A reliable fault detection and classification scheme based on wavelet transform and ensemble of SVM for microgrid protection

With the advent of distributed generation in the power distribution network due to rising concern towards the green energy, the power generation scenario has witnessed a gradual transition from large sized power plants to the micro-sized distributed generating units involving the large penetration of distributed energy resources (DER). The associated protection problems due to distinct operating characteristics of DERs and loads with different operational dynamics associated with dual operating modes, demand for efficient protection technique. In this regard, this paper presents a wavelet and ensemble of SVM classifier based scheme to perform fault detection/classification considering non-linear load under dual mode of microgrid. The proposed scheme utilizes standard deviation of the approximate coefficients obtained during feature extraction process through DWT as the input feature to train the ensemble of SVMs. The test results of the proposed scheme against varying fault parameters and the performance comparison with single SVM classifier clearly reveal the effectiveness of the developed scheme for providing accurate and reliable protection to the microgrid.