Data-driven real-time fuel cetane estimation and control design for multifuel UAVs

Anuj Pal,Andrew Cornelius,Zongxuan Sun,Kenneth Kim,Chol-Bum Mike Kweon
DOI: https://doi.org/10.1016/j.apenergy.2024.123336
IF: 11.2
2024-05-11
Applied Energy
Abstract:Multifuel compression ignition (CI) engines have been proven to be useful for the aviation industry, especially for powering unmanned aerial vehicles (UAVs). To operate multifuel UAVs at desired operating conditions, an appropriate control strategy is needed that can adapt based on changing fuel properties. Therefore, having knowledge of fuel properties at all times is critical. However, the challenge comes in identifying fuel property in real-time when fuel switching happens during engine operation. The current work proposes a novel data-driven control algorithm for estimating fuel property (cetane number) and adapting the control strategy in real-time. Gaussian process regression (GPR) is used for modeling the multifuel CI engine, and feedforward control is developed to correct the control inputs based on the estimated fuel cetane property. For feedforward control development, model inversion of data-driven GPR is performed. Based on the estimated fuel property and using the feedforward table, the control strategy gets adjusted in real-time within one combustion cycle to avoid the system performance from drifting away beyond specified bounds. The entire approach was implemented and validated on an engine test bed. Fuel switching is performed in real time, and the results demonstrated the capability of the algorithm to maintain the desired performance.
energy & fuels,engineering, chemical
What problem does this paper attempt to address?