Multi-objective intake strategy optimization for elliptical rotor engines with unique intake system

The elliptical rotor engine (ERE) suffers from limited intake performance at high rotational speeds due to its unique rotary intake process and complex port and chamber geometry. To address this issue efficiently, this study develops an optimization framework to rapidly predict intake performance and identify improved intake-port design strategies with substantially reduced computational cost. First, this study used prior knowledge and CFD results to extract key intake process parameters, effectively representing the irregular variations in port flow capacity. Subsequently, an ensemble neural network model was developed using data from 60 CFD simulations to predict volumetric efficiency and pumping loss. Coupled with full factorial design, the ERE intake strategy was optimized under various design preferences. The results indicate that intake control parameters derived from port geometry and effective flow area accurately capture port flow capacity, enabling accurate predictions of intake performance. The ensemble model exhibits superior regression and generalization capabilities compared to alternative algorithms, with prediction R 2 exceeding 0.99 and errors below 3% for all output parameters. Parametric sensitivity analysis reveals a trade-off between volumetric efficiency and pumping loss, necessitating divergent optimization strategies. Optimization results show the efficiency-oriented strategy enhances volumetric efficiency by 11.95% and indicated power by 6.61%. The pumping loss-oriented strategy decreases single-cycle intake loss to 1.45 J/cycle. The balanced strategy achieves comprehensive performance enhancement—boosting volumetric efficiency by 11.25% and power by 11.81% without increasing baseline losses. This study provides a theoretical foundation and a practical approach for the rapid design and performance optimization of ERE intake systems.