Model-based temperature monitoring of the internal combustion engine of a lightweight long-endurance fixed-wing UAV

The overheating of internal combustion engines tailored for UAV applications is avoided during high-power flight phases (e.g., climb), by applying step-climb manoeuvres to repeatedly cool down the engine. This cyclic condition, over successive missions, can accumulate relevant latent performance degradations that can persist for long time before scheduled maintenance. This study deals with the modelling of cylinder head temperature of internal combustion engines for UAV applications, aiming to facilitate an accurate temperature model-based monitoring from which infer unscheduled maintenance. Starting from thermodynamics first principles and applying simplifying assumptions, a 0D dynamic model of a cylinder head temperature, comprehensive of dependencies on throttle position, angular speed, altitude and UAV advance speed, is realized. First the model parameters and functions are identified via a particle-swarm optimization algorithm by minimizing the mean square error between the model output and the experimental data, subsequently, the model is validated leveraging data coming from another flight mission. The results highlight that the proposed model, executable for real-time application, is quite accurate, presenting a mean error lower than 3% and peak error lower than 10%. Based on the experimentally validated model, it is developed a model-based algorithm able to predict temperature deviation of the engine with reference its nominal conditions.