steps of: and carrying out band-pass filtering treatment on the real-time vibration signal of the rolling bearing, removing low-frequency noise, calculating to obtain a root mean square value of the vibration signal of the rolling bearing to be diagnosed, and drawing an envelope spectrum through Hilbert transformation.
In the present invention, the establishment of the rolling bearing digital twin body preferably comprises the steps of: selecting the model of the rolling bearing, and obtaining different selected physical parameters of the rolling bearing, wherein the physical parameters comprise the supporting rigidity of the bearing outer ring and the bearing seatk h And bearing damping c; establishing a four-degree-of-freedom dynamic model of the bearing system; calculating the rigidity and damping of the bearing system; establishing a peeling fault time-varying displacement excitation model and a timely displacement excitation function; establishing a multi-degree-of-freedom dynamic differential equation; and solving a multi-degree-of-freedom dynamic differential equation by using a fixed-step length four-order Dragon-Gregory tower method to obtain the vibration response of the rolling bearing with the peeling fault.
In the invention, a four-degree-of-freedom dynamic model of the bearing system is established, and preferably comprises the following steps: each is put intoThe inner ring and the shaft of the rolling bearing are regarded as a whole, and the total mass of the inner ring and the shaft is m i A representation; the outer ring of the rolling bearing is fixed on the bearing seat, and the total mass of the rolling bearing is m o A representation; the method comprises the steps that a rigid and damped elastic support exists between a shaft and a bearing seat, the contact between a rolling body of a rolling bearing and an outer ring raceway and an inner ring raceway also has the rigid and damped elastic support, the rolling body of the rolling bearing is also expressed by a spring-damping model, and a plane coordinate system, namely a vertical X direction and a horizontal Y direction, is established; based on the plane motion of the outer ring and the inner ring of the rolling bearing in the radial direction, totally 4 degrees of freedom are adopted, and a four-degree-of-freedom dynamic model of the bearing system is established;
the four-degree-of-freedom dynamic model diagram of the bearing system in the method provided by the invention is shown in figure 2.
In the present invention, the stiffness and damping of the bearing system preferably comprises: support stiffness of shaft and inner race of rolling bearingk s Support stiffness of bearing housing and outer race of rolling bearingk h Contact stiffness of rolling bearingkDamping of a shaft and an inner ring of a rolling bearingc s Damping of bearing blocks and outer rings of rolling bearingsc h Damping of rolling bearingc。
In the invention, a spalling fault time-varying displacement excitation model and a time-varying displacement excitation function are established, and preferably the method comprises the following steps: according to the Hertz contact theory, the contact deformation of the rolling body and the rollaway nest is caused to change at the bearing peeling fault position, so that the contact force is changed, the abnormal vibration of the bearing is further induced, and a peeling fault time-varying displacement excitation model is established aiming at the time-varying displacement excitation generated by the peeling fault of the outer ring of the bearing; the invention provides a method, wherein a bearing spalling fault time-varying displacement excitation model diagram is shown in figure 3, W in figure 3 is the width of spalling defect, d is the diameter of rolling body, omega s The rotation angular velocity of the inner ring; then, the formula for establishing the time-varying displacement excitation function of the peeling fault of the outer ring of the bearing is as follows:
in the method, in the process of the invention,H d for time-varying displacement excitation of the outer race spalling failure,H max for the maximum additional displacement created by the spalling defect,θ d for the range angle of the defective area,θ j is the firstjThe time-varying position angle of the individual rolling bodies,θ i and (5) peeling the initial angle of the defect for the outer ring raceway.
In the invention, the multi-degree-of-freedom dynamic differential equation preferably further comprises an outer ring dynamic differential equation of the rolling bearing; the formula of the outer ring kinematic differential equation of the rolling bearing is preferably as follows:
in the method, in the process of the invention,m i for the total mass of the bearing inner race and the shaft,c s is the damping of the inner ring of the bearing and the shaft,k s for the bearing inner race and the supporting rigidity of the shaft,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、And->Respectively is provided with an inner ring atXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the outer ringXAnd (3) withYSpeed in the direction.
In the invention, the multi-degree-of-freedom dynamic differential equation preferably further comprises an outer ring dynamic differential equation of the rolling bearing; the formula of the outer ring kinematic differential equation of the rolling bearing is preferably as follows:
in the method, in the process of the invention,m o is the total mass of the outer ring of the bearing and the bearing seat,c h is the damping of the outer ring of the bearing and the bearing seat,k h the support rigidity of the bearing outer ring and the bearing seat is adopted,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and (3) with、/>And->、/>And->Respectively is at the outer ringXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the inner ringXAnd (3) withYSpeed in the direction.
In the invention, a dynamic differential equation of multiple degrees of freedom is solved by using a fixed-step length four-order Dragon lattice tower method to obtain the vibration response of the rolling bearing with the peeling fault, and the method preferably comprises the following steps: determining the supporting rigidity of the bearing outer ring and the bearing seat in the inner ring and outer ring kinematic differential equation (namely the multi-degree-of-freedom dynamic differential equation) of the bearing systemk h And solving the physical parameters of the bearing damping c by using a fixed-step length four-order Dragon lattice tower method to obtain the vibration response of the rolling bearing with the peeling fault.
