Damage detection through nonparametric models using Kautz filters

Abstract

The goal of this paper is to present an approach to detect structural changes by using nonparametric models. The impulse response functions (IRFs) of mechanical systems in healthy conditions are identified through the sum of convolution expanded on an orthonormal basis. The Kautz filters with multiple poles optimized are implemented as an efficient orthonormal basis to identify nonparametric reference models. Identifying the IRFs of a smart beam with PZTs actuators/sensors coupled is used to illustrate the necessary procedures. The damages are simulated as a loss of bolts, nuts, and washers attached near a PZT sensor. A statistical procedure of a hypothesis test is also used to confirm the smart beam’s structural state. The experimental results achieved show that it is possible to detect the structural changes with statistical confidence.