Comparison of Computational-Model and Experimental-Example Trained Neural Networks for Processing Speckled Fringe Patterns

Tile responses of artificial neural networks to experimental and model generated inputs are compared for detection of damage in twisted fan blades using electronic holography. Tile training set inputs, for this work, are experimentally generated cllaracteristic pattenls of tile vibrating blades. Tile outputs are damage flag indicators or second derivatives of tile sensitivity vector projected displacement vectors from a finite element model. Artificial neural networks have been trained in tile past with computational model generated training sets. This approacll avoids tile difficult inverse calculations traditionally used to compare interference fringes with tile models. But tile high modeling standards are hard m acllieve, even with fan blade finite element models.