Fault Diagnosis and Condition Evaluation for Soft Start of a Kind of Typical Induction Motor

Induction motors usually use soft starters with three-phase six anti-parallel thyristors as the driving device to achieve motor soft start and reduce energy consumption. Due to the half-wave rectification effect of the thyristor, the stator voltage and current are very likely to generate many harmonics at the moment of commutation, which is the main source of motor failure, will bring losses to the motor, and even affect the reliable operation of the motor. Aiming at this problem, taking the thyristor-driven induction motor system as the research object, the circuit mathematical model of the object is established, the typical fault data set is constructed in the soft-start process, and the system operating state is evaluated by the spectral characteristics. A novel fault prediction model based on convolution neural network is proposed, the calculation results are verified by computer simulation, and the final experiment proves that this method is effective to predict fault type in most situation.