Partial least squares modelling for dynamic overhead line ratings

Due to the deregulation of the power supply industry and the continuing need for network expansion, there is pressure to maximize utilization of the existing transmission and distribution networks. Compared to the traditional static line rating, a dynamic line rating provides a more accurate methodology to determine the ampacity of the overhead line in real time. Taking the wind cooling effect into account allows extra power to be accommodated. Line-rating methodologies developed based on IEEE and CIGRE standards are widely employed for dynamic line ratings. However, there are a number of parameters that need to be chosen for each line which can lead to errors in the temperature/ampacity prediction. In comparison, the statistical technique, partial least squares (PLS) regression, only requires the monitored data to be collected. The model coefficients are determined by processing the data sets over a period, such as one year. In this paper, both physical (CIGRE) and statistical (PLS) models are analyzed using the data collected from field testing, and the comparisons of the two models are also demonstrated.