Multi-objective optimization of skirt profile for air-cooled pistons based on adaptive ensemble multi-kernel Kriging and an improved hybrid WOA-DE algorithm

Abstract

Operating under severe thermo-mechanical coupled loads, air-cooled diesel engine pistons are highly susceptible to secondary motion instability, piston slap, and exacerbated friction. To address this multi-objective performance trade-off, this study investigates a representative air-cooled diesel engine piston as the research object and proposes an improved multi-objective optimization method for the piston skirt profile. First, a high-fidelity multibody dynamics (MBD) model was constructed via AVL EXCITE. By incorporating thermo-mechanical coupling and mixed lubrication mechanisms, it resolves the transient dynamic and tribological responses. Second, based on Latin hypercube sampling, an adaptive ensemble multi-kernel Kriging (AEMKK) surrogate model was constructed to replace time-consuming simulations, achieving reliable predictions for strongly nonlinear indicators including peak radial displacement, peak slap kinetic energy, and friction mean effective pressure (FMEP). Subsequently, an improved multi-objective whale optimization algorithm with differential evolution (MO-WOA-DE) was developed by introducing adaptive perturbation and a crowding-distance-based pruning strategy into the optimization process. These modifications were intended to maintain population diversity and global search capability while improving local refinement and non-dominated solution distribution. Compared with the original design, the optimized skirt profile reduced the peak radial displacement and peak slap kinetic energy by 22.7% and 67.7%, respectively, while FMEP changed marginally from 0.3370 to 0.3331 MPa. This indicates that lateral stability and slap suppression were substantially improved without an appreciable friction-loss penalty. Overall, the proposed optimization framework provides a feasible approach for coordinating dynamic stability, NVH behavior, and frictional performance in the refined design of piston skirt profiles under severe thermo-mechanical coupling conditions.