基于高斯过程回归的船舶DMCC发动机整机性能优化

打开文本图片集

中图分类号：U664.121；TP18 文献标志码：A

Abstract： In response to the NOx emission peak phenomenon in medium and high load conditions under the propulsion characteristics in diesel engines，as well as the urgent need to reduce fuel consumption due to rising fuel prices，this study adjusts multiple control parameters of a diesel/methanol compound combustion（DMCC）engine to achieve simultaneous reductions in NOx emission and the brake specific fuel consumption （BSFC）under the premise of ensuring the power performance. To avoid the increased cost caused by large-scale experiments，predictive models for NOx volume fraction，BSFC，and indicated power of the DMCC engine are established based on Gaussian process regression. These models are then combined with the non-dominated sorting genetic algorithm- I （NSGA-II） to optimize NOx volume fractionand BSFC.The Pareto front solutions obtained are further analyzed using the technique for order preference by similarity to an ideal solution （TOPSIS）to find the optimal control parameter combination. Finally，the optimal control parameters are calibrated into the electronic control unit and compared with the original engine data.The results show that the predictive models based on Gaussan process regression achieve a goodness of fit greater than O.95 and a root mean square error of less than 1，indicating good consistency and accuracy. Compared to the parameters before optimization （the original engine conditions），using the optimal control parameters obtained by NSGA- I results in a 74.5% reduction in NOx emissions to just 3.47g/（kW⋅h） ，and an average BSFC reduction of 6.7% to 203.5g/（kW⋅h） ：

Key words： marine diesel engine; diesel/methanol compound combustion; Gaussian process regression ; non-dominated sorting genetic algorithm- I （NSGA- II）； technique for order preference by similarity to an ideal solution（TOPSIS）

0 引言

航运业是化石燃料消耗和温室气体排放的主要来源之一，因此船用发动机必须严格遵守日益严格的排放法规[1]。（剩余9886字）