Chelidze D., Liu M.

A new method for damage tracking and identification in variable operating conditions is described. The method is based on dynamical systems perspective on damage evolution, where damage is viewed as evolving in a hierarchical dynamical system with distinct, two time-scales. Fast-time dynamics is related to structural vibrations that can be measured directly. Slow-time dynamics describes damage evolution driven by fast-time variables and causes drifts in structural system parameters. Fast-time vibration measurements are used to develop damage tracking feature vectors based on phase space warping concept. A smooth orthogonal decomposition (SOD) is used to identify linear subspace of the feature space related to the changes in operating conditions. Damage tracking and identification is performed in a feature subspace that is not sensitive to the changes in operating conditions. For this the SOD is applied to the feature vectors projected onto this subspace. The method is validated using an experimental system were a fatigue damage is introduced into a vibrating cantilever beam. The changing operating condition is simulated by perturbing a nonlinear potential field of the beam with two electromagnets. The electromagnets are powered by a programmable power supply and their voltages are altered in a manner to introduce a two-dimensional changing operating condition. The results show that the damage mode can be separated from the operating changes and can be used for damage prognosis.