In the invention, a sample set is input into a MATLAB regression learner to select a wide neural network for training, and a trained wide neural network model is obtained, preferably comprising the following steps: dividing a sample set into a test set and a training set according to a preset proportion; inputting the training set into a MATLAB regression learner to select a wide neural network for training, so as to obtain a trained wide neural network; and verifying the validity of the fault diagnosis effect of the trained wide neural network by using the test set.
In the present invention, a sample set with a preset ratio of preferably 20% is used as the test set, and a sample set with 80% is used as the training set.
In the present invention, the validity of the failure diagnosis effect of the trained wide neural network is verified by using the test set, preferably comprising the steps of:
and inputting the test set into a wide neural network model for training, and verifying the effectiveness of the fault diagnosis effect of the trained wide neural network.
In the present invention, the support rigidity including the bearing outer race and the bearing housing is updated in real timek h And repeating the steps (3) - (5) of the rolling bearing digital twin body after the physical parameters of the bearing damping c, preferably comprising the following steps of:
(S-1) updating the support rigidity including the bearing outer race and the bearing housing with real timek h After the physical parameters of the bearing damping c, the digital twin body simulation of the rolling bearing obtains the vibration response of the rolling bearing under the normal state and the fault state, and then the simulation obtains the fault width of the rolling bearing and the root mean square of the vibration signal, and the root mean square is taken as an updated sample set;
(S-2) inputting the updated sample set obtained in the step (S-1) into a regression learner to select a wide neural network for training, so as to obtain an updated wide neural network model;
and (S-3) performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the updated wide neural network obtained in the step (S-2) to obtain a fault state of the rolling bearing to be diagnosed.
The invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes any digital twin-based bearing fault diagnosis method in the technical scheme when executing the computer program.
The technical solutions of the present invention will be clearly and completely described in the following in connection with the embodiments of the present invention. It will be apparent that the described embodiments are only some, but not all, embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
FIG. 1 is a block diagram of a digital twin body of a rolling bearing according to the present invention, specifically: the rolling bearing test bed parameters of the physical space are mapped to the virtual space in real time to establish a rolling bearing digital twin body, wherein the rolling bearing digital twin body comprises physical parameters such as mass, size, rigidity and damping, a sample set is simulated through the rolling bearing digital twin body, the sample set variable comprises an acceleration response root mean square value and the fault width of a bearing, the sample set is input into a MATLAB regression learner to select a wide neural network for training, a trained wide neural network model is obtained, at the moment, the acceleration root mean square value acquired by the rolling bearing test bed of the physical space is input into the wide neural network model, the fault width can be diagnosed, and corresponding maintenance suggestions such as equipment shutdown and bearing replacement are provided for operators.
Fig. 4 is a flowchart of a rolling bearing fault diagnosis method according to an embodiment of the present invention, specifically:
(1) Preprocessing the collected real-time vibration signals of the rolling bearing to obtain root mean square and envelope spectrum of the real-time vibration signals;
(2) According to the physical parameters of different rolling bearings, establishing a rolling bearing digital twin body;
(3) Utilizing the rolling bearing digital twin body simulation obtained in the step (2) to obtain the vibration response of the rolling bearing under the normal state and the fault state, and then simulating to obtain the fault width of the rolling bearing and the root mean square of vibration signals, and taking the root mean square as a sample set;
(4) Inputting the sample set obtained in the step (3) into a MATLAB regression learner to select a wide neural network for training, so as to obtain a wide neural network model;
(5) Performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the wide neural network model obtained in the step (4) to obtain a fault state of the rolling bearing to be diagnosed;
updating the supporting rigidity of the rolling bearing digital twin body obtained in the step (2) comprising the bearing outer ring and the bearing seat in real timek h And the physical parameters of the bearing damping c until the difference between the amplitude of the main peak of the envelope spectrum of the vibration response obtained by the digital twin body simulation of the rolling bearing after the real-time update and the amplitude of the main peak of the envelope spectrum of the real-time vibration signal in the step (1) is the minimum value, the steps are as follows:
(S-1) utilizing real-time updated rolling bearing digital twin body simulation to obtain vibration response of the rolling bearing under normal state and fault state, and then simulating to obtain the fault width of the rolling bearing and root mean square of vibration signals, and taking the root mean square as an updated sample set;
(S-2) inputting the updated sample set obtained in the step (S-1) into a MATLAB regression learner to select a wide neural network for training, so as to obtain an updated wide neural network model;
and (S-3) performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the updated wide neural network model obtained in the step (S-2) to obtain a fault state of the rolling bearing to be diagnosed.
Example 1
The bearing fault diagnosis method based on digital twin comprises the following steps of:
the vibration acceleration collectors are arranged in the x and y directions of the bearing seat, real-time vibration signals of the rolling bearing to be diagnosed, namely outer ring acceleration signals of the rolling bearing to be diagnosed, are collected, then band-pass filtering processing is carried out on the real-time vibration signals, low-frequency noise is removed, root mean square values of the vibration signals of the rolling bearing to be diagnosed are obtained through calculation, and an envelope spectrum is drawn through Hilbert transformation;
the rolling bearing to be diagnosed is marked as a bearing 1; the outer ring fault width of the bearing 1 is w=1 mm, the fault depth is s=1 mm, and the fault length is l=4 mm;
(2) According to the physical parameters of the rolling bearing 1, the supporting rigidity of the outer ring and the bearing seat of the bearing is includedk h And (3) bearing damping c two variable parameters, and establishing a rolling bearing digital twin body.
