An evidence theory-based validation method for models with multivariate outputs and uncertainty

Researchers usually rely on simulations to predict the response of complex systems, we recognize that the models that underlie these simulations are never perfect. Model validation is a crucial ingredient in simulation credibility assessment. Multivariate responses under uncertainty often exist in complex simulation model, and the corresponding validation problem is not be solved effectively based on the existing validation methods. Hence, this paper presents a new validation method based on evidence theory for simulation model under uncertainty. For analyzing the extent of agreement between simulation outputs and experimental observations under uncertainty, the data features of system responses under uncertainty are extracted primarily. Next, the validation data such as large sample, small sample, data features, and expert opinions are represented as evidence theory. Then the traditional evidence distance method is improved to measure the agreement extent of simulation outputs and experimental observations. The proposed method is verified through an application example on validation of a simulation model about the terminal guidance stage of flight vehicle to illustrate their validity and potential benefits.