正则化学习算法在软测量建模中的应用

Regularization Learning Algorithm Applied for Soft-sensing Modeling

软测量技术的核心是建立软测量模型。基于过程可测信息集建立软测量模型即逼近建模过程是不适定的。以径向基函数神经网络作为软测量模型,在软测量建模中引入正则化学习算法。以广义交叉验证作为正则化参数估计方法,讨论了径向基函数神经网络软测量逼近建模的全局与局部正则化学习算法,给出的实例说明了其有效性。

The core of soft-sensing techniques is soft-sensing modeling. It is ill-posed that the design of soft-sensing model based on measured data in process control. Using RBF （ Radial Basis Function） network as the soft-sensing model, it＇s natural to introduce regnlarization learning algorithm. The Regularization parameters are estimated by GCV （ Generalized Cross Validation）in the paper. Global regnlarization and local regnlarization learning algorithms of soft sensing modeling based on RBF network are introduced respectively. The effectiveness is demonstrated in the case study.