Abstract:

Accurately accounting for the errors in meteorological parameter predictions is essential for the precise forecasting of dynamic current-carrying capacity in overhead power transmission lines. Statistical and computational analyses have revealed for the first time the distinct distribution characteristics of meteorological parameter prediction errors under varying forecasted weather conditions and environmental contexts. Existing methods for probabilistic load capacity forecasting of overhead lines fail to consider the impact of these two critical factors, leading to challenges in achieving accurate predictions. Addressing this gap, the issue of meteorological parameter prediction error analysis is formulated as a problem of solving for the conditional distribution of errors, influenced by both forecasted meteorological conditions and the environment. Incorporating Sklar's theorem, its associated Copula function, and non-parametric kernel density estimation, a novel approach to determining the conditional distribution of prediction errors is established. Further, a new methodology for probabilistic forecasting of transmission line load capacity that integrates the conditional distribution of meteorological parameter prediction errors is proposed, using Monte Carlo sampling techniques. Comparative computational analysis has demonstrated that, relative to two conventional approaches, the proposed method significantly enhances the coverage of prediction intervals by 5.51 and 1.99 percentage points, and concurrently reduces the normalized average width of these intervals by 7.86 and 3.62 percentage points. These improvements confirm the method's heightened accuracy and practicality.

Key words: overhead transmission line, current-carrying capacity, meteorological parameters, prediction errors, conditional distribution