Prediction of thermal behaviour of Switched Reluctance Machine using regression technique

In the new century, thermal analysis has received more attention to ensure the design of more efficient motors in terms of loading conditions, cost and size. From the design to its manufacture, temperature rise is considered as an important parameter in any electrical machine. This has also a significant effect on the long term stability of the machine. Conventionally, the thermal analysis has been carried out using Lumped parameter Thermal Model or Finite Element Analysis. All the proposed techniques depend on the computation of losses and thermal resistances which is a difficult task. Hence based on the real time temperature measurement carried out on a 8/6 pole Switched Reluctance Machine at different loading conditions, a new technique is suggested to predict the thermal behaviour using the statistical tool such as Regression analysis. Using these polynomial equations, temperature can be predicted at any operating condition without wastage of power and time.