Estimation ofLossCoefficients ofNonlinear Rubber UsingIterative Ho Filter

For automobile industry, rubber iswidely usedfor noise isolation andvibration reduction. However, duetoits distributed andnonlinear characteristics, itishardtoprecisely estimate its characteristics suchastheloss coefficient whichisde- fined asthetangent ofthephasedelay between thefundamental components ofthestrain andthestress undersinusoidal driving. Moreover, evenusing atruncated finite-dimensional model, with rubber's nonlinearity, resonance ofthetesting mechanical system, andmeasurement noise, optimal estimation oftheloss coefficient byusing Kalmanfilter isnotfeasible inthepresence ofthese uncertainties andnon-Gaussian disturbances. Therefore, HI,,, filter isapplied inthis papertorobustly estimate theloss co- efficient fromthestate-space perspective. Asastate-space model forrepresenting asinusoidal signal haseigenvalues ontheunit circle, themeasured dataisfirst processed byimposing asuitable exponential decayinordertoensurethestability oftheH" filter. Moreover, duetofinite datalength, aniterative Hoofilter isdeveloped toimprove theaccuracy ofparameter estimates. Ateachiteration, estimation ofdisturbances byusing theH,0 filter isfirst performed byapplying thepreviously estimated components ofthedesired signal. Thenarobust estimation ofthe desired signal ismadewithrespect tothemeasured signal which issubtracted bytheestimated disturbance. Bothsimulation study andexperimental testareconducted toverify theperformance oftheproposed iterative H,,filter. IndexTermsRubber, H,,filter, Losscoefficient, Iterative learning algorithm