Digital twin enhancement of real-time predictive EMS for FCEV truck operating under varying conditions

• Feedback updating digital twins enhance real-time predictive energy management. • Physics based correction of feedback training data against powertrain dynamics. • DT training saves 5.3–8.7 % H 2 across longhaul, regional delivery, suburban missions. • 0.42 % added H 2 savings at 1500 m, 35°C over optimized EMS using ambient cautious DTs. • Detailed HD FCEV HIL modeling with mechanical, electrical, thermal dynamics and CAN. The energetic behaviour of fuel cell electric trucks (HD FCEVs) can get significantly affected by the upcoming route, speed, payload, ambient temperature and even altitude. This can make the optimized energy management less effective in yielding maximum possible efficiency benefit from the actual powertrain. The result is increased H 2 consumption and inaccurate range estimation, which impacts economic and logistics feasibility for the FCEV operators. The current work demonstrates digital twin (DT) enhancement of a real-time predictive energy management system (P-EMS) to improve H 2 efficiency, vehicle states forecast and range estimation across diverse operating conditions. The precise DT feedback characterization of actual subsystems’ steady-state behaviour against ongoing powertrain transients and its real-time deployment for effective P-EMS remains relatively unexplored. The proposed approach benefits from real-time data and evolving machine learning DTs of the fuel cell system, DC/DC converter, battery, electric drive, traction chain, road resistance and cooling auxiliary load, to precisely predict the FCEV’s future mission specific behaviour. Feedback trained DTs are further used to re-optimize fuel-electric power split tailored to the learnt hybrid system operation. The paper also describes FCEV modelling approach for both feedback updating DTs inside predictive EMS and detailed CAN connected hardware-in-the-loop emulation (HIL). Compared to optimized EMS using averaged prediction models, feedback characterized DTs could decrease H 2 consumption of the HIL emulated FCEV by 5.3 %, 5.8 % and 8.7 % on VECTO long haul, regional delivery and suburban missions, respectively. Furthermore, deviating behaviour under demanding high altitude-high temperature scenarios (e.g., 1500 m, 35°C) could be better managed with up to 0.42 % added H 2 savings over ground level optimized EMS, by characterizing for ambient specific hybrid operation under such extreme conditions.