A hybrid high frequency pulse and pattern recognition method for water tree detection in long distance underground cables

The demand for wind energy has been on the rise. To satisfy this demand, power utilities have begun to construct an increasing number of wind generation facilities. Many of facilities are located in areas that require long distance underground cables connections to the grid. Maintaining these cables requires techniques to identify and locate potential problems such as water tree. However, many traditional detection techniques are not designed for long cables; therefore, new method must be developed to remotely and non-destructively locate and measure water tree development in long underground cables. By using a combination of high frequency pulse and pattern recognition, the new method can establish a system of benchmark frequencies corresponding to specific water tree capacitance values for a given piece cable. The new method works best for long cables and small water tree capacitances. It supplements tradition water tree detection techniques, such as travelling wave and partial discharge.