With artificial neural network approach estimate the friction materials performance

The aim of this thesis is modelling the time-consuming friction performance of brake pads with artifical neural network and estimate them with minimal eror rate. With the result of estimation, is to create an assessment model. By this method, the friction parameters which are determined in a long time are determined in much shırter time and at a lover cost. In the first stage, artificial neural networks (Artificial Neural Networks) and previously tested performance of brake pads were examined. The data of friction parameters that were obtained from 18 samples which are experimented 3 timesfor each content were put in arithmetic meanand had a training of artifical neural network in the second stage, based on the data of previously tested frication parameters 19 sample content were prepared and friction parameter was estimated with trained artifical neural network. At the final stage, samples of which friction parameters were estimated were produced, for each sample, the tests were repeated 3 times and their friction parameters were put in arithmetic mean, and the constistency between them and estimated friction parameters were examined.