Automated Recognition of Wood Damages Using Artificial Neural Network

Quantitative analysis of wood damages, which influence the reliability and safety of wood structure, was studied by artificial neural networks. It was proved by experiment that three different degree damages of wood can be recognized by neural network in acoustic emission (AE) testing when reasonable neural network was chosen, efficient training case was constructed, reasonable parameter and training means were decided. The results showed that the artificial neural network has excellent non-linear ability of solution. And the method provides an efficient approach to the identification and quantification of wood damages.