Fisher Information–Based Optimal Sensor Locations for Structural Identification of Nonclassically Damped Systems

Abstract

The investigation of optimal sensor location problems in the field of vibration-based structural identification is predominantly performed for classically damped systems. In this study, a methodology is introduced to obtain the optimal sensor locations for identifying the structural (modal as well as physical) parameters of a nonclassically damped system. The optimal sensor locations are obtained by maximizing the determinant of the estimated Fisher information matrix out of all possible sensor combinations, such that the measurements would have maximum information about the parameters of interest. To deal with the inherently complex characteristics of the eigenvalues and mode shapes of nonclassically damped structures, the formulation is carried out in the state-space framework. To illustrate the approach numerically, primary-secondary systems exhibiting nonclassical damping behavior have been used. The methodology is also validated with experimental data from shake-table tests conducted on a primary-secondary structure. It is shown that the uncertainty in the estimated parameters is minimum when the sensors are placed at the optimal locations.