The establishment of the digital twin body of the rolling bearing is a key step of the digital twin-based bearing fault diagnosis method provided by the invention. The rolling bearing digital twin body can dynamically simulate the rolling bearing through a numerical calculation and simulation technology, and can simulate the abrasion and faults of the rolling bearing; the establishment of the rolling bearing digital twin body needs to take the actual working environment and the running condition of the bearing into consideration, so that the accurate rolling bearing digital twin body is established. The invention considers a plurality of coupling factors of the bearing system and accurately simulates the vibration response signals of the spalling fault bearing.
The step of establishing the rolling bearing digital twin body is as follows:
(1) selecting the model of the rolling bearing and obtaining different selected rollersSupport stiffness of bearing outer ring and bearing seat of dynamic bearingk h And a physical parameter of the bearing damping c;
(2) establishing a four-degree-of-freedom dynamic model of the bearing system;
dynamic modeling of a bearing system by using a spring-damping model, wherein the inner ring and the shaft of each rolling bearing are regarded as a whole, and the total mass of the inner ring and the shaft is m i A representation; the outer ring of the rolling bearing is fixed on the bearing seat, and the total mass of the rolling bearing is m o A representation; the method comprises the steps that a rigid and damped elastic support exists between a shaft and a bearing seat, the contact between a rolling body of a rolling bearing and an outer ring and an inner ring of a raceway also has the rigid and damped elastic support, the rolling body of the rolling bearing is also expressed by a spring-damping model, and a plane coordinate system, namely a vertical X direction and a horizontal Y direction, is established; based on the plane motion of the outer ring and the inner ring of the rolling bearing in the radial direction, 4 degrees of freedom are taken as a total, and a four-degree-of-freedom dynamic model of the bearing system is built.
A four-degree-of-freedom dynamics model diagram of the bearing system is shown in FIG. 2.
(3) Calculating stiffness and damping in a bearing system, the stiffness and damping in the bearing system comprising: the method comprises the following steps of supporting rigidity ks of a shaft and a bearing inner ring, supporting rigidity kh of a bearing seat and a bearing outer ring, contact rigidity k of a bearing, damping cs of the shaft and the bearing inner ring, damping ch of the bearing seat and the bearing outer ring and damping c of the bearing;
(4) establishing a peeling fault time-varying displacement excitation model and a timely displacement excitation function;
according to the Hertz contact theory, the peeling fault position of the rolling bearing can cause the contact deformation of the rolling body of the rolling bearing and the raceway of the rolling bearing to change, thereby causing the change of contact force and further inducing the abnormal vibration of the rolling bearing, and aiming at the time-varying displacement excitation generated by the peeling fault of the outer ring of the rolling bearing, the invention establishes a peeling fault time-varying displacement excitation model as shown in figure 3, wherein W is the width of the peeling defect in figure 3, d is the diameter of the rolling body, omega s And finally, establishing a time-varying displacement excitation function of the peeling fault of the outer ring of the bearing for the rotation angular velocity of the inner ring as follows:
in the method, in the process of the invention,H d for time-varying displacement excitation of the outer race spalling failure,H max for the maximum additional displacement created by the spalling defect,θ d for the range angle of the defective area,θ j is the firstjThe time-varying position angle of the individual rolling bodies,θ i and (5) peeling the initial angle of the defect for the outer ring raceway.
(5) Establishing a multi-degree-of-freedom dynamic differential equation;
according to the Newton's second law of motion and the established four-degree-of-freedom dynamics model of the bearing system, establishing a multi-degree-of-freedom dynamics differential equation;
the multi-degree-of-freedom dynamic differential equation comprises a rolling bearing inner ring dynamic differential equation; the formula of the rolling bearing inner ring kinematic differential equation is as follows:
in the method, in the process of the invention,m i for the total mass of the bearing inner race and the shaft,c s is the damping of the inner ring of the bearing and the shaft,k s for the bearing inner race and the supporting rigidity of the shaft,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、And->Respectively is provided with an inner ring atXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the outer ringXAnd (3) withYSpeed in the direction.
The multi-degree-of-freedom dynamic differential equation also comprises an outer ring dynamic differential equation of the rolling bearing; the formula of the outer ring kinematic differential equation of the rolling bearing is as follows:
in the method, in the process of the invention,m o is the total mass of the outer ring of the bearing and the bearing seat,c h is the damping of the outer ring of the bearing and the bearing seat,k h the support rigidity of the bearing outer ring and the bearing seat is adopted,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、/>And->Respectively is at the outer ringXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the inner ringXAnd (3) withYSpeed in the direction.
(6) Solving a multi-degree-of-freedom dynamic differential equation by using a fixed-step length four-order Dragon-Gregorian tower method to obtain the vibration response of the spalling fault rolling bearing, wherein the steps are as follows: two parameters in the differential equation of motion of the inner ring and the outer ring of the bearing system are determined: support stiffness of bearing outer race and bearing housingk h And solving the bearing damping c by using a fixed-step length four-order Dragon lattice tower method to obtain the vibration response of the spalling fault rolling bearing.
