Multi-objective differential evolution algorithm for stochastic system identification

The last decade has witnessed rapid developments in structural system identification methodologies based on intelligent algorithms, which are formulated as multi-modal optimization problems. However, these deterministic methods more or less ignore uncertainties, such as modeling errors and measurement errors, that are inevitably involved in the system identification problem of civil-engineering structures. A new stochastic structural identification method is proposed that takes into account parametric uncertainties in the parameters of building structures. The proposed method merges the advantages of the multi-objective differential evolution optimization algorithm for the non-domination selection strategy and the probability density evolution method for incorporating parametric uncertainties. The results of simulations on identifying the unknown parameters of a structural system demonstrate the feasibility and effectiveness of the proposed method.