Validation experiment design of degradation models for composite materials with interval uncertainties

Abstract

The degradation model for composite materials with interval uncertainties is a mathematical representation of the residual properties in the degradation process. And validation experiments are necessary to validate the match between the prediction model and the actual degradation process. Generally, an economic validation experiment plan is essential for engineering practice. As the key to model validation, a novel model metric is established based on the overlap of the interval processes, which can accurately characterize the agreement between the prediction model and the experimental data. Since the new metric utilizes an unbiased estimation method to quantify the interval bounds of experimental data, it can reduce the need for experimental sample size. Furthermore, a validation experiment design optimization model is developed using the prior information obtained from the predictive model, where the number and distribution of observation instants, and the number of experimental samples are the design variables. To solve the optimization problem effectively, a collaborative optimization algorithm is proposed to obtain the optimal distribution of observation instants and sample size. Finally, an existing degradation model of composite materials is utilized to verify the feasibility and effectiveness of the presented experimental design method in engineering applications.