About 40% of induction motor faults are caused by faults related to stator. So stator winding life estimation is an important part of induction motor life determination. This study aims to estimate motor insulation life by presenting a method based on online measurement of leakage current and predicting its future trend by neural network. Online leakage current measurement using three differential CTs with high accuracy and presentation of simple insulation model is one of the novel methods used for motors insulations condition monitoring which provides us with useful information about online insulation resistance, insulation capacitance and dissipation factor. This study utilizes this simple insulation model and novel method. The ability of the method to determine insulation condition was confirmed through simulation and practical tests. In addition to aging factors, sudden events like faults are of the factors that causes insulations quality loss and decrease in remaining life. In order to separate data related to mentioned factors and data related to aging factors an algorithm was presented. This algorithm is essential for accurate prediction of insulation condition. Leakage current and its changes related to it are good indications of insulation aging; therefore leakage current prediction makes it possible to predict insulation condition and remaining life estimation. Because of the difference in environmental and aging factors, like temperature, moisture, power supply condition and etc. and also insulation type and nominal power in different motors, in order for the suggested method to be comprehensive, prediction of changes is accomplished using an adaptive method. In this regard a special neural network was designed. putting information obtained from leakage current method and model, data separator algorithm and predictor neural network, a novel method for insulation condition prediction and life estimation was gained.finally in order to validate this neural network, leakage current changes related to a 3000 HP, 4160 V motor’s insulation aging were tested which shows very good agreement with experimental results. Keywords: Insulation, Life Estimation, Insulation Remaining Life, Leakage Current, Neural Network
