Approaches For Experimental Identification Of Loss Factors

Damping is of great importance for structural design and dynamic characteristics prediction.In fact,experiment is the best way to obtain loss factors.In this dissertation,research on identification of modal loss factors and frequency-band loss factor for linear system and damping for stiffness-nonlinear system is carried out.A new approach for modal loss factor estimation is introduced based on Gause-Newton iteration.A mount of data points near the natural frequency are used so that it is able to eliminate effection of test error on damping identification.The approach is classified into three methods,namely direct fittig method,square fitting method,reciprocal fitting method,according to the types of fitting function.Noise immunity for the proposed approach is better than the improved half-power bandwidth method.Effection of frequency spacing,number of data points beside the natural frequency,and test repeating time on the error is analyzed which indicates that direct fitting method and reciprocal fitting method are prior for loss factor estimation because the error for the two methods is smaller and repeating time is less than the other two methods.The direct and reciprocal fitting methods can improve the accuracy if the signal-to-noise ratio is between 10dB and 20dB compared with improved half-power bandwidth method.Research on frequency-band loss factor identification is carried out in which process loss factor is introduced to describe frequency-band loss factor during decay process.The estimated error obtained by initial decay rate method cannot be neglected if there is a mode with lower natural frequency and larger loss factor in the band.Hence,a new approach,named last decay rate separation method(LDRSM),is introduced to obtain frequency-band loss factor.A simplified LDRSM is proposed which can simplify the identifying process and obtain similar loss factor to the total LDRSM.Both simulation and experiment show that the proposed total and simplified LDRSM are able to obtain frequency-band loss factor accurately within 10% error.The problem of loss factor estimate for vibration decay curve with ‘segmentations' for decay method is solved by LDRSM.At last,after damping characteristics of stiffness-nonlinear system is analyzed,an integral method based on Particle Swarm Optimization is introduced,which is of strong noise immunity and within 10% average error even if the signal-to-noise ratio is as low as 10dB.The error decreases with increased signal-to-noise ratio.Resarch on modal loss factor identification is carried out and the estimate errors of the proposed fitting methods,especially direct fitting method and reciprocal fitting method,are all smaller than the improved half-power bandwidth method.The direct and reciprocal fitting methods are of small estimate error and test repeating time so that it is helpful for engineering application.LDRSM is introduced to identify frequency-band loss factor.For the modal loss factor estimate of stiffness-nonlinear system,the proposed integral method used many data points near the natural frequency which eliminate the effect of experimental error on the loss factor estimate,and the result is stable and of good noise immunity so that it is a good choice for loss factor identification.