Generalized Profile Function Model Based on Neural Networks

Abstract

Generalized profile function model (GPFM) provides approximations of the individual models (individual stem profile models) of the objects using only two basic measurements. In this paper it is shown that this GPFM can be successfully derived by using artificial computational intelligence, that is, neural networks. GPFM is obtained as a mean value of all the available normalized individual models. Generation of GPFM is performed by using the basic dataset, and verification is done by using the validation data set. Statistical properties of the original, measured data and estimated data based on the generalized model are presented and compared. Testing of the obtained GPFM is performed also by the regression analysis. The obtained correlation coefficients between the real data and the estimated data are very high, 0.9946 for the basic data set, and 0.9933 for the validation dataset.