(3) And (3) obtaining vibration responses of the rolling bearing in the X/Y directions under the normal state and the fault state by utilizing the rolling bearing digital twin body simulation obtained in the step (2), and obtaining root mean square of fault width and vibration signals of 100 groups of rolling bearings 1 by simulation, wherein the root mean square is taken as a sample set, and the sample set is shown in a table 1.
The root mean square value is a characteristic parameter of the state of the rolling bearing, can reflect the total vibration amount of the rolling bearing, can effectively reflect the overall degradation degree of the bearing, and has good stability.
TABLE 1 sample set
Root mean square value | Fault width | Root mean square value | Fault width | Root mean square value | Fault width | Root mean square value | Fault width |
2.033 | 0.05 | 17.462 | 1.55 | 193.0418 | 3.05 | 699.9337 | 4.55 |
2.0331 | 0.1 | 19.5258 | 1.6 | 204.1839 | 3.1 | 722.9434 | 4.6 |
2.0331 | 0.15 | 21.77 | 1.65 | 215.7199 | 3.15 | 746.3937 | 4.65 |
2.0334 | 0.2 | 24.1992 | 1.7 | 227.664 | 3.2 | 770.2421 | 4.7 |
2.0341 | 0.25 | 26.8225 | 1.75 | 240.0209 | 3.25 | 794.502 | 4.75 |
2.0358 | 0.3 | 29.6524 | 1.8 | 252.7964 | 3.3 | 819.1366 | 4.8 |
2.0389 | 0.35 | 32.6926 | 1.85 | 265.973 | 3.35 | 844.1514 | 4.85 |
2.0448 | 0.4 | 35.9529 | 1.9 | 279.5611 | 3.4 | 869.4954 | 4.9 |
2.055 | 0.45 | 39.4457 | 1.95 | 293.5657 | 3.45 | 895.2141 | 4.95 |
2.0718 | 0.5 | 43.1804 | 2 | 307.9777 | 3.5 | 921.2707 | 5 |
2.099 | 0.55 | 47.1627 | 2.05 | 322.815 | 3.55 | - | - |
2.1431 | 0.6 | 51.4057 | 2.1 | 338.0434 | 3.6 | - | - |
2.2115 | 0.65 | 55.9228 | 2.15 | 353.6702 | 3.65 | - | - |
2.3138 | 0.7 | 60.7155 | 2.2 | 369.6992 | 3.7 | - | - |
2.4608 | 0.75 | 65.7993 | 2.25 | 386.1261 | 3.75 | - | - |
2.6636 | 0.8 | 71.1795 | 2.3 | 402.9098 | 3.8 | - | - |
2.9324 | 0.85 | 76.8653 | 2.35 | 420.0739 | 3.85 | - | - |
3.2766 | 0.9 | 82.8631 | 2.4 | 437.6185 | 3.9 | - | - |
3.7039 | 0.95 | 89.1825 | 2.45 | 455.5202 | 3.95 | - | - |
4.2196 | 1 | 95.8347 | 2.5 | 473.7925 | 4 | - | - |
4.8316 | 1.05 | 102.8204 | 2.55 | 492.4171 | 4.05 | - | - |
5.5434 | 1.1 | 110.1676 | 2.6 | 511.4159 | 4.1 | - | - |
6.3607 | 1.15 | 117.8605 | 2.65 | 530.7753 | 4.15 | - | - |
7.2902 | 1.2 | 125.9178 | 2.7 | 550.5174 | 4.2 | - | - |
8.3366 | 1.25 | 134.3514 | 2.75 | 570.6501 | 4.25 | - | - |
9.506 | 1.3 | 143.1698 | 2.8 | 591.1621 | 4.3 | - | - |
10.8066 | 1.35 | 152.3649 | 2.85 | 612.0776 | 4.35 | - | - |
12.246 | 1.4 | 161.9456 | 2.9 | 633.4188 | 4.4 | - | - |
13.8287 | 1.45 | 171.9222 | 2.95 | 655.1533 | 4.45 | - | - |
15.5646 | 1.5 | 182.2847 | 3 | 677.3341 | 4.5 | - | - |
(4) Inputting the sample set obtained in the step (3) into a MATLAB regression learner to select a wide neural network for training, and obtaining a trained wide neural network model by the following steps: dividing a sample set into a test set and a training set according to a preset proportion; the training set variables comprise root mean square values and fault widths of bearings, the fault widths are taken from 0.05mm to 5mm every 0.05mm, 100 groups of numerical values are taken, 100 groups of root mean square values are obtained through bearing digital twin body simulation of 100 groups of fault widths, the training set is input into a MATLAB regression learner to select a wide neural network for training, a trained wide neural network model is obtained, the root mean square error of the trained wide neural network is 0.076501 (the root mean square error is used for measuring the deviation between an observed value (true value) and a predicted value), and the error requirement is met. At the moment, inputting a root mean square value into the wide neural network model, and obtaining the effectiveness of the fault diagnosis effect of the wide neural network which is obtained by the corresponding fault width verification training through the test set;
a sample set with a preset proportion of preferably 20% is used as a test set, and a sample set with a preset proportion of preferably 80% is used as a training set;
(5) Performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by utilizing the optimal neural network obtained in the step (4), and obtaining that the fault width of the rolling bearing 1 to be diagnosed is 1.9455mm;
updating the supporting rigidity of the bearing outer ring and the bearing seat of the rolling bearing digital twin body obtained in the step (2) in real timek h The damping c parameter of the bearing until the difference value between the amplitude of the main peak of the envelope spectrum of the vibration response obtained by the digital twin body simulation of the rolling bearing after the real-time updating and the amplitude of the main peak of the envelope spectrum of the real-time vibration signal in the step (1) is the minimum value, the steps are as follows;
