Sensor placement for time-domain modal parameter estimation

Abstract

A new method of sensor placement for modal identification is presented. The sensor placement technique utilizes the variance of estimates derived from a perturbation analysis of a time-domain parameter estimation method to determine appropriate locations of sensors. An efficient scheme is developed to implement the backward elimination scheme for sensor placement. The performance of this method is compared with two alternatives: a process of maximizing the determinant of the Fisher information matrix corresponding to the target modal partitions and a process of minimizing the trace of the covariance matrix. The effect of finite element model error on this placement technique is investigated. Numerical examples demonstrate the efficiency and effectiveness of the new approach.