불인지 불확실성을 고려한 CAE 해석 모델의 예측 정확도 개선

In the design of passenger vehicles in the automotive industry, virtual testing is widely used. However, building a highly-predictive CAE model is still challenging. In this paper, to build a highly-predictive CAE model, several issues in statistical model validation approach are studied with an example of the steering column in cockpit modules: (1) selecting unknown input variables, (2) calibrating unknown parameters, and finally (3) checking the validity of CAE model using area metric based hypothesis testing. It is expected that the study helps to enhance the predictive capability of CAE models.