【摘要】 为了减小车用汽油机空燃比传输延迟对空燃比控制精度的影响,提出一种基于Elman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,为了提高过渡工况空燃比预测精度,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。对HL495发动机过渡工况实验数据进行学习,采用梯度算法对Elman神经网络的权值进行调整。研究结果表明:采用该方法能精确预测过渡工况空燃比,预测模型的最大误差小于1%,平均误差小于0.5%。该预测模型可用于实现车用汽油机过渡工况空燃比的精确控制,提高车用汽油机过渡工况排放性能。更多还原

【Abstract】 A multi-step predictive model of air fuel ratio was provided to overcome the influence of air fuel ratio transmission delay on air fuel ratio control accuracy. Input vector of neural network multi-step predictive model was determined by the maths model of air fuel ratio,and derivation of air fuel ratio reflecting the air fuel ratio tendency was included within input vector to improve the prediction accuracy in transient conditions.The simulation was accomplished using experiment data of HL495 gasoline engine,and weight values of Elman neural networks were adjusted by gradient algorithm.The results show the multi-step predictive model can be used to predict accurately air fuel ratio during transient condition and maximum error of prediction model is below 1% and average error is below 0.5%.The model can be used to accurately control air fuel ratio and improve the emission for gasoline engine in transient conditions.更多还原