Singularity detection for phase-to-ground fault in non-effective grounded power networks based on multi-frequency bands analysis

The cubic B-spline wavelet packet transform is adopted to detect the singularity of single-phase-to-ground fault signals. By means of such transform, the maximal module in wavelet domain is obtained. The characteristic frequency band of each line, in which the maximal module is most concentrated, is chosen according to the view of maximum energy. For each case that the characteristic frequency band of each line lies in or out a same one, two types of numerical fault detecting criteria are proposed. The proposed method overcomes the blindness in determining the feasible frequency band that conventional methods cannot avoid. The effectiveness and practicality of the method proposed are demonstrated via field test results.