Panayiotis Moutis
November 2022
It is always a joy to join (albeit remotely) the co-organizer of the , my old student, the very hard-working and inspiring researcher . On, Nov. 21st at 14:00 15:00 (London time) I will be presenting my at the IET Generation, Transmission & Distribution journal on the solution of the AC OPF with the machine learning tool of top-down heuristically inducted binary decision trees (hiBDT). I strongly urge you to register and follow the Group鈥檚 webinar series . They also neatly keep recordings of the webinars they have previously organized ; great resource and a nice overview of recent developments in the space.
I will be discussing the theoretical guarantees (and some apparent implications) of a feasible space search with hiBDT recursively tightening the bounds/intervals of control variables towards a global optimum. The IEEE PES Task Force鈥檚 Power Grid library will be briefly presented as a crucial benchmark in assessing AC OPF solvers, relaxations, etc. The efficiency of the proposed hiBDT method will be posed as an open issue requiring considerations of 鈥渉ot starting鈥 and how to effectively search the AC OPF feasible space. Lastly, the recursive variable bound tightening with hiBDT that progressively improves the dispatch cost will be discussed as a feature of the method to robustly price the commitment of renewables unaffected by their volatility.