An Intelligent Estimator for Transient Overvoltages Study during Induction Motors Starting

This paper deals with transient overvoltage phenomenon which is occurred during induction motors (IMs) starting. This power quality (PQ) disturbance can damage motors’ dielectric insulation and affect the locally connected other loads. First, effective parameters on these overvoltages are identified. Then, an artificial neural network (ANN) is proposed to evaluate them. The most common structures, i.e. multilayer perceptron (MLP) and radial basis function (RBF) are adopted to train the ANN. The MLP structure is trained with the six learning algorithms, including backpropagation (BP), delta-bar-delta (DBD), extended delta-bar-delta (EDBD), directed random search (DRS), quick propagation (QP), and levenbergmarquardt (LM). The results show the effectiveness of proposed approach to predict accurate value of overvoltage peak. Based on performed comparison among all developed ANNs, it is proven that LM and EDBD algorithms have best performance for this goal.