Air system optimization coupled with electric supercharger matching of a two-stroke aircraft engine based on machine learning and NSGA-III

Abstract

To increase the flight altitude of small unmanned aerial vehicles (UAVs), achieving power recovery at high altitudes through supercharging is crucial. A spark ignition two-stroke engine model was developed in GT-POWER and calibrated using experimental data. Engine performance with an electric supercharger was analyzed. The coupling mechanism of the intercooler in the intake system, the exhaust resonance pipe in the exhaust system, and the electrically supercharged two-stroke engine were studied. The results demonstrated that the supercharged two-stroke engine with the intercooler successfully maintained pwer at alitudes between 1,000 and 6,000 m without degradation, despite the decrease in power-to-weight ratio due to the intercooler. The nondominated sorting genetic algorithm-III (NSGA-III) and machine learning were used to optimize geometry parameters of the exhaust resonance pipe. Brake specific fuel consumption (BSFC), compressor power, and exhaust temperature were selected as optimization objectives. The Pareto solution set revealed significant tradeoff relationships among the objectives. In the Pareto solution set, the optimal values of BSFC, compressor power consumption, and exhaust gas temperature are 389.6 [g/(kW·h)], 1.31 kW, and 604.1 K, respectively. The optimized exhaust resonance pipe can maintain engine and compressor performance in a wide altitude range, allowing the engine to maintain power at higher altitudes.