Determination of optimal sensor locations for system identification based on maximum likelihood approach

초록

The paper presents an algorithm for determining optimal sensor locations (OSL) based on the maximum likelihood (ML) approach, which is applicable to modal system identification (SI) methods. The best estimate of unknown parameters can be obtained by maximizing the likelihood of the occurrence of the measurements relative to the prediction. Under assumptions on measurement noise and variance of measured response, the ML approach results in a least squares minimization, which is similar to the formula for the applied modal SI algorithm. By applying the Cramer-Rao inequality, a Fisher information matrix (FIM) in terms of the probability density function of measurements is obtained from a lower bound of the estimation error. Each column of the obtained FIM is defined by a vector of mode-shape sensitivity with respect to each unknown system parameter. Simulation studies have been carried out to examine the proposed OSL algorithm.