Examination of Sensor Distribution Schemes for Vibration Estimation

The general outset in vibration analysis is that a sensor distribution is optimal if the utility of the deployed system is maximized; with system referring to, for example, an identification system if the sensors are deployed to facilitate modal parameter estimation or a monitoring system if damage detection is the target. While numerous studies have been presented in the context of system identification and structural health monitoring, few studies have addressed the sensor placement task for vibration estimation (also known as virtual sensing). Therefore, the present paper offers an examination of this by adapting four schemes---developed and widely used for optimal sensor placement in system identification and structural health monitoring---for placing sensors to estimate unmeasured vibrations from a limited number of output sensors. More specifically, we test the performance of the effective independence (EI) method, the driving point residue (DPR) method, a combination of the two (the EI-DPR method), and the kinetic energy (KE) method in the context of estimating unmeasured vibrations of a structural system of engineering interest.