An intelligent fault detection and classification technique based on variational mode decomposition-CNN for transmission lines installed with UPFC and wind farm

Abstract

The output voltage and current signals of grid-connected doubly-fed induction generator (DFIG)-based wind turbine generators (WTGs) are significantly modulated during grid disturbances. Consequently, the performance of conventionally used distance relaying-based transmission line (TL) protection schemes are affected. The problem becomes further complicated when the interconnected TLs are compensated with unified power flow controller (UPFC). In this paper, an intelligent relaying scheme is proposed for efficient detection and classification of faults in such crucial TLs. In the proposed method, the suitable features are extracted from the locally measured current signals using variational mode decomposition (VMD). The main advantage of VMD is that it can extract the different frequency components of a faulted current signal adaptively with fast convergence. The extracted gray-scale images through VMD are utilized further by the deep convolution neural network (CNN) classifier for effective classification of faults. The efficacy of the proposed relaying algorithm is evaluated on 12000 fault and 120 non-fault cases generated on a system comprising of DFIG-wind farm and UPFC compensation using MATLAB/Simulink® and real time digital simulation platform (OP4510). The fast fault detection time (¡10 ms), high accuracies in the fault detection and classification (100% and 99.86%) justify the merits of the proposed relaying scheme.