摘要
高压共轨柴油机可控燃油喷射参数的增加,在使对燃烧的控制更加灵活的同时也带来标定和优化工作量显著增加的问题。为适应高效率的需要,提出并研究了基于模型的标定优化,即采用神经网络在一些工况点上建立模型,再通过自适应神经模糊推理系统(ANFIS)进行插值,将模型由这些工况点扩展到所需工况空间。模型精度由对象、建模所用数据量及模型参数调整共同决定。试验在一台六缸高压共轨柴油机上进行。理论分析和试验结果表明:该方法可以在保证精度的同时有效减少标定优化的试验工作量。
The large number of controllable fuel injection parameters in diesel engines equipped with high pressure common-rail fuel injection systems makes the combustion control more flexible, but also increases the calibration and optimization workloads. A higher efficiency, model-based calibration and optimization method was developed. Neural network was used to build subsidiary models at some separated operating condition points, and then adaptive network-based fussy inference system ANFIS was used to expand the model to a continuous engine operating range. The model accuracy depends on the modeling object, the amount of experimental data used to build the model and the model structure parameter setting. The experiment was carried out on a 6-cylinder high pressure common rail diesel engine. Analytical and experimental results show that the method can obviously reduce the workloads with an identical accuracy level compared with the conventional methods.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2004年第11期1524-1527,1535,共5页
Journal of Tsinghua University(Science and Technology)
基金
国家"九七三"重点基础研究项目(2001CB209205)
关键词
柴油机
高压共轨
预喷射
神经网络
自适应神经模糊推理系统(ANFIS)
diesel engines
high pressure common rail
pilot injection
neural network
adaptive network-based fussy inference system ANFIS
