Real-time dynamic thermal rating evaluation of overhead power lines based on online adaptation of Echo State Networks

To assist utilities in utilizing the overhead power lines more effectively and thus to optimize the utilization of the existing system, it is important to know how to accurately assess the real-time dynamic overload current capacity of lines down to a `per span´ level of granularity. Accurate prediction of the conductor temperature ahead of time subject to various conductor overload conditions is the most critical and challenging step when determining the line dynamic thermal rating. An Echo State Network (ESN) based identifier has been demonstrated to identify the overhead conductor thermal dynamics under different weather conditions in a batch learning mode with good accuracy. Through the use of the ESN model, the prediction of conductor temperature can be obtained easily, which in turn helps determine the line dynamic rating in real time. This paper proposes a Sliding-Window (SW) based online learning algorithm to obtain the online adaptation of the ESN-based thermal dynamics identifier to any new/changed ambient weather conditions along the overhead conductor on a continuous base. Both simulation and experimental results are presented to validate the performance of the proposed algorithm. This method requires only temperatures and line current as inputs and its simplified calculation makes it an attractive and cost effective solution to real-time implementation.