A support vector regression (SVR)-based method for dynamic load identification using heterogeneous responses under interval uncertainties

Abstract

A support vector regression (SVR)-based method is presented in this paper, which aims to reconstruct the uncertain dynamic load using heterogeneous responses. Essentially, the technology of load identification can be regarded as an issue of nonlinear regression. Considering the incomplete limitation of the single response signal, the complicated relationship between the dynamic load and heterogeneous responses for a specific structure may be determined through some support vectors in simulated training samples based on the SVR algorithm. In view of multi-source uncertainties, an interval dimension-by-dimension (D–D) method is investigated, which can approximate the load changing concerning the each-dimensional uncertain parameter by Legendre orthogonal polynomials. Then, the dynamic load boundaries can be calculated through the maxima/minima of the approximation polynomial in the time history. The validity and specificity of the proposed methodology are clarified by 3 numerical examples. In addition, the results indicate that the proposed method can be utilized to identify the interval of dynamic load with outstanding accuracy and efficiency.