The performance improvement of condition monitoring for induction motor based on neural network

With the development of electrical machines in specific working conditions, the significant increase of power energy losses take place while their thermal capability is constantly decreased. The thermal overheating and cycling degrades the integrity of the materials used for stator winding insulation, resulting in acceleration of thermal aging. A new approach for induction motor temperature monitoring based on wavelet transform and neural network is presented. The wavelet transform is a well-suited tool for analyzing high-frequency transients in the presence of low-frequency components. The wavelet network has been successfully applied to nonlinear static function approximation and classification, and dynamical system modeling. The stator and rotor resistance are identified on-line, and then the temperature is calculated according to the principle that the metal resistance depends on its temperature. The evolutionary algorithm is used to fulfill the network parameter identification, reducing the size of the training set by selecting the most relevant features. The simulation results approve that the proposed method is effective for temperature condition monitoring of induction motor.