A study of induction motor stator flux observer based on radial basis function network

A new method which uses RBF neural network to observe stator flux is presented in this paper. This method uses RBF neural network to reconstruct a variable cutoff frequency stator flux estimator which based on voltage model with amplitude and phase compensation. Experimental results show that the method can achieve a more accurate observation of stator flux of induction motor when stator voltage frequency change and load change, and simplify the stator flux observer´s structure, improve adaptive capability of the stator flux observer without the problem of DC offset and initial phase.