基于BP神经网络的车用汽油机过渡工况空燃比多步预测模型

Multi-step Predictive Model for Air/Fuel Ratio of Gasoline Engine at Transient Conditions Based on Back Propagation Neural Network

为克服车用汽油机空燃比传输延迟对空燃比控制精度的影响,提出了一种基于BP神经网络的空燃比多步预测模型。通过对空燃比数学模型的分析,确定神经网络空燃比多步预测模型的输入向量,同时为提高空燃比预测精度,在神经网络输入向量中增加反映空燃比变化趋势的导数信息。以HL495发动机过渡工况试验数据进行仿真,结果表明该方法能精确预测过渡工况空燃比。

A multi-step predictive model for air/fuel ratio of gasoline engine at transient conditions is presented. By analyzing the model, the input vectors for neural network-based multi-step predictive model are determined. Meanwhile, the input vectors include the derivatives of air/fuel ratio for increasing the prediction accuracy of air/fuel ratio. The results well agree with experiment data at transient conditions of HL495 engine, showing high accuracy of prediction model.