Islamic Republic of IRAN

In this paper, we study an induction motor identification in all states and conditions whether transient or steady using Elman neural network. Induction motors have highly nonlinear dynamic behaviours where the parameters vary with time and operating conditions. These nonlinear dynamic behaviours make difficult the identification of induction motor. While many applications such as control [1] need an accurate identification of induction motor, therefore having an appropriate identification seems to be necessary. Here a recurrent neural network introduced by Elman [2] which has the ability of learning temporal patterns as well as spatial ones is employed for induction motor identification. Our experiments show that using Elman recurrent neural network for identification could achieve high degree of accuracy in all states and conditions