Rolling bearing full-life state monitoring method based on digital twinning
Technical Field
The invention relates to the technical field of bearing health monitoring, in particular to a digital twin-based rolling bearing full-life state monitoring method.
Background
As a new technology serving for fault prediction and health management, a digital twin can realize system health management, fault diagnosis and intelligent maintenance, which is firstly proposed by the university of Michigan in America in a product life cycle management course and then is used as a key technology of intelligent manufacturing, the digital twin is widely concerned by students and enterprises at home and abroad and is developed and practiced, Shangguand proposes a satellite system fault diagnosis and health monitoring method based on the digital twin, which utilizes signal processing and data mining to obtain implicit information of a system operation state, provides a reasonable fault diagnosis algorithm and maintenance strategy on the basis of data fusion, Tao constructs a five-dimensional digital twin model commonly used for complex equipment, can effectively carry out data interaction fault diagnosis and maintenance design, and GrievesM proposes how to synchronously connect physical product data with information contained in a virtual product, the method has the advantages that a digital factory is simulated by using the information, how products are manufactured is predicted, long-term reliable operation of product equipment is ensured, Zhuang provides an intelligent production management and control framework based on digital twins, information acquisition of a physical assembly shop, construction of a data twins assembly model and provision of prediction service and management for a satellite assembly shop by the built model are achieved, Benjamin provides a comprehensive reference model and outlines application of the reference model in geometric transformation management, a concept framework is provided, industrial engineering design is enriched, Tuegel EJ utilizes the digital twins model to calculate local damage and material state evolution of an airplane, and therefore structural integrity of the airplane is guaranteed by predicting the structural life of the airplane;
the bearing is taken as a key part in mechanical equipment, the running state of a rolling bearing has important influence on the service life and the whole performance of complex machinery and mechanisms, the bearing dynamics research is usually inclined to establish a simulation model in a virtual space, the whole-service-life bearing is rarely analyzed, the interaction research of the virtual space and a physical space is lacked, Gupta researches the motion of a retainer and the performance of the bearing based on the relative position relation among all parts, on the basis of the model proposed by Gupta, Cao proposes a dynamic model of local defect vibration of a cylindrical roller bearing and researches the vibration response of a single-defect, multi-defect and composite fault bearing, Ahmadi AM overcomes the limitation that rolling elements are taken as point mass, the fault defect shape is taken as a rectangular pit in consideration of the influence of Hertz contact force and damping, the method improves the accuracy of predicting low-frequency and high-frequency events, liu breaks away from the limitation of a traditional rectangular fault defect model and researches a dynamic model considering the coupling of time-varying displacement excitation and time-varying contact stiffness excitation of lubrication traction, Kulkarni PG provides a dynamic model for predicting the vibration behavior of a bearing under the influence of local defects and simulates the influence of radial load, defect size, position and the like on a bearing time domain and a bearing frequency domain, however, the above article is only limited to the dynamic analysis of a normal bearing or a fault bearing in a virtual space, and the bidirectional connection between a real space and a virtual space in a full life cycle is not established.
Disclosure of Invention
The invention aims to provide a rolling bearing full-life state monitoring method based on digital twinning, so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme: a rolling bearing full-life state monitoring method based on digital twinning comprises the following steps: step one, signal acquisition; step two, establishing a virtual dynamic model; step three, mapping and matching; step four, parameter prediction; step five, simulating signals;
in the first step, collecting physical environment full-life vibration acceleration data for later use through a bearing test bed in a system level;
in the second step, after the data collected in the first step, a virtual dynamic model of the rolling bearing is established by considering time-varying fault displacement excitation and lubrication;
in the third step, a segmentation searching method according to the change of the fault evolution speed is provided to search for matched fault dynamic response, and the relation between the vibration signal and the defect size and the relation between the vibration signal and the comprehensive equivalent stiffness are revealed by the vibration data in the first step and the vibration data generated by the virtual model in the second step through data connection matching in the unit level, so that a bearing digital twin model is obtained;
in the fourth step, the bearing fault width change and the comprehensive equivalent stiffness change in the degradation stage obtained in the third step are matched to obtain an evolution rule of a full life cycle through a radial basis function neural network;
in the fifth step, after the life cycle evolution rule is obtained from the fourth step, the predicted defect displacement excitation and stiffness evolution trend is input into a rolling bearing life cycle dynamic model to obtain a life cycle vibration acceleration signal, the obtained data is mapped into corresponding data of a physical space, then a bearing digital twin model is obtained according to the establishment, and the model is used for detecting the life cycle state of the bearing.