(the parameters of the rolling bearing digital twin body are changed in real time due to the influence of an actual physical model, the size of the rolling bearing defects in the physical space can influence the rolling bearing digital twin body of the bearing.) according to the size of the rolling bearing defects in the diagnosed physical space, the supporting rigidity of the bearing outer ring and the bearing seat of the rolling bearing digital twin body is updated and corrected in real time by introducing time-varying displacement excitation of the corresponding outer ring spalling faultk h And (3) dynamically updating the physical parameters of the bearing damping c in real time within a limited range, so as to change the amplitude of the main peak of the envelope spectrum of the vibration response obtained by the simulation of the real-time updated digital twin bodies of the rolling bearings, and stopping updating the parameters of the digital twin bodies of the rolling bearings obtained in the step (2) in real time when the difference between the amplitude of the main peak of the envelope spectrum of the real-time vibration signal obtained in the step (1) and the amplitude of the main peak of the envelope spectrum is the minimum (namely, the error of the amplitude of the main peak of the envelope spectrum is reduced).
Fig. 5 is a comparison chart of envelope spectrum images before parameter updating in the embodiment of the present invention, and fig. 6 is a comparison chart of envelope spectrum images after parameter updating in the embodiment of the present invention, as can be seen from fig. 5 and 6, by updating the supporting rigidity kh of the bearing outer ring and the bearing seat of the rolling bearing digital twin body and the bearing damping c in real time, the difference between the envelope spectrum main peak amplitude of the vibration response obtained by the simulation of the rolling bearing digital twin body after updating and the envelope spectrum main peak amplitude of the real-time vibration signal in the step (1) is the minimum, so as to ensure that the high-fidelity vibration response is obtained.
Repeating the steps (3) - (5) by using the real-time updated rolling bearing digital twin body, wherein the steps are as follows:
(S-1) obtaining vibration responses of the rolling bearings in the X/Y directions under the normal state and the fault state by utilizing digital twin body simulation of the rolling bearings after real-time updating, and obtaining the fault width and root mean square of vibration signals of the rolling bearings of 100 groups after updating by simulation, wherein the root mean square is taken as an updated sample set, and the updated sample set is shown in a table 2.
Table 2 updated sample set
Root mean square value | Fault width | Root mean square value | Fault width | Root mean square value | Fault width | Root mean square value | Fault width |
2.7468 | 0.05 | 39.1977 | 1.55 | 302.4507 | 3.05 | 835.8053 | 4.55 |
2.7469 | 0.1 | 42.9478 | 1.6 | 316.5044 | 3.1 | 856.0093 | 4.6 |
2.747 | 0.15 | 46.9506 | 1.65 | 330.9009 | 3.15 | 876.2606 | 4.65 |
2.7472 | 0.2 | 55.7411 | 1.7 | 345.609 | 3.2 | 896.548 | 4.7 |
2.7477 | 0.25 | 60.5529 | 1.75 | 360.632 | 3.25 | 916.8654 | 4.75 |
2.749 | 0.3 | 65.6465 | 1.8 | 375.9738 | 3.3 | 937.2041 | 4.8 |
2.7515 | 0.35 | 76.7321 | 1.85 | 391.6281 | 3.35 | 957.5571 | 4.85 |
2.7564 | 0.4 | 82.7354 | 1.9 | 407.5597 | 3.4 | 977.9127 | 4.9 |
2.7651 | 0.45 | 89.0534 | 1.95 | 423.7862 | 3.45 | 998.275 | 4.95 |
2.7798 | 0.5 | 89.0534 | 2 | 440.3036 | 3.5 | 1018.612 | 5 |
2.8043 | 0.55 | 95.6931 | 2.05 | 457.0895 | 3.55 | - | - |
2.8445 | 0.6 | 102.6636 | 2.1 | 474.1544 | 3.6 | - | - |
2.9077 | 0.65 | 109.9648 | 2.15 | 491.4757 | 3.65 | - | - |
3.0035 | 0.7 | 117.621 | 2.2 | 509.0587 | 3.7 | - | - |
3.1436 | 0.75 | 125.6132 | 2.25 | 526.8799 | 3.75 | - | - |
3.3404 | 0.8 | 133.9567 | 2.3 | 544.9439 | 3.8 | - | - |
3.6064 | 0.85 | 142.6596 | 2.35 | 563.2386 | 3.85 | - | - |
3.9535 | 0.9 | 151.7304 | 2.4 | 581.7401 | 3.9 | - | - |
4.392 | 0.95 | 161.1609 | 2.45 | 600.44 | 3.95 | - | - |
4.9293 | 1 | 170.9597 | 2.5 | 619.3319 | 4 | - | - |
5.5745 | 1.05 | 181.1262 | 2.55 | 638.379 | 4.05 | - | - |
7.2081 | 1.1 | 191.649 | 2.6 | 657.5966 | 4.1 | - | - |
9.34 | 1.15 | 202.5389 | 2.65 | 676.9565 | 4.15 | - | - |
10.6069 | 1.2 | 213.7789 | 2.7 | 696.4429 | 4.2 | - | - |
13.5794 | 1.25 | 225.3727 | 2.75 | 716.0636 | 4.25 | - | - |
17.1807 | 1.3 | 237.3298 | 2.8 | 735.7923 | 4.3 | - | - |
21.4749 | 1.35 | 249.6492 | 2.85 | 755.6282 | 4.35 | - | - |
23.9043 | 1.4 | 262.3306 | 2.9 | 775.5552 | 4.4 | - | - |
29.3646 | 1.45 | 275.3577 | 2.95 | 795.5712 | 4.45 | - | - |
32.4169 | 1.5 | 288.7328 | 3 | 815.6531 | 4.5 | - | - |
