Intelligent Optimization Based on a Virtual Marine Diesel Engine Using GA-ICSO Hybrid Algorithm

Abstract

Considering the trade-off relationship between brake specific fuel consumption (BSFC), combustion noise (CN) and NOx emission, it is a difficult task to optimize them simultaneously in a marine diesel engine. In order to overcome this problem, a novel genetic algorithm and improved chicken swarm optimization (GA-ICSO) hybrid algorithm was proposed, where the enhanced Levy flight and adaptive self-learning factor were introduced in this algorithm. Computational comparisons between GA-ICSO and other effective optimization algorithms were performed using four standard test functions, validating the improvements in both accuracy and stability for GA-ICSO. Furthermore, a predictive engine model based on a phenomenological approach was developed and validated. This model coupled the proposed algorithm for the optimization of a marine diesel engine. In the optimization process, five control parameters were selected as design variables, including injection timing (IT), intake cam phasing (ICP), intake valve closing (IVC), intake temperature and pressure. Results show that, a lower objective value can be obtained by GA-ICSO than other widely used optimization algorithms for all the operating conditions. Besides, by comparing the results between the optimal generations and baselines, it could be found that, under the condition of 50%, 75% and 100%load, CN is reduced by 10.7%, 4.9% and 3.9%, NOx is decreased by 15%, 31% and 33%, and BSFC is suppressed by 10.8%, 13.3% and 9.5%, respectively. Finally, heat release rates, noise spectrums, cylinder pressures and temperatures were all employed to discuss the optimization results of a marine diesel engine under different working conditions.