Preferably, in the second step, in order to study the dynamic characteristics of the rolling bearing, assuming that the outer ring of the bearing is a rigid body and is fixedly supported, the inner ring is fixedly connected with the rotating shaft and rotates counterclockwise, considering the slip phenomenon between the rolling body and the inner and outer rings, considering the bearing lubrication influence, neglecting the influence of the inertia force of the rolling body, considering the rolling body as nonlinear spring damping, and loading the external radial load to the bearing through the center of the rotating shaft and further loading the external radial load to the simplified spring damping system in the vertical direction.
Preferably, in the second step, as a deformation amount caused by the fault, when the bearing is in a healthy stage, the deformation amount is 0, and when the bearing is in a fault stage, a fault model is established to maximally meet the actual condition of the fault of the bearing, and the generated time-varying displacement excitation is specifically calculated as:
wherein theta isdIs a fault zone span angle, theta0To the fault starting angle, HdmaxTo be rolledThe maximum amount of deflection released by the body upon failure of the outer ring is as follows:
the rolling deformation amount is:
the deformation of the outer ring is as follows:
wherein W is the fault width, rbIs the roller radius, r0Is the outer raceway radius;
predicting the evolution defect through a rolling bearing full life cycle vibration signal and a radial basis function neural network, and obtaining the contact deformation of the ith rolling body as follows:
Ai=xsinθi+ycosθi-(Cr+hii+hoi+ H) equation 5
Wherein x is the horizontal displacement of the bearing, y is the vertical displacement of the bearing, CrIs the radial clearance of the bearing, hiiIs the thickness of a central oil film between the ith rolling body and the inner ring, hoiThe thickness of a central oil film between the ith rolling body and the outer ring is calculated as follows:
wherein U is a dimensionless speed parameter, G is a dimensionless material parameter, e is a load parameter, k is a load coefficient, RixIs the equivalent radius of the rolling body and the inner ring in the x direction, RoxThe equivalent radius of the rolling body and the outer ring in the x direction is shown.
Preferably, in the fourth step, the bearing fault width and the comprehensive equivalent stiffness in the degradation stage and the maximum value of the sample signal are input into the radial basis function neural network to obtain the evolution law of the bearing fault width and the comprehensive equivalent stiffness in the whole life cycle.
Preferably, in the fifth step, the dynamic model of the full life of the rolling bearing is as follows:
wherein m is the equivalent mass of the bearing, x is the horizontal displacement of the bearing, y is the vertical displacement of the bearing, c is the equivalent damping of the bearing,
and
axial and radial forces to which the bearing is subjected, F
HxAnd F
HyThe contact force components of the rolling elements and the inner and outer rings are respectively shown, and the calculation formula is as follows:
wherein KtTo combine equivalent stiffness, thetaiIs the angular position of the ith rolling element, n is the load deformation index, gammaiThe parameters for determining whether the ith roller contacts the raceway are defined as:
preferably, in the third step, the method for establishing the bearing digital twin model comprises the following steps:
1) inputting: degraded bearing vibration sample set at N moments
Where Δ t denotes the separation of two adjacent samples, a set of simulated signal samples
The maximum expected error beta, the group number k of each stage, the fault width evolution step delta and the comprehensive equivalent stiffness evolution step tau;
2) calculating initial fault width W of fault germination stage11Initial stiffness K at fault initiation11Initial failure width W of failure degradation stage21Initial stiffness K at the fault degradation stage21Initial fault width W at near failure stage of fault31Initial stiffness K at near failure stage of failure31;
3) Inputting the grouped defect widths and the comprehensive equivalent stiffness in the fault initiation stage, the fault degradation stage and the fault near failure stage into formulas 1-9 to obtain a group of numerical values of which the error is minimum by comparing the simulation signal peak value and the experimental acceleration signal peak value, and assuming that the real fault size and the comprehensive equivalent stiffness of the bearing at the moment are the group of numerical values.
