PREDICTION AND ANALYSIS OF EXPERIMENTAL RESULTS OF HOT WIRE ANEMOMETRY BY USING OF NEURAL NETWORK

Neural network is a powerfull tool to predict the behavior and response of systems in engineering area. In present work, the time dependent results of experimental investigations have processed with a neural network model. Neural network model was Feed-Forward and has used the Levenberg-Marquardt algorithm as training function. The results of one and two wire probe of hot wire anemometry in form instantaneous voltages with coincidence them with average velocity values and change to velocity vectors were used as input data for training of neural network. The network was training in one and two conditions and the results were compared with experimental data. The results showed that the network results are as good agreement with accurate results. Also, a model was defined for interpolating and predicting of middle situations.