Estimation of Thermal Load in Structural Members from Measured Accelerations

Abstract

Presented herein is a new computational strategy to eliminate the thermal effect in measured responses. The internal forces of a structure and its thermal load components are estimated alternately at each time step based on a recursive least-squares algorithm. The noise effect in the measured data and the estimated force in previous steps are removed with the Kalman filter. This approach directly models the thermal effects on the dynamic responses without the temperature information. This approach adopts force identification technique which has been researched extensively in the last two decades. The effectiveness of the proposed approach is validated via numerical studies with a bridge structure with or without noise in the measured data. The thermal load in the structural components can be estimated accurately with the proposed strategy with less than 8% error when there is 10% noise in the measured responses of the structure.