(S-2) inputting the updated sample set obtained in the step (S-1) into a MATLAB regression learner to select a wide neural network for training, wherein the steps are as follows: dividing the updated sample set into an updated test set and an updated training set according to a preset proportion; the updated training set variables comprise root mean square values and fault widths of bearings, the fault widths are taken once every 0.05mm from 0.05mm to 5mm, 100 groups of numerical values are taken, 100 groups of root mean square values are obtained through bearing digital twin body simulation of 100 groups of fault widths, the updated training set is input into an MATLAB regression learner to select a wide neural network for training, a new wide neural network model with root mean square error of 0.073510 is obtained after training, and error requirements are met; at this time, inputting a root mean square value into a new wide neural network model to obtain a corresponding fault width;
and (S-3) performing fault diagnosis on the root mean square of the vibration signal of the rolling bearing to be diagnosed by utilizing the new wide neural network model obtained in the step (S-2), so as to obtain the fault width of the rolling bearing 1 to be diagnosed of 1.5521mm, wherein the actual fault width of the known bearing 1 is 1mm, and compared with the fault width before parameter updating, the error is reduced by 0.3934mm.
Example 2
The rolling bearing to be diagnosed is marked as a bearing 2, and fault diagnosis is carried out on the bearing 2 according to the method of the embodiment 1; the outer ring failure width of the bearing 2 is w=2 mm, the depth of the failure is s=1 mm, and the length of the failure is l=4 mm.
Example 3
The rolling bearing to be diagnosed is marked as a bearing 3, and fault diagnosis is carried out on the bearing 3 according to the method of the embodiment 1; the outer ring failure width of the bearing 3 is w=3 mm, the depth of the failure is s=1 mm, and the length of the failure is l=4 mm.
Example 4
The rolling bearing to be diagnosed is marked as a bearing 4, and fault diagnosis is carried out on the bearing 4 according to the method of the embodiment 1; the outer ring failure width of the bearing 4 is w=2 mm, the depth of the failure is s=1 mm, and the length of the failure is l=4 mm.
Fig. 7 is a comparison chart of actual fault widths of the bearings 1 to 4 in the embodiments 1 to 4 and fault widths of the rolling bearings 1 to 4 diagnosed before and after updating parameters of the rolling bearing digital twin bodies in real time, as can be seen from fig. 7, after the parameters of the rolling bearing digital twin bodies are updated, the diagnosis errors of the rolling bearing digital twin bodies can be reduced.
In conclusion, the method provided by the invention effectively solves the problem that the bearing of the industrial equipment lacks historical fault data and cannot carry out fault diagnosis through the historical data information; the digital twin body parameters are updated in real time, so that high-fidelity vibration response is obtained, accurate identification of the fault width of the bearing under the variable speed working condition is realized, and reliability of a diagnosis result is effectively improved.
The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.
Claims (10)
1. The bearing fault diagnosis method based on digital twin is characterized by comprising the following steps of:
(1) Preprocessing the acquired real-time vibration signals of the rolling bearing to be diagnosed to obtain root mean square and envelope spectrum of the real-time vibration signals;
(2) According to the physical parameters of different rolling bearings, including the supporting rigidity of the outer ring and the bearing seatk h The method comprises the steps of (1) establishing a rolling bearing digital twin body by two variable parameters of bearing damping c;
(3) Utilizing the rolling bearing digital twin body simulation obtained in the step (2) to obtain the vibration response of the rolling bearing under the normal state and the fault state, and then simulating to obtain the fault width of the rolling bearing and the root mean square of vibration signals, and taking the root mean square as a sample set;
(4) Inputting the sample set obtained in the step (3) into an MATLAB regression learner to select a wide neural network for training, and obtaining a trained wide neural network model;
(5) Performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the wide neural network model obtained in the step (4) to obtain a fault state of the rolling bearing to be diagnosed;
updating the supporting rigidity of the rolling bearing digital twin body obtained in the step (2) comprising the bearing outer ring and the bearing seat in real timek h And the physical parameters of the bearing damping c until the difference between the amplitude of the main peak of the envelope spectrum of the vibration response obtained by the digital twin body simulation of the rolling bearing after the real-time updating and the amplitude of the main peak of the envelope spectrum of the real-time vibration signal obtained in the step (1) is the minimum value;
repeating the steps (3) - (5) by using the rolling bearing digital twin body updated in real time.
