News Center

Picture Name

Shanghai Shang Shaft Bearing Quality Inspection Institute Co., Ltd.

Picture Name

Address: 301 Shangxue Road, Jiading District, Shanghai

Research on Fault and Life of Large Rolling Bearing

2021-09-30 13:47

The large size of large rolling bearings brings a series of problems in manufacturing, installation, operation and maintenance:Lack of systematic design and manufacturing technology, design and manufacturing accuracy is not easy to guarantee, heat treatment is difficult, lubrication quality is not easy to guarantee, installation accuracy is not easy to control, sealing trapping, etc. Under the action of complex working conditions such as heavy load, partial load and variable load, the local bearing strength will significantly increase and change frequently. The uneven temperature field and lubrication flow field of the bearing will further worsen the working state of the bearing, resulting in premature local wear, damage and failure..

The definition of rolling bearing life can be dividedfor 3 stages:

1) Fatigue life, refers to the bearing at a given cyclic loadconditions to the occurrence of fatigue spalling, usually according to the S-The N curve is obtained;


2) Residual fatigue life, refers to the bearing in a given cycle.After a period of cumulative operation under load conditions, continue under that condition.Time from operation to peeling;



3) Residual fault life, refers to the bearing in realContinued operation after a fault occurs for a period of cumulative operation under international operating conditionsTime to failure. The failure form of large heavy-duty bearings has a specialIn many cases, failure leads to end of lifeIs the conventional fatigue life problem, that is, the fault residual life index is more.For important.


Typical Failure Modes


At present, the common engineeringWind turbine turntable bearingThe failure modes are mainlyBroken ferrule,Cage fracture,Cage worn,Grease leakage of sealing ringWaitMain bearing of roadheaderThe main faults areRupture of gear ring and seal failureWait

railway wagon bearingThere are two main types of failure: 1) due to the periodic effect of extrusion and friction load on the bearing, the tissue below the raceway surface fatigue and produce micro cracks, with the gradual expansion of micro cracks finally producepeeling from the inside out2) becauseOverheating due to poor sealing and deterioration of lubricationFailure (Figure 9), accounting for about 72% of the total failure, which is the key reason for the shortening of the actual service life of railway freight car bearings, and even lead to major accidents such as shaft cutting when the failure is aggravated.

In addition, the axial cracks on the raceway and rolling body surfaces and the subsurface ofwhite etching crack(White Etching Crack , WEC) is considered to be the root cause of premature bearing failure, wind power gearbox bearings often have such failures, the research in this area has been bearing industry, engineering and academia and other fields of researchers attach great importance.Regardless of the manner of expression, the bearing'sEarly failure always follows the short board principle:WhenPoor lubrication or contactThe surface is not smoothWhen the contact surface is short, the bearing failure is often caused by the table.Surface wear, etc.; when the lubrication is good, the secondary surface is short board, shaftFailure is often caused by material defects or stress concentrations at the secondary surface.caused. Large size bearings are prone to large local deformation, guide.This leads to severe wear and local stress concentrations, which in turn leads to bearing failure.barrier, the occurrence of early failure.


Research Status of Bearing Life

2.1 Foreign Related Research

Foreign research on the fatigue life of rolling bearings began earlier,In particular, SKF,FAG and NSK and other large bearing enterprises andNASA and other scientific research institutions, has formed a set of relatively perfect theoryOn. at present,L-P are commonly used rolling bearing fatigue life theorylife theory, I-H life theory, Zaretsky life theory.


1947 to 1952, literature [70] based on Weibull's theory proposedThe L-P life theory, through the continuous revision of production practice, the most.Finally adopted by ANSI/ABMA/ISO and other standards. 1985 [71]On the basis of L-P life theory, the ultimate fatigue shearThe concept of shear stress is introduced into the study of fatigue life of rolling bearings,A new theory of rolling bearing fatigue life is proposed, andIn 1990, through the study of turbine engine main shaft ball bearings, the finalThe I-H life theory was established.Literature [73]By scrollingA large number of studies of bearing materials and rolling body fatigue life data, divided.Consider the stress-life coefficient and the Weibull coefficient, re-establish.bearing failure probability function and proposed a new bearing fatigue lifetheory of life. In theoretical analysis, L-P life theory and I-Hlife-span theories are all based on the Weibull distribution, while Zaretsky, on the other hand, re-established the Weibull distribution function. 2012, Literature [74]Ignore bearing support and ring deformation on rolling body uploadThe static strength theory is introduced into the bearing fatigue life.model, the fatigue life of large slewing bearings is obtained, and the ISO junctionThe comparison of the results verifies the accuracy of the model. 2016, Literature [75]Based on the NREL design guide for wind turbine pitch,Yaw bearing standard, 2.5 MW wind calculation using Romax softwareThe life of the motor unit pitch and yaw bearings, the results show that the useThe NREL guide calculates the life value than using the Romax software.The value is more conservative. Although the NREL design guidelines and Romax analysisThere are some differences in the results, but since the NREL design guidelines haveAdvantages of simplicity and conservative output characteristics for pitchingInitial design of bearings and yaw bearings.


