Partial discharge recognition using a neural network

A new analyzing technique for partial discharges using a neural network and a Whitehead partial discharge model is proposed. The teacher's data for the neural network are discharge inception phase data calculated from characteristic values of the discharge, such as discharge inception voltage, statistical time lag or residual voltage after discharge. This method is applied to a typical void discharge with or without UV illumination and their characteristic values are calculated from the measured discharge inception phase data.<
>