An application of artificial neural network for classification of timber's durability / Noorsalzatul Azura Zakaria

This project present the application of Artificial Neural Network (ANN) in attempt to identify the timber's durability. MATLAB version 6.5.1 has been used as the software. There are 75 training data and 55 testing data used for this project. The data are from the timber testing that has been carried out in the Wood Technology laboratory in Universiti Teknologi MARA, Shah Alam. Five factors that affecting the durability of timber are modulus of elasticity (MOE), modulus of rupture (MOR), compression parallel of grain, compression perpendicular of grain and shear strength. This five parameters are use as an input parameters for classification of timber's durability. Six network functions from Back-Propagation is used, there are traingd, traindx, traingdm, traingda, trainrp and trainlm training function have been compared to obtained the best model for classification of timber's durability. The best result produced will determine the most suitable model for classification of timber's durability.