Modelling the thrust developed by a sector motor via artificial neural networks

This paper is aimed at characterising the behaviour of the thrust developed by a sector motor under transient and steady state operating conditions. If the machine is fed with variable frequency voltages and different loading conditions have to be considered, the determination of the appropriate factors to correct the machine equations turns out to be a quite involved task. Using an equivalent electric circuit may be complicated as well. In order to overcome this difficulty in modelling the machine thrust is characterised in this work via artificial neural networks. The research comprises simulation of the machine behaviour and laboratory experimentation as well. The output variable to be estimated by the neural network is the thrust that is available in the machine shaft. When fed with the data acquired in the laboratory, the neural network learning process was able to produce the appropriate network topology so that the results from it were in good agreement with the actual data.