With the mean value model for gasoline engine brought forward by Hendricks as a base, the air fuel ratio model, oxygen sensor model and PI controller are added to establish a new system simulation model, which is validated by bench test. The fault in response delay of oxygen sensor is simulated with this model to investigate the influences of different degrees of failure on the fuel injection law and emissions of engine. Meanwhile a virtual oxygen sensor model is proposed and a network model is built based on Elman neural network theory. The network model is then trained and tested with model outputs as the training samples. The results show that the model can well predict air-fuel ratio signal, which is then used to perform compensation control under malfunction status of oxygen sensor, and the fuel injection law based on Elman neural network and virtual oxygen sensor is in accordance with normal fuel injection law, meeting the needs of practical air-fuel ratio control.%在Hendricks提出的汽油机平均值模型的基础上增加了空燃比模型、氧传感器模型和PI控制器,建立了系统仿真模型,并通过台架试验进行了验证.在系统仿真模型上模拟了氧传感器的响应延迟故障,研究其不同故障程度对发动机喷油规律和排放的影响.同时提出了一种基于Elman神经网络的虚拟氧传感器,根据Elman神经网络的基本理论构建了网络模型,以模型输出作为网络的训练样本,并对该网络模型进行了训练和测试.结果表明,该模型能较好地预测空燃比信号,并利用预测信号进行氧传感器故障状态下的补偿控制;基于Elman神经网络和虚拟氧传感器信号的喷油规律与正常状态下的喷油规律一致,能满足实际空燃比控制需求.
展开▼
