摘要
针对混动变速器液压系统主油路电磁阀油压-电流特性曲线受系统复杂输入影响,传统修正方法控制精度不足的问题,文章提出一种基于神经网络的电磁阀电流修正控制方法。以驱动电流、油泵转速、目标压力及油液温度为输入,构建神经网络模型,从而实现对电磁阀特性曲线偏移量的精准预测。该方法可显著提升油压控制精度,实车动态试验结果表明:换挡过程压力控制平均绝对误差与均方根误差分别降低80%和44%,在挡过程压力控制平均绝对误差与均方根误差分别降低81%和79%。
To address the issue of insufficient control accuracy in conventional correction methods arising from complex system inputs influencing the pressure-current characteristic curve of main oil circuit solenoid valves in hybrid transmission hydraulic systems,a neural network-based correction control method for solenoid valve current is proposed.Utilizing drive current,oil pump speed,target pressure,and oil temperature as inputs,a neural network model is constructed to achieve precise prediction of characteristic curve deviations.The proposed method significantly enhances oil pressure control accuracy.Dynamic vehicle test results demonstrate that:during gear shifting,the mean absolute error(MAE)and root mean square error(RMSE)of pressure control are reduced by 80%and 44%,respectively;during steady-state in-gear operation,the MAE and RMSE are reduced by 81%and 79%,respectively.
作者
田华
杨逸风
哈迪
TIAN Hua;YANG Yifeng;HA Di
出处
《上海汽车》
2026年第4期30-34,共5页
Shanghai Auto
