基于Atkinson循环发动机的燃油消耗率优化方法对比研究

发布时间：2018-07-15 20:06

【摘要】：针对利用GT-Power详细模型耦合遗传算法(方案1)优化Atkinson循环发动机燃油消耗率时,存在公认的不易收敛且计算缓慢的问题,提出了神经网络简化模型耦合遗传算法(方案2)进行全局优化并与方案1进行比对.方案1利用GT-Power搭建某Atkinson循环发动机详细仿真模型,应用Heywood公式建立爆震预测模型,并耦合遗传算法对Atkinson循环发动机燃油消耗率进行优化;方案2则利用拉丁超立方算法采集4 500个实验点,将GT-Power详细模型及爆震模型简化为神经网络模型,通过简化模型耦合遗传算法进行优化.研究结果表明:利用方案2可以将Atkinson循环发动机的实际燃油消耗率最多降低4.6%,且仿真优化结果相对实际优化结果的最大误差率仅为7.3%,同时相对于方案1仿真优化时间最大可节省322倍.因此,采用方案2替代方案1用于Atkinson循环发动机燃油消耗率的快速全局优化是切实可行的方法.
[Abstract]:In order to optimize the fuel consumption rate of Atkinson cycle engine using GT-Power detailed model coupling genetic algorithm (scheme 1), it is recognized that it is not easy to converge and the calculation is slow. A neural network simplified model coupled genetic algorithm (scheme 2) is proposed for global optimization and compared with scheme 1. Scheme 1 uses GT-Power to build a detailed simulation model of a Atkinson cycle engine, establishes a detonation prediction model by using Heywood formula, and optimizes the fuel consumption rate of Atkinson cycle engine by coupling genetic algorithm. Scheme 2 uses Latin hypercube algorithm to collect 4 500 experimental points, simplifies GT-Power detailed model and knock model into neural network model, and optimizes by simplified model coupling genetic algorithm. The results show that the actual fuel consumption rate of Atkinson cycle engine can be reduced by the maximum of 4.6 by using option 2, and the maximum error rate of simulation optimization results is only 7.3 compared with the actual optimization results, and the simulation optimization is relative to scheme 1. The maximum time saving can be 322 times. Therefore, it is feasible to use alternative scheme 1 in the fast global optimization of fuel consumption rate of Atkinson cycle engine.
【作者单位】：湖南大学汽车车身先进设计制造国家重点实验室;重庆理工大学汽车零部件先进制造技术教育部重点实验室;
【基金】：国家高技术研究发展计划(863计划)项目(2012AA111703) 湖南省研究生科研创新项目(CX2015B088)~~
【分类号】：TP18;U464

【参考文献】