Modeling, emission characterization and multi-objective optimization of a confined piston linear generator

Abstract

The confined piston linear generator (CPLG) combines the controllability of conventional internal combustion engines (ICEs) with the high energy conversion efficiency (ECE) of free-piston linear generator, making it a promising energy conversion device. In this study, a coupled modeling framework was developed by integrating a dq0-coordinate permanent magnet synchronous linear generator (PMSLG) model in MATLAB/Simulink with an ICE model in GT-Power. Based on this framework, the effects of key operating parameters on emission characteristics were systematically analyzed. A multi-objective particle swarm optimization algorithm was then applied to identify the optimal operating conditions. After evaluating five regression models, Gaussian process regression (GPR) was selected as the surrogate model. A hybrid weighting method combining the analytic hierarchy process and criteria importance through intercriteria correlation was used to assess the trade-offs among power performance, economy, and emissivity. Compared with the baseline case, the optimized CPLG achieved a 15.4% increase in power, an 8.2% improvement in ECE, and a 34.5% reduction in emission objective value. The GPR surrogate model demonstrated high predictive accuracy, with a maximum prediction error of 3.88%.