2. The digital twin-based bearing fault diagnosis method according to claim 1, wherein the real-time vibration signal of the rolling bearing to be diagnosed in step (1) is acquired by a vibration acceleration sensor; the pretreatment comprises the following steps: and carrying out band-pass filtering treatment on the real-time vibration signal of the rolling bearing, removing low-frequency noise, calculating to obtain a root mean square value of the vibration signal of the rolling bearing to be diagnosed, and drawing an envelope spectrum through Hilbert transformation.
3. The digital twin based bearing fault diagnosis method according to claim 1, wherein the step (2) of establishing a rolling bearing digital twin body comprises the steps of: selecting the model of the rolling bearing, and obtaining different selected physical parameters of the rolling bearing, wherein the physical parameters comprise the supporting rigidity of the bearing outer ring and the bearing seatk h And bearing damping c; establishing a four-degree-of-freedom dynamic model of the bearing system; calculating the rigidity and damping of the bearing system; establishing a peeling fault time-varying displacement excitation model and a timely displacement excitation function; establishing a multi-degree-of-freedom dynamic differential equation; and solving a multi-degree-of-freedom dynamic differential equation by using a fixed-step length four-order Dragon-Gregory tower method to obtain the vibration response of the rolling bearing with the peeling fault.
4. A digital twin based bearing fault diagnosis method according to claim 3, characterised in that the four degrees of freedom dynamics model of the bearing system is built comprising the steps of: the inner ring and the shaft of each rolling bearing are regarded as a whole, and the total mass of the inner ring and the shaft is m i A representation; the outer ring of the rolling bearing is fixed on the bearing seat, and the total mass of the rolling bearing is m o A representation; the method comprises the steps that a rigid and damped elastic support exists between a shaft and a bearing seat, the contact between a rolling body of a rolling bearing and an outer ring and an inner ring of a raceway also has the rigid and damped elastic support, the rolling body of the rolling bearing is also expressed by a spring-damping model, and a plane coordinate system, namely a vertical X direction and a horizontal Y direction, is established; based on the plane motion of the outer ring and the inner ring of the rolling bearing in the radial direction, 4 degrees of freedom are taken as a total, and a four-degree-of-freedom dynamic model of the bearing system is built.
5. A digital twin based bearing fault diagnosis method according to claim 3, wherein the multiple degree of freedom dynamic differential equation comprises an inner ring dynamic differential equation of a rolling bearing;
the formula of the inner ring kinematic differential equation of the rolling bearing is as follows:
in the method, in the process of the invention,m i for the total mass of the bearing inner race and the shaft,c s is the damping of the inner ring of the bearing and the shaft,k s for the bearing inner race and the supporting rigidity of the shaft,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、And->Respectively is provided with an inner ring atXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the outer ringXAnd (3) withYSpeed in the direction.
6. The digital twin based bearing fault diagnosis method according to claim 3 or 4, wherein the multiple degree of freedom dynamic differential equation further comprises an outer ring dynamic differential equation of the rolling bearing;
the formula of the outer ring kinematic differential equation of the rolling bearing is as follows:
in the method, in the process of the invention,m o is the total mass of the outer ring of the bearing and the bearing seat,c h is the damping of the outer ring of the bearing and the bearing seat,k h the support rigidity of the bearing outer ring and the bearing seat is adopted,cfor the damping of the bearing,N b for the number of rolling bodies,β j for the judgment coefficient of whether the rolling bodies generate contact force with the roller path,F kj for the contact force between the rolling bodies and the raceway,θ j is the firstjThe time-varying position angle of the individual rolling bodies,and->、/>And->、/>And->Respectively is at the outer ringXAnd (3) withYVibration acceleration, speed and displacement in direction, +.>And->Is arranged at the inner ringXAnd (3) withYSpeed in the direction.
7. The digital twin-based bearing fault diagnosis method according to claim 1, wherein in the step (4), the sample set is input into a MATLAB regression learner to select a wide neural network for training, and a trained wide neural network model is obtained, and the method comprises the following steps: dividing a sample set into a training set and a testing set according to a preset proportion; inputting the training set into a MATLAB regression learner for training to obtain a trained wide neural network model; the test set is used for verifying the effectiveness of the fault diagnosis effect of the trained wide neural network.
8. The digital twin based bearing fault diagnosis method according to claim 7, wherein the validation of the fault diagnosis effect of the trained wide neural network using the test set comprises the steps of: and inputting the test set into a wide neural network model for training, and verifying the effectiveness of the fault diagnosis effect of the trained wide neural network.
