An evidence-based approach to damage location on an aircraft structure

This paper discusses the use of evidence-based classifiers for the identification of damage. In particular, a neural network approach to Dempster–Shafer theory is demonstrated on the damage location problem for an aircraft wing. The results are compared with a probabilistic classifier based on a multi-layer perceptron (MLP) neural network and shown to give similar results. The question of fusing classifiers is considered and it is shown that a combination of the Dempster–Shafer and MLP classifiers gives a significant improvement over the use of individual classifiers for the aircraft wing data.