Damage diagnosis in drive-lines using response-based optimization

Abstract

Abstract

This paper proposes an inverse vibration analysis approach for the prediction of element level damage in automobile drive-lines. The time-domain response signatures available along the drive-line at a known engine excitation are employed to predict the location and extent of damage. Damage signatures are expressed in terms of unknown element stiffness reduction coefficients and are evaluated by minimizing a residual force at every time instant. The residual objective function is minimized using genetic algorithms, and unknown stiffness reduction coefficients corresponding to the current state of response are predicted. The methodology is illustrated with torsional vibration response data obtained from finite element simulation.