A novel hybrid deep learning approach including combination of 1D power signals and 2D signal images for power quality disturbance classification

摘要: Abstract As a result of the widespread use power electronic equipment and increase in consumption, importance effective energy policies smart grid begins to increase. Nonlinear loads other electric systems are considered as main reason for quality disturbance. Distortions signal shape due disturbance cause decrease total efficiency. The proposed hybrid convolutional neural network method consists 1D structure 2D structure. features acquired by these two architectures classified using fully connected layer, which is traditionally used classifier architectures. Power signals processed their original form. Then converted into images network. Then, feature vectors generated networks combined. Finally, this combined vector layer. well suited nature processing. It novel approach that covers steps an expert examining signal. framework compared with state-of-the-art classification methods literature. While method's performance relatively high methods, computational complexity almost same.