A hybrid deep learning and ensemble learning mechanism for damaged power line detection in smart grids

Globally, 80% of the world population use electricity as a prime energy source. Government and private organizations face many challenges in providing efficient power facilities to their customers due to over-population and exponential increase in electricity demands. Furthermore, the abrupt damages in transmission lines pose another big barrier in the form of reliable and safer power transmissions. These line damages become more severe when the transmission infrastructure spans thousands of kilometers. Mostly, it results in life loss (humans and cattle), destruction of homes and crops, over-costing, etc. To address these problems, a hybrid deep learning mechanism is proposed in this research work that can accurately identify the damages in the transmission lines. This model consists of convolution neural network (CNN) and support vector machine (SVM) where CNN is used for the classification damaged power-line images, while SVM for the identification and calculating the severity of damaged power-lines using statistical information. Applicability of the model is validated using UAVs and other performance metrics such as accuracy, precision, F-score, error-rate, simulation time, area under curve values, and True-False values. The proposed model outperformed by generating a high recognition rate of 95.57% for the identification of damaged power-lines. The implications of this research work include no humans and cattle life loss, no extra transmission lines management and checkup costs, no destruction of homes crops, etc.