Detection and Classification of Power Quality Disturbances Using Variational Mode Decomposition and Deep Learning Networks

Nowadays, electronic gadgets become part of everyone's life, and as a drawback, it injects power quality disturbance (PQD) into the power system, which results in maloperation of the device, affects system stability, and reduces the life of the machine. It is somewhat difficult to detect and identify what type of power quality disturbance is occurred in the power system network. Due to the data complexity, it’s highly difficult to separate PQDs data from other data like normal system data and data during faults. In this paper, power quality disturbance data is generated as per IEEE standards and this data is processed by the variational mode decomposition (VMD) to extract hidden features from the PQD signals. These features are transformed into 3D image vectors, which are used to train the convolutional neural network (CNN). In this work, the study is made on VMD parted with various pre-trained CNNs and compared their results with developed CNN models, to find the optimum classifier.