怠速和停机数据拟合的摩擦扭矩在线学习算法
Online Learning Algorithms of Diesel Engine Friction Torque Based on Data Fitting of Idle and Stop Process

针对发动机摩擦扭矩因为磨损、老化等发生缓慢变化的问题，发展了一种基于怠速和停机工况数据拟合的 柴油机摩擦扭矩在线学习算法．建立了一种简化的摩擦扭矩模型；设计了怠速和停机工况摩擦扭矩观测算法，采用 噪声抑制能力强的最速跟踪微分器优化了停机阶段的摩擦扭矩估计；利用怠速和停机观测的摩擦扭矩实现了摩擦扭 矩模型的在线修正，并在柴油机电控单元(ECU)中成功实现．发动机台架试验结果表明，通过摩擦扭矩基础模型的 在线学习，更新后的摩擦扭矩与倒拖试验数据基本吻合，摩擦扭矩估计偏差为2～7,N？m．
The friction torque of the engine changes with the growth of service life of diesel engine．To solve this problem，an online learning algorithm of diesel friction torque based on data fitting of idle and stopping process was developed．A simplified model of friction torque was built as the basis of correction．The current state of friction torque was observed in idle and stopping process．The estimation of friction torque during stopping process was optimized by steepest tracking differentiator of high noise suppression ability．The online correction of the base model was realized using the data obtained in idle and stopping process in electronic control unit(ECU)．Engine bench test proves that through this algorithm，friction torque of updated model and data from drag test are basically the same， and the deviation is about 2―7,N？m