稀燃汽油机瞬态空燃比的滑模-神经网络控制

Sliding Mode-neural Network Control for Transient Air-fuel Ratio of Lean Burn Gasoline Engine

提出了稀燃汽油机空燃比滑模-神经网络控制方案,采用滑模变结构控制对稀燃工况的空燃比进行反馈控制,采用神经网络实现对进气量的精确预测和油膜动态特性的前馈补偿,进而实现对稀燃发动机瞬态空燃比的精确控制。采用自行开发的电控系统,在一台稀燃发动机上进行了瞬态空燃比实验。实验结果表明:节气门急速变化时空燃比超调可以控制在1个空燃比单位以内,调整时间不超过3s;转速越低,节气门变化越剧烈;空燃比超调越大,调整时间越长。

To meet the special requirement of the transient air-fuel ratio control for lean burn engine, a sliding mode-neural network control scheme was presented. The sliding variable structure control strategy has been provided as feedback to control the air-fuel ratio. At the same time, the neural network method was applied to build the intake air flow observer and realize the instantaneous fuel compensation as feedforward. By way of employing the engine control system that developed by ourselves, the lean burn engine air-fuel ratio control experiments were carried out on a lean burn gasoline engine. The experiments results showed that the maximal overshoot of air-fuel ratio is one AFR unit and the longest transient period is less than 3 seconds when the throttle body moves rapidly. The more slowly the engine worked, the more rapidly the throttle body moved~ the bigger the overshoot of air-fuel ratio was, the longer the transient period would be.