摘要
为了使双燃料发动机满足日益严苛的排放法规要求,同时获得更高的经济效益,有必要对发动机进行全工况综合优化,以获得排放和燃油消耗的折衷.首先基于长短期记忆(LSTM)神经网络建立了发动机NO_(x)排放和燃油消耗率(BSFC)的预测模型,然后将所建模型与NSGA-Ⅱ算法结合,对NO_(x)排放及BSFC进行优化,并获得最优Pareto前沿解集以及决策变量的最佳控制参数组合.最后将最优控制参数组合标定至电子控制单元(ECU)中进行试验验证,结果表明:优化后的NO_(x)排放平均下降了76.4%,BSFC平均下降了3.5%,且NO_(x)排放满足IMO Tier-Ⅲ的限制要求.
In order to enable a dual-fuel marine engine to meet increasingly stringent emission regulations and obtain higher economic benefits,it is necessary to comprehensively optimize full operating performances of the engine to obtain a better compromise between emissions and fuel consumption.First of all,based on the long shortterm memory(LSTM)neural network,a prediction model for engine-out NO_(x) and brake specific fuel consumption(BSFC)was established,and then the built model was combined with the NSGA-Ⅱalgorithm to optimize NO_(x) and BSFC and to obtain the corresponding optimal Pareto frontier solution set.Finally,the optimal control parameter combination was calibrated to the electronic control unit(ECU)for experimental verification.The experimental results show that the optimized NO_(x) emissions decrease by 76.4%on average,the BSFC decreases by 3.5%on average,and the NO_(x) emissions meet the limit requirements of IMO Tier-Ⅲ.
作者
许朵
姚崇
马骋
宋恩哲
Xu Duo;Yao Chong;Ma Cheng;Song Enzhe(College of Power and Energy Engineering,Harbin Engineering University,Harbin 150001,China)
出处
《内燃机学报》
EI
CAS
CSCD
北大核心
2022年第5期403-411,共9页
Transactions of Csice
关键词
双燃料发动机
预测建模
多目标优化
dual-fuel engine
prediction modeling
multi-objective optimization
