摘要
利用AVL-Fire建立了高压共轨柴油机燃烧系统仿真模型,通过台架试验验证仿真模型的基础上,分析了混合气浓度场和温度场动态分布特性并将燃烧过程划分为预混合和扩散燃烧分界点,提出燃烧始点、预混合燃烧速率、扩散燃烧速率及燃烧持续期等量化评价参数。建立了预测量化评价参数的人工神经网络(ANN)模型,预测分析发动机控制参数对放热规律评价参数的影响。研究结果表明:当采用"5-18-4"型神经网络结构模型并采用trainlm算法时,预测的鲁棒性、响应性和收敛精度均良好,所创建的ANN模型可以很好地预测放热规律。
A combustion simulation model is built with AVL-Fire,and the dynamic distribution characteristics of the mixture concentration field and temperature field are analyzed by verification of the simulation results on test bench.The start point of combustion,premixed rate,diffusion rate and combustion duration are taken as the evaluation parameters.The effect of the control parameters on Heat Release Rate (HRR)at different stage is analyzed.An Artificial Neural Network (ANN)model is established for quantitative prediction of the evaluation parameters.The structure and of the ANN algorithm is confirmed.The effects of the injection parameters and intake parameters on the evaluation parameters are analyzed and predicted.The results show that,when the structure 5-18-4 ANN with trainlm algorithm are adopted,the robustness,responsiveness and convergence precision are good,indicating the ANN model is a powerful tool to predict HRR.
作者
江涛
林学东
李德刚
杨淼
汤雪林
JIANG Tao;LIN Xue-dong;LI De-gang;YANG Miao;TANG Xue-lin(State Key Laboratory of Automotive Simulation and Control,J ilin University,Changchun 130022,China)
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2018年第6期1747-1754,共8页
Journal of Jilin University:Engineering and Technology Edition
基金
吉林省科技发展计划项目(20150101032JC)
吉林省发改委产业技术研究与开发项目(2014Y093)
关键词
动力机械及工程
人工神经网络
控制参数
放热规律
量化评价参数
量化预测
power machinery and engineering
artificial neural network (ANN)
control parameter heat release rate (HRR)
quantitative evaluation parameter
quantitative prediction