9. The digital twin based bearing fault diagnosis method according to claim 1, wherein the step (5) is repeated by using the digital twin bodies of the rolling bearing updated in real time, and the method comprises the following steps:
(S-1) utilizing real-time updated rolling bearing digital twin body simulation to obtain vibration response of the rolling bearing under normal state and fault state, and then simulating to obtain the fault width of the rolling bearing and root mean square of vibration signals, and taking the root mean square as an updated sample set;
(S-2) inputting the updated sample set obtained in the step (S-1) into a regression learner to select a wide neural network for training, so as to obtain an updated wide neural network;
and (S-3) performing fault diagnosis on root mean square of vibration signals of the rolling bearing to be diagnosed by using the updated wide neural network obtained in the step (S-2) to obtain a fault state of the rolling bearing to be diagnosed.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the digital twinning-based bearing fault diagnosis method according to any one of claims 1 to 9 when executing the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310472921.7A CN116183231B (en) | 2023-04-28 | 2023-04-28 | Bearing fault diagnosis method based on digital twin |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310472921.7A CN116183231B (en) | 2023-04-28 | 2023-04-28 | Bearing fault diagnosis method based on digital twin |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116183231A CN116183231A (en) | 2023-05-30 |
CN116183231B true CN116183231B (en) | 2023-07-11 |
Family
ID=86446615
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310472921.7A Active CN116183231B (en) | 2023-04-28 | 2023-04-28 | Bearing fault diagnosis method based on digital twin |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116183231B (en) |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FI112972B (en) * | 1998-07-15 | 2004-02-13 | Abb Research Ltd | Assessment of bearing condition |
WO2019061006A1 (en) * | 2017-09-26 | 2019-04-04 | Schaeffler Technologies AG & Co. KG | Bearing failure diagnosis method and device, readable storage medium, and electronic device |
CN110991004A (en) * | 2019-11-06 | 2020-04-10 | 安徽大学 | Bearing fault diagnosis method, terminal equipment and computer storage medium |
KR102344852B1 (en) * | 2019-11-20 | 2021-12-28 | 한국수력원자력 주식회사 | Digital Twin-based Prediction and Diagnostic Device for Pump Bearing Systems |
EP4160337A1 (en) * | 2021-09-30 | 2023-04-05 | Siemens Aktiengesellschaft | System and method for estimating remaining useful life of a bearing |
CN114492511A (en) * | 2021-12-31 | 2022-05-13 | 中云开源数据技术(上海)有限公司 | Fault diagnosis method based on digital twinning |
CN114755012A (en) * | 2022-04-08 | 2022-07-15 | 河南工业大学 | Virtual-real interactive sliding bearing test monitoring system based on digital twinning |
CN115563853A (en) * | 2022-08-30 | 2023-01-03 | 武汉理工大学 | Rolling bearing fault diagnosis method based on digital twinning |
CN115700363B (en) * | 2022-11-07 | 2023-08-08 | 南京工业大学 | Method and system for diagnosing faults of rolling bearing of coal mining machine, electronic equipment and storage medium |
-
2023
- 2023-04-28 CN CN202310472921.7A patent/CN116183231B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN116183231A (en) | 2023-05-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Cui et al. | HVSRMS localization formula and localization law: Localization diagnosis of a ball bearing outer ring fault | |
Cui et al. | Fault severity classification and size estimation for ball bearings based on vibration mechanism | |
CN112487584A (en) | Dynamics-based rolling bearing digital twin modeling method | |
CN107704695B (en) | Full-size quantitative diagnosis method for defects of outer ring of rolling bearing | |
Reddy et al. | Application of artificial neural networks for identification of unbalance and looseness in rotor bearing systems | |
Huang et al. | Novel morphological scale difference filter with application in localization diagnosis of outer raceway defect in rolling bearings | |
CN113496061A (en) | SOM network clustering electromechanical device bearing fault analysis method based on transfer learning and manifold distance | |
CN106650146A (en) | Online quantitative recognition analysis method for loose fault of bearing support of rotor system | |
CN115563853A (en) | Rolling bearing fault diagnosis method based on digital twinning | |
Espinoza-Sepulveda et al. | Theoretical validation of earlier developed experimental rotor faults diagnosis model | |
CN113916535B (en) | Bearing diagnosis method, system, equipment and medium based on time frequency and CNN | |
Doguer et al. | Vibration analysis using time domain methods for the detection of small roller bearing defects | |
CN115023598A (en) | Method for determining the remaining useful life of a bearing | |
CN116183231B (en) | Bearing fault diagnosis method based on digital twin | |
CN109684711A (en) | A kind of turboshaft engine is pneumatically connected more rotor coupled vibration analysis methods | |
CN117034669A (en) | Method and system for predicting residual life of rolling bearing of rotary machine | |
CN116070090A (en) | Rolling bearing fault sample generation method based on digital twin model | |
Hamadache et al. | Improving signal-to-noise ratio (SNR) for inchoate fault detection based on principal component analysis (PCA) | |
CN115270342A (en) | Multi-degree-of-freedom dynamic modeling method for deep groove ball bearing with spalling fault | |
CN115809595A (en) | Digital twin model construction method reflecting rolling bearing defect expansion | |
JP2015031626A (en) | Device for monitoring state of rolling bearing | |
Kadarno et al. | Vibration analysis of defected ball bearing using finite element model simulation | |
Mustafa et al. | Experimental research on machinery fault simulator (MFS): A review | |
Jamadar et al. | Dimensional amplitude response analysis of vibrations produced by defects in rolling contact bearings | |
CN114997046B (en) | Dynamic simulation guided domain countermeasure bearing fault diagnosis method |
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 |