王正江,郑风才,姚国仲,等.柴油机电动废气再循环阀建模与参数识别[J].内燃机工程,2025,46(6):30-40.
柴油机电动废气再循环阀建模与参数识别
Modeling and Parameter Identification of Electric Exhaust Gas Recirculation Valve of a Diesel Engine
DOI：10.13949/j.cnki.nrjgc.2025.06.004
关键词:柴油机  废气再循环  神经网络  遗传算法  参数识别
Key Words:diesel engine  exhaust gas recirculation（EGR）  neural network  genetic algorithm（GA）  parameter identification
基金项目:云南省科技厅重大科技专项项目（202302AC080002）
作者单位E-mail
王正江* 昆明理工大学 云南省内燃机重点实验室，昆明 650500 wangzhengjiang@kust.edu.cn 
郑风才 昆明理工大学 云南省内燃机重点实验室，昆明 650500 569462436@qq.com 
姚国仲* 昆明理工大学 云南省内燃机重点实验室，昆明 650500 yaoguozhong@kust.edu.cn 
王贵勇 昆明理工大学 云南省内燃机重点实验室，昆明 650500 wangguiyong@kust.edu.cn 
摘要点击次数: 253
全文下载次数: 150
摘要:通过对柴油机电动废气再循环（exhaust gas recirculation， EGR）阀的结构和工作原理进行分析，建立了EGR阀驱动电机模型和工作机构模型。针对EGR阀驱动电机转矩常数和阻转矩阻尼系数难以直接测量的问题，提出了一种基于神经网络和遗传算法优化的参数识别方法。通过优化，EGR阀驱动电机的转矩常数优化值趋于一个常值，而阻转矩阻尼系数随驱动占空比和电机角速度呈非线性变化。在所确定的转矩常数和阻转矩阻尼系数预测模型下，驱动电机的电流平均绝对误差和角速度平均绝对误差分别为1.21×10-3 A和4.27 rad/s，具有较好的预测精度。在所有参数确定后，对EGR阀模型进行了试验验证。结果表明：在开环控制下EGR阀模型的开度偏差平均值为0.52%；在闭环控制下EGR阀模型瞬态控制过程的上升时间和超调量的平均绝对误差（mean absolute error， MAE）分别为3.22×10-3 s和1.97%，均方根误差（root mean square error， RMSE）分别为4.01×10-3 s和2.3%，平均绝对百分比误差（mean absolute percentage error， MAPE）分别为7.88%和14.05%，决定系数R2均为0.96。结合驱动电机模型和工作机构模型建立的EGR阀模型能够正确地表征EGR阀的电气特性和运动特性，研究结果可为EGR阀的控制研究提供理论依据。
Abstract:The exhaust gas recirculation（EGR） valve driving motor model and the working mechanism model were established by analyzing the structure and working principle of the diesel engine electric EGR valve. A parameter identification method using a neural network and genetic algorithm（GA） was developed to address the issue of difficulty in measuring torque constant and damping torque coefficient directly during EGR valve modeling. Through optimization， the torque constant optimization value of the EGR valve driving motor tended to a constant value， and the damping torque coefficient changed nonlinearly with duty ratio and motor angular velocity. Under the determined torque constant and torque damping coefficient prediction model， the average absolute error of current and angular velocity of the driving motor were 1.21×10-3 A and 4.27 rad/s， respectively， with good prediction accuracy. After all the parameters were determined， the EGR valve model was verified by experiments. The results show that the opening deviation of the EGR valve model is 0.52% on average under the open-loop control. Under the closed-loop control， the average absolute errors of the rise time and overshoot of the EGR valve model in the transient control process are 3.22×10-3 s and 1.97%， respectively. The root mean square errors of the rise time and overshoot are 4.01×10-3 s and 2.3%， respectively. And the average absolute percentage errors of the rise time and overshoot are 7.88% and 14.05%， respectively， and the decision coefficient is 0.96. The EGR valve model established by combining the driving motor model and the working mechanism model can correctly characterize the electrical characteristics and motion characteristics of the EGR valve. The research results can provide a theoretical basis for the control research of the EGR valve.
查看全文  HTML   查看/发表评论