Diagnostics of Ball Bearings in Varying-Speed Motors by Means of Artificial Neural Networks

This paper deals with the diagnostics of ball bearings in direct-drive motors by means ofArtificial Neural Networks (ANN). Direct-drive motors are becoming commonly usedin automatic machines, e.g. in the field of packaging, since these motors are easilydriven by the control system to perform polynomial profiles of motion avoiding thepresence of gears train or cams between the motor and the load. An ordinary task of themotor involves continuous changes of the shaft speed and a cyclic inversion of itsrotating direction. The continuous change of rotational speed of the motor represent themain drawback in terms of diagnostics of the ball bearing, since the large part ofalgorithms proposed in the literature need a constant rotation frequency of the motor toidentify fault frequencies in the spectrum. In this paper the use of Artificial NeuralNetworks overcomes the constant-speed limits and they are proven to be a powerful toolto diagnose the health of ball bearing even in variable-speed applications