As can be seen from the above theory,between interacting partsThe contact stress-strain characteristics affect the fatigue failure mode of the bearing.Main reason. Therefore, scholars have carried out based on elastic dynamics.Theoretical fatigue failure mechanism study:Literature [76] based on projectile flowThe theory of force lubrication first obtains three sub-surfaces of two elastic bodies in line contact.Principal stress and shear stress in the coordinate direction, positive pressure on the contact surfacemechanical relation to friction; literature [77] study found that the subsurfaceThe maximum shear stress gradually moves to the surface with fatigue accumulation, dueStress set due to installation damage, surface inclusions and lubrication contaminationIn, the crack that is initiated at the maximum stress of the subsurface is accelerated to the table.surface expansion, the formation of pockmarks, spalling and other faults; literature [78] based on projectileThe theory of sex dynamics further optimizes the design of the fatigue of ordinary; the literature [79] established the sphere and the channel based on the elastic-plastic theoryThe contact finite element model is obtained for the first time.Force distribution; literature [80] for large three-row cylindrical roller bearings, examinationConsidering the geometric parameters of the bearing, the internal dynamic load of the bearing is stress distribution, the contact fatigue failure of the raceway as the shaft.The criterion of life.


InBased on the analysis of bearing failure mechanism,ScholarWeFurther completed by considering the impact of specific factors on bearing life.Good bearing life model; literature [81] introducing a Archard wear modeland Lemaitre damage models, combined with accelerated life testingThe use of a small number of material characterization tests for bearing life prediction is proposed.reliable method; literature [82] studies the process of precision hard turning, grindingThe effect of the surface integrity of the machined bearing ring on the fatigue life.response; literature [83] based on the creep mechanism of the surface damage integral, the axisAbnormal heat generation of bearings caused by local friction is introduced into bearing life.calculation; literature [84] based on experimental analysis of high load and vibrationThe change of wear signal and friction torque under the condition of swing motion.The degradation model that can characterize the residual life of the bearing is proposed.Xian [85] proposed a method to simultaneously consider operating condition parameters and condition monitoringWeibull Accelerated Failure Time Regression (WAFTR) Model of SignalsBearing life prediction; literature [86] proposes to use raceway surface rawA new method for calculating the basic life of rolling contact with the core of the probability of survival.Law.


2.2 domestic related research

The relevant research in China mainly considers different factors on largeThe fatigue life model of rolling bearings is revised.: Literature [87] BaseThe fatigue life of large rolling bearings is analyzed in statistical theory.its reliability, the results show thatFatigue life of large rolling bearings is subjectWeibull distribution with three parameters, for its life and reliability evaluation.for theoretical guidance; literature [88] considering bearing lubrication and clearance changes,The theory of bearing life calculation is revised; literature [89] introducesThe yaw, pitch bearing in the axial force, radial force and overturning moment total.Under the same effect, the calculation is considered.Yaw, pitch bearing,Fatigue life of outer ringThe overall fatigue life of the bearing; literature [90] needleThe stress characteristics of double row four-point contact ball bearings are considered.Bearing stress distribution of axial load, radial load and overturning momentThe calculation method is based on the L-P life theory to calculate the equivalent dynamic load.and bearing life under dynamic load rating.


Research Status of Bearing Condition Evaluation and Residual Life Prediction


Bearing condition evaluation and residual life prediction are carried out bearing vision.maintenance, timely detection of problems, reduce economic losses of the mainway.


3.1 state evaluation

Based onBearing fault diagnosis and prediction of vibration signal is to judgeCommon methods of bearing degradation state: Literature [4] on Bearing Fault LettersThe fast Fourier transformation of the intrinsic mode function of the number is obtained.intended diagnostic effect; literature [5] based on multi-domain features, improved distanceseparation discriminant technique and improved fuzzy adaptive resonance theory mapping (A-daptive Resonance Theory Map,ARTMAP)提出了一A New Intelligent Fault Diagnosis Method for Bearing with Robustness and Stabilitymethod; literature [6] utilized principal component analysis (Principle Compo-nent Analysis,PCA) and artificial neural networks (ArtificialNeural Network,ANN) combined with the rescaled range method and de-The trend fluctuation analysis method is 3 under different speed and external load conditions.The results show that the four types of bearing faults are diagnosed.combination of methods can effectively carry out fault monitoring and diagnosis.


