A statistical approach for sizing an aircraft electrical generator using extreme value theory

Résumé

The sizing of aircraft electrical generators mainly depends on the electrical loads installed in the aircraft. Currently, the generator capacity is estimated by summing the critical loads, but this method tends to overestimate the genera-tor capacity. A new method to challenge this approach is to use the electrical consumption recorded during ﬂights and study the distribution of operational ratios between the actual consumption and the theoretical maximum consump-tion then size the future aircraft generators by applying a ratio to the theoretical value. This paper focuses on the application of extreme value theory on these operational ratios to estimate the maximal capacity utilization of a generator. A real data example is provided to illustrate the approach and estimate extreme quantiles and the right endpoint of the distribution of the ratios together with their approximate conﬁdence interval in the nominal conﬁguration. In all situ-ations the right endpoint is proven to be ﬁnite and does not depend on the use procedures. This approach shows that ELA overestimates the maximal perma-nent consumption by 20% with error level of 10−3 in the nominal conﬁguration.

Mots-clés

Electrical load analysis; Aeronautic electrical system; Generalized Pareto distribution; Quantile estimation; Endpoint estimation; Diagnostics for threshold selection;