Comparison of classifying the material mechanical properties by using k-Nearest Neighbor and Neural Network Backpropagation
Abd Rahim Intan Maisarah, Fauziah Mat, Yaacob Sazali · 2011
This paper present a development of a system with non-destructive testing on the material to define the mechanical properties of material. The experimental and testing of the material mechanical properties using vibration technique could determine the natural frequencies, the damping ratio and mode shapes of the structure. However, in this study, we only considering the natural frequencies and its amplitude of the material as the input data needed for training. As an extension for the study, the input data tested with various method of classifier. The k-Nearest Neighbor classifier and artificial neural network with Levenberg-Marquardt Backpropagation are developed to work as a system to classify the materials tested according to their mechanical properties. The result from the classification system shows that k-NN is giving the accuracy of 99.79783 % with the k value of 1 and in the other hand, Levenberg-Marquardt Backpropagation is giving the best classification rate of 99.86%.