Compared with the prior art, the invention has the beneficial effects that: the invention provides a digital twin model with high fidelity, high reliability and high precision characteristics, which can generate equivalent information of a virtual space and a physical entity, the model can capture and store the measurement information of a physical entity model of a bearing, can predict the fault size and the comprehensive equivalent stiffness of a bearing in the whole life cycle, further monitor the state of the bearing, perform information interaction on the physical characteristics and the degradation information of the entity model and a dynamic model of the virtual bearing, evaluate the change of the fault width and the change of the comprehensive equivalent stiffness of the bearing, realize the real-time update and the dynamic evolution of the production, management and operation and maintenance in the whole life cycle of the bearing, improve the precision and the efficiency of the fault prediction and the health management and improve the digitization degree of the whole life cycle of the whole mechanical equipment.
Drawings
FIG. 1 is an analytical model diagram of a rolling bearing of the present invention, wherein (a) is a structural diagram and (b) is a simplified diagram of a spring damping system;
FIG. 2 is a schematic enlarged view of a partial defect of the outer ring of the present invention;
FIG. 3 is an architectural diagram of a digital twin model of the present invention;
FIG. 4 is a block diagram of a neural network of the present invention;
FIG. 5 is a flow chart of the method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
Referring to fig. 1-5, an embodiment of the present invention is shown: a rolling bearing full-life state monitoring method based on digital twinning comprises the following steps: step one, signal acquisition; step two, establishing a virtual dynamic model; step three, mapping and matching; step four, parameter prediction; step five, simulating signals;
in the first step, collecting physical environment full-life vibration acceleration data for later use through a bearing test bed in a system level;
wherein in the second step, after the data collected in the first step, a virtual dynamic model of the rolling bearing is established by considering time-varying fault displacement excitation and lubrication, in order to research the dynamic characteristics of the rolling bearing, an outer ring of the bearing is assumed to be a rigid body and fixedly supported, an inner ring is fixedly connected with a rotating shaft and rotates anticlockwise, the phenomenon of slipping between the rolling body and the inner ring and the outer ring are considered, the lubricating effect of the bearing is considered, the influence of the inertia force of the rolling body is ignored, the rolling body is considered to be nonlinear spring damping, an external radial load is loaded on the bearing through the center of the rotating shaft and further loaded to a simplified spring damping system in the vertical direction to serve as the deformation caused by faults, when the bearing is in a healthy stage, the deformation is 0, when the bearing is in a fault stage, a fault model is established to be maximally consistent with the actual condition of the bearing fault, and the generated time-varying displacement excitation is specifically calculated as follows:
wherein theta isdIs a fault zone span angle, theta0For the starting angle of the fault, HdmaxThe maximum deformation of the rolling elements released by the outer ring failure is as follows:
the rolling deformation amount is:
the deformation of the outer ring is as follows:
wherein W is the fault width, rbIs the roller radius, r0Is the radius of the outer raceway;
predicting the evolution defect through a rolling bearing full life cycle vibration signal and a radial basis function neural network, and obtaining the contact deformation of the ith rolling body as follows:
Ai=xsinθi+ycosθi-(Cr+hii+hoi+ H) equation 5
Wherein x is the horizontal displacement of the bearing, y is the vertical displacement of the bearing, CrIs the radial clearance of the bearing, hiiIs the thickness of a central oil film between the ith rolling body and the inner ring, hoiThe thickness of a central oil film between the ith rolling body and the outer ring is calculated as follows:
wherein U is a dimensionless speed parameter, G is a dimensionless material parameter, e is a load parameter, k is a load coefficient, RixIs the equivalent radius of the rolling body and the inner ring in the x direction, RoxThe equivalent radius of the rolling body and the outer ring in the x direction;
