Adaptive local-learning models for synchrophasor-based dynamic thermal rating

Estimating the real temperature of electrical components and, consequently, their load capability is becoming of vital importance in modern power systems. In this domain, the large data-sets acquired by Wide Area Monitoring Systems (WAMSs) could play a strategic role in assessing the Dynamic Thermal Rating (DTR) of critical lines. Although several estimation techniques have been proposed for WAMS-based DTR, different open problems should be solved for their large-scale deployment. In this context, the identification of the complex and hidden relationship ruling the input/output mapping between the phasors and the line temperature is one of the main limitations to address. To solve this issue, a data-driven technique based on the local-learning theory is proposed. Numerical results for a realistic case study are presented in order to assess the performance of the proposed technique.