Data–driven calibration of power conductors thermal model for overhead lines overload protection

The increasing complexity of transmission networks may cause a significant rise in the load flows during and after serious system disturbances. In this context, accurate thermal rating assessment of overhead lines is essential to maximise infrastructure utilisation, while ensuring a reliable functioning of the power networks. Thermal assessment demands reliable simulation models to provide short and long term predictions of thermal behaviour. Although many physical based models are available nowadays, the incomplete knowledge of many parameters drastically narrows their range of application and their effectiveness. Typical sources of uncertainty are the non-stationary load and the fluctuating operating conditions. This paper proposes a gradient-based data driven technique for calibrating a thermal dynamic model on the basis of observed measures. The approach relies on the computation of the dynamics of the sensitivity of the solution of a differential equation to variations of the parameters. The approach is assessed by calibrating the IEEE thermal model of an overhead power conductor, on the basis of a real dataset recorded under a variety of operating and weather conditions