基于最小二乘支持向量机的传感器非线性动态系统辨识  

Identification for Nonlinear Dynamic System of Transducer Based on Least Squares Support Vector Machine

作　　者：吴德会[1] 

机构地区：[1]九江学院电子工程系

出　　处：《计量学报》2008年第3期226-230,共5页Acta Metrologica Sinica

基　　金：国家自然科学基金(70272032)

摘　　要：讨论了一种基于最小二乘支持向量机的非线性动态传感器系统辨识方法,并给出了相应的推导过程和学习算法。首先,将传感器的非线性动态系统分解为静态非线性子环节和动态线性子环节串联——Hammerstein模型;然后,建立类似线性的中间模型,通过该模型能将Hammerstein模型的非线性传递函数转换为等价的类线性形式;再通过LS-SVM线性回归算法求取中间模型参数;最后推导出中间模型参数与Hammerstein模型参数之间的关系,并通过该关系反演出原传感器系统的Hammerstein模型参数,实现传感器非线性动态辨识。仿真与实际传感器系统辨识的实验结果均表明该方法可行。An identification method for nonlinear dynamic system of the transducer based on least squares support vector machine （LS-SVM） is discussed and the corresponding deduction processes and learning algorithm are also addressed. Firstly, the original nonlinear dynamic system of transducer is supposed to be expressed by a nonlinear static subunit followed by a linear dynamic subunit--Hammerstein model. Secondly, the intermediate linear model is established, by which the nonlinear transfer function of Hammerstein model could be convert to the same form as linear one. Thirdly, the coefficients of the intermediate model are gotten through LS-SVM linear regression algorithm. Finally, relations between the coefficients of intermediate linear model and Hammerstein model are derived, through which the original nonlinear system of transducer is identified. Simulations and experimental results show the identification method for nonlinear dynamic system of transducer is effective.

关 键 词：计量学 传感器 非线性动态系统 辨识 最小二乘支持向量机 

分 类 号：TB931[一般工业技术—计量学]