Uncertain Damage Identification Methods Based on Residual Force Vector under the Influence of Measurement Noise
Qinghe Shi,Shuwei Qian,Zhenxian Luo,Zhongliang Cao,Chen Yang,Kejun Hu,Wenqin Han
DOI: https://doi.org/10.1016/j.ijnonlinmec.2024.104732
IF: 3.336
2024-01-01
International Journal of Non-Linear Mechanics
Abstract:This paper proposes probabilistic and interval methods, grounded in the residual force vector technique, to quantify measurement uncertainties and effectively integrate them into the identification process. In the probabilistic method, the measurement noise is assumed to be a zero-mean normal distribution, while the stiffness parameters of structural elements are assumed to follow a normal distribution, represented by mean values and standard deviations. In the interval method, the measurement noise is modelled as interval numbers, and the interval perturbation technique is employed to derive interval bounds for elemental stiffness parameters. A two-step model updating strategy is implemented, separately addressing pristine and damaged structures, to mitigate modeling errors. By using the results of probabilistic damage identification, a novel damage index termed damage expectation capable of capturing both the extent and probability of damage is introduced. Furthermore, an interval damage index is introduced to quantify the damage based on interval damage identification results. The impact of noise levels on the statistical properties of uncertain damage indices is also investigated. Numerical examples reveal that deterministic indexes exhibit a monotonic increase in average values with increasing noise levels. Particularly, as noise levels rise, the standard deviation of the combination index initially surpasses the deterministic damage index but eventually diminishes. Through the proposed methodology, structural damage can be identified even in the presence of uncertainties. The efficacy of the proposed damage identification technique is also validated by experimental data.