PSI - Issue 52

Dong Xiao et al. / Procedia Structural Integrity 52 (2024) 667–678

669

Dong Xiao et al. / Structural Integrity Procedia 00 (2023) 000–000

3

2. Methodology

2.1. Impact identification method

In this study, a novel impact identification method is developed based on the FEM and surrogate-assisted e ffi cient global optimisation. The FEM model is used to evaluate the impact responses, with the impact characteristics, such as impact locations and impact force, serving as inputs. The objective is to find the optimal impact characteristics that minimize the response di ff erences between the simulated and target impacts. This minimization is achieved through the surrogate-assisted e ffi cient global optimisation method. The e ffi cient global optimisation method utilizes a Kriging meta-model to establish the relationship between the impact characteristics and the response di ff erences. In this study, The impact characteristics of interest are impact location and impact force. In the Kriging meta-model, these two impact properties are known as the design variables. Besides, the combinations of impact location, impact force are referred to as samples. By infill sampling on the Kriging meta-model based on the generalized expected improvement criterion and simulating the impacts corresponding to the infilled samples, the Kriging meta-model and the optimum are updated iteratively. And impact location, impact force of target impact is identified with minimal response di ff erences. Additionally, to expedite the convergence of the optimisation, local surrogates for the impact location and impact peak force are constructed, which adaptively bound the design variables within a feasible range. The optimisation process terminates when the di ff erences in the responses are below a predetermined threshold or when the maximum number of iterations is reached. The proposed impact identification method is illustrated in Figure 1, depicting a flowchart of the various steps involved. The method can be divided into three main parts: sample initialization, impact simulations to obtain the impact responses, surrogate updating and infill sampling to update the identified minimum and generate samples adaptively.

Fig. 1: Impact identification method