| Title |
Capacity Sizing of Axle Driven Generation Power Supply for Freight Train Reefer Containers |
| Authors |
김주욱(Joouk Kim) ; 곽민호(Minho Kwak) ; 임지원(Jiwon Lee) ; 박영(Young Park) |
| DOI |
https://doi.org/10.5370/KIEE.2026.75.2.403 |
| Keywords |
Reefer container; Self-generation power system; Deep learning; Axle generator |
| Abstract |
This paper proposes an axle driven self generation system with an Energy Storage System (ESS) to supply power to refrigerated containers on freight trains. Instead of sizing the generator and ESS from a conservative worst case peak load, a probabilistic design based on a Long Short-Term Memory (LSTM) quantile model is adopted. The model predicts short term reefer load profiles and provides design values at selected quantiles, in particular the 95th percentile (P95). Using these results, capacity trends for peak load, generator rating and ESS energy are derived as functions of the consist size. For 20 and 33 car trains, the P95 based design reduces the required generator rating by about 30 percent compared with a deterministic peak load design, while the ESS covers occasional demand spikes and short stops without external power. The proposed approach therefore avoids excessive overdesign, yet maintains a specified reliability level for reefer power supply in long freight consists. |