Energy loss prediction in nonoriented materials using machine learning techniques: A novel approach

Traditional extrapolation performed by machine designers for energy loss estimation results in the decrease of the overall efficiency of electrical machines. Therefore, state-of-the-art techniques need to be developed in order to accurately predict the energy loss in electrical machines for their improved performance. To this end, machine learning techniques have been employed to predict accurate energy loss at different frequencies and induction levels under rotational conditions. Such types of flux exist near the teeth of the stator in synchronous machines. In transformers, rotational flux arises at the bends and corners of the stators. It was observed that the random forest machine learning algorithm has the least mean square error and as such is the most suited algorithm, which can be used for the accurate prediction of energy loss in nonoriented materials.