Low-energy-consumption structural identification with switching piezoelectric semi-active input

Herein, we propose a structural system identification method that uses semi-active inputs generated from piezoelectric transducers. Previous onsite structure identification methods consumed significant amounts of energy to generate input signals and thus could not be actively implemented. The proposed method consumes less energy to generate input signals during identification. Two problems limit the use of semi-active inputs for identification purposes: (1) a control strategy to generate semi-active inputs suitable for identification does not exist, and (2) the polarity of the semi-active input cannot be freely controlled. To solve the first issue, we constructed a semi-active input generation strategy that matches the polarities of semi-active inputs with target signals appropriate for identification. To address the second problem, the mechanism by which polarity control is possible and impossible is clarified based on an analytical solution of a semi-active input generation circuit. A data calibration method focusing on sign similarity was proposed to improve the identification accuracy when polarity control cannot be implemented. The proposed semi-active identification method successfully identified the frequency response of a structure using 98% lesser energy than that used by conventional active methods. The proposed method was validated through simulations and experiments. The proposed method achieved accurate identification and can be used for onsite structural health monitoring.