«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《内燃机学报》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2022年06

页码:

543-551

栏目:

出版日期:

2022-11-25

文章信息/Info

Title:

Active Disturbance Rejection-Based Self-Optimizing Control of SI-CAI Hybrid Combustion for Gasoline Engines

作者:

宋康; 谢辉

天津大学 内燃机燃烧学国家重点实验室，天津 300350

Author(s):

Song Kang; Xie Hui

State Key Laboratory of Engines，Tianjin University，Tianjin 300350，China

关键词:

汽油机; 混合燃烧; 火花点火-可控自燃(CAI); 主动抗扰控制; 自学习因子

Keywords:

gasoline engine; hybrid combustion; spark-ignited-controlled auto-ignition; active disturbance rejection control; self-learning factors

分类号:

TK411.2

文献标志码:

A

摘要:

火花点火-可控自燃(SI-CAI)混合燃烧对边界条件非常敏感，针对其燃烧相位和放热过程的瞬态控制难题，笔者提出一种基于燃烧模型的前馈、即时扰动观测反馈和累积观测校正相结合的主动抗扰自趋优控制算法．首先，通过引入点火角修正因子和模型参数自学习因子，发展了具备自学习能力的SI-CAI混合燃烧预测模型，并构建了CA50、IMEP和过量空气系数fa的前馈控制算法．其次，为补偿模型的偏差和外部环境条件的随机干扰，提出了基于扩张状态观测器的扰动即时观测算法．最后，为不断校正模型参数自学习因子，改善前馈效果，提出基于递推最小二乘法的累积观测器．对提出的控制算法进行仿真和台架验证，结果表明：IMEP可在3～5个循环内完成阶跃，该过程中CA50波动幅度不高于4°CA，fa偏差小于0.08，满足控制要求．

Abstract:

The spark-ignited-controlled auto-ignition(SI-CAI)hybrid combustion is highly sensitive to boundary conditions. To achieve smooth transient control of combustion timing and heat release process，an active disturbance rejection self-optimizing solution was proposed，which consists of a combustion model-based feedforward controller，a short-term disturbance observer and a long-term disturbance observer. First，a control-oriented SI-CAI hybrid combustion prediction model was developed by introducing an ignition timing correct factor and self-learning factors. A feedforward controller of CA50，IMEP and excessive air coefficient fa was also developed. Then，a disturbance rejection controller based on extended state observer was adopted in order to compensate for the model deviation and random interference of environment conditions. Finally，an online model parameter learning controller based on recursive least square method was introduced in order to continuously correct the self-learning factors of model parameters and improve the feedforward effect. The proposed controller was validated in simulation and engine bench test. The results show that the actual IMEP converges to its desired value within 3—5 cycles during step response test with a CA50 deviation of less than 4°CA and fa deviation of less than 0.08.

常用功能

更新日期/Last Update: 2022-11-25