Heterogeneous data fusion for impact force identification in truss structures

Abstract

Civil engineering structures can undergo serious damage due to impact forces. But accurate and rapid identification of impact force is quite challenging because its measurement is difficult and location is unpredictable. This study proposes a novel approach for the complete identification of impact force including its location and time history. The proposed method combines an augmented Kalman filter (AKF) and Genetic algorithm (GA) for accurate identification of impact force. In AKF unknow force is included in the state vector and estimated in conjunction with the states. First, the location of impact force is statistically determined in the way to minimize the AKF response estimate error at measured locations, assumed co-variance values are used in AKF at this stage. These values are assumed based on a few analyses in which force location is assumed to be known. Then, GA is applied to optimize the error co-variances by minimizing the error between measured and estimated structural response. Once optimized co-variances are obtained, the exact time history of impact force can be constructed using AKF. Numerical example of a truss is considered to validate the efficacy of proposed approach. Strain and acceleration measurements are used as input for the AKF. Both modelling error and measurement noise are considered in the analysis to simulate the actual field conditions.