Delamination detection in composite laminates using ensemble of surrogates

Abstract

The use of surrogates to model complex engineering problems such as the vibration based delamination detection in composite laminates involves uncertainty especially in regions with high error and areas where the predictions made by the surrogates vary widely. Moreover, the choice and suitability of the surrogate for this problem is dependent on the nature of the problem. Therefore in this work the absolute errors of different surrogate modeling techniques are analyzed and the standard deviation of predictions made by the surrogate model is used to quantify the areas of uncertainty in the design space. The current delamination detection problem in composite laminates, the locations of damage, size of the damage are the variables of interest and the change in natural frequency being the response parameter is analyzed numerically and experimentally to model this prediction problem.