基于Hammerstein模型的MAF传感器的动态非线性建模  

MAF Sensor Dynamic Non-linear Modeling Based on Hammerstein Model

作　　者：张媛媛[1] 徐科军[2] 张进[2] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院 [2]合肥工业大学电气与自动化工程学院,合肥230009

出　　处：《电子测量与仪器学报》2008年第4期68-73,共6页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金资助项目(编号:60474057)

摘　　要：改进了热膜式空气质量流量(MAF)传感器的静动态标定实验。基于Hammerstein模型描述MAF传感器。采用多项式回归分析建立其静态非线性环节的模型。针对动态线性环节,分别采用基于ARMAX、ARX、OE的模型进行建模。通过误差平方和的比较,确定利用基于输出误差(OE)模型的预报误差法所建立的模型的精度最高。再分别根据不同幅值输入时的实验数据,建立相应的多个模型,比较误差平方和,最终确定误差平方和最小的模型为传感器的模型。该模型精度高且适应性好,实现了MAF传感器在不同工况下模型的统一。Static and dynamic experiments of MAF sensor are improved in this article. Hot-film mass air flow sensor was described as Hammerstein model. Polynomial regression analysis was used to model the static non-linear part of Hammerstein model. Dynamic linear part of Hammerstein model was described as ARMAX model, ARX model or OE model. Through comparing their sums of squared error, it was concluded that prediction error algorithm-based OE model has the best precision. Then several models were established according to different experiment data, and MAF sensor model is selected through comparing their sums of squared error. This model features high precision and good adaptability, and the uniform MAF sensor mathematical model is realized under different working conditions.

关 键 词：HAMMERSTEIN模型 热膜式空气质量流量传感器 预报误差法 最小二乘法 多项式回归分析 

分 类 号：U463.6[机械工程—车辆工程]