Input force identification using Kalman filter techniques: application to soil-pile interaction

An identification method for estimating the time varying excitation force acting on a structural system based on its response measurement is presented in this study. The method employs the simple Kalman filter to establish a regression model between the residual innovation and the input excitation forces. Based on the regression model, a recursive least-squares estimator is proposed to identify the input excitation forces incorporating with the measurement noise and the modeling error. In applying the method, the ambient vibration measurement of a structural system was used first. The stochastic subspace identification is applied to estimate the system matrix "A" and the measurement matrix "C". Then the Kalman filter with a recursive estimator is applied to determine the input excitation forces. The dynamic excitation forces are estimated from the measured structural responses by an inverse algorithm while least-square method with a recursive estimator is employed to update the estimation in the sense of real-time computation. Verification of the method with numerical simulation through MIMO system is conducted first. Identification of soil forces during the shaking table test of soil-pile interaction is also demonstrated.