Distance relaying for transmission line using support vector machine and radial basis function neural network

The proposed technique consists of preprocessing the fault current signal samples using discrete wavelet transform to yield the change in energy (cc) and standard deviation (sd) at the appropriate level of decomposition of fault current and voltage signal for faulty phase identification and fault location determination. After feature extraction (ce and sd) from fault current signal, support vector machine (SVM) is used for decision of fault or no-fault on any phase or multiple phases of the transmission line. The ground detection is done by a proposed indicator 'index' with a threshold value. Once the faulty phases are identified, the fault location from the relaying point can be accurately estimated using RBFNN (radial basis function neural network) with recursive least square algorithm. For fault location both current and voltage signals are preprocessed through wavelet transform to yield change in energy (cc) and standard deviation (sd) which are used to train and test the RBFNN to provide fault location from the relaying point accurately. The combined SVM and RBFNN based technique is tested for faults with wide range of operating conditions and provides accurate results for fault classification and location determination, respectively. (c) 2007 Elsevier Ltd. All rights reserved.