However, the bearing failures involved in the above results areartificialPrefabrication failure, and bearing actual operationappeared in the processFault notcharacterIt is impossible to effectively divide the running state of the bearing. Therefore, it is necessaryThe bearing parts in the running-in process should be analyzed to obtain the trueReal bearing running state. The literature [91] in the study of the working subsurface shapeOn the basis of the direct analysis of the appearance, the rolling bearing damage degradation process.Divided into 3 stages of running-in, smooth and unsteady; literature [92-93]FurtherThe damage degradation process is divided into running, smooth, and damage.5 stages of life, damage evolution, and damage development, and the existing research.The research and test work are summarized and combed for the failure and life.relationship analysis puts forward new ideas, and also points out the reasons in the actual process.The form of the barrier signal is different from the simulation results and needs to be different for different grinding.The loss evolution process establishes specific fault diagnosis, evaluation and prediction methods.Law.


3.2  residual life prediction

The existing bearing life prediction is dividedbased on physical modelandBaseData-drivenThe 2 class methods. Physical model-based approachThe bearing failure mechanism is analyzed by elastic fracture mechanics and fatigue crack expansion theory, the bearing life model is established, the effective residual life of the bearing is predicted by the model, and the prediction accuracy is improved by correcting the model parameters.Document [94-95] first proposed a bearing fatigue crack growth model, and document [96] extended the model from one dimension to two dimensions. Document [97] pointed out through analysis that the traditional crack growth model could not meet the life calculation of bearings with fragmentation problems. Document [98] proposed a bearing residual life prediction method based on crack growth considering nonlinear random problems. In addition, the literature [99] proposed a contact fatigue failure life prediction method for bearings with fragmentation initiation. The above-mentioned physical model-based bearing residual life prediction is obtained by the life design criteria based on material fatigue damage, and does not take into account the loss of life caused by bearing failure and its degradation, until failure, and the corresponding fault residual life.


Bearing life prediction method based on artificial intelligence and statistics belongs.Data-driven approach. among them,The artificial intelligence method uses the optimization combination of intelligent algorithms to fit the obtained data, obtains the evolution law of the bearing life process, and predicts the bearing life by setting a reasonable threshold.. Literature [100] proposes a bearing residual life prediction method based on support vector machine (SupportVectorMa-chine ,SVM), which trains SVM using simulation and experimental data respectively, and the results show that the trained SVM can predict bearing failure time; Literature [101] combines correlation vector machine with logistic regression to propose a new prediction methodEvaluate failure degradation and remaining life, and verify the prediction effect using simulation and test data of a single set of bearings. Neural network is an intelligent algorithm widely used in prediction technology: literature [102-103] respectively uses Back Propagation Neural Network (BPNN) to predict the residual life of bearings; Literature [104] is based on Weibull distribution and artificial neural network to study bearing fault prediction and health management, and proposes an accurate bearing residual life prediction method. Literature [105] Based on vibration signals, gaussi-ans Hidden Markov Model (GHMM) is used for bearing feature extraction and residual life prediction. Literature [106] uses time domain feature parameters to train artificial ant clustering to obtain bearing performance degradation state, and uses Hidden Markov Model (HMM) to predict bearing life. Literature [107] is based on time domain, frequency domain and wavelet packet decomposition features, the hidden Markov model is used for bearing fault diagnosis and remaining effective life prediction; the literature [108] uses the hidden semi-Markov model (HSMM) to obtain the state of the bearing and the probability of state transition through real-time data processing to complete the bearing remaining life prediction. Semi-MarkovModel.


Statistical method based on bearing state and response data for bearing operationThe trend is predicted to obtain the residual life distribution curve of the bearing..However, due to the limitations of the working environment and testing technology, the operating state of the bearing cannot be directly measured, and the performance degradation process of the bearing can only be indirectly reflected by the response parameters related to the operating state of the bearing (such as vibration signal and temperature). Because the bearing state degradation process involves multiple physical quantities, a single measurement and characteristic parameters can not fully reflect the bearing degradation state, it is necessary to integrate multiple physical quantities to evaluate the bearing performance degradation state, and introduce the life model to predict the remaining life. Moreover, the failure and destruction of large rolling bearings often come from faults, and the end of life caused by material fatigue is almost invisible, but there is no relevant report on the prediction of the remaining life of large rolling bearings considering faults. Therefore,The life prediction of large rolling bearings should be carried out on the basis of multi-physical quantity testing, combined with fault diagnosis, fault failure mode and degradation mechanism..