Online diagnosis of power systems insulation condition in wind farms

Abstract

Wind power has developed quickly in recent decades. The reliability of wind farms is strongly determined by the insulation conditions of all the assets in the power production and distribution system, which contains the wind turbines, power cables and accessories, switchgears, transformers, etc. The online partial discharge test has been proven to be an effective diagnostic approach to evaluate the insulation condition in multiple assets in power systems, which has the benefits of no power outage, nondestructive, and it has a high accuracy in defect recognition and fault localization. The challenge of applying online partial discharge testing in wind farms is due to the high interferences from the inverters, from which the impulsive voltage has similar bandwidth as that of partial discharge signals. Conventional filtering approaches have very limited effects in removing the interferences from the inverters. In this paper, the T-F Map filtering approach is applied in the online partial discharge testing, in which each acquired pulse is characterized by its equivalent pulse width (T) and pulse bandwidth (F). The discharge pulses from the same sources tend to gather into clusters in the T-F Map. With this method, the partial discharge pulses can be effectively discriminated from the interferences from the inverter. This paper also shows some experiences of diagnosis for wind farms based on online partial discharge testing.