in the third step, a segmented searching method according to the change of the fault evolution speed is provided to find out the matched fault dynamic response, the vibration data in the first step and the vibration data generated by the virtual model in the second step are connected and matched in a unit layer through data to reveal the relation between the vibration signal and the defect size and the relation between the vibration signal and the comprehensive equivalent stiffness, so as to obtain a bearing digital twin model, and the establishment method of the bearing digital twin model comprises the following steps:
1) inputting: degraded bearing vibration sample set at N moments
Where Δ t denotes the separation of two adjacent samples, a set of simulated signal samples
The maximum expected error beta, the group number k of each stage, the fault width evolution step delta and the comprehensive equivalent stiffness evolution step tau;
2) calculating initial fault width W of fault germination stage11Initial stiffness K at fault initiation11Initial failure width W of failure degradation stage21Initial stiffness K at the fault degradation stage21Initial fault width W at near failure stage of the fault31Initial stiffness K at near failure stage of failure31;
3) Inputting the grouped defect widths and the comprehensive equivalent stiffness in the fault initiation stage, the fault degradation stage and the fault near failure stage into formulas 1-9 to obtain a group of numerical values with the minimum error between the simulation signal peak value and the experimental acceleration signal peak value, and assuming that the real fault size and the comprehensive equivalent stiffness of the bearing at the moment are the group of numerical values;
in the fourth step, the bearing fault width change and the comprehensive equivalent stiffness change in the degradation stage obtained by matching in the third step are used for obtaining an evolution rule of the full life cycle through the radial basis function neural network, and the bearing fault width, the comprehensive equivalent stiffness and the maximum value of the sample signal in the degradation stage are input into the radial basis function neural network to obtain the bearing fault width and the comprehensive equivalent stiffness evolution rule under the full life cycle;
in the fifth step, after obtaining the life cycle evolution rule from the fourth step, inputting the predicted defect displacement excitation and stiffness evolution trend into a rolling bearing life cycle dynamic model to obtain a life cycle vibration acceleration signal, mapping the obtained data into corresponding data of a physical space, then obtaining a bearing digital twin model according to establishment, detecting the life cycle state of the bearing by using the model, wherein the life cycle evolution rule of the rolling bearing is as follows:
wherein m is the equivalent mass of the bearing, x is the horizontal displacement of the bearing, y is the vertical displacement of the bearing, c is the equivalent damping of the bearing,
and
axial and radial forces to which the bearing is subjected, F
HxAnd F
HyThe contact force components of the rolling elements and the inner and outer rings are respectively calculated according to the following formula:
wherein KtTo combine equivalent stiffness, thetaiIs the angular position of the ith rolling element, n is the load deformation index, gammaiThe parameters for determining whether the ith roller contacts the raceway are defined as:
based on the above, the invention has the advantages that the invention provides a digital twin model with high fidelity, high reliability and high precision characteristics, which can generate equivalent information of a virtual space and a physical entity, the model can capture and store the measurement information of the physical entity model of the bearing, can predict the fault size and the comprehensive equivalent stiffness of the bearing in the whole life cycle, further monitor the state of the bearing, perform information interaction on the physical characteristics and the degradation information of the entity model and the dynamic model of the virtual bearing, evaluate the fault width change and the comprehensive equivalent stiffness change of the bearing, realize the real-time update and the dynamic evolution of the production, management and operation and maintenance in the whole life cycle of the bearing, improve the precision and the efficiency of fault prediction and health management, and improve the digitization degree of the whole life cycle of the whole mechanical equipment.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.