Forecasting dynamic line rating to enhance transmission line utilization in South India using temporal convolutional networks

Abstract

Dynamic Line Rating (DLR) has gained significant attention in recent years due to its potential to optimize the utilization of transmission lines, improve system reliability, and reduce renewable energy curtailment. Therefore, forecasting the DLR accurately is essential. This research article proposes a novel approach for forecasting the DLR using the Temporal Convolutional Network (TCN). Here, TCN forecasts the weather parameters at the selected locations along the line. These forecasted weather parameters are utilized to compute the day-ahead DLR. The proposed method is applied to a transmission line in the southern Indian grid. TCN-based DLR computation method is compared with benchmark deep learning and machine learning models widely used in forecasting applications. The comparison shows that the proposed DLR computation method performs better than the other models. Incorporating weather forecasting using TCN into the line rating calculations enables a more accurate assessment of available transmission capacity under varying climatic conditions. DLR ensures 120% more ampacity than static line rating over 80% of the time.