The magnetic inrush current and internal fault types recognition in transformer based on probabilistic neural network

A new method based on classified function of probabilistic neural network (PNN) is presented to distinguish transformer magnetic inrush current and internal faulted types. At first, it uses PSCAD/EMTDC to simulate kinds of the transformer states, extracted and made pretreatment about simulated data, then built a model of PNN. Through training and learning the model by different spread values it can distinguish transformer magnetic inrush current and kinds of the internal faulted types, thus the veracity of magnetic inrush current and internal faulted types recognition in transformer based on PNN can be proved, simulation results and dynamic test results indicate that this technique is effective under different fault conditions, it has better foreground for engineering application.