基于渐近法的废气氧传感器Hammerstein模型辨识

Hammerstein model identification of exhaust gas oxygen sensor based on ASYM method

为了解决与发动机空燃比控制相关的废气氧(EG0)传感器精确建模问题,基于渐近(ASYM)法辨识了EGO传感器的Hammerstein模型。模型的非线性部分用静态实验数据拟合,动态线性部分的辨识分为三步。先估计一个高阶ARX模型,然后依据渐近准则(ASYC)找出最佳频率响应估计的模型阶次,再采用极大似然(ML)法估计降阶后的模型参数。通过残差分析、交叉验证和模型误差模型(MEM)测试,将得到的ASYM模型与输出误差(OE)模型和Box-Jenkins(BJ)模型进行比较。结果表明,基于ASYM法的Hammerstein模型能够更精确地捕获EGO传感器的频域动态特性,并且用ASYM法能够量化模型的频域误差上限以评价建模精度。

In order to solve the precise modeling problem of exhaust gas oxygen （EGO） sensor relating to engine air fuel ratio control, the Hammerstein model of EGO sensor is identified based on asymptotic （ASYM） method. The nonlinear part of the model is fitted using static experimental data. The identification of the dynamic linear part is divided into three steps. Firstly, a high order ARX model is estimated. Then, the model order for best frequency response estimation is obtained using the asymptotic criterion （ASYC）. Next, the parameters of the order-reduced model are estimated based on the maximum likelihood （ML） method. The ASYM model is compared with the output error （OE） model and Box-Jenkins （B J） model through residual analysis, cross validation and model error model （MEM） test. Results show that the dynamical characteristics of the EGO sensor in frequency domain can be more accurately captured by the Hammerstein model based on ASYM method ; and the upper error bound of the model in frequency domain can be quantified using ASYM method for modehng accuracy evaluation.