Fault prognosis for rotating electrical machines monitoring using recursive least square

Nowadays there is a strong competition between companies in order to develop new solutions to keep competitive on the market, e.g. by reducing the production costs. Since the maintenance costs are substantial share of the production costs, companies need to plan the maintenance costs effectively. Deterioration prognosis of electric machine can decrease the maintenance costs by minimizing the loss of production due to machine breakdown and avoiding the overstocking of spare parts. This paper proposes a prognostic system to predict the wear condition of rotating electrical machines. Experimentations have been carried out on a laboratory experimental system, which is equipped with analog-to-digital converters, a microcontroller and two accelerometers in order to acquire vibration measurements and predict the machine degradation trend. Tests are carried out on induction motors using an electrical motor test bench to show the performances.