Identification of Elastic Properties Based on Belief Function Inference

In this paper, we consider parameter identification from measurement fields in an uncertain environment. An approach based on the theory of belief functions is developed to take into account all possible sources of information. Information from measurements is described through likelihood-based belief functions, while consonant random sets are used to handle prior information on the model parameters. Next, we construct the posterior random set by combining measurement and prior information using Dempster’s rule. To summarize the posterior random sets, we propose to find the minimal-area region in the parameter space, whose belief and plausibility values exceed given thresholds. This approach was applied to identify the elastic properties of a 2D plate